[574206]

Thats my understanding too.
Rob's DAC collects the digital data(clocked by the source) and stores the bits in fifo order in a buffer in the FPGA.
Then using its own clock cycles processes them further.

Any input jitter fluctuations dont matter anymore.

Its kind of like filling a bathtub(buffer) while turning open and close the crane(jitter).. but the drain is set to a fixed opening so the water comes out with a stable constant.

I guess I just need time to digest all the recent intelligence.

So I'm guessing from cheap to high-end clocks, the jitter fluctuations don't matter in all cases.

Like I mentioned, jitter was never really a concern as overall SQ with high-end clocks but that's a positive note if it don't matter anymore.

In any case, I'll eventually have two off mains Music sources over time. A State of the Art Clocked Toslink and a un-clocked USB optical powered by Ultra Caps 5V which I use for gaming currently but will convert to Music over time.

But it's nice to know since jitter is becoming a thing. In any case, just hoping to learn and grow for the next Chord iteration of products to not be caught so off guard.

 

[Pulcino]

First, many thanks to Sir Rob Watts for building such solid products. To be able to feed 16-bit instead of 32-bit Toslink into my Hugo₂ is amazing. I believe most DACs nowadays need to convert to 32-bit before proceeding. And to be able to handle Clock Frequencies near the lowest common denominator. I don't run the standard 44 kHz Clocks like mainstream products / streamers, I run the highest-performance reference low phase noise super low MHz Frequency optimised purely for Redbook use. I believe in High Quality Well-Mastered Sources, not High Resolution Sources. Too many HiRes/DSD/etc/ gimmicks, I just want purity. I like to keep it simple because I have a niche use case of being CIEM-only, Redbook-only, Toslink-only, etc. and keeping my chain within a certain "Goldilocks"-zone. I don't want to get too crazy and out of control which can easily happen in the Summit-Fi world. It's also amazing that pre-emphasis and de-emphasis are built into Chord DACs. I have not had a chance to test, but since I'm a Redbook fanatic, I will try one day.

Jitter is a very small part. It's about the quality of the source and attempting to re-create what the Artist originally intended. For I, since I was an 80s kid, I try to re-create a Time Machine to take me back to that early 80s era. With the help of high performance state of the art external clocks and Chord DACs, it gets me there. I can close my eyes and the sound is Analog as can be given current technology and it takes me back to that 80s era. But you can only do this when your entire chain is super transparent then feeding it well-mastered sources from original 80s recording before sound processing became a thing. I exclusive run only WAV as it's audibly transparent once your chain gets to a different level. If you run WAV on a Apple Dongle, there's not much difference. The higher up your chain in getting super transparent, the more you notice the little things can make a big difference.

For example, I only use specific Masterings such a this (the CD design is easily recognisable by Purists):



To be able to listen to these Artists in a well-mastered format as purely and faithfully possible is a dream. So much media out there that were never well-mastered back in the day makes me sad. It just makes me appreciate more, but the costs of well-mastered media is rising.

Imagine if you had a Goto CD Player that plays Redbook really well. You don't worry about Jitter. You might worry about upgrading the clock in that CD Player for the highest performance possible. You treat the CD Player independent of the DAC. I don't understand the obsession with everything having to revolve around the Chord DAC instead of focusing on the chain as a whole. For I, the Chord DAC is just a conduit to translate from a high performance digital player source to it's fullest. If you feed the Chord DAC 128Kbs MP3 Audio File, it's going to translate that the best it can. If you feed it a highly-optimised clocked source that faithfully reproduces what the Artist intended, the Chord DAC will translate that to it's fullest. GIGO (Garbage In, Garbage-Out). For I, I built a Time Machine to bring me back to early 80s so I couldn't be happier.

A clocked source to a FPGA DAC is more a dcS philosophy and implementation than Chord philosophy. That's why 95%+ of dCS users purchase a Clock along with their dCS DAC as one could not live with just a basic dCS DAC alone once experiencing the Clock effect. I just try to combine both worlds to Max my CIEM experience. There are no chains that are built specifically for CIEMs in mind so I'm glad to build the most robust possible in this modern age. I'm trying to simulate old school well-mastered sources with modern technology to the fullest extent. Any DAC > Hugo₂ is overkill in power and having to throw an amplifier in the mix, so Hugo₂ pretty perfect for my use case. It helps clarify my chain because I'll definitely go crazy with a DAC > Hugo₂. Of course, there's nothing wrong with a dCS alone or a Chord DAC alone.







