Overall Isolation - network, USB, and power

[gstew]

As I said recently on this thread, I try to be an empiricist with an open mind. I have ideas and theories, yes, but I endeavor never to ignore data to fit a preconceived dogmatic point of view.

But this did not come easily.

Like many other computer scientists and network engineers on CA, it took me a while to get my head around the idea of audio SQ differences in the supposedly data-only parts of the reproduction chain (network, USB, etc).

The epiphany for me was reading @JohnSwenson 's posts and recognizing that the audio effects of digital data transmission are due, not to the data transmission per se, but to the analog noise in leakage loops that can flow across the chain, ultimately polluting the analog sections of the chain, and from there, degrade the SQ your ears hear. That is a simplification, I know, but it gets to the heart of the matter.

So even though the science of isolation techniques is still evolving, this epiphany has helped me to accept the notion of tweaks for SQ improvement in areas where conventional engineering wisdom would say it does not matter.

Rajiv,

 

I would suggest that ‘conventional engineering wisdom’ is generally engineering dumbed down for us laymen. What really frustrates me are those who know some conventional engineering wisdom that insert themselves in threads such as these where people are actually grappling with the appropriate issues and finding resolutions and they say "It is all bunk, you cannot be hearing what you say you hear"... while they have no intention to try these things themselves... or any openess or ability to hear the differences we experience.

 

Luckily we have none here at this time!

 

All of this experimenting and different situations providing different results has been keeping my poor little brain occupied recently. It went into overdrive a couple of days ago from these posts by Barrows:

 

Question: Is everyone here making sure their routers/switches/NAS etc. are connected to different circuit branches (preferably on the opposite phase) from their audio systems?

I am very hesitant to try Ethernet cables from Audioquest, as they make no effort to insure equal wire lengths by virtue of their construction. For example, BJC 6 and 6a both use Belden bonded pair technology, which assures that all individual wires are of equal length, and hence have equal TD. If one must use an "audiophile" cable, i would suggest trying Wireworld, as their cable construction does guarantee equal length wires.

 

And

 

On FMCs, are there any real Networking experts here? Electrical to optical converters do add jitter, so it may be that adding an FMC will contribute to a higher error rate on Ethernet transmission. A higher error rate (Ethernet errors are of course corrected by re-sends) will likely result in higher processor activity in the µRendu, and more spiky noise.

I would love to know (and this can be confirmed by network analysis) for sure if the addition of an FMC results in a higher error rate in Ethernet transmission.

 

And this triggered a flash-back to some of the comments by John Swenson on what spawned the Regen. I don’t know how many here followed that, basically he had been trying to understand the reasons that a DAC with a well-designed power supply and galvanic isolation / reclocking of the I2S just before it went into the DAC chip still exhibited sonic differences depending on the source and source configuration. Here are some of those key posts:

 

http://www.computeraudiophile.com/f10-music-servers/mac-mini-version-computer-audiophile-pocket-server-music-server-step-step-17666/index6.html#post368326

 

http://www.computeraudiophile.com/f6-dac-digital-analog-conversion/universal-serial-bus-industry-standard-cables-connectors-and-communications-protocols-between-computers-and-electronic-devices-isolator-advice-needed-23475/#post399803

 

http://www.computeraudiophile.com/f6-dac-digital-analog-conversion/are-all-asynchronous-universal-serial-bus-industry-standard-cables-connectors-and-communications-protocols-between-computers-and-electronic-devices-chips-implementations-created-equal-19819/#post309021

 

http://www.computeraudiophile.com/f27-uptone-audio-sponsored/uptone-audio-regen-22803/index13.html#post418669

 

http://www.computeraudiophile.com/f27-uptone-audio-sponsored/uptone-audio-regen-22803/index27.html#post428008

 

http://www.computeraudiophile.com/f27-uptone-audio-sponsored/uptone-audio-regen-22803/index27.html#post428114

 

http://www.computeraudiophile.com/f27-uptone-audio-sponsored/uptone-audio-regen-22803/index28.html#post428509

 

http://www.computeraudiophile.com/f27-uptone-audio-sponsored/uptone-audio-regen-22803/index29.html#post430128

 

http://www.computeraudiophile.com/f27-uptone-audio-sponsored/uptone-audio-regen-22803/index37.html#post435504

