Reality Quest: Review of the AudioWise OPTO•DX (and how clean power makes everything better)
With the broad availability of mScaler with the Chord Hugo mScaler, some wonderful products are coming available to really showcase the revolutionary impact of mScaler. In the early days, us Chord Blu2 owners were limited to DIY options and experimentation. Today, not only does Hugo mScaler have improved RF suppression (vs Blu2), but there are now several very credible commercial options for having a turnkey high-hygiene DX dual BNC connection between mScaler and your Chord DAC of choice.
The latest is something I’ve been waiting for for a long time: the OPTO•DX from AudioWise (https://audiowise-canada.myshopify.com). Unlike other options that focus on filtering to minimize RF on BNC cables, the OPTO•DX goes right to what I always thought would be the end game: true optical and electrical isolation between mScaler and your DAC.
Dan (@dmance) was kind enough to reach out and provide me with one of the first prototype units for testing, with the ask that I give feedback into development (which was a lot of fun), and that I share my unbiased and independent impressions with the community once the unit is ready to ship (which I am delighted to do).
In the interest of disclosure, I have no commercial relationship with Dan, and am sharing impressions as an audiophile without any commercial conflict or bias or economic benefit to me. I have only recently finished running through my full test matrix with a final preproduction unit (Dan tells me these preproduction units are identical to the production units he’s going to launch with). At long last Dan is ready to publicly disclose the product, so I’m finally able to publicly share my experiences with it.
Since this is fundamentally a new approach to one of the most critical links for anyone with a mScaler, I put it through the full battery of tests and configurations, and learned a heck of a lot along the way. Because of the criticality of clean power to digital hygiene, I also took a long and winding and very productive detour into DC power hygiene and AC power hygiene.
Please forgive the long write up, but having gone through all this, I’m convinced that the various test scenarios are interesting not just to determine if something may be useful to others, but also because they give insight to what factors can impact the remarkable quality of mScaled music, and what knobs are available to tune them up.
For those that care less about the method than the results, I have included a TL;DR findings summary at the end of each section, and I have rolled up my interpretation of my findings at the very end. I also have my current recipe (based on my findings) summarized there.
Background on Digital Hygiene
I know that digital hygiene in audio can be a contentious topic (bits are bits, right?). Rather than debate mechanics, I am sharing my experiences in my home with my equipment with my ears and my biases. Almost certainly yours will vary. Hopefully my experiences are helpful to you, but it not, that’s OK too. As a disclaimer, I’m a hobbyist not an expert. I am sharing my experiences and learnings as I try to figure things out, and I am eager to hear the experiences of fellow travelers. If something below is of potential interest to you, give it a go, and let us know what you hear and learn!
My efforts in digital hygiene have been driven by the following core principle: any digital hygiene optimization must start and end with your DAC. If it doesn’t somehow someway impact your DAC (even if you have no idea how or why), it won’t impact sound quality. If one is starting with bit-perfect sources (and if you’re not, fix that), these bit-perfect bits are not affected at all by any digital optimizations (if bits were being changed, you would hear very audible pops clicks and drop outs).
For me, the core question for digital hygiene and optimization is how to best preserve the clock, ground, and reference power of the DAC? Put another way, what can be done to the digital chain to minimize anything being injected to or induced in the DAC that impacts clock, ground, or reference power in the DAC?
I’ve be surprised again and again in this journey, but for bit perfect digital audio streams, I have to take as ground truth that unless something on the digital side is impacting clock, ground, or reference voltage in your DAC, you are not going to hear it. The impact may be direct, or it may be indirect through several intermediate components, over wires, or over through the air. If you hear a difference with bit perfect audio streams, something is be transmitted to or induced in your DAC that is impacting the core references: clock, ground, reference voltage.
Couple rules of thumb that I try to apply when thinking about this stuff:
Am I confident if you buy the same cables and ferrites and power supplies and everything I did, that you'd hear the same thing I hear in your setup? Not at all! Your RFI world is potentially WAY different than mine.
Am I confident you’d consider what you hear significant or important or worthwhile? Not at all, we all have different expectations and sensitivities to sound, with different thresholds for what something is worth to us and how much expense and hassle we are willing to put up with to get it.
Paradoxically, every time you step up the performance ladder (better DAC, better speakers, etc), it is easier and easier to hear things that are dragging performance down. The fantastic news is that with the incredible performance of the Chord stack, you have world class equipment that is refined enough to easily hear these differences.
Enough blather…let’s get to the OPTO•DX, and take a trip into the digital hygiene rabbit hole.
So Tell Me About the OPTO•DX
The OPTO•DX takes dual BNC in from the Hugo mScaler on the “dirty” side (the TX unit). It converts that dual BNC signal to optical signals, and then has dual fiber optic outputs that connect the TX unit to a second independently powered unit on the “clean” side (the RX unit). The RX unit then converts the optical signal back to electrical, and connects to your Chord DAC (Chord DAVE in my case) via dual BNC cables. Each of the units is independently powered by external 12V supplies, and in separate boxes to minimize capacitive coupling between the “dirty” and “clean” sides.
Essentially, the units together give you true 100% optical isolation between your mScaler and DAC, with independent power for the “dirty” and “clean” sides. In effect, it creates both a signal and electrical air gap between everything upstream of the DAC and the DAC, in theory providing the ultimate in digital hygiene at the most critical stage immediately before your DAC.
In my setup, I connect my Hugo mScaler to the dirty side (TX), and connect the clean side (RX) to my DAVE, which I then use to directly drive my high efficiency Voxativ 9.87 speakers.
When evaluating the OPTO•DX, I tested it with a variety of switching 12V supplies, batteries, and advanced linear supplies on both sides of the OPTO•DX. I also experimented with different DC cables, and with a variety of BNC cable configurations (stock, ferrited, etc) on both the dirty and clean sides.
Once I settled on an optimal configuration between the HMS and DAVE, I then went back and removed some of the digital tweaks that I’ve been experimenting for these past several months. The question I wanted to answer is whether the OPTO•DX could basically eliminate the need for upstream digital tweaks and optimizations, by cleaning up the signals right before the DAVE. That is, how well does the OPTO•DX deliver on our ultimate dream of a single digital “moat” that takes care of everything, so we can get rid of all the upstream digital and power crap.
I was very very excited to test the OPTO•DX in my system. With this configuration, it is in theory possible to basically isolate my DAC (Chord DAVE) from all upstream electrical and RF noise, and isolate the DAVE electrically by connecting it (and the clean side of the OPTO•DX) to a dedicated power circuit with isolation transformer, etc. In theory, we’re getting to ultimate hygiene levels. But how does theory map to real listening experiences?
TL;DR - Pretty damn well!
To test the theory, I explored basically three different areas to see what impact the OPTO•DX has, and how best to configure it. They are (from most obvious to most esoteric):
DX Optimization - Baseline
For the baseline, I started with Hugo mScaler -> Stock Chord BNC -> DAVE vs HMS -> FCoS -> DAVE, to repeat the experiment I did with Blu2 and DAVE in my big Blu2 review (see my Blu2 review for specifics on the BNC cables and ferrites in my Ferrited Cable of Shame, what I heard, music I use for critical listening, etc)
As I’ve reported previously, even with Hugo mScaler, there is (to my mind) a significant SQ lift when using the FCoS, although it is much less of a lift than I remember when I had my Blu2. I confirmed this difference is still there with my current optimized digital chain (it is), and used this as my sound quality baseline.
TL;DR - Even with Hugo mScaler, I hear a clear and welcome sound quality improvement when using heavily ferrited BNC cables (but less of a lift than I remembering hearing with Chord Blu2)
DX Optimization - Ferrites vs OPTO•DX
Next, I inserted OPTO•DX into the chain, but with generic switching power supplies on each of the clean and dirty sides (optical isolation, but no power hygiene). I used stock Chord BNC cables to connect the Hugo mScaler to the OPTO•DX dirty side, and stock Chord BNC cables to connect the OPTO•DX clean side to my DAVE.
Compared with FCoS going from HMS direct to DAVE, OPTO•DX with stock cables and stock SMPS power had a clear and unambiguous sound quality lift (all the stuff we’ve come to expect as you clean up RF with ferrites, just more of it). This was a pleasant surprise, and gave me hope that I could finally part with my FCoS!
Next, I swapped in my FCoS on the clean side of OPTO•DX to connect to my DAVE, and compared to the stock Chord BNC cables from clean side of OPTO•DX to DAVE. Essentially, was anything still getting through the OPTO•DX that the ferrites would clean up? In this configuration, I gave a very slight edge (mainly in depth queues) to using the OPTO•DX with the FCoS. (Unfortunately) There is still some residual something that the FCoS is cleaning up, but there is more to that story…
TL;DR - OPTO•DX out of the box is a clear step up from heavily ferrited BNC cables, but the combination of stock OPTO•DX and heavily ferrited BNC cables is better still.
Sidebar - DC Power Quality Is Surprisingly Important For Digital Audio
All of the above experiments were with stock 12V SMPS’s, with stock DC cables. I have been finding that quality DC power can have a huge impact with digital audio, so my next step was to explore the sensitivity of the OPTO•DX to DC power quality, and to see whether the OPTO•DX reduced or eliminated the impact oF DC power upstream in the digital chain.
As background, I have on order a custom Paul Hynes SR7 DC power supply with three dual regulated DC rails. This promises to be the ultimate audiophile DC supply, but it could be years before Paul gets around to building mine. As I’m waiting, I have been experimenting with various DC power configurations to see what improvements I can have while I wait, and to learn more about the impact of DC power supply quality on sound quality.
Currently, I have on hand stock 12V SMPS’s, a couple PowerAdd Pilot2 portable laptop batteries with variable DC output (https://smile.amazon.com/gp/product/B015OAJFOC), a couple Uptone Audio UltraCap LPS 1.2 linear power supplies (https://uptoneaudio.com/products/ultracap-lps-1-2), and several AA battery packs (https://smile.amazon.com/gp/product/B01GEBW7Q8).
Having the assortment has been invaluable in tracking down power and noise issues because they each emphasize a different aspect of power quality. By narrowing down sound quality differences between these different types of power sources, it shines a light on what aspects of power quality seem to have the largest impact. These key differences are:
I generally have found the AA battery packs great for helping understand where mains noise may be impacting your system, but I have also found that quality voltage regulation (and related to that, dynamic current capacity) has a huge impact on digital equipment, and that these AA packs just don’t cut it. I generally use the AA battery packs more for quick and dirty diagnostics (which they are awesome for), and don’t use them for critical listening tests. As I step through the tests outlined below, I am using combinations of standard switching power supplies, the PowerAdd Pilot2 batteries, and the Uptone LPS 1.2 supplies.
TL;DR - DC power supply quality really really matters. Having a variety of supplies available is really helpful to identify what aspect of power supply quality (mains isolation, voltage stability, low noise, etc.) is most impactful at different parts of your audio chain.
OPTO•DX Optimization - What impact Does DC Power Quality Have On The OPTO•DX?
My first set of power experiments were test stock SMPS vs PowerAdd Pilot2 vs Uptone LPS 1.2 connected to the “clean” side of the OPTO•DX (the side closest to the DAC). During these tests, I used a stock SMPS on the OPTO•DX “dirty” side (the side closest to the Hugo mScaler). I heard a big step up in sound quality with the PowerAdd Pilot2 over SMPS, and an even bigger step up with the LPS 1.2. The clean side of the OPTO•DX really benefits from cleaner/better power.
Next, with an LPS 1.2 on the clean side, how does power quality to the dirty side of the OPTO•DX impact sound quality? Powering the “dirty” side with a PowerAdd Pilot2 gave me a nice SQ lift over SMPS on the “dirty” side (again, “better” being all the things we hear as we clean up RF on the BNC lines…better depth and height imaging, more detail, etc). Better power to the dirty side matters, implying that power may be introducing some signal modulation that traverses the fiber link to the clean side.
I strongly suspect 2 LPS 1.2’s (one each for the dirty and clean sides) will be another step up, but I don’t have a spare LPS 1.2 to experiment with this configuration right now. I have a Paul Hynes SR4 on order, and when that arrives, I’ll be able to bring my 2nd LPS 1.2 to the OPTO•DX party and verify one way or the other.
Regardless, it is clear that clean stable power makes a big difference with the OPPO•DX. If your budget allows, I highly recommend that you at least audition a LPS 1.2 in your setup (@Superdad has a VERY generous 30 day no questions asked return policy)
TL;DR - The Uptone Audio LPS 1.2 is a great addition to the OPTO•DX, and you should definitely audition it (or something similar) to see what it does for you. Improved DC power has a significant impact on the “clean” downstream side of OPTO•DX, and a lesser but still very welcome impact on the “dirty” upstream side of the OPTO•DX. For the OPTO•DX quality voltage regulation seems to be more important than mains isolation.
OPTO•DX Optimization - What impact Do Shielded DC Cables Have?
So the above tests were stock DC cables. Is there a benefit to using short shielded DC cables with high quality conductors and connectors? To test, I left the PowerAdd Pilot2 on the dirty side, and switched the DC cable from the LPS 1.2 to the clean side to a Ghent Audio DC-GAC4 Gotham GAC-4 shielded DC cable (https://ghentaudio.com/part/dc-gac4.html).
With the Ghent DC cable, I heard a VERY nice lift (this was surprising). I had heard from people I trust that DC cables can have a big impact, but this was my first test of that since I had received my Ghent cables. Color me impressed!
I then switched to my fancy new Ghent Custom Neotech 7N 18awg DC splitter-cable (also JSSG360 shielded) to feed both the clean and dirty sides of the OPTO•DX from the same LPS 1.2. In this configuration, the sound was perhaps a big darker than the PowerAdd Pilot2 to the dirty side and LPS 1.2 to the clean side. I had to go back and forth a couple times, but I currently prefer the same LPS 1.2 feeding both units with the Ghent shielded splitter cable.
I’m very much a novice at this DC cable stuff, just starting down this particular corner of the rabbit hole. It is unclear to me whether the DC cable shielding is more impactful, or the quality of the conductors, or the connectors, or some combination of all the above. Sounds great though, so doing a future deep dive into DC cables is now on the “to test” list.
TL;DR - A quality well shielded DC cable has a surprisingly positive impact on sound quality. If you only have one high quality DC supply, get a splitter and power both sides of the OPTO•DX with same quality power supply and the same quality DC cable (the lift from quality voltage regulation to both sides is more than the loss from any leakage path between the sides).