I get the same results as a "dCS external clock + dCS DAC" with my "State of the Art external clock + Chord DAC". Sure, the basic Chord DAC can stand on it's own, but "Clock+Chord DAC" takes me to another world (Time Machine) like it would with a dCS chain.

I experience the same as dCS Marketing speak:



When an MSI laptop with Toslink OUT is used, jitter is never discussed. When someone is using a Toslink Streamer with Average basic Clocks, jitter is never discussed. I think jitter is mentioned only as a defense mechanism for something unknown. For these other Toslink solutions, all I see are cheap embedded clocks and solutions. Why not use high performance clocks instead of the cheap stuff on your Toslink source? dCS does it. I also combine with a high performance realtime low latency Audiophile OS and the transient response is taken to new levels. The Chord DAC definately does it's job and can handle a modern high performance feed. So it can be a bit hypocritical to mention jitter when ANY Toslink source also employs clocks, but I understand that this can be such an unknown territory. Just follow dCS's example. dCS users don't mention jitter as they know they are using high quality not AliExpress gear.

This chain performs so well, I started investing in Ultra Capacitors to slowly upgrade the Power in the next few years. Of course, they are passive power supplies that don't run off Mains. I would never use Active Power Supplies with Chord DACs. I don't believe in gimmicks like LPS and 1TB files for each track / song, I believe in solid fundamentals. I'll upgrade the power on the Clocks @16V to Ultra Caps with 9000 Farad one day.

For now, I'm focusing on the world's best 5V @ 3000 Farad power supply:







I'll build 3.3V, 5V @ 3000 Farads this Summer and 12V, 15.5V @ 9000 Farads in the upcoming years. I wish I could build a 12V for you Summit-Fi guys this year, but Ultra Caps get pretty expensive. For this solution, the Mains is always plugged-in to recharge the UltraCaps when low. So maybe 3 minutes a day running 24/7 depending on use. It then goes back to full-time optical isolation mode which 100% cuts off from Mains during performance use. I think it's dynamic up to 1000Amps, so it's will handle dynamic swings beautifully. Solid Core wiring of course... I don't cut corners. Zero regulators for 3.3V, 5V, 12V and 15.5V, just pure goodness.

For clocks, it's about keeping it's composure even with high dynamism and fast attacks. It's about placing vocals and instruments in a properly spaced SoundStage. The high performance clocks do their job and the Chord DAC does it's job. I would only be concerned about jitter on cheap Toslink clocks / embedded clocks like MSI Laptops, Chromecast Audio, AliExpress devices, Average Hi-Fi Streamers, etc. With a dCS clock or high performance clocks, jitter is about political not performance.

This Soundstage is not much of an exaggeration as Spatial Audio is properly placed within an large Soundstage. Depending on the Mastering of course. I run Japanese Masterings too as a good amount have low noise floor. But one only experiences this once their chain is tight. The precision, layered Bass I did not expect with external clocks, but it's welcomed.

Another advantage is being to able to listen @ super low volumes with zero to very low noise floor and zero fatigue while retaining all the details. So for ear health reasons alone, this chain is easily justifiable for my use case. Average HiFi clocks are bright and fatiquing, state of the art are just smooth and clean.

Hi, would you be willing to share your clock design with us? Guess all DIY fans in this forum would love to see it.

 

[574206]

Hi, would you be willing to share your clock design with us? Guess all DIY fans in this forum would love to see it.

Maybe in September, it's not my design but it's an Italian design. I'm out of energy and my whole main purpose on being back at this site is to complete my Ultra Capacitor Power @5V for my Hugo₂ which I've been running 24/7/365 since COVID started. I hope to report success, but the Ultra Capacitor needs 3 minutes off Mains to charge so I don't know how that plays with the Hugo₂ since it needs a constant charge with zero interruption to stay in 24/7 mode (forget the tech). I don't know that when it switches back and forth between Mains if it will interrupt the charging process. Also, my chain is completely disassembled ATM.

I can give you a sneak preview though.

These are owners of two major Audiophile companies, so they likely run 6 to 7 digit chains. It's very difficult to 'Wow' them as they heard it all.

But they are Amateurs, for I, my jaw dropped for the 'Wow' moment in .5 second. For them, it took a few notes. Amateurs. I also questioned to myself if I really owned this chain. That this is a dCS quality chain, how can I possibly own and experience this just for CIEMS? I'm not worthy.

I am running the best performance clock at single digit MHz. They are running a clock in the 20s MHz for Ethernet. It's a $5000 Network Distribution device. They replaced an average clock with a SOTA (State of The Art) clock. My Infrastructure is fully optimised around SOTA clock. There's is a replacement so my dedicated SPDIF source should have a larger difference. It bring tears to my eyes reading as I know exactly what they are experiencing, but I'm on a CIEM budget, not the unlimited budget of Audiophile CEOs.