 

http://www.computeraudiophile.com/f27-uptone-audio-sponsored/uptone-audio-regen-22803/index55.html#post447479

 

http://www.computeraudiophile.com/f26-sonore-sponsored/sonore-microrendu-27389/index12.html#post518167

 

I know that the Ethernet protocol does not function the same way as USB. There are of course similarities (both are transmitting/receiving an electromagnetic signal, in the case of USB always over a wire, in the case of Ethernet sometimes, both have PHY layers at the receiving end to best extract the data from the received signal, etc). But they do function in VERY different ways… and I am not much more than a slightly-educated layman on both.

 

But re-reading John’s comments, I suspect that approaching it the way he did that resulted in a whole new class of audio products, the USB signal regenerator (starting with the Regen, but moving on from there) MIGHT be the way that would produce the better solutions.

 

Sadly, I’m not sure any of us here are the ones to do that and John is otherwise occupied at the moment. But I suggest keeping John’s USB transmission issues and resolution analysis in the back of our minds (as Barrows appears to have) in looking at this.

 

Given that, here’s the rules of thumb that have driven my network configuration:

 

1. Getting the data in to the endpoint (uRendu, Aries, SMS-200, Raspberry-Pi, whatever) with the least amount of electrical noise created in that endpoint will produce the best sonic results.
    
   
2. Achieving that means getting only the best-formed Ethernet packets to the endpoint’s receiver and only the packets needed (extra packets, poor signal waveform, etc means either more electrical noise receiving or ignoring them or re-transmits or more packets per transmission)
    
   
3. Reducing noise generated in the Ethernet transmission process improves SQ. That means clean and quiet power supplies both feeding AND inside of the gear, good transmission processes / media (cables), galvanic isolations (signal transformers, fiber media, WiFi), etc.
    
   
4. Clean and quiet power supplies generally means linear supplies and linear regulators, but that is not a hard and fast rule. Breaking the AC leakage loop just before the Ethernet endpoint AND the DAC with off-grid batteries or something like an Uptone LPS-1 or Vinnie Rossi’s supplies also reduces noise.
    
   
5. Good transmission media (cables, fiber) will help #2 & #3. At least reasonably good cables are a must, with shorter likely better. Remember, Ethernet (like USB) is a system designed to be robust in its operation… bad transmission rarely causes data loss, but the processes used to make the system robust in the face of transmission problems WILL cause additional electrical noise.
    
   
6. Different devices performing seemingly the same function may sound wildly different due to their implementation AND also a particular device may produce differing results in different setups for very understandable (but hard to identify in many cases) reasons.
    
   
7. Having a setup where Ethernet packets not relevant to your music feed reach your endpoint will cause additional electrical noise and reduce sonics. NO HUBS!
    
   
8. EVERYTHING has tradeoffs. As Barrows suggested, adding something like FMCs may reduce signal fidelity at the endpoint. The specific gear must be tested in your setup!
    
   
9. EVERY setup is different. For example, your endpoint may have a more robust receiver setup with a better designed and functioning power system that minimizes the effect of issues in the Ethernet signal. Mine may be just the opposite. What works in my setup may not work as well or at all in yours.
   

 

Thoughts?

 

Greg in Mississippi

[gstew]

Greg,

 

Some of your points may be a bit too conservative, but in general I agree.

 

Again, it is not the digital transmission function that is the issue, but rather it is the good old fashioned analog noise that rides along that impacts SQ.

 

Unlike USB, data reclocking or regeneration is not an issue on Ethernet.

 

Sent from my iPhone using Computer Audiophile

 

Rajiv,

 

First I hope you've been having a very fun and informative weekend. I'm very curious if you'll have anything to report.

 

Second, I'm very curious to hear more on what points of mine might be a bit too conservative. I was basing my analysis on the thought that the ethernet protocol had some similarities to USB and that some of the same thoughts and ideas used to maximize SQ out of USB connections could be applied (with appropriate adaptations) to ethernet:

 

<SNIP>

I know that the Ethernet protocol does not function the same way as USB. There are of course similarities (both are transmitting/receiving an electromagnetic signal, in the case of USB always over a wire, in the case of Ethernet sometimes, both have PHY layers at the receiving end to best extract the data from the received signal, etc). But they do function in VERY different ways… and I am not much more than a slightly-educated layman on both.