With OPTO•DX + Clean Power, Do Ferrites Matter Any More?
As a next test, I took my optimum power configuration (LPS 1.2 feeding both dirty and clean sides of the OPTO•DX, using the Ghent DC splitter cable), and experimented with the stock Chord BNC cable vs FCoS from the clean side to the DAVE. With clean power to the OPTO•DX, is there anything left for the ferrites on the FCoS to filter out?
If there was a difference, it was very very very slight (I can’t say there was). Once you have solid power sources for the OPTO•DX, it looks like there is nothing left to clean up! (at least anything that I can hear) I may finally be able to retire the FCoS!
TL;DR - With clean power to the OPTO•DX, I no longer hear a benefit with a heavily ferrited BNC cables vs the stock Chord BNC cables. My Ferrited Cable of Shame is now retired!
With the OPTO•DX, Do Upstream Tweaks Still Make A Difference?
Now it’s time for an extended detour in the wild and wooly realm of digital tweaks and optimization. The question for me is if one has “ideal” signal isolation going into the DAC, do any of the upstream digital tweaks and optimizations that I’ve accumulated over the years still matter? They all mattered quite a bit when I first put them into place, but are they obsoleted by the OPTO•DX?
Beware, here is where the true dragons dwell, with lots and lots of different ways to poke the dragons. Look to the summary at the end of each section if it gets too wooly.
As a baseline, I currently run the following for the rest of my digital chain:
Impact of USB Signal Integrity/Signal Regeneration
Previously, I had done a lot of experiments with large buffer processes on the NUC end point (using Squeezelite). Squeezelite is configurable enough to let me have maximum network caching (songs preload to memory in a second or so) and maximum in memory buffers between the end point process and the USB output.
This had a huge positive impact on SQ vs stock Roon Bridge (which has minimal buffering), but is extremely glitchy in practice (lots of song skipping, and playback freezes when Roon core gets confused dealing with the large buffers). Even though large buffer Squeezelite drives me nuts, the large buffers configuration highlighted the SQ that was possible, so a lot of my digital tweaking has been toward getting to that reference sound quality level without Roon getting all glitchy and skippy.
First, the Lush^2 shielded USB cable had a very positive impact, but there was still a long way to go to close the SQ gap between Roon Bridge and large buffer Squeezelite. With the Lush^2 in place, I next pivoted to USB isolation and signal regeneration to see if it was possible to get the SQ benefit of large buffer Squeezelite with Roon Bridge by cleaning up the USB signal before it got to the Hugo mScaler.
To experiment with this, I purchased an Uptone Audio ISO REGEN to audition in my system, and by chance got a second one to audition as well. When I added one ISO REGEN (with clean power from an LPS 1.2), sound quality improved quite a bit, but there was still a gap between Roon Bridge and large buffer Squeezelite.
When I added a second ISO REGEN in series (also with clean power from an LPS 1.2), the gap between Roon Bridge and large buffer Squeezelite virtually disappeared(!) Clearly cleaning up the USB signal going into HMS and injecting clean power into Hugo mScaler via the USB Vbus was cleaning up a lot of the upstream sins in my chain. I have been happily running this configuration as my normal listening configuration.
So with OPTO•DX in the chain, what happens if I remove the ISO REGEN’s and take USB direct from the NUC to the HMS? Is the OPTO•DX able to clean up whatever the ISO REGEN’s were cleaning up?
Alas, mostly yes, but not completely. I still prefer to have the ISO REGEN’s in the chain, but the gap is quite a bit less than what I was hearing before OPTO•DX. A very promising step forward, but there is still SQ to be revealed on the digital side of of the OPTO•DX moat.
If I didn’t have two ISO REGEN’s already, would I invest them to get this extra step up? Value (as always) is in the ear of the beholder, but I would be tempted, but it would be a tough call (who am I kidding, of course I would
TL;DR - Even with the OPTO•DX in place, USB isolation and regeneration still makes an audible improvement in sound quality, but the gap has significantly narrowed.
Impact of USB Vbus Isolation
So given the above, is the incremental benefit from the LPS 1.2 powered ISO REGENs more from the USB signal isolation and regeneration they provide, or from the ISO REGEN providing clean power on USB Vbus to the Hugo mScaler USB port, or both? Remember, Hugo mScaler uses USB Vbus to power part of its USB input circuitry. This means the USB input on Hugo mScaler is a power vector into HMS that you need to pay attention to. With the ISO REGEN, the LPS 1.2 is energizing the on board precision voltage regulators, which in turn are providing the Vbus +5V on the USB output of the ISO REGEN, providing very high quality Vbus +5V to the USB input to the Hugo mScaler.
To test whether USB signal regeneration or high quality Vbus is having a bigger impact with Hugo mScaler, on a lark (and with very little research) I bought a iFi Defender to test. This is a USB dongle which passes through USB signals, but has a separate micro USB interface on the side that can be used to inject external power into the USB cable (vs passing through Vbus and ground from the USB source).
For this experiment, I powered the iFi Defender with +5V from the PowerAdd Pilot2. This allowed me to inject clean power into the Vbus of the HMS directly (NUC End point to iFi Defender w/ PowerAdd Pilot2 to Lush^2 USB cable to HMS) with no ISO REGENs in the chain.
So how does it sound compared to have two LPS 1.2 powered ISO REGEN’s in series? The 2x LPS 1.2 powered ISO REGEN’s in series win hands down. Signal regeneration still matters.
Although it is a harder test to do with a single LPS 1.2 (to switch between configurations I need to switch the voltage on the LPS 1.2 from 7V to 5V), do things change when I powered the iFi Defender with +5V from the LPS 1.2?
First I compared the Lush^2 straight from my NUC to my Hugo mScaler, vs iFi Defender added to the Lush^2 (no ISO REGEN) with PowerAdd Pilot2 @ 5V, vs iFi Defender with LPS 1.2 @ 5V. The iFi Defender powered by the LPS 1.2 is definitely a non-subtle step up from the iFi Defender powered by the PowerAdd Pilot 2 (both were clear step up from a direct USB connection from NUC to HMS). The Hugo mScaler definitely appreciates clean power on the USB Vbus.
So how does the iFi Defender + LPS 1.2 injecting clean power on the USB Vbus compare to 2x LPS 1.2 powered ISO REGEN’s in series injecting clean power on the USB Vbus and regenerating the USB signal twice? No question, there is additional lift with the ISO REGENs. Both USB signal integrity and clean power on the Vbus are having a positive impact on sound quality. When running Roon Bridge on my NUC, USB regeneration and isolation still matter (which is consistent with hearing such a profound difference between large buffer Squeezelite and Roon Bridge when connected directly to the Hugo mScaler…USB signal integrity is having some impact on HMS, although I can only guess as to what the mechanism may be)
So is the second ISO REGEN helping because of additional signal regeneration, or by injecting clean Vbus into the next ISO REGEN, or both? Can clean power on USB Vbus let me go down to a single ISO REGEN in my USB chain? Let’s test that by comparing two ISO REGEN’s in series, vs 1 ISO REGEN with clean PowerAdd Pilot2 power on the USB Vbus (from iFI Defender) input to the ISO REGEN, vs 1 LPS 1.2 powered ISO REGEN.
2 ISO REGEN’s (LPS 1.2 powered) in series still clearly have the best sound quality, followed by 1 ISO REGEN (LPS 1.2) with a clean powered iFi Defender (PowerAdd Pilot2), followed by a single ISO REGEN (LPS 1.2). The ISO REGEN clearly benefits from clean Vbus. With two ISO REGEN’s in series, the LPS 1.2 power to the first ISO REGEN is powering the high quality voltage regulators that are feeding high quality Vbus to the USB input circuitry of the 2nd ISO REGEN.
Alas, the 2 ISO REGEN’s in series are staying in my USB chain, at least until someone finds a way to get the sound quality benefit of large buffer Squeezelite with the usability and convenience of Roon Bridge (alas, I’m too locked in to Roon to give up my Roon Core server, even if it refuses to play nice with large buffer Squeezelite).
TL;DR - Power on the USB input to Hugo mScaler has an audible impact on sound quality. Paying attention here will help isolate your HMS from upstream components. USB signal integrity is somehow having an impact on sound quality.
Impact of Mains Isolation
So what if I power the Hugo mScaler with a PowerAdd Pilot2? That would eliminate any mains connection on the Hugo mScaler and OPTO•DX (remember, the Vbus on the USB input to the HMS is being powered from the ISO REGEN, with is being powered by a mains disconnected Uptone LPS 1.2). That would completely isolate DAVE as the one component with a mains connection.
First I compared powering the Hugo mScaler with the stock Hugo mScaler Chord switching supply (15V) vs the PowerAdd Pilot2 at 12V. In this configuration, I prefer the HMS stock switching supply (slightly better spatial resolution and depth resolution). However, when I switched the PowerAdd Pilot2 to 16V, I clearly preferred the PowerAdd Pilot2 @ 16V to the HMS stock supply.
These are surprising findings for me, and a reminder to always listen with an open mind. Higher voltage seems to matter to HMS, enough so that the added benefit of 15V with SMPS outweighed the benefit from mains isolation at 12V.
Alas, the LPS 1.2 does not have enough current output to drive the HMS, so I have to wait for my Paul Hynes SR4 later this summer (and someday a double regulated rail on my Paul Hynes SR7) before I can properly test powering the HMS with a more serious linear power supply.
TL;DR - Clean isolated power to the Hugo mScaler makes a tangible difference in sound quality. You will need a higher capacity supply to properly feed the HMS, but I suspect sound quality will handsomely scale as you scale power supplies.
Impact of End Point Isolation (Optical)
Let’s keep marching back the digital chain. My end point is a NUC7i7DNKE in a fanless Akasa Plato X7D case, with 16GB of memory, and an Intel Optane M.2 2280 16GB PCI-e 3.0 x2 NVMe Memory Module where I keep boot image. I boot and run AudioLinux to memory, with a highly stripped down OS. That keeps power consumption low enough that I can actually run the NUC off of a PowerAdd Pilot2 battery or LPS 1.2(!). That will take mains isolation all the way back to the ethernet port on the NUC.
My NUC also has HDMI out. Using a Kanex HDMI Digital Audio Extractor, I can convert HDMI digital audio (up to 48kHz) to TOSLINK optical, and connect my NUC to the Hugo mScaler via TOSLINK. I have a lot of high resolution music content, so this is less than optimal for me, but it is a useful experiment to compare optical input with a battery powered HMS (complete electrical isolation) to my more typical USB to ISO REGEN (x2) path with a battery powered HMS.
Essentially, the experiment with the Kanex HDMI optical configuration let’s me assess how material is any crap that is getting into the HMS via the USB interface, and is it worthwhile to scrub that interface even more?
So in this experiment, I have HDMI out of the NUC going to a PowerAdd Pilot2 powered Kanex HDMI digital audio extractor, which is then connected by TOSLINK to a PowerAdd Pilot2 powered (16V) Hugo mScaler. I also have USB out from the NUC going through a shielded Lush^2 cable to 2x Uptone ISO REGEN’s in series (powered by +7V from a LPS 1.2 with Ghent Audio DC shielded splitter cable) going to the USB input on the HMS. On Roon, I am playing to both the optical and USB interfaces NUC as a grouped zone, so it is very easy to switch between the inputs on the HMS.
So with USB and optical connected at same time, I find it very difficult to hear a difference, but I give the slight edge to USB.
But crap could be getting through the USB interface and impacting optical playback. How does optical sound when I physically disconnect USB? When I physically disconnect the USB cable from the HMS, it is very easy to tell difference between optical and USB. With optical, I’m hearing more of the usual benefit when RF is managed down (sense of space, clarity of notes), but something seems a bit off that I can’t put my finger on (vs USB). Perhaps too forward? More stress or less control while I listen? Even though optical by itself has more of I’m looking for in these digital hygiene experiments, I find myself preferring to listen to USB.
So all this was with the Kanex powered by +5V from the PowerAdd Pilot2. What happens with different power sources to the Kanex? With a stock 5V SMPS, the Kanex (vs the PowerAdd Pilot2) has a comparatively dulled and flattened soundstage. Once again, power is making a difference with an optical transducer.
The big revelation is when I compare the Kanex with PowerAdd Pilot2 vs Kanex with Uptone LPS 1.2 (5V). With the LPS 1.2, the impact is significant and immediate. That stress and awkwardness I’m was hearing with optical before is gone. Spatial resolution has gone up again, and almost all of the stress is gone. This is now the best my system has ever sounded (at least with my 44/16 Redbook content). Clearly the lowly Kanex is benefiting from better power regulation!
So all these tests were with Redbook content. I’ve resisted going down the all optical path because of my high resolution content. Given the great sound quality of the LPS 1.2 powered Kanex (!), can I live with the limitations of optical? To assess, I tried some favorite 192/24 Chesky binaural content via USB vs isolated optical with USB disconnected (Roon automatically downsamples 192/24 content to 48kHz). Unfortunately, I still prefer USB. I also evaluated some favorite Sound Liaison single mic 352/24 content (again, automatically downsampled by Roon to 48kHz). I again clearly preferred USB. To confirm that my hearing hadn’t drifted, I went back to some favorite Redbook content (no downsampling by Roon Core), and again preferred isolated optical.
Ug. I most definitely do not want different digital chains depending on what music I want to listen to. I need to do some research and figure out if there is LinuxFu to configure the HDMI output of my NUC to max out at 96/24 or even 192/24, and repeat these experiments. That is a project for another day.
TL;DR - It is easier to get to high quality sound with the optical inputs to the Hugo mScaler (if you can live with lower resolution content), but there are different gremlins with optical (vs USB) that I can hear but don’t yet understand. Even for optical sources, power stability seems to matter. There is more to explore here.
Impact of End Point Isolation (Power)
So with the rest of the downstream chain in an optimal configuration (OPTO•DX with LPS 1.2, HMS on PowerAdd PIlot2, NUC outputting via HDMI to a LPS 1.2 powered Kanex with optical to HMS, 44/16 Redbook content so I don’t drive myself crazy), how does running my end point NUC with a PowerAdd Pilot2 at 19V compare with the stock SMPS at 19V?