I'll be back in September. My USB connector parts are here, so I need to focus on DIY in August.

 

[AnalogEuphoria]

Ok I’ve said my bit but you might want to tone it down a bit because this whole external clock obsession is taking over the thread.

 

[AnalogEuphoria]

I have a question for @Rob Watts

In your opinion, for two channel systems what amplifier topology would be best suited for your DAC design? I know you recently said class D is a big no for preserving the transients. I have been reading about class G which is interesting but not used by many.

 

[Rob Watts]

Class AB is absolutely fine when running with my second order analogue noise shaper topology, as the OP stage distortion is completely eliminated - no change in distortion (discounting a small change in 2nd harmonic) when loaded or unloaded. If efficiency is important, then Class G (when PSU voltages are changed in large steps) or even better Class H (when PSU changes with high resolution tracking the required audio OP voltage) are fundamentally better than Class D, which creates transient timing errors and subsequent poor sound quality.

The upsides to Class G/H is much better efficiency - and you could bias the OP stage into Class A too (not that I would ever need that). The downsides to Class G/H is cost. The audio downside to Class G is that large steps in PSU voltage that is audio related will create HF errors that I expect would be audible. Class H doesn't have this problem, but there is always a risk that modulating power rails even linearly with Class H may have SQ issues, so it needs to be done with considerable care. One alternative, that would certainly not have any SQ problems, is to modulate the power rails with volume only, and that wouldn't degrade SQ and would improve efficiency as most people listen at low volumes anyway.

 

[miketlse]

Class AB is absolutely fine when running with my second order analogue noise shaper topology, as the OP stage distortion is completely eliminated - no change in distortion (discounting a small change in 2nd harmonic) when loaded or unloaded. If efficiency is important, then Class G (when PSU voltages are changed in large steps) or even better Class H (when PSU changes with high resolution tracking the required audio OP voltage) are fundamentally better than Class D, which creates transient timing errors and subsequent poor sound quality.

The upsides to Class G/H is much better efficiency - and you could bias the OP stage into Class A too (not that I would ever need that). The downsides to Class G/H is cost. The audio downside to Class G is that large steps in PSU voltage that is audio related will create HF errors that I expect would be audible. Class H doesn't have this problem, but there is always a risk that modulating power rails even linearly with Class H may have SQ issues, so it needs to be done with considerable care. One alternative, that would certainly not have any SQ problems, is to modulate the power rails with volume only, and that wouldn't degrade SQ and would improve efficiency as most people listen at low volumes anyway.

What sort of increase in cost are we talking about for Class H, I am talking ballpark figures (eg 5% or 500%) and not any hard numbers that anyone would hold you too.
Also but I suspect that you may not be authorised to talk about, have chord explored Class H for any of their amplifiers?

 

[AnalogEuphoria]

Class AB is absolutely fine when running with my second order analogue noise shaper topology, as the OP stage distortion is completely eliminated - no change in distortion (discounting a small change in 2nd harmonic) when loaded or unloaded. If efficiency is important, then Class G (when PSU voltages are changed in large steps) or even better Class H (when PSU changes with high resolution tracking the required audio OP voltage) are fundamentally better than Class D, which creates transient timing errors and subsequent poor sound quality.

The upsides to Class G/H is much better efficiency - and you could bias the OP stage into Class A too (not that I would ever need that). The downsides to Class G/H is cost. The audio downside to Class G is that large steps in PSU voltage that is audio related will create HF errors that I expect would be audible. Class H doesn't have this problem, but there is always a risk that modulating power rails even linearly with Class H may have SQ issues, so it needs to be done with considerable care. One alternative, that would certainly not have any SQ problems, is to modulate the power rails with volume only, and that wouldn't degrade SQ and would improve efficiency as most people listen at low volumes anyway.

I was curious because Arcam has a class G design that always runs in class A in low to medium volumes but uses a second rail putting it in to AB when more power is required. Seems like a good solution for people that have efficient to normal sensitivity speakers and don’t listen too loud. Mostly being in class A all time would mean preserving the low level details that can be scrubbed out by crossover distortion. After all I think most of us would agree that small signals is a big part of what makes Hifi sound so good.

 

[ChrisGB]

I was curious because Arcam has a class G design that always runs in class A in low to medium volumes but uses a second rail putting it in to AB when more power is required. Seems like a good solution for people that have efficient to normal sensitivity speakers and don’t listen too loud. Mostly being in class A all time would mean preserving the low level details that can be scrubbed out by crossover distortion. After all I think most of us would agree that small signals is a big part of what makes Hifi sound so good.

