Holography and Rube Goldberg Power Supply 2.0
I mentioned in a previous post that I like to start out with an enquiry and then apply it to my set up which leads to trying out various things...
I was wondering if the Wi-Fi signal from my Comcast/Xfinity cable modem affected either the dedicated local power or Ethernet feeding my audio at close proximity in the equipment closet ....
I don't have a direct answer for that but this led to some discoveries that improved my streamed audio by about 10% along with streamed video clarity by some noticeable amount. Then leading to more significant SQ improvements as described in this post.
The Equipment Closet
I learned that both my Supra AND JCAT Signature Gold Ethernet cables have shields connected end-to-end. I use two Ethernet cable outputs from the cable modem feeding two TP Link Fiber Converters, one dedicated to the CX and another to the house router, each separately located about 50' away. I used to have two dedicated LPSs, then went to two iPower SMPS - a 12V Elite with DC earth ground opened to the Cable Modem and a two-prong 9V iPower X for the two Fiber Repeaters. My testing sequence was flawed as after discovering both my Supra and JCAT Signature Gold have end-to-end ground connectivity for the shield, I cut the metal housing that fits into the connector on one end of one Supra. I should have waited for the power supply update which put the cable modem AND the two fiber repeaters on the same ground reference and THEN tested end-to-end shield connectivity first vs. shield open at one end...
Before that though, I could not hear a difference between the Supra and the $$$JCAT feeding the TP Link to the CX during streaming. Fast-Forwarding a little,
I noticed on one end of the Antipodes Audio [AA] supplied CX/EX Ethernet cable that I could see wire strands through only one of the clear plastic connectors. I drilled a 1/32" hole to try a continuity test where sticking a wire through the hole connected to my continuity tester provided momentary continuity to the side of that shell but not to the shell on the other end. I'm assuming that the wire I poked through could not make perfect contact to the strands inside the shell. This implies to me that this is a drain wire for a shield that is only connected at one end. I tried sticking a pin through the surface of the cable to see if it would connect to the shield but was unable to get continuity. I did not receive an answer from AA when I inquired about its cable construction and to see if they had any of these cables laying around that I could buy.
I tested the JCAT and the original AA Ethernet jumper that came with the CX/EX between the B-Side [galvanically ungrounded] of the EtherRegen to the EX and found the AA supplied cable [with wire strands end at the EX] was 95% as good as the JCAT.
The 10% audible improvement came after implementing a 24V iPower Elite [leaving DC Earth Ground intact] feeding two TI DC-DC converters - one set to 12V to the Cable Modem and the other custom-resistor-optioned to 9V to the two Fiber Repeaters. Although the common ground feeding everything is proper practice and likely contributed to
some of the SQ improvement, I believe the DC-DC converter after the SMPS is the crowning improvement taking the 3% iPower SMPS-only improvement to 10%.
It would have been proper due diligence if I had compared both-ended shield vs. single-ended shield after the DC-DC converters after providing common DC ground to both the Cable Modem AND the Fiber Repeaters. Also, if I had on hand similar length UTP-6 to test vs. shielded. 20-20 hindsight...
On to the more juicy stuff...
EX and Rube Goldberg 2.0
I then enquired as to the degree EMI from the EX mother board could influence the USB signal or power within the mother board chamber or even leaking out above the perforated top metal plate where I mounted the TI USB galvanic repeater and its DC-DC converters. Up to this point, I had temporary unshielded generic computer cable/pin converters jiggered from the mother board USB connector to the USB-B connector on the TI USB Repeater.
The original AA USB connectorized USB cable is a miniature 10-pin unshielded wired connector from the mother board spliced via a shielded in-between cable to a larger 10-pin unshielded wired connector to the dual USB ports on the back of the CX/EX. That cable is 50% unshielded..... [That was painful to read wasn't it?
]
In short, I mounted a ~ 6" x 6"metal plate to the underside of the 3/16" thick aluminum plate that covers the motherboard enclosure and double mesh shielded what ended up as just the D+/D- USB leads from the miniature 10-pin USB mother board connector and directly soldered them to the TI USB Repeater board. This produced a noticeable improvement in calm that I don't have a % magnitude for.