This is actually a more difficult A/B test, since it takes about a minute for the end point NUC to boot up and load the OS to memory. However, after swapping back and forth a couple times, I’d give a very slight edge to the PowerAdd Pilot2 (over the SMPS to the NUC). Mains isolation may be making a very slight impact on the NUC, but it is a VERY difficult difference for me to hear through the HDMI/Kanex/Optical path.
How about for the USB path? Does running the NUC off the PowerAdd Pilot2 battery close the gap between USB and isolated optical? Alas no. With the NUC powered by PowerAdd Pilot2, I heard approximately the same difference in sound quality between USB and optical with both plugged in (very slight), and approximately the same larger difference in favor fo the isolated HDMI/Kanex/LPS 1.2 path (with USB disconnected) over USB.
With all the other hygiene in the chain, I’m not hearing a material impact to power changes on the NUC. This is an interesting result for me. Several months ago, when I did some quick and dirty experiments with power on my NUC (prior to me adding all the other down stream optimizations), I did hear a clear improvement when using a PowerAdd Pilot2 to power the NUC. It is a reminder that every change one makes to the chain changes how everything else in the chain impacts sound quality (there are lots of different ways to filter out things that are adversely impacting the DAC).
TL;DR - Different power to my NUC didn’t seem to have a material impact (even though it had a noticeable impact before I had the OPTO•DX in the chain). As one optimizes different pieces of the chain, you need to go back and test previous learnings to see if they still apply.
How About A Purely Isolated End Point? (Network, Power, Optical)
I have (briefly) posted before about how impressed I am with the Chromecast Audio as a Roon end point (alas, it sucks serious backside as a Chromecast end point). It remains my go to recommendation for civilians looking to get to great sound quickly. For ~$40 used, you get a little puck of a device that connects to the network via WiFi (no crude coming down on ethernet lines), is externally powered so you can use clean power (clean power made a huge difference in my initial experiments several months ago), has TOSLINK audio out (up too 96/24 content with Roon with no downsampling), and Roon streams it bit perfect content and behaves impeccably with it. What’s not to like?
So how does CCA compare with isolated optical from the Kanex, with all the other optimizations still in place? (SMPS powered NUC, PowerAdd Pilot2 powered HMS, LPS 1.2 powered OPTO•DX, going to DAVE). In this case, HMS has 2 optical inputs, so it is very easy to toggle between the NUC + Kanex path and the Chromecast Audio path. (I should offer the caveat that while it is easy to switch between optical sources, my ear is still not used to optimizing an optical path…please forgive me for not quite being sure how to express the differences, or why I have a preference one way pro another…the differences are definitely different than what I’m accustomed to and I’m still getting my head around them)
So to start, how does the optimized NUC + LPS 1.2 powered Kanex HDMI path compare to the CCA powered by stock USB wall wart? With Redbook content, the Kanex HDMI path sounds smoother and more relaxed to my ear, with a more even (less sloppy?) tone. I easily prefer the LPS 1.2 powered Kanex to the stock Chromecast Audio.
When powering the Chromecast Audio with a PowerAdd Pilot 2 (5V), with Redbook content, the LPS 1.2 power Kanex is still noticeably smoother than the Chromecast Audio, but the gap has been reduced quite a bit. Switching to high res content, the PowerAdd Pilot2 powered Chroomecast Audio does an audibly better job with high res content (Roon downsamples to 96kHz with the CCA vs 48kHz with the NUC HDMI output) but still has stress compared to the LPS 1.2 powered Kanex. I still prefer the Kanex. This is making me that much more eager to figure out how to get my NUC to output 96/24 or even 192/24 to the HDMI output to the Kanex, so I can get the best of both worlds with high res content.
So how does the (very) lowly Chromecast Audio do as a Roon end point when powered by an Uptone LPS 1.2 at 5V, compared to the optimized NUC (SMPS @ 19V) + Kanex (LPS 1.2 at 5V) combo? CCA is very nice, but Redbook is still a bit more relaxed through the Kanex. Even with high res content, I still prefer the NUC+Kanex (even though the high resolution content is more detailed through the Chromecast Audio). I am suspecting that the output clock in the Kanex may be better than the output clock in the CCA? I’m going to need some additional components to audition to be able to triangulate what I’m hearing.
Regardless of the above, I stand by my recommendation that every Roon user should have a Chromecast Audio available as an optical end point into their Hugo mScaler. The quality per dollar ratio is completely off the charts, and it is a great baseline to have available as you experiment with with other end point options. Color me impressed that the CCA is able to scale to this level.
TL;DR - The Chromecast Audio is one of the best value Roon endpoints for the money you can get, and every mScaler owner should have one to experiment with. While better power definitely makes it better (way better actually), its performance ceiling is below what you can get with other approaches.
Impact of Network Isolation, Master Femto Clocks, Goosed Servers, etc (TBD)
VERY high on my list is to eval the Uptone Audio EtherRegen, when Alex (@Superdad) and team finally release it. That device promises to create a moat on the ethernet side going into the end point. If that moat is effective, I hope to be able to ignore everything upstream of the end point, and limit the part of my digital chain that I need to pay attention to to my end point and HMS (with the OPTO•DX providing the moat between HMS and DAVE). In theory, the stuff of digital hygiene dreams. Alas, EtherRegen is not available yet, and I have been resisting any investment in optimizing my network and server until it become available. Can’t wait to put EtherRegen (or similar network isolation solutions) to the test!
So how about fancy femto master clocks, goosed up servers, goosed up power cords and cables, etc? I have no idea. I’ve been intentionally avoiding all that stuff until the network isolation options shake out, and people with all these things already can report out on whether they still make a big difference.
Until then, I’m going to shift my attention back to the analog side (faster/better transducers, headphones, etc), and get back to my home theatre tinkering and see how I can easily integrate a high quality surround setup into this kit. No rest for the weary, but it will keep me busy (and distracted) until my Paul Hynes SR7 arrives, and I get to go through all this power stuff all over again.
TL;DR - There is a whole world of network hygiene, clock hygiene, server optimizations, mechanical isolation, and goodness knows what else. I am resisting diving into that end of the pool for now, but I am weak, and I will succumb at some point.
With A Newly Optimized Digital Chain, Does That Change Anything With The OPTO•DX?
Recall that at the start of this process, I did my evaluation of the OPTO•DX with my then preferred digital chain (NUC to USB through 2x ISO REGENs, Hugo mScaler on SMPS, etc). With the new chain (isolated optical via a LPS-powered Kanex HDMI link, Hugo mScaler on PowerAdd Pilot2 battery at 16V, etc), do my findings on how to optimally power and configure the OPTO•DX still hold?
First I started with comparing both the dirty and clean sides of the OPTO•DX with stock 12V SMPS’s, vs using a single Uptone LPS 1.2 supply with a Ghent Audio shielded splitter DC cable to power both dirty and clean sides. The LPS 1.2 clearly had better sound quality (enhanced soundstage, less stress, more natural, etc.) No surprises so far.
I then listened with the stock SMPS on the “dirty” side, and the LPS 1.2 connected with the Ghent Audio DC cable only to the “clean” side. To my surprise (and contrary to my findings when the Hugo mScaler was being powered by the Chord SMPS), the sound quality was a significant step up from the same LPS 1.2 powering both sides. Huh.
As a final test, I powered the “dirty” side with a PowerAdd Pilot2 battery at 12V, with the LPS 1.2 on the “clean” side. Another significant step up, and the best I’ve ever heard my system. I am now more convinced than every that a second LPS 1.2 to power the “dirty” side of OPTO•DX will be a very worthwhile thing to try.
Clearly, disconnecting the Hugo mScaler from mains (or removing the Chord SMPS) had a positive impact for me. In the new optimal configuration, I have an optical connection from my NUC to the HMS, HMS is running on battery (mains disconnected), “dirty” side of the OPTO•DX is on battery, and a mains disconnected Uptone LPS 1.2 is powering the “clean” side of the OPTO•DX. The DAVE is the last mains connected man standing, and it sounds glorious. But wait, that’s not quite true. As I’ll detail below, there is one more wrinkle to this story, but this time, on the analog side of the chain.
TL;DR - Always revisit your core findings and assumptions - when you change something upstream, what may be optimal downstream may change. In my case, disconnecting my Hugo mScaler from mains and SMPS by going to battery led to better sound by dedicating my LPS 1.2 to only the clean side of the OPTO•DX.
Are There Other DC Power Options?
Based on the urgent and consistent prodding of many audiophiles I know, I placed an order for a Paul Hynes SR7 several months go. My custom SR7 build will have three dual regulated (DR) rails, which basically is the ultimate DC supply. Based on my recent experiments with the LPS 1.2, this may turn out to be one of the better lifetime audio investment decisions I have made (Paul is no longer taking SR7 orders, but does offer spots in line when others cancel). Alas, the wait for my SR7 could be anywhere between 4 months (yeah right) and 4 years to never. It is a crap shoot if and when mine gets built.
In the meantime, besides the AA batteries, laptop batteries, and the awesome Uptone Audio LPS 1.2 that I have been experimenting with, are there other options available as well?
I currently have a Paul Hynes SR4 on order, just to get a sense of what that next level of supply will do in my system. It has more current output capacity than the LPS 1.2, so it will allow me to mix and match and experiment with more configurations. Uptone Audio has their own higher current capacity (and dual rail) JS-2 supply. Once I have my system (and ear) better tuned, I may track down a unit for a proper audition.
But how about something cheaper and more accessible than the LPS 1.2, JS-2, or Paul Hynes supplies? The DC voltage regulation gold standard is the LT-3045 voltage regulator. The best implementations cascade these in series to improve voltage regulation (using ultra clean DC power to power a LT-3045 chip to make even cleaner DC power), and have several of these regulators in parallel to improve current output. I’m a novice at this stuff, but ebay is full of DIY or mass market LT-3045 offerings at very reasonable prices. I have not experimented with any of these yet, but reviews have been mixed.
However, I am eyeing the products at http://www.ldovr.com/default.asp In particular, the single regulated and dual regulated supplies ($95 and $145, respectively). I am also tempted to experiment with the raw boards (~$40) as a retrofit to the OPTO•DX to clear up a LPS 1.2 for the rest of my chain.
In theory, any LT-3045-based supply will have much better regulation than my PowerAdd Pilot2’s. The question is where in my chain will good enough high quality voltage regulation be good enough (vs even better voltage regulation with LPS 1.2 or some future Paul Hynes rail). If a basic LT-3045 supply is good enough (for the Kanex or OPTO•DX for example), that will free up a LPS 1.2 or future-precious Paul Hynes rail for some other component where they can have more impact.
Then there is the whole question of DC cables, shielding options, and grounding options (shunting ground back to a high quality ground, in a star configuration). I have no idea where that journey will go, or how worthwhile it will be, but I was surprised enough by the impact of my Ghent Audio DC cables that I’m now definitely paying attention.
If folks have experiences to share with these or other supplies (and cables), please share! DC power is clearly impactful enough that having a spectrum of available sources will go a long way to optimizing my chain without breaking the bank (or waiting decades for Paul Hynes rails). I’m developing a working theory (aka, educated guess) of what is going on, and I’m eager to try some additional configurations (and read about other’s experiences) to test and develop that theory.
TL;DR - DC power makes a huge impact, and there are a lot of options out there. I’m still a novice at this stuff with very limited experience, and eager to hear what others have learned and experienced.
So What About Clean Power on the Analog Side?
(NOTE: AC power is dangerous and can do serious harm to you, your equipment, your home, and your family…please use a licensed electrician for any AC work in your home, and ensure that everything is done consistent with local electrical codes…no hobby is worth putting yourself, your family, or your home at risk)
With all the upside for clean DC on the digital side, the obvious question is how does clean power impact the analog side of the chain? (eg, DAVE, and in my case, the bass units on my Voxativ 9.87’s)
Several months ago, after some gentle prodding from Roy (@romaz), I actually had a couple dedicated 30A circuits installed to my listening room, with really heavy duty electrical wiring (10-2 Romex) and high conductivity outlets. I use one circuit for my digital chain, and the other for my analog chain (DAVE + the class AB amps in my Voxativ bass units). Since these lines go directly back to my main electrical panel, it has a direct connection to home ground, with a breaker between it and all the other crud in my house electrical system.
It cleaned up a HUGE number of RF gremlins for me (wow!). Once I had this added clarity and resolution, it made it that much easier to hear the remaining digital hygiene gremlins in my system. In the experiments I describe above, it only takes a very brief listen to hear most of the changes, and I no longer have that time of day variability to deal with (the differences are very consistent and repeatable and very clear…makes iterating and experimenting MUCH faster and more effective)
Not everyone has a home setup that lends itself to dedicated electrical circuits for your audio setup. However, if you're close enough to your electrical panel, you can side step a LOT of the chasing down digital hygiene annoyances by spending a very modest amount on dedicated lines. Highly highly highly recommended, if practical for you.
Given this win, I took a leap to experiment some more with AC power hygiene. I recently purchased a used Topaz model 91002-31 ultra isolating isolation transformer. This beast of a transformer has 2500VA output capacity (basically 2500W) so more than enough to drive all my “clean” side kit. The theory is to have two huge transformer coils to buffer any spikes or ripples or other crap on the AC line. For background on this new toy see:
https://www.head-fi.org/threads/top...ansformers-for-affordably-clean-power.857448/
Initially, I heard even more of the clean power awesome (definitely worthwhile), but my unit had an annoying 60Hz hum and ran quite hot. Sometimes the hum was very strong, and other times almost silent (based on my reading, the level of buzz/heat is reflective of the crap the transformer is buffering out). Thus began a love/hate relationship with the Topaz, and day dreams of hiding it in my crawl space or behind a wall (alas, both impractical in my home).
Eventually, I tired of it, and decided to rewire the Topaz in a balanced configuration. Rather than hot being 120V and neutral being a nominal 0V (in the US, our 120VAC single phase home wiring typically has neutral and ground connected at the electrical panel), in a balanced configuration, hot is positive phase 60V, and neutral is negative phase 60V (with ground at 0V). In this configuration, the Topaz all of sudden was near silent, and ran MUCH cooler. Except for being a hideous block of metal, I now have nothing but love for the Topaz (CAUTION: Balanced power has particular rules for wiring, and not all devices are safe with balanced power…make sure your electrician knows how to work with balanced power, and make sure you’re not introducing ground faults between neutral and ground, and that all your power switches are double poled so they operate on both neutral and hot!)