I have one, it runs up to 20W in class A, 120W in class A/B. Coupled with 92db/W speakers, it exhibits both finesse and muscularity. I have auditioned and listened to a fair few amps over the years and this one is natural sounding, very detailed, but no hint of graininess, brightness or harshness, unless the music calls for it. Imaging / soundstage is good and it doesn't heat up my listening room!

 

[miketlse]

I have one, it runs up to 20W in class A, 120W in class A/B. Coupled with 92db/W speakers, it exhibits both finesse and muscularity. I have auditioned and listened to a fair few amps over the years and this one is natural sounding, very detailed, but no hint of graininess, brightness or harshness, unless the music calls for it. Imaging / soundstage is good and it doesn't heat up my listening room!

Must be better than the Arcam Solo Neo.
When I use that as a power amplifier for the Mojo, it removes all the clarity and musicality.
Rob has spoilt us, second best is no longer good enough.

 

[ChrisGB]

Must be better than the Arcam Solo Neo.
When I use that as a power amplifier for the Mojo, it removes all the clarity and musicality.
Rob has spoilt us, second best is no longer good enough.

Yeah, it's a different kind of device altogether.

I've always been quite sensitive to amplifiers that seem to add "false" detail to the sound. I've always called it "grain" as it's an artificial texture around sounds that you wouldn't hear on the original instruments. It's almost like certain sounds become etched away from their background. It's all very impressive unless you know how the instruments are supposed to sound. It kind of mirrors the visual effect where film grain gives an image extra contrast and pop. Some very big names have this quality to my ears.

Reading this thread and in particular, Rob's assessment of the effects of how noise floor modulation affects the sound, I'm wondering if this is present in amplifiers that I perceive to be grainy?

 

[miketlse]

Reading this thread and in particular, Rob's assessment of the effects of how noise floor modulation affects the sound, I'm wondering if this is present in amplifiers that I perceive to be grainy?

It wouldn't surprise me.
Other designers have little incentive to be ruthless about electrical noise and RFI, if they help fool owners into thinking that their hifi is far better (and more sellable by dealers) than it is.

 

[GuiltyRocker]

I have a question for @Rob Watts

In your opinion, for two channel systems what amplifier topology would be best suited for your DAC design? I know you recently said class D is a big no for preserving the transients. I have been reading about class G which is interesting but not used by many.

Check out the Parasound JC amps, they run class A to 15-25 Watts, then class AB, they are truly amazing with Chord DACs. Parasound amps have that meaty full sound that I find really complements Chord.

 

[Kentajalli]

Class AB is absolutely fine when running with my second order analogue noise shaper topology, as the OP stage distortion is completely eliminated - no change in distortion (discounting a small change in 2nd harmonic) when loaded or unloaded. If efficiency is important, then Class G (when PSU voltages are changed in large steps) or even better Class H (when PSU changes with high resolution tracking the required audio OP voltage) are fundamentally better than Class D, which creates transient timing errors and subsequent poor sound quality.

The upsides to Class G/H is much better efficiency - and you could bias the OP stage into Class A too (not that I would ever need that). The downsides to Class G/H is cost. The audio downside to Class G is that large steps in PSU voltage that is audio related will create HF errors that I expect would be audible. Class H doesn't have this problem, but there is always a risk that modulating power rails even linearly with Class H may have SQ issues, so it needs to be done with considerable care. One alternative, that would certainly not have any SQ problems, is to modulate the power rails with volume only, and that wouldn't degrade SQ and would improve efficiency as most people listen at low volumes anyway.

going back to my original post, and how we arrived here, another question if I may:
- OK we have established that class D is not good for our use. But I suggested it for its n
1 - efficiency 2- small size 3- low cost , in order to get a possible affordable TT3 that could drive real speakers.
- I understand that you prefer class H or G. I take it class G is cheaper to implement (right?), but I doubt it would be small and light.
- Class H , would cost the most, and will be smaller and lighter.
If I am correct, then Class G implementation would turn our hypothetical TT3 rather large and heavy, and class H would make it expensive and not so affordable!
Is there no way of having our cake and afford to eat it too.
Perhaps a redesigned RW version of a cost effective class D that wouldn't butcher the sound?
Just thinking (& dreaming) out loud . . .

 

[rkt31]

I use tt2+ hms directly to ahb2 amp. Now there are cheaper options available too like topping LA90, one if the best measuring amps.