Since adding the metal plate blocking the open slots and larger ribbon cable openings, the EX temperature indicator has remained unchanged between 51C - 53C on average. I did read that the Celeron quad-core is capable of running between 100-105 degrees C for demanding gaming applications but the cooler the processor, the longer the life span.
The real paradigm change came from adding a third 5V DC-DC converter off the 15V Elite SMPS to the input of the TI USB Regen, bypassing the mother board's 5V USB regulator feed. The holography expanded by 60% on top of previous improvements. This was far greater than the factory G3 upgrade that I'd put at 25% improvement at best.
The theme here is that eliminating in-between linear regulators and going direct to load with the DC-DC converter output has 100% of the time made a paradigm level improvement to the sound.
Example 1 - SingXer SU-2 DDI:
iPower X SMPS into Sparkos or Belleson 5V regulator = 15%.
iPower X SMPS to 5V TI DC-DC converter = 60% on top of the original 15%
The 5V feeds a bank of six or so 3.3V linear regulators in the SU-2.
I feel really confident that if there were enough real estate, and I had the necessary SMD soldering skills, eliminating the 3.3V regulators would take the SU-2 significantly higher. I table that for a potential future Rube project.
Example 2 - AGD DAC:
Replacing integrated AC Regen with iPower X SMPS to digital section = 15%
Adding single 5V TI DC-DC converter after iPower X to feed digital section = additional 15%
Adding 3.3V DC-DC converters [bypassing 3.3V linear regulators] directly feeding L&R channels' CPLD Processor which controls the discrete R-Ladders = at least 20% precision improvement over all previous improvements.
Back to the EX
- Added a 5VDC-DC converter off the 15V iPower Elite SMPS to replace the mother board 5V USB supply regulator - originally going to the EX back panel 5V USB output connector - but now going to the input of the TI USB Galvanic Repeater. There was over a 50% improvement in SQ emphasized by very noticeable holography expansion. It went well beyond the G3 factory PS upgrade for both CX and EX which I pegged at up to 25% improvement.
1) 15V iPower Elite [adjacent right-G3 DC-DC converter from factory update]
2) Enclosed and isolated iPower X SMPS for TI USB galvanic repeater output [3]
3) TI USB Galvanic USB Regen [feeds Innuos Phoenix]
4) TI DC-DC 5V Converter fed by [2] isolated iPower X for galvanic output
5) Latest 5V DC-DC converter feeding TI USB Regen input - replacing 5V USB power from mother board regulator
On to the
Innuos Phoenix USB Regen...
The Phoenix has two Sean Jacobs Design 5V supplies, one dedicated to the OCXO and the second to the USB chip. Both the OCXO and chip nomenclatures are covered which I left in place.
There was ample space inside the chassis to add a 9V iPower X encased in a sealed, grounded aluminum enclosure as well as a TI DC-DC converter optioned for 5V.
I replaced the 5V LPS feed to the USB chip and the holography expanded by at least another 60%.
Listening to Chicago's "Beginnings" beforehand, the opening guitar strumming showed up around 115 degrees with a secondary appearance emphasized around 15 degrees in the soundstage.
Now the strumming is contiguously holographic and evenly distributed within a hemisphere-like setting.
I was reading the Celeron Quad manual which emphasized importance of the power supply design to provide instantaneous power for the processor to deliver maximum performance. John Swenson in his "DIY DC Cables" post emphasized how digital circuits do not draw power in a steady-state manner and thus the importance of minimizing DC cable inductance using star-quad to deliver transient current. AA tutorials emphasize "fast" power supplies [G3 and G4].
While I can't provide scientific measurements, the listening at least shows positive correlation.
For me, this implies additional SQ runway constrained only by real estate within the chassis and soldering skills depending on how far I'm willing to take it.
Sorry if the run-on sentences and detail made it difficult to breeze through. FWIW.