Based on the positive experience with my analog chain, I bought a second Topaz, and have it wired in balanced mode for my digital chain, also on a dedicated circuit. Once I have completely optimized my digital chain, I’ll do real A/B tests on the digital side, but for now, big win on the analog side, and TBD (if any) win the digital side.
So all that is well and good, but what is that last wrinkle I mentioned about power hygiene on the analog side of the chain? (remember, by definition, the DAC lives in both the digital chain and the analog chain, and is impacted by both).
As I sat back with a big grin on my face listening to the best my system has every sounded, I reassessed my system based on what I’ve learned. It turns out I have two additional components that are mains connected, but they are in my analog chain (and connected in two different ways to my DAVE, so they definitely need my attention). Each of my Voxativ Pi bass units are connected to the XLR outputs of my DAVE, and have a built in 200W class A/B amp that are plugged into the same balanced isolation transformer as my DAVE, on the same dedicated 30A circuit I use for my analog chain.
As a final sanity check, repeated some listening tests with the Voxativ Pi Bass units unplugged from mains and from my DAVE. In this scenario, DAVE is the only mains connected device (through an isolation transformer to a dedicated electrical circuit), with Hugo mScaler mains isolated by a battery and optically and mains isolated on both input and output. This is now the most isolated I can make DAVE, with my best DC power dedicated to feeding the final component before Hugo mScaler (the Kanex HDMI audio extractor), and my best DC power dedicated to feeding the final component before DAVE (the “clean” side of OPTO•DX).
As a baseline, I listened in my above new optimal configuration, but with the bass units plugged in and connect to DAVE, turned on, but with the volume controls on the bass units turned all the way down. It took a while to acclimate to the new tone and find some passages in test tracks that were particularly challenging. Once I thought I had a good handle on it, I unplugged the bass units from the wall and removed the connections to my DAVE, and listened again.
With the bass units removed from my system, I heard a very small decrease in harshness and sibalance on some particularly challenging passages in some recordings (we’re talking slight here). Huh.
I reran the listening tests, but this time with the bass units plugged in and on, but with the connectors disconnected from the DAVE. Basically, is the vector for the noise back through the the XLR/RCA cables connecting my DAVE to the bass units, or back through the mains AC cables to the DAVE via DAVE’s mains connection?
With the bass units disconnected from the DAVE, but plugged into the wall and on, the slight increase in harshness was back. Clearly I have some work to do to insulate mains leakage from my bass units back to my DAVE. Huh^2.
As a next to last experiment, I moved my bass units to my dedicated digital outlets, leaving my DAVE and balanced isolation transformer as the only things plugged into my dedicated analog outlets. Now that little bit of extra harshness was gone again. DAVE appreciates having a moat on all sides, even the AC side!
As a last experiment, I moved my bass units to my dedicated analog circuit outlet, but connected at the wall, with DAVE connected to my balanced isolation transformer connected to the same outlet. Basically is the balanced isolation transformer capable of blocking any nasties from getting from my bass units to my DAVE?
Alas, only partially yes. I’m still hearing a hint of the harshness I heard before, but much decreased (I have to listen very carefully to catch it). I’m getting the best isolation by having the Voxativ bass units on a completely different circuit, and isolating DAVE on its own circuit behind an isolation transformer.
So how the heck is stuff getting through a big ass balanced isolation transformer from a class AB amp? (not even class D!!!). Even if I’m only hearing it on some extremely difficult passages where recording quality is already on edge, the fact that I can hear it at all is mind blowing, and reenforces how every little thing and every little vector is worth paying attention to.
There is a lot more to be said about my journey (so far) in analog chain optimization, but that is for some future opus. The TL;DR: get rid of EVERYTHING you possibly can, get the most transparent and ideal DAC you can (Chord DAVE), get the cleanest power you can, and get the fastest and most dynamic transducer you can (Voxativ 4D in my case, but I’m dying to hear the new 4phi drivers).
TL;DR - Component minimalism and clean power are ULTRA impactful on the analog part of your system. Anything you can do to isolate and minimize your analog components and get them clean power will be handsomely rewarded. Isolate your DAC mains as much as you possibly can. Much more to talk about on the analog side, but alas this write up is already too damn long.
Dang, This Is Tough to Follow…Can You Draw a Picture?
Note: there are some redundant options shown here (for example I usually power the HMS with the stock supply, but many of the above experiments are with the HMS powered by a PowerAdd Pilot2 battery. Similarly, if using optical in on the HMS, be sure to disconnect the USB path to make sure the HMS is electrically isolated. Lastly, when the “dirty” side of the OPTO•DX is powered by PowerAdd Pilot2, only the “clean” side is being powered by the LPS 1.2.
So What Did I Learn?
When it comes to isolating your digital chain from your analog chain, there is nothing better than an optical air gap
So What Is My Current Recipe?
With the caveat that noise and noise management is COMPLETELY environment and equipment dependent, here is my current recipe for digital nirvana with Chord Hugo mScaler and a Chord DAC.
I will be purchasing an OPTO•DX day one when Dan posts them up on his site. It is now a core part of my system, and Dan can’t have my preproduction units back until I receive my production unit. I recommend the OPTO•DX without reservations to all mScaler owners. With it, I am finally able to retire my Ferrited Cable of Shame, and no longer have to explain that monster to people that visit my home (“What the hell is that thing?” “It filters out RF that impacts the ground plane of my DAC” “I’m sorry, what’s a dack?”).
If your budget allows, I strongly recommend that you at least audition a premium DC power supply (ideally mains disconnected) like the Uptone Audio UltraCap LPS 1.2 and see what sort of difference it makes for you. Alex offers a very generous 30 day fully money back policy. Do yourself a favor and take him up on it and see if it works for you (I have no commercial relationship with Uptone Audio…I just love the work Alex and John are doing, and respect the stand up way they run their company). I’m sure there are even better power options out there. When I (finally) get my Paul Hynes supplies, I’m looking forward to repeating these tests to see how this scale with power quality.
I am far from an expert on DC cables, but if the LPS 1.2 makes a huge difference for you, get a high quality shielded DC cable from folks like Ghent Audio (price is right too). Again, I’m sure there are other great options out there. Given the lift I heard with the Ghent cables, that is definitely on the list for future experimentation.
I’m finding what a lot of others have already discovered: you can never have too many ultra high quality DC rails, even the lowliest digital equipment benefits from better DC power, there are always difficult Sophie’s Choices about which component gets the best power, and all this makes the excruciating wait for precious Paul Hynes SR7 DR rails 100x more painful.
If you’re serious about optimizing digital audio, get an OPTO•DX to optically isolate your Chord DAC from HMS, and as many high quality DC cables and power supplies as you can afford, and start experimenting to see what makes the biggest difference for you.
My next step is to take my above learnings and reassess how I lay out, power, and connect my equipment (are there less expensive power supplies that are good enough? Can I lose the batteries but keep the SQ? Will even better power supplies elevate things more? etc). With that done and system stabilized, the itch will return, and it will be time to see if there is even more magic to be found by diving ever deeper into the rabbit hole.
Whew…
As I reread the above, I find myself shaking my head in awe at the amazing audio journey I’ve been on the past several years. As is clear from my Blu2 review, I used to think that mScaled music was the ultimate, and there was no way things could get better. Boy was I wrong. In ways known and unknown, the rest of my system was holding it back.
Since mScaler finally gave me an absolute reference, seemingly every month or so, I’m once again hearing the best sound quality I’ve ever heard. Amazingly (and wonderfully), I’m finding that ultra fast transducers, absolute minimalism in your analog chain (no speaker cross overs, no amps, etc), obsession over power quality and stability, and relentless digital hygiene continue to reveal more and more (and MORE) magic and emotion in the music. The mScaler sets an absolute reference. As such, it handsomely rewards supporting equipment that takes away as little as possible from that reference.
Getting past all the equipment and speaking to the music, the humbling thing for me is the realization that all this amazing emotion and magic was in the recorded music all along, but that hadn’t been heard and experienced except by those precious few that were in the studio for the original recordings. What an unexpected blessing and gift to be able to experience music like this!
All that being said, the more I learn and hear, the more optimistic and excited that there is even more to learn and hear. Incredible stuff, and a true gift to be able to experience this level of artistry in our living rooms.
Addendum
9/14/19
To the question of whether I have tried my new Paul Hynes SR4 in this setup, I have. SR4-19 (@15V) has a huge positive impact powering the HMS (vs PP2 battery). It also has a huge positive SQ impact when powering the end point NUC @ 19V, with the HMS on PP2 battery. I slightly prefer the NUC End Point on PP2 @ 19V + SR4 @ 15V on HMS combo, to the NUC on SR4 @ 19V and HMS on PP2 @ 16V combo. Alas, I don't have 2 SR4's to see what the marginal lift is by running both the end point and HMS on SR4's.
My current setup is end point NUC on PP2 battery at 19V, USB out (via Lush^2 USB cable) to ISO Regen (on LPS 1.2) to USB out (via USPCB) to a Matrix Audio XSPDIF 2 DDC (powered via USB from the ISO Regen, so LPS 1.2 sourced power) to optical toslink to HMS. I am also powering the TX side of ODX with a PP2 battery, and the RX side of ODX with a LPS 1.2.
9/15/19
My NUC on Paul Hynes SR4 had a huge impact, almost as much as when I had the SR4 on my HMS. I was surprised.
My core theory is that anything touching reference voltage or ground plane on a DAC is audible. Anything that does not can not (assuming bit perfect). Anything we hear can come from the outside, or be induced in the DAC. Somehow, power to the NUC was impacting what was getting to my DAC (through USB regenerators and DDC to optical toslink)
Here is my working theory:
Perfect power supply does the following:
Based on this, the internal loads on the device can self-induce noise on the ground and reference voltage planes, and somehow those are getting propagated to the DAC, either through direct connection or getting induced in the DAC. It can be over the air (RFI) or something to do with signal integrity in the optical signal that is inducing noise in the DAC. Could the receiver chip in the DAC be inducing noise in the DAC based on the signal integrity? Don't know, but two ISORegen USB regenerators in series sound clearly better than 1 in my
By having the best PS you can get for anything close to the DAC (including the DAC), when there is a big current demand, it protects the device from having the lights dim, which whacks reference voltage and ground plane.
Anything touching reference voltage or ground plane on a DAC is audible (ditto clock, but I haven't done anything with clocks yet) You will not hear anything does not (assuming transmission is bit perfect). The maddening thing is trying to come up with theories for how things way away from the DAC could be influencing the DAC.
With the broad availability of mScaler with the Chord Hugo mScaler, some wonderful products are coming available to really showcase the revolutionary impact of mScaler. In the early days, us Chord Blu2 owners were limited to DIY options and experimentation. Today, not only does Hugo mScaler have improved RF suppression (vs Blu2), but there are now several very credible commercial options for having a turnkey high-hygiene DX dual BNC connection between mScaler and your Chord DAC of choice.
The latest is something I’ve been waiting for for a long time: the OPTO•DX from AudioWise (https://audiowise-canada.myshopify.com). Unlike other options that focus on filtering to minimize RF on BNC cables, the OPTO•DX goes right to what I always thought would be the end game: true optical and electrical isolation between mScaler and your DAC.
Dan (@dmance) was kind enough to reach out and provide me with one of the first prototype units for testing, with the ask that I give feedback into development (which was a lot of fun), and that I share my unbiased and independent impressions with the community once the unit is ready to ship (which I am delighted to do).
In the interest of disclosure, I have no commercial relationship with Dan, and am sharing impressions as an audiophile without any commercial conflict or bias or economic benefit to me. I have only recently finished running through my full test matrix with a final preproduction unit (Dan tells me these preproduction units are identical to the production units he’s going to launch with). At long last Dan is ready to publicly disclose the product, so I’m finally able to publicly share my experiences with it.
Since this is fundamentally a new approach to one of the most critical links for anyone with a mScaler, I put it through the full battery of tests and configurations, and learned a heck of a lot along the way. Because of the criticality of clean power to digital hygiene, I also took a long and winding and very productive detour into DC power hygiene and AC power hygiene.
Please forgive the long write up, but having gone through all this, I’m convinced that the various test scenarios are interesting not just to determine if something may be useful to others, but also because they give insight to what factors can impact the remarkable quality of mScaled music, and what knobs are available to tune them up.
For those that care less about the method than the results, I have included a TL;DR findings summary at the end of each section, and I have rolled up my interpretation of my findings at the very end. I also have my current recipe (based on my findings) summarized there.
Background on Digital Hygiene
I know that digital hygiene in audio can be a contentious topic (bits are bits, right?). Rather than debate mechanics, I am sharing my experiences in my home with my equipment with my ears and my biases. Almost certainly yours will vary. Hopefully my experiences are helpful to you, but it not, that’s OK too. As a disclaimer, I’m a hobbyist not an expert. I am sharing my experiences and learnings as I try to figure things out, and I am eager to hear the experiences of fellow travelers. If something below is of potential interest to you, give it a go, and let us know what you hear and learn!
My efforts in digital hygiene have been driven by the following core principle: any digital hygiene optimization must start and end with your DAC. If it doesn’t somehow someway impact your DAC (even if you have no idea how or why), it won’t impact sound quality. If one is starting with bit-perfect sources (and if you’re not, fix that), these bit-perfect bits are not affected at all by any digital optimizations (if bits were being changed, you would hear very audible pops clicks and drop outs).
For me, the core question for digital hygiene and optimization is how to best preserve the clock, ground, and reference power of the DAC? Put another way, what can be done to the digital chain to minimize anything being injected to or induced in the DAC that impacts clock, ground, or reference power in the DAC?
I’ve be surprised again and again in this journey, but for bit perfect digital audio streams, I have to take as ground truth that unless something on the digital side is impacting clock, ground, or reference voltage in your DAC, you are not going to hear it. The impact may be direct, or it may be indirect through several intermediate components, over wires, or over through the air. If you hear a difference with bit perfect audio streams, something is be transmitted to or induced in your DAC that is impacting the core references: clock, ground, reference voltage.
Couple rules of thumb that I try to apply when thinking about this stuff:
- Anything in your chain impacts everything downstream (what it is plugged into), and anything in your chain (annoyingly) impacts everything upstream (what plugs into it). It can also impact things through the air (WiFi base station broadcasting 2.4GHz), and be impacted from the air. Some of these impacts don’t matter at all, some matter a lot. Study and understand all these potential vectors in and out of your DAC, and decide which ones you want to test and attack or eliminate.
- Everything in your chain has internal goodies that can be impacted by nasties, and these internal goodies can induce nasties that impact itself and other pieces of kit (there is a reason the mScaler is a separate box from the Chord DACs...the FPGA is a beast of a processor). Study and understand these potential sources and sinks, and decide which ones you want to test and attack or eliminate.
- Draw a diagram of your setup, with all the wires in and out (mine is at the end of this review). If anything has a metal path to your DAC (even through your home electrical wiring) pay attention and think how crap through that connection could impact your DAC. If anything is a power line, think about how power levels bouncing around or power noise could impact your DAC. Once you start sleuthing, you’ll be surprised how audible things are, how nasties can hop through many steps to impact your DAC, and how insidious noise can be.
- The thing with RFI and high end audio is that what works and what doesn't work is INCREDIBLY situation dependent: your home setup, what else is happening in your house and neighborhood, what equipment is on or off, etc. Just because something works (or doesn’t work) for someone else doesn’t mean it will work (or not work) for you. All these reports and experience are tactics on the tool belt, not a recipe to win a battle. My recommendation is when you read about experiences with shielded cables or ferrites or power supplies or anything else, take it as input for things to try and see if it matters in your setup. If you don't hear a difference, fantastic! If you do, even better still.
Am I confident if you buy the same cables and ferrites and power supplies and everything I did, that you'd hear the same thing I hear in your setup? Not at all! Your RFI world is potentially WAY different than mine.
Am I confident you’d consider what you hear significant or important or worthwhile? Not at all, we all have different expectations and sensitivities to sound, with different thresholds for what something is worth to us and how much expense and hassle we are willing to put up with to get it.
Paradoxically, every time you step up the performance ladder (better DAC, better speakers, etc), it is easier and easier to hear things that are dragging performance down. The fantastic news is that with the incredible performance of the Chord stack, you have world class equipment that is refined enough to easily hear these differences.
Enough blather…let’s get to the OPTO•DX, and take a trip into the digital hygiene rabbit hole.
So Tell Me About the OPTO•DX
The OPTO•DX takes dual BNC in from the Hugo mScaler on the “dirty” side (the TX unit). It converts that dual BNC signal to optical signals, and then has dual fiber optic outputs that connect the TX unit to a second independently powered unit on the “clean” side (the RX unit). The RX unit then converts the optical signal back to electrical, and connects to your Chord DAC (Chord DAVE in my case) via dual BNC cables. Each of the units is independently powered by external 12V supplies, and in separate boxes to minimize capacitive coupling between the “dirty” and “clean” sides.
Essentially, the units together give you true 100% optical isolation between your mScaler and DAC, with independent power for the “dirty” and “clean” sides. In effect, it creates both a signal and electrical air gap between everything upstream of the DAC and the DAC, in theory providing the ultimate in digital hygiene at the most critical stage immediately before your DAC.
In my setup, I connect my Hugo mScaler to the dirty side (TX), and connect the clean side (RX) to my DAVE, which I then use to directly drive my high efficiency Voxativ 9.87 speakers.
When evaluating the OPTO•DX, I tested it with a variety of switching 12V supplies, batteries, and advanced linear supplies on both sides of the OPTO•DX. I also experimented with different DC cables, and with a variety of BNC cable configurations (stock, ferrited, etc) on both the dirty and clean sides.
Once I settled on an optimal configuration between the HMS and DAVE, I then went back and removed some of the digital tweaks that I’ve been experimenting for these past several months. The question I wanted to answer is whether the OPTO•DX could basically eliminate the need for upstream digital tweaks and optimizations, by cleaning up the signals right before the DAVE. That is, how well does the OPTO•DX deliver on our ultimate dream of a single digital “moat” that takes care of everything, so we can get rid of all the upstream digital and power crap.
I was very very excited to test the OPTO•DX in my system. With this configuration, it is in theory possible to basically isolate my DAC (Chord DAVE) from all upstream electrical and RF noise, and isolate the DAVE electrically by connecting it (and the clean side of the OPTO•DX) to a dedicated power circuit with isolation transformer, etc. In theory, we’re getting to ultimate hygiene levels. But how does theory map to real listening experiences?
TL;DR - Pretty damn well!
To test the theory, I explored basically three different areas to see what impact the OPTO•DX has, and how best to configure it. They are (from most obvious to most esoteric):
- Different combinations of BNC cables
- Stock Chord BNC cables vs my DIY ferrited cable of shame (FCoS) that I describe in great detail in my long Blu2 review (https://www.head-fi.org/showcase/chord-blu-mk-ii-digital-cd-transport.22848/reviews#review-19675)
- Different power supplies and DC cables to the clean and dirty side of the OPTO•DX
- Stock switching supplies, AA batteries, laptop batteries, precision linear supplies, etc.
- Whether other digital tweaks upstream of HMS were still needed
- USB regenerators, optimized NUC end points, cables, etc.
DX Optimization - Baseline
For the baseline, I started with Hugo mScaler -> Stock Chord BNC -> DAVE vs HMS -> FCoS -> DAVE, to repeat the experiment I did with Blu2 and DAVE in my big Blu2 review (see my Blu2 review for specifics on the BNC cables and ferrites in my Ferrited Cable of Shame, what I heard, music I use for critical listening, etc)
As I’ve reported previously, even with Hugo mScaler, there is (to my mind) a significant SQ lift when using the FCoS, although it is much less of a lift than I remember when I had my Blu2. I confirmed this difference is still there with my current optimized digital chain (it is), and used this as my sound quality baseline.
TL;DR - Even with Hugo mScaler, I hear a clear and welcome sound quality improvement when using heavily ferrited BNC cables (but less of a lift than I remembering hearing with Chord Blu2)
DX Optimization - Ferrites vs OPTO•DX
Next, I inserted OPTO•DX into the chain, but with generic switching power supplies on each of the clean and dirty sides (optical isolation, but no power hygiene). I used stock Chord BNC cables to connect the Hugo mScaler to the OPTO•DX dirty side, and stock Chord BNC cables to connect the OPTO•DX clean side to my DAVE.
Compared with FCoS going from HMS direct to DAVE, OPTO•DX with stock cables and stock SMPS power had a clear and unambiguous sound quality lift (all the stuff we’ve come to expect as you clean up RF with ferrites, just more of it). This was a pleasant surprise, and gave me hope that I could finally part with my FCoS!
Next, I swapped in my FCoS on the clean side of OPTO•DX to connect to my DAVE, and compared to the stock Chord BNC cables from clean side of OPTO•DX to DAVE. Essentially, was anything still getting through the OPTO•DX that the ferrites would clean up? In this configuration, I gave a very slight edge (mainly in depth queues) to using the OPTO•DX with the FCoS. (Unfortunately) There is still some residual something that the FCoS is cleaning up, but there is more to that story…
TL;DR - OPTO•DX out of the box is a clear step up from heavily ferrited BNC cables, but the combination of stock OPTO•DX and heavily ferrited BNC cables is better still.
Sidebar - DC Power Quality Is Surprisingly Important For Digital Audio
All of the above experiments were with stock 12V SMPS’s, with stock DC cables. I have been finding that quality DC power can have a huge impact with digital audio, so my next step was to explore the sensitivity of the OPTO•DX to DC power quality, and to see whether the OPTO•DX reduced or eliminated the impact oF DC power upstream in the digital chain.
As background, I have on order a custom Paul Hynes SR7 DC power supply with three dual regulated DC rails. This promises to be the ultimate audiophile DC supply, but it could be years before Paul gets around to building mine. As I’m waiting, I have been experimenting with various DC power configurations to see what improvements I can have while I wait, and to learn more about the impact of DC power supply quality on sound quality.
Currently, I have on hand stock 12V SMPS’s, a couple PowerAdd Pilot2 portable laptop batteries with variable DC output (https://smile.amazon.com/gp/product/B015OAJFOC), a couple Uptone Audio UltraCap LPS 1.2 linear power supplies (https://uptoneaudio.com/products/ultracap-lps-1-2), and several AA battery packs (https://smile.amazon.com/gp/product/B01GEBW7Q8).
Having the assortment has been invaluable in tracking down power and noise issues because they each emphasize a different aspect of power quality. By narrowing down sound quality differences between these different types of power sources, it shines a light on what aspects of power quality seem to have the largest impact. These key differences are:
- The AA battery packs have perfect electrical isolation and no switching noise, but terrible voltage regulation (pro tip: you can adjust voltage output with AA battery packs by adding and removing AA batteries, and using a paper clip to jumper to the output terminal)
- The PowerAdd Pilots have good electrical isolation minimal switching noise (it is fully disconnected from main, but has internal electronics), and good but not great voltage regulation
- The Uptone LPS 1.2 has great isolation (it has a unique internal battery switching mechanism so it is always disconnected from mains), and fantastic voltage regulation if you operate away from its max current limits
I generally have found the AA battery packs great for helping understand where mains noise may be impacting your system, but I have also found that quality voltage regulation (and related to that, dynamic current capacity) has a huge impact on digital equipment, and that these AA packs just don’t cut it. I generally use the AA battery packs more for quick and dirty diagnostics (which they are awesome for), and don’t use them for critical listening tests. As I step through the tests outlined below, I am using combinations of standard switching power supplies, the PowerAdd Pilot2 batteries, and the Uptone LPS 1.2 supplies.
TL;DR - DC power supply quality really really matters. Having a variety of supplies available is really helpful to identify what aspect of power supply quality (mains isolation, voltage stability, low noise, etc.) is most impactful at different parts of your audio chain.
OPTO•DX Optimization - What impact Does DC Power Quality Have On The OPTO•DX?
My first set of power experiments were test stock SMPS vs PowerAdd Pilot2 vs Uptone LPS 1.2 connected to the “clean” side of the OPTO•DX (the side closest to the DAC). During these tests, I used a stock SMPS on the OPTO•DX “dirty” side (the side closest to the Hugo mScaler). I heard a big step up in sound quality with the PowerAdd Pilot2 over SMPS, and an even bigger step up with the LPS 1.2. The clean side of the OPTO•DX really benefits from cleaner/better power.
Next, with an LPS 1.2 on the clean side, how does power quality to the dirty side of the OPTO•DX impact sound quality? Powering the “dirty” side with a PowerAdd Pilot2 gave me a nice SQ lift over SMPS on the “dirty” side (again, “better” being all the things we hear as we clean up RF on the BNC lines…better depth and height imaging, more detail, etc). Better power to the dirty side matters, implying that power may be introducing some signal modulation that traverses the fiber link to the clean side.
I strongly suspect 2 LPS 1.2’s (one each for the dirty and clean sides) will be another step up, but I don’t have a spare LPS 1.2 to experiment with this configuration right now. I have a Paul Hynes SR4 on order, and when that arrives, I’ll be able to bring my 2nd LPS 1.2 to the OPTO•DX party and verify one way or the other.
Regardless, it is clear that clean stable power makes a big difference with the OPPO•DX. If your budget allows, I highly recommend that you at least audition a LPS 1.2 in your setup (@Superdad has a VERY generous 30 day no questions asked return policy)
TL;DR - The Uptone Audio LPS 1.2 is a great addition to the OPTO•DX, and you should definitely audition it (or something similar) to see what it does for you. Improved DC power has a significant impact on the “clean” downstream side of OPTO•DX, and a lesser but still very welcome impact on the “dirty” upstream side of the OPTO•DX. For the OPTO•DX quality voltage regulation seems to be more important than mains isolation.
OPTO•DX Optimization - What impact Do Shielded DC Cables Have?
So the above tests were stock DC cables. Is there a benefit to using short shielded DC cables with high quality conductors and connectors? To test, I left the PowerAdd Pilot2 on the dirty side, and switched the DC cable from the LPS 1.2 to the clean side to a Ghent Audio DC-GAC4 Gotham GAC-4 shielded DC cable (https://ghentaudio.com/part/dc-gac4.html).
With the Ghent DC cable, I heard a VERY nice lift (this was surprising). I had heard from people I trust that DC cables can have a big impact, but this was my first test of that since I had received my Ghent cables. Color me impressed!
I then switched to my fancy new Ghent Custom Neotech 7N 18awg DC splitter-cable (also JSSG360 shielded) to feed both the clean and dirty sides of the OPTO•DX from the same LPS 1.2. In this configuration, the sound was perhaps a big darker than the PowerAdd Pilot2 to the dirty side and LPS 1.2 to the clean side. I had to go back and forth a couple times, but I currently prefer the same LPS 1.2 feeding both units with the Ghent shielded splitter cable.
I’m very much a novice at this DC cable stuff, just starting down this particular corner of the rabbit hole. It is unclear to me whether the DC cable shielding is more impactful, or the quality of the conductors, or the connectors, or some combination of all the above. Sounds great though, so doing a future deep dive into DC cables is now on the “to test” list.
TL;DR - A quality well shielded DC cable has a surprisingly positive impact on sound quality. If you only have one high quality DC supply, get a splitter and power both sides of the OPTO•DX with same quality power supply and the same quality DC cable (the lift from quality voltage regulation to both sides is more than the loss from any leakage path between the sides).
With OPTO•DX + Clean Power, Do Ferrites Matter Any More?
As a next test, I took my optimum power configuration (LPS 1.2 feeding both dirty and clean sides of the OPTO•DX, using the Ghent DC splitter cable), and experimented with the stock Chord BNC cable vs FCoS from the clean side to the DAVE. With clean power to the OPTO•DX, is there anything left for the ferrites on the FCoS to filter out?
If there was a difference, it was very very very slight (I can’t say there was). Once you have solid power sources for the OPTO•DX, it looks like there is nothing left to clean up! (at least anything that I can hear) I may finally be able to retire the FCoS!
TL;DR - With clean power to the OPTO•DX, I no longer hear a benefit with a heavily ferrited BNC cables vs the stock Chord BNC cables. My Ferrited Cable of Shame is now retired!
With the OPTO•DX, Do Upstream Tweaks Still Make A Difference?
Now it’s time for an extended detour in the wild and wooly realm of digital tweaks and optimization. The question for me is if one has “ideal” signal isolation going into the DAC, do any of the upstream digital tweaks and optimizations that I’ve accumulated over the years still matter? They all mattered quite a bit when I first put them into place, but are they obsoleted by the OPTO•DX?
Beware, here is where the true dragons dwell, with lots and lots of different ways to poke the dragons. Look to the summary at the end of each section if it gets too wooly.
As a baseline, I currently run the following for the rest of my digital chain:
- Roon server on a highly optimized NUC server - real time OS (Audio Linux), everything running in memory, real time priority and dedicated CPUs to the audio processes, ethernet bridged to a NUC end point to isolate ethernet traffic, etc.
- Roon bridge on a highly optimized NUC end point - real time stripped to the minimum OS, everything in memory, real time priority to USB interface and audio processes, fanless case, etc.
- Lush^2 shielded USB cable (JSSG360 shielded)
- Two Uptone Audio ISO REGEN USB regenerators in series (https://uptoneaudio.com/products/iso-regen), connected by Uptone Audio USPCB USB boards to each other and my Hugo mScaler. Both ISO REGENs are powered by a Uptone Audio LPS 1.2 via a generic DC splitter cable, fed by Ghent GAC4 shielded DC cable (alas, I only have one high quality shielded splitter DC cable, and I found it has a bigger impact on the OPTO•DX).
Impact of USB Signal Integrity/Signal Regeneration
Previously, I had done a lot of experiments with large buffer processes on the NUC end point (using Squeezelite). Squeezelite is configurable enough to let me have maximum network caching (songs preload to memory in a second or so) and maximum in memory buffers between the end point process and the USB output.
This had a huge positive impact on SQ vs stock Roon Bridge (which has minimal buffering), but is extremely glitchy in practice (lots of song skipping, and playback freezes when Roon core gets confused dealing with the large buffers). Even though large buffer Squeezelite drives me nuts, the large buffers configuration highlighted the SQ that was possible, so a lot of my digital tweaking has been toward getting to that reference sound quality level without Roon getting all glitchy and skippy.
First, the Lush^2 shielded USB cable had a very positive impact, but there was still a long way to go to close the SQ gap between Roon Bridge and large buffer Squeezelite. With the Lush^2 in place, I next pivoted to USB isolation and signal regeneration to see if it was possible to get the SQ benefit of large buffer Squeezelite with Roon Bridge by cleaning up the USB signal before it got to the Hugo mScaler.
To experiment with this, I purchased an Uptone Audio ISO REGEN to audition in my system, and by chance got a second one to audition as well. When I added one ISO REGEN (with clean power from an LPS 1.2), sound quality improved quite a bit, but there was still a gap between Roon Bridge and large buffer Squeezelite.
When I added a second ISO REGEN in series (also with clean power from an LPS 1.2), the gap between Roon Bridge and large buffer Squeezelite virtually disappeared(!) Clearly cleaning up the USB signal going into HMS and injecting clean power into Hugo mScaler via the USB Vbus was cleaning up a lot of the upstream sins in my chain. I have been happily running this configuration as my normal listening configuration.
So with OPTO•DX in the chain, what happens if I remove the ISO REGEN’s and take USB direct from the NUC to the HMS? Is the OPTO•DX able to clean up whatever the ISO REGEN’s were cleaning up?
Alas, mostly yes, but not completely. I still prefer to have the ISO REGEN’s in the chain, but the gap is quite a bit less than what I was hearing before OPTO•DX. A very promising step forward, but there is still SQ to be revealed on the digital side of of the OPTO•DX moat.
If I didn’t have two ISO REGEN’s already, would I invest them to get this extra step up? Value (as always) is in the ear of the beholder, but I would be tempted, but it would be a tough call (who am I kidding, of course I would
TL;DR - Even with the OPTO•DX in place, USB isolation and regeneration still makes an audible improvement in sound quality, but the gap has significantly narrowed.
Impact of USB Vbus Isolation
So given the above, is the incremental benefit from the LPS 1.2 powered ISO REGENs more from the USB signal isolation and regeneration they provide, or from the ISO REGEN providing clean power on USB Vbus to the Hugo mScaler USB port, or both? Remember, Hugo mScaler uses USB Vbus to power part of its USB input circuitry. This means the USB input on Hugo mScaler is a power vector into HMS that you need to pay attention to. With the ISO REGEN, the LPS 1.2 is energizing the on board precision voltage regulators, which in turn are providing the Vbus +5V on the USB output of the ISO REGEN, providing very high quality Vbus +5V to the USB input to the Hugo mScaler.
To test whether USB signal regeneration or high quality Vbus is having a bigger impact with Hugo mScaler, on a lark (and with very little research) I bought a iFi Defender to test. This is a USB dongle which passes through USB signals, but has a separate micro USB interface on the side that can be used to inject external power into the USB cable (vs passing through Vbus and ground from the USB source).
For this experiment, I powered the iFi Defender with +5V from the PowerAdd Pilot2. This allowed me to inject clean power into the Vbus of the HMS directly (NUC End point to iFi Defender w/ PowerAdd Pilot2 to Lush^2 USB cable to HMS) with no ISO REGENs in the chain.
So how does it sound compared to have two LPS 1.2 powered ISO REGEN’s in series? The 2x LPS 1.2 powered ISO REGEN’s in series win hands down. Signal regeneration still matters.
Although it is a harder test to do with a single LPS 1.2 (to switch between configurations I need to switch the voltage on the LPS 1.2 from 7V to 5V), do things change when I powered the iFi Defender with +5V from the LPS 1.2?
First I compared the Lush^2 straight from my NUC to my Hugo mScaler, vs iFi Defender added to the Lush^2 (no ISO REGEN) with PowerAdd Pilot2 @ 5V, vs iFi Defender with LPS 1.2 @ 5V. The iFi Defender powered by the LPS 1.2 is definitely a non-subtle step up from the iFi Defender powered by the PowerAdd Pilot 2 (both were clear step up from a direct USB connection from NUC to HMS). The Hugo mScaler definitely appreciates clean power on the USB Vbus.
So how does the iFi Defender + LPS 1.2 injecting clean power on the USB Vbus compare to 2x LPS 1.2 powered ISO REGEN’s in series injecting clean power on the USB Vbus and regenerating the USB signal twice? No question, there is additional lift with the ISO REGENs. Both USB signal integrity and clean power on the Vbus are having a positive impact on sound quality. When running Roon Bridge on my NUC, USB regeneration and isolation still matter (which is consistent with hearing such a profound difference between large buffer Squeezelite and Roon Bridge when connected directly to the Hugo mScaler…USB signal integrity is having some impact on HMS, although I can only guess as to what the mechanism may be)
So is the second ISO REGEN helping because of additional signal regeneration, or by injecting clean Vbus into the next ISO REGEN, or both? Can clean power on USB Vbus let me go down to a single ISO REGEN in my USB chain? Let’s test that by comparing two ISO REGEN’s in series, vs 1 ISO REGEN with clean PowerAdd Pilot2 power on the USB Vbus (from iFI Defender) input to the ISO REGEN, vs 1 LPS 1.2 powered ISO REGEN.
2 ISO REGEN’s (LPS 1.2 powered) in series still clearly have the best sound quality, followed by 1 ISO REGEN (LPS 1.2) with a clean powered iFi Defender (PowerAdd Pilot2), followed by a single ISO REGEN (LPS 1.2). The ISO REGEN clearly benefits from clean Vbus. With two ISO REGEN’s in series, the LPS 1.2 power to the first ISO REGEN is powering the high quality voltage regulators that are feeding high quality Vbus to the USB input circuitry of the 2nd ISO REGEN.
Alas, the 2 ISO REGEN’s in series are staying in my USB chain, at least until someone finds a way to get the sound quality benefit of large buffer Squeezelite with the usability and convenience of Roon Bridge (alas, I’m too locked in to Roon to give up my Roon Core server, even if it refuses to play nice with large buffer Squeezelite).
TL;DR - Power on the USB input to Hugo mScaler has an audible impact on sound quality. Paying attention here will help isolate your HMS from upstream components. USB signal integrity is somehow having an impact on sound quality.
Impact of Mains Isolation
So what if I power the Hugo mScaler with a PowerAdd Pilot2? That would eliminate any mains connection on the Hugo mScaler and OPTO•DX (remember, the Vbus on the USB input to the HMS is being powered from the ISO REGEN, with is being powered by a mains disconnected Uptone LPS 1.2). That would completely isolate DAVE as the one component with a mains connection.
First I compared powering the Hugo mScaler with the stock Hugo mScaler Chord switching supply (15V) vs the PowerAdd Pilot2 at 12V. In this configuration, I prefer the HMS stock switching supply (slightly better spatial resolution and depth resolution). However, when I switched the PowerAdd Pilot2 to 16V, I clearly preferred the PowerAdd Pilot2 @ 16V to the HMS stock supply.
These are surprising findings for me, and a reminder to always listen with an open mind. Higher voltage seems to matter to HMS, enough so that the added benefit of 15V with SMPS outweighed the benefit from mains isolation at 12V.
Alas, the LPS 1.2 does not have enough current output to drive the HMS, so I have to wait for my Paul Hynes SR4 later this summer (and someday a double regulated rail on my Paul Hynes SR7) before I can properly test powering the HMS with a more serious linear power supply.
TL;DR - Clean isolated power to the Hugo mScaler makes a tangible difference in sound quality. You will need a higher capacity supply to properly feed the HMS, but I suspect sound quality will handsomely scale as you scale power supplies.
Impact of End Point Isolation (Optical)
Let’s keep marching back the digital chain. My end point is a NUC7i7DNKE in a fanless Akasa Plato X7D case, with 16GB of memory, and an Intel Optane M.2 2280 16GB PCI-e 3.0 x2 NVMe Memory Module where I keep boot image. I boot and run AudioLinux to memory, with a highly stripped down OS. That keeps power consumption low enough that I can actually run the NUC off of a PowerAdd Pilot2 battery or LPS 1.2(!). That will take mains isolation all the way back to the ethernet port on the NUC.
My NUC also has HDMI out. Using a Kanex HDMI Digital Audio Extractor, I can convert HDMI digital audio (up to 48kHz) to TOSLINK optical, and connect my NUC to the Hugo mScaler via TOSLINK. I have a lot of high resolution music content, so this is less than optimal for me, but it is a useful experiment to compare optical input with a battery powered HMS (complete electrical isolation) to my more typical USB to ISO REGEN (x2) path with a battery powered HMS.
Essentially, the experiment with the Kanex HDMI optical configuration let’s me assess how material is any crap that is getting into the HMS via the USB interface, and is it worthwhile to scrub that interface even more?
So in this experiment, I have HDMI out of the NUC going to a PowerAdd Pilot2 powered Kanex HDMI digital audio extractor, which is then connected by TOSLINK to a PowerAdd Pilot2 powered (16V) Hugo mScaler. I also have USB out from the NUC going through a shielded Lush^2 cable to 2x Uptone ISO REGEN’s in series (powered by +7V from a LPS 1.2 with Ghent Audio DC shielded splitter cable) going to the USB input on the HMS. On Roon, I am playing to both the optical and USB interfaces NUC as a grouped zone, so it is very easy to switch between the inputs on the HMS.
So with USB and optical connected at same time, I find it very difficult to hear a difference, but I give the slight edge to USB.
But crap could be getting through the USB interface and impacting optical playback. How does optical sound when I physically disconnect USB? When I physically disconnect the USB cable from the HMS, it is very easy to tell difference between optical and USB. With optical, I’m hearing more of the usual benefit when RF is managed down (sense of space, clarity of notes), but something seems a bit off that I can’t put my finger on (vs USB). Perhaps too forward? More stress or less control while I listen? Even though optical by itself has more of I’m looking for in these digital hygiene experiments, I find myself preferring to listen to USB.
So all this was with the Kanex powered by +5V from the PowerAdd Pilot2. What happens with different power sources to the Kanex? With a stock 5V SMPS, the Kanex (vs the PowerAdd Pilot2) has a comparatively dulled and flattened soundstage. Once again, power is making a difference with an optical transducer.
The big revelation is when I compare the Kanex with PowerAdd Pilot2 vs Kanex with Uptone LPS 1.2 (5V). With the LPS 1.2, the impact is significant and immediate. That stress and awkwardness I’m was hearing with optical before is gone. Spatial resolution has gone up again, and almost all of the stress is gone. This is now the best my system has ever sounded (at least with my 44/16 Redbook content). Clearly the lowly Kanex is benefiting from better power regulation!
So all these tests were with Redbook content. I’ve resisted going down the all optical path because of my high resolution content. Given the great sound quality of the LPS 1.2 powered Kanex (!), can I live with the limitations of optical? To assess, I tried some favorite 192/24 Chesky binaural content via USB vs isolated optical with USB disconnected (Roon automatically downsamples 192/24 content to 48kHz). Unfortunately, I still prefer USB. I also evaluated some favorite Sound Liaison single mic 352/24 content (again, automatically downsampled by Roon to 48kHz). I again clearly preferred USB. To confirm that my hearing hadn’t drifted, I went back to some favorite Redbook content (no downsampling by Roon Core), and again preferred isolated optical.
Ug. I most definitely do not want different digital chains depending on what music I want to listen to. I need to do some research and figure out if there is LinuxFu to configure the HDMI output of my NUC to max out at 96/24 or even 192/24, and repeat these experiments. That is a project for another day.
TL;DR - It is easier to get to high quality sound with the optical inputs to the Hugo mScaler (if you can live with lower resolution content), but there are different gremlins with optical (vs USB) that I can hear but don’t yet understand. Even for optical sources, power stability seems to matter. There is more to explore here.
Impact of End Point Isolation (Power)
So with the rest of the downstream chain in an optimal configuration (OPTO•DX with LPS 1.2, HMS on PowerAdd PIlot2, NUC outputting via HDMI to a LPS 1.2 powered Kanex with optical to HMS, 44/16 Redbook content so I don’t drive myself crazy), how does running my end point NUC with a PowerAdd Pilot2 at 19V compare with the stock SMPS at 19V?
This is actually a more difficult A/B test, since it takes about a minute for the end point NUC to boot up and load the OS to memory. However, after swapping back and forth a couple times, I’d give a very slight edge to the PowerAdd Pilot2 (over the SMPS to the NUC). Mains isolation may be making a very slight impact on the NUC, but it is a VERY difficult difference for me to hear through the HDMI/Kanex/Optical path.
How about for the USB path? Does running the NUC off the PowerAdd Pilot2 battery close the gap between USB and isolated optical? Alas no. With the NUC powered by PowerAdd Pilot2, I heard approximately the same difference in sound quality between USB and optical with both plugged in (very slight), and approximately the same larger difference in favor fo the isolated HDMI/Kanex/LPS 1.2 path (with USB disconnected) over USB.
With all the other hygiene in the chain, I’m not hearing a material impact to power changes on the NUC. This is an interesting result for me. Several months ago, when I did some quick and dirty experiments with power on my NUC (prior to me adding all the other down stream optimizations), I did hear a clear improvement when using a PowerAdd Pilot2 to power the NUC. It is a reminder that every change one makes to the chain changes how everything else in the chain impacts sound quality (there are lots of different ways to filter out things that are adversely impacting the DAC).
TL;DR - Different power to my NUC didn’t seem to have a material impact (even though it had a noticeable impact before I had the OPTO•DX in the chain). As one optimizes different pieces of the chain, you need to go back and test previous learnings to see if they still apply.
How About A Purely Isolated End Point? (Network, Power, Optical)
I have (briefly) posted before about how impressed I am with the Chromecast Audio as a Roon end point (alas, it sucks serious backside as a Chromecast end point). It remains my go to recommendation for civilians looking to get to great sound quickly. For ~$40 used, you get a little puck of a device that connects to the network via WiFi (no crude coming down on ethernet lines), is externally powered so you can use clean power (clean power made a huge difference in my initial experiments several months ago), has TOSLINK audio out (up too 96/24 content with Roon with no downsampling), and Roon streams it bit perfect content and behaves impeccably with it. What’s not to like?
So how does CCA compare with isolated optical from the Kanex, with all the other optimizations still in place? (SMPS powered NUC, PowerAdd Pilot2 powered HMS, LPS 1.2 powered OPTO•DX, going to DAVE). In this case, HMS has 2 optical inputs, so it is very easy to toggle between the NUC + Kanex path and the Chromecast Audio path. (I should offer the caveat that while it is easy to switch between optical sources, my ear is still not used to optimizing an optical path…please forgive me for not quite being sure how to express the differences, or why I have a preference one way pro another…the differences are definitely different than what I’m accustomed to and I’m still getting my head around them)
So to start, how does the optimized NUC + LPS 1.2 powered Kanex HDMI path compare to the CCA powered by stock USB wall wart? With Redbook content, the Kanex HDMI path sounds smoother and more relaxed to my ear, with a more even (less sloppy?) tone. I easily prefer the LPS 1.2 powered Kanex to the stock Chromecast Audio.
When powering the Chromecast Audio with a PowerAdd Pilot 2 (5V), with Redbook content, the LPS 1.2 power Kanex is still noticeably smoother than the Chromecast Audio, but the gap has been reduced quite a bit. Switching to high res content, the PowerAdd Pilot2 powered Chroomecast Audio does an audibly better job with high res content (Roon downsamples to 96kHz with the CCA vs 48kHz with the NUC HDMI output) but still has stress compared to the LPS 1.2 powered Kanex. I still prefer the Kanex. This is making me that much more eager to figure out how to get my NUC to output 96/24 or even 192/24 to the HDMI output to the Kanex, so I can get the best of both worlds with high res content.
So how does the (very) lowly Chromecast Audio do as a Roon end point when powered by an Uptone LPS 1.2 at 5V, compared to the optimized NUC (SMPS @ 19V) + Kanex (LPS 1.2 at 5V) combo? CCA is very nice, but Redbook is still a bit more relaxed through the Kanex. Even with high res content, I still prefer the NUC+Kanex (even though the high resolution content is more detailed through the Chromecast Audio). I am suspecting that the output clock in the Kanex may be better than the output clock in the CCA? I’m going to need some additional components to audition to be able to triangulate what I’m hearing.
Regardless of the above, I stand by my recommendation that every Roon user should have a Chromecast Audio available as an optical end point into their Hugo mScaler. The quality per dollar ratio is completely off the charts, and it is a great baseline to have available as you experiment with with other end point options. Color me impressed that the CCA is able to scale to this level.
TL;DR - The Chromecast Audio is one of the best value Roon endpoints for the money you can get, and every mScaler owner should have one to experiment with. While better power definitely makes it better (way better actually), its performance ceiling is below what you can get with other approaches.
Impact of Network Isolation, Master Femto Clocks, Goosed Servers, etc (TBD)
VERY high on my list is to eval the Uptone Audio EtherRegen, when Alex (@Superdad) and team finally release it. That device promises to create a moat on the ethernet side going into the end point. If that moat is effective, I hope to be able to ignore everything upstream of the end point, and limit the part of my digital chain that I need to pay attention to to my end point and HMS (with the OPTO•DX providing the moat between HMS and DAVE). In theory, the stuff of digital hygiene dreams. Alas, EtherRegen is not available yet, and I have been resisting any investment in optimizing my network and server until it become available. Can’t wait to put EtherRegen (or similar network isolation solutions) to the test!
So how about fancy femto master clocks, goosed up servers, goosed up power cords and cables, etc? I have no idea. I’ve been intentionally avoiding all that stuff until the network isolation options shake out, and people with all these things already can report out on whether they still make a big difference.
Until then, I’m going to shift my attention back to the analog side (faster/better transducers, headphones, etc), and get back to my home theatre tinkering and see how I can easily integrate a high quality surround setup into this kit. No rest for the weary, but it will keep me busy (and distracted) until my Paul Hynes SR7 arrives, and I get to go through all this power stuff all over again.
TL;DR - There is a whole world of network hygiene, clock hygiene, server optimizations, mechanical isolation, and goodness knows what else. I am resisting diving into that end of the pool for now, but I am weak, and I will succumb at some point.
With A Newly Optimized Digital Chain, Does That Change Anything With The OPTO•DX?
Recall that at the start of this process, I did my evaluation of the OPTO•DX with my then preferred digital chain (NUC to USB through 2x ISO REGENs, Hugo mScaler on SMPS, etc). With the new chain (isolated optical via a LPS-powered Kanex HDMI link, Hugo mScaler on PowerAdd Pilot2 battery at 16V, etc), do my findings on how to optimally power and configure the OPTO•DX still hold?
First I started with comparing both the dirty and clean sides of the OPTO•DX with stock 12V SMPS’s, vs using a single Uptone LPS 1.2 supply with a Ghent Audio shielded splitter DC cable to power both dirty and clean sides. The LPS 1.2 clearly had better sound quality (enhanced soundstage, less stress, more natural, etc.) No surprises so far.
I then listened with the stock SMPS on the “dirty” side, and the LPS 1.2 connected with the Ghent Audio DC cable only to the “clean” side. To my surprise (and contrary to my findings when the Hugo mScaler was being powered by the Chord SMPS), the sound quality was a significant step up from the same LPS 1.2 powering both sides. Huh.
As a final test, I powered the “dirty” side with a PowerAdd Pilot2 battery at 12V, with the LPS 1.2 on the “clean” side. Another significant step up, and the best I’ve ever heard my system. I am now more convinced than every that a second LPS 1.2 to power the “dirty” side of OPTO•DX will be a very worthwhile thing to try.
Clearly, disconnecting the Hugo mScaler from mains (or removing the Chord SMPS) had a positive impact for me. In the new optimal configuration, I have an optical connection from my NUC to the HMS, HMS is running on battery (mains disconnected), “dirty” side of the OPTO•DX is on battery, and a mains disconnected Uptone LPS 1.2 is powering the “clean” side of the OPTO•DX. The DAVE is the last mains connected man standing, and it sounds glorious. But wait, that’s not quite true. As I’ll detail below, there is one more wrinkle to this story, but this time, on the analog side of the chain.
TL;DR - Always revisit your core findings and assumptions - when you change something upstream, what may be optimal downstream may change. In my case, disconnecting my Hugo mScaler from mains and SMPS by going to battery led to better sound by dedicating my LPS 1.2 to only the clean side of the OPTO•DX.
Are There Other DC Power Options?
Based on the urgent and consistent prodding of many audiophiles I know, I placed an order for a Paul Hynes SR7 several months go. My custom SR7 build will have three dual regulated (DR) rails, which basically is the ultimate DC supply. Based on my recent experiments with the LPS 1.2, this may turn out to be one of the better lifetime audio investment decisions I have made (Paul is no longer taking SR7 orders, but does offer spots in line when others cancel). Alas, the wait for my SR7 could be anywhere between 4 months (yeah right) and 4 years to never. It is a crap shoot if and when mine gets built.
In the meantime, besides the AA batteries, laptop batteries, and the awesome Uptone Audio LPS 1.2 that I have been experimenting with, are there other options available as well?
I currently have a Paul Hynes SR4 on order, just to get a sense of what that next level of supply will do in my system. It has more current output capacity than the LPS 1.2, so it will allow me to mix and match and experiment with more configurations. Uptone Audio has their own higher current capacity (and dual rail) JS-2 supply. Once I have my system (and ear) better tuned, I may track down a unit for a proper audition.
But how about something cheaper and more accessible than the LPS 1.2, JS-2, or Paul Hynes supplies? The DC voltage regulation gold standard is the LT-3045 voltage regulator. The best implementations cascade these in series to improve voltage regulation (using ultra clean DC power to power a LT-3045 chip to make even cleaner DC power), and have several of these regulators in parallel to improve current output. I’m a novice at this stuff, but ebay is full of DIY or mass market LT-3045 offerings at very reasonable prices. I have not experimented with any of these yet, but reviews have been mixed.
However, I am eyeing the products at http://www.ldovr.com/default.asp In particular, the single regulated and dual regulated supplies ($95 and $145, respectively). I am also tempted to experiment with the raw boards (~$40) as a retrofit to the OPTO•DX to clear up a LPS 1.2 for the rest of my chain.
In theory, any LT-3045-based supply will have much better regulation than my PowerAdd Pilot2’s. The question is where in my chain will good enough high quality voltage regulation be good enough (vs even better voltage regulation with LPS 1.2 or some future Paul Hynes rail). If a basic LT-3045 supply is good enough (for the Kanex or OPTO•DX for example), that will free up a LPS 1.2 or future-precious Paul Hynes rail for some other component where they can have more impact.
Then there is the whole question of DC cables, shielding options, and grounding options (shunting ground back to a high quality ground, in a star configuration). I have no idea where that journey will go, or how worthwhile it will be, but I was surprised enough by the impact of my Ghent Audio DC cables that I’m now definitely paying attention.
If folks have experiences to share with these or other supplies (and cables), please share! DC power is clearly impactful enough that having a spectrum of available sources will go a long way to optimizing my chain without breaking the bank (or waiting decades for Paul Hynes rails). I’m developing a working theory (aka, educated guess) of what is going on, and I’m eager to try some additional configurations (and read about other’s experiences) to test and develop that theory.
TL;DR - DC power makes a huge impact, and there are a lot of options out there. I’m still a novice at this stuff with very limited experience, and eager to hear what others have learned and experienced.
So What About Clean Power on the Analog Side?
(NOTE: AC power is dangerous and can do serious harm to you, your equipment, your home, and your family…please use a licensed electrician for any AC work in your home, and ensure that everything is done consistent with local electrical codes…no hobby is worth putting yourself, your family, or your home at risk)
With all the upside for clean DC on the digital side, the obvious question is how does clean power impact the analog side of the chain? (eg, DAVE, and in my case, the bass units on my Voxativ 9.87’s)
Several months ago, after some gentle prodding from Roy (@romaz), I actually had a couple dedicated 30A circuits installed to my listening room, with really heavy duty electrical wiring (10-2 Romex) and high conductivity outlets. I use one circuit for my digital chain, and the other for my analog chain (DAVE + the class AB amps in my Voxativ bass units). Since these lines go directly back to my main electrical panel, it has a direct connection to home ground, with a breaker between it and all the other crud in my house electrical system.
It cleaned up a HUGE number of RF gremlins for me (wow!). Once I had this added clarity and resolution, it made it that much easier to hear the remaining digital hygiene gremlins in my system. In the experiments I describe above, it only takes a very brief listen to hear most of the changes, and I no longer have that time of day variability to deal with (the differences are very consistent and repeatable and very clear…makes iterating and experimenting MUCH faster and more effective)
Not everyone has a home setup that lends itself to dedicated electrical circuits for your audio setup. However, if you're close enough to your electrical panel, you can side step a LOT of the chasing down digital hygiene annoyances by spending a very modest amount on dedicated lines. Highly highly highly recommended, if practical for you.
Given this win, I took a leap to experiment some more with AC power hygiene. I recently purchased a used Topaz model 91002-31 ultra isolating isolation transformer. This beast of a transformer has 2500VA output capacity (basically 2500W) so more than enough to drive all my “clean” side kit. The theory is to have two huge transformer coils to buffer any spikes or ripples or other crap on the AC line. For background on this new toy see:
https://www.head-fi.org/threads/top...ansformers-for-affordably-clean-power.857448/
Initially, I heard even more of the clean power awesome (definitely worthwhile), but my unit had an annoying 60Hz hum and ran quite hot. Sometimes the hum was very strong, and other times almost silent (based on my reading, the level of buzz/heat is reflective of the crap the transformer is buffering out). Thus began a love/hate relationship with the Topaz, and day dreams of hiding it in my crawl space or behind a wall (alas, both impractical in my home).
Eventually, I tired of it, and decided to rewire the Topaz in a balanced configuration. Rather than hot being 120V and neutral being a nominal 0V (in the US, our 120VAC single phase home wiring typically has neutral and ground connected at the electrical panel), in a balanced configuration, hot is positive phase 60V, and neutral is negative phase 60V (with ground at 0V). In this configuration, the Topaz all of sudden was near silent, and ran MUCH cooler. Except for being a hideous block of metal, I now have nothing but love for the Topaz (CAUTION: Balanced power has particular rules for wiring, and not all devices are safe with balanced power…make sure your electrician knows how to work with balanced power, and make sure you’re not introducing ground faults between neutral and ground, and that all your power switches are double poled so they operate on both neutral and hot!)
Based on the positive experience with my analog chain, I bought a second Topaz, and have it wired in balanced mode for my digital chain, also on a dedicated circuit. Once I have completely optimized my digital chain, I’ll do real A/B tests on the digital side, but for now, big win on the analog side, and TBD (if any) win the digital side.
So all that is well and good, but what is that last wrinkle I mentioned about power hygiene on the analog side of the chain? (remember, by definition, the DAC lives in both the digital chain and the analog chain, and is impacted by both).
As I sat back with a big grin on my face listening to the best my system has every sounded, I reassessed my system based on what I’ve learned. It turns out I have two additional components that are mains connected, but they are in my analog chain (and connected in two different ways to my DAVE, so they definitely need my attention). Each of my Voxativ Pi bass units are connected to the XLR outputs of my DAVE, and have a built in 200W class A/B amp that are plugged into the same balanced isolation transformer as my DAVE, on the same dedicated 30A circuit I use for my analog chain.
As a final sanity check, repeated some listening tests with the Voxativ Pi Bass units unplugged from mains and from my DAVE. In this scenario, DAVE is the only mains connected device (through an isolation transformer to a dedicated electrical circuit), with Hugo mScaler mains isolated by a battery and optically and mains isolated on both input and output. This is now the most isolated I can make DAVE, with my best DC power dedicated to feeding the final component before Hugo mScaler (the Kanex HDMI audio extractor), and my best DC power dedicated to feeding the final component before DAVE (the “clean” side of OPTO•DX).
As a baseline, I listened in my above new optimal configuration, but with the bass units plugged in and connect to DAVE, turned on, but with the volume controls on the bass units turned all the way down. It took a while to acclimate to the new tone and find some passages in test tracks that were particularly challenging. Once I thought I had a good handle on it, I unplugged the bass units from the wall and removed the connections to my DAVE, and listened again.
With the bass units removed from my system, I heard a very small decrease in harshness and sibalance on some particularly challenging passages in some recordings (we’re talking slight here). Huh.
I reran the listening tests, but this time with the bass units plugged in and on, but with the connectors disconnected from the DAVE. Basically, is the vector for the noise back through the the XLR/RCA cables connecting my DAVE to the bass units, or back through the mains AC cables to the DAVE via DAVE’s mains connection?
With the bass units disconnected from the DAVE, but plugged into the wall and on, the slight increase in harshness was back. Clearly I have some work to do to insulate mains leakage from my bass units back to my DAVE. Huh^2.
As a next to last experiment, I moved my bass units to my dedicated digital outlets, leaving my DAVE and balanced isolation transformer as the only things plugged into my dedicated analog outlets. Now that little bit of extra harshness was gone again. DAVE appreciates having a moat on all sides, even the AC side!
As a last experiment, I moved my bass units to my dedicated analog circuit outlet, but connected at the wall, with DAVE connected to my balanced isolation transformer connected to the same outlet. Basically is the balanced isolation transformer capable of blocking any nasties from getting from my bass units to my DAVE?
Alas, only partially yes. I’m still hearing a hint of the harshness I heard before, but much decreased (I have to listen very carefully to catch it). I’m getting the best isolation by having the Voxativ bass units on a completely different circuit, and isolating DAVE on its own circuit behind an isolation transformer.
So how the heck is stuff getting through a big ass balanced isolation transformer from a class AB amp? (not even class D!!!). Even if I’m only hearing it on some extremely difficult passages where recording quality is already on edge, the fact that I can hear it at all is mind blowing, and reenforces how every little thing and every little vector is worth paying attention to.
There is a lot more to be said about my journey (so far) in analog chain optimization, but that is for some future opus. The TL;DR: get rid of EVERYTHING you possibly can, get the most transparent and ideal DAC you can (Chord DAVE), get the cleanest power you can, and get the fastest and most dynamic transducer you can (Voxativ 4D in my case, but I’m dying to hear the new 4phi drivers).
TL;DR - Component minimalism and clean power are ULTRA impactful on the analog part of your system. Anything you can do to isolate and minimize your analog components and get them clean power will be handsomely rewarded. Isolate your DAC mains as much as you possibly can. Much more to talk about on the analog side, but alas this write up is already too damn long.
Dang, This Is Tough to Follow…Can You Draw a Picture?
Note: there are some redundant options shown here (for example I usually power the HMS with the stock supply, but many of the above experiments are with the HMS powered by a PowerAdd Pilot2 battery. Similarly, if using optical in on the HMS, be sure to disconnect the USB path to make sure the HMS is electrically isolated. Lastly, when the “dirty” side of the OPTO•DX is powered by PowerAdd Pilot2, only the “clean” side is being powered by the LPS 1.2.
So What Did I Learn?
When it comes to isolating your digital chain from your analog chain, there is nothing better than an optical air gap
- No matter how much you filter on an electrical connection, there is always a residual something that can impact your DAC
- Put an optical air gap as close to your DAC as you possibly can
- Yes, bits are bits, but you can help your DAC immensely by making sure that only the bits get into it (and not all the crap that can come along with them)
- The AudioWise OPTO•DX is the digital optical moat I’ve been waiting for since I first got my Chord Blu2 to pair with my Chord DAVE…bravo Dan!
- The bang for the buck for putting a moat as close as possible to your DAC can’t be underestimated (cleans up a LOT of upstream sins)
- Even with the moat, I’m still some hearing small differences in upstream tweaks (although MUCH smaller than before)…they’re still worthwhile (for me), but you’re sitting pretty close to pinnacle audio with Chord Hugo mScaler and OPTO•DX
- Even a crappy little battery AA battery pack is a great diagnostic aid to see if investing in DC power hygiene will make a difference (hint: it really does)
- Closer you are to your DAC, more important better DC power is
- Great voltage regulation really matters (a LOT), and seems to have a bigger impact than mains isolation (on most kit)
- Have power supplies that are a LOT beefier than you need really matters (surprisingly true)
- Being disconnected from mains really matters (AC leakage, ground loops, etc), but high quality and stable voltage regulation seems to matter a lot more
- I’m surprised how much of a difference well shielded DC cables make (keep them as short as possible)…lot more to learn here
- If your listening space is remotely close to your electrical panel, put in some high capacity dedicated power lines, with beefy 10g wiring (WOW!)
- Isolation transformers are pretty cool, and protect the mains side of your analog chain
- Isolating your DAC on its own dedicated circuit is even better still
So What Is My Current Recipe?
With the caveat that noise and noise management is COMPLETELY environment and equipment dependent, here is my current recipe for digital nirvana with Chord Hugo mScaler and a Chord DAC.
- Use the optical input to your Hugo mScaler for the quickest path to highest quality audio. If you have the patience, USB input can be better still, but harder to get it to pinnacle levels. Even with optical, source quality (and power quality on source) matters, and I have no idea why. I am using a Kanex HDMI digital audio extractor connected to a NUC streaming endpoint, powered by an Uptone Audio LPS 1.2 DC supply. For quick and dirty, a Chromecast Audio puck ($30 used) powered by a clean DC supply is an amazing optical Roon endpoint for the Hugo mScaler. My next step is to elevate my USB chain based on these findings, so I can get more out of my higher resolution content.
- For Hugo mScaler, get the best DC supply you can possibly (most stable ground and reference DC voltage, fast with lots of reserves, ideally mains isolated if you can), pair it with a high quality DC cable (as short as practical, preferably shielded), and operate the DC supply at 15V and as far away from its current limits if you can (you don’t want to be straining the supply…you want it to have plenty of reserves to react to anything that is impacting the ground or reference voltages). I am using a PowerAdd Pilot2 battery (at 16V), but I have a Paul Hynes SR4 (and eventually a Paul Hynes SR7 DR rail) on order.
- Use an OPTO•DX to optically isolate the dual BNC output from your Hugo mScaler from your Chord DAC. Ideally, separately power the “dirty” and “clean” sides with high quality DC 12V power (mains isolated makes a difference), with the better supply (if you have to choose) on the “clean” side. The Uptone Audio LPS 1.2 is an excellent choice. DC cable quality makes a huge difference here as well.
- Isolate your Chord DAC (electrically and optically) as much as you possibly can. I have a dedicated electrical circuit, balanced isolation transformer, and no amplifier attached to my Chord DAVE (I direct drive single driver high efficiency Voxativ 9.87 speakers). If you have a Chord DAC that takes DC input power, I suspect you’ll get a win by mains isolating the DAC and having cleaner DC power as well (not verified by me).
- If you make any meaningful change in your DC chain, go back and revisit the other pieces. You may be surprised that the noise profile of your system has changed, and that there are different optimal configurations to be found.
I will be purchasing an OPTO•DX day one when Dan posts them up on his site. It is now a core part of my system, and Dan can’t have my preproduction units back until I receive my production unit. I recommend the OPTO•DX without reservations to all mScaler owners. With it, I am finally able to retire my Ferrited Cable of Shame, and no longer have to explain that monster to people that visit my home (“What the hell is that thing?” “It filters out RF that impacts the ground plane of my DAC” “I’m sorry, what’s a dack?”).
If your budget allows, I strongly recommend that you at least audition a premium DC power supply (ideally mains disconnected) like the Uptone Audio UltraCap LPS 1.2 and see what sort of difference it makes for you. Alex offers a very generous 30 day fully money back policy. Do yourself a favor and take him up on it and see if it works for you (I have no commercial relationship with Uptone Audio…I just love the work Alex and John are doing, and respect the stand up way they run their company). I’m sure there are even better power options out there. When I (finally) get my Paul Hynes supplies, I’m looking forward to repeating these tests to see how this scale with power quality.
I am far from an expert on DC cables, but if the LPS 1.2 makes a huge difference for you, get a high quality shielded DC cable from folks like Ghent Audio (price is right too). Again, I’m sure there are other great options out there. Given the lift I heard with the Ghent cables, that is definitely on the list for future experimentation.
I’m finding what a lot of others have already discovered: you can never have too many ultra high quality DC rails, even the lowliest digital equipment benefits from better DC power, there are always difficult Sophie’s Choices about which component gets the best power, and all this makes the excruciating wait for precious Paul Hynes SR7 DR rails 100x more painful.
If you’re serious about optimizing digital audio, get an OPTO•DX to optically isolate your Chord DAC from HMS, and as many high quality DC cables and power supplies as you can afford, and start experimenting to see what makes the biggest difference for you.
My next step is to take my above learnings and reassess how I lay out, power, and connect my equipment (are there less expensive power supplies that are good enough? Can I lose the batteries but keep the SQ? Will even better power supplies elevate things more? etc). With that done and system stabilized, the itch will return, and it will be time to see if there is even more magic to be found by diving ever deeper into the rabbit hole.
Whew…
As I reread the above, I find myself shaking my head in awe at the amazing audio journey I’ve been on the past several years. As is clear from my Blu2 review, I used to think that mScaled music was the ultimate, and there was no way things could get better. Boy was I wrong. In ways known and unknown, the rest of my system was holding it back.
Since mScaler finally gave me an absolute reference, seemingly every month or so, I’m once again hearing the best sound quality I’ve ever heard. Amazingly (and wonderfully), I’m finding that ultra fast transducers, absolute minimalism in your analog chain (no speaker cross overs, no amps, etc), obsession over power quality and stability, and relentless digital hygiene continue to reveal more and more (and MORE) magic and emotion in the music. The mScaler sets an absolute reference. As such, it handsomely rewards supporting equipment that takes away as little as possible from that reference.
Getting past all the equipment and speaking to the music, the humbling thing for me is the realization that all this amazing emotion and magic was in the recorded music all along, but that hadn’t been heard and experienced except by those precious few that were in the studio for the original recordings. What an unexpected blessing and gift to be able to experience music like this!
All that being said, the more I learn and hear, the more optimistic and excited that there is even more to learn and hear. Incredible stuff, and a true gift to be able to experience this level of artistry in our living rooms.
Addendum
9/14/19
To the question of whether I have tried my new Paul Hynes SR4 in this setup, I have. SR4-19 (@15V) has a huge positive impact powering the HMS (vs PP2 battery). It also has a huge positive SQ impact when powering the end point NUC @ 19V, with the HMS on PP2 battery. I slightly prefer the NUC End Point on PP2 @ 19V + SR4 @ 15V on HMS combo, to the NUC on SR4 @ 19V and HMS on PP2 @ 16V combo. Alas, I don't have 2 SR4's to see what the marginal lift is by running both the end point and HMS on SR4's.
My current setup is end point NUC on PP2 battery at 19V, USB out (via Lush^2 USB cable) to ISO Regen (on LPS 1.2) to USB out (via USPCB) to a Matrix Audio XSPDIF 2 DDC (powered via USB from the ISO Regen, so LPS 1.2 sourced power) to optical toslink to HMS. I am also powering the TX side of ODX with a PP2 battery, and the RX side of ODX with a LPS 1.2.
9/15/19
My NUC on Paul Hynes SR4 had a huge impact, almost as much as when I had the SR4 on my HMS. I was surprised.
My core theory is that anything touching reference voltage or ground plane on a DAC is audible. Anything that does not can not (assuming bit perfect). Anything we hear can come from the outside, or be induced in the DAC. Somehow, power to the NUC was impacting what was getting to my DAC (through USB regenerators and DDC to optical toslink)
Here is my working theory:
Perfect power supply does the following:
- Insulates device from electrical cruft from outside world (these will impact ground and reference voltage in the device, which may work down the the chain)
- Prevents electrical cruft from device getting to outside world (do no harm)
- Prevents ground loops (corrupts ground plane, which will impact everything connected to that ground plane)
- Has perfect voltage regulation (low ripple, stability) (reference voltage needs to be stable and reference)
- Has huge current capacity and fast response to changing loads (don't want instantaneous current draw to impact ground or reference voltage)
Based on this, the internal loads on the device can self-induce noise on the ground and reference voltage planes, and somehow those are getting propagated to the DAC, either through direct connection or getting induced in the DAC. It can be over the air (RFI) or something to do with signal integrity in the optical signal that is inducing noise in the DAC. Could the receiver chip in the DAC be inducing noise in the DAC based on the signal integrity? Don't know, but two ISORegen USB regenerators in series sound clearly better than 1 in my
By having the best PS you can get for anything close to the DAC (including the DAC), when there is a big current demand, it protects the device from having the lights dim, which whacks reference voltage and ground plane.
Anything touching reference voltage or ground plane on a DAC is audible (ditto clock, but I haven't done anything with clocks yet) You will not hear anything does not (assuming transmission is bit perfect). The maddening thing is trying to come up with theories for how things way away from the DAC could be influencing the DAC.