 

But re-reading John’s comments, I suspect that approaching it the way he did that resulted in a whole new class of audio products, the USB signal regenerator (starting with the Regen, but moving on from there) MIGHT be the way that would produce the better solutions.

 

Sadly, I’m not sure any of us here are the ones to do that and John is otherwise occupied at the moment. But I suggest keeping John’s USB transmission issues and resolution analysis in the back of our minds (as Barrows appears to have) in looking at this.<SNIP>

 

Then John comes along and posts some very relevant information on how the Ethernet mechanism works at a physical level in this thread:

 

http://www.computeraudiophile.com/f22-networking-networked-audio-and-streaming/ethernet-cables-which-are-most-important-31052/index3.html

 

My read of his comments suggests there might be something to making some parallels between optimization of USB and of Ethernet:

 

<SNIP>

First off, Ethernet PHYs

I have worked in the semiconductor field for 33 years, during that time I have worked on over 50 PHYs for many different standards including many Ethernet PHYs. As a matter of fact there is a good probability that your computer has a PHY I worked on. I know a little bit about the subject.

 

Ethernet PHYs DO draw different amounts of current depending on the signal integrity of the signal being fed into the receiver. There is a LOT of both analog and digital circuitry trying to extract the bits out of that noisy piece of wire. When these circuits are running full bore (to extract bits from very noisy wires) they use a LOT of power. Part of my job has been to design power networks in the chip that can try and deliver clean power to those analog circuits. The advent of battery powered devices has put a premium on low power consumption of every chip in a device, so the PHYs are very carefully designed to use just the bare minimum of circuitry necessary to extract the bits. The result is the power varies radically with the signal integrity of the received signal.

 

When testing these PHYs we actually use a $700 cable to give the PHY the best signal integrity possible in order to find the baseline low power usage. These $700 cables use very expensive materials and extremely high precision geometry to have the least possible degradation on the signal. We then test with all kinds of more "normal" cables and lengths to see what happens in the real world.

 

What happens is that the BER stays almost the same, the chip is doing its job right and adjusting its internal circuitry to handle increasing signal degradation, which causes increased power draw and noise on power supplies and ground planes. At some point the BER skyrockets, it is interesting that what usually causes this is the PLLs that generate the timing for the PHY start outputing too much jitter as the noise on the PS and planes increases. One solution is to include an internal linear regulator on chip for the PLLs, which works, but increases the power dissipation of the chip. The customers don't like this.

 

The result of this modern heavy focus on power consumption HAS produced Ethernet PHYs which produce significantly varying power consumption depending on small changes in signal integrity.

 

On the issue of EMI coming from the cable to other parts of the system.

It is not so much the 125MHz symbol rate, but the PACKET rate that matters. In audio over Ethernet the packet rate frequently winds up right smack dab in the middle of the audio band where the human ear is quite sensitive. It is possible for the 125MHz to wind up getting into analog systems and going through non-linear parts of the circuit which can lead to small amounts of the packet frequency winding up in the audio signal. It probably is pretty low intensity and systems will vary radically on how sensitive they are to this, but I don't think it is wise to completely discount it.

 

On leakage currents and Ethernet cabling, I have no idea what is going on there. I have had no time to do any testing and won't for a long time. My lab is going to be packed up and in storage for many months so I won't be able to do any testing on this subject. I don't know enough about this subject at this time to make any comment on it.

 

On expensive Ethernet cables from audio cable makers, I'm fairly skeptical that they know what they are doing. I DO know what it takes to make a REALLY good cable, and what I have seen from the audio cable companies does not even come close to this. If these cables wind up sounding better I think it is much more likely to be an accident than any form of exceptional cable construction. And BTW making a cable with VERY low signal degradation DOES cost a lot of money, maybe the companies that make these should look into the audiophile market!

<SNIP>

Well I guess that is about it for now.

 

John S.

 

After trying to digest this, I look again at my rules of thumb and don't see anything I'd pull... if anything I'd make a few stronger AND add some.

 

So again, I'm really curious to hear what struck you as too conservative.

 

THANKS!

 

Greg in Mississippi

[gstew]

Really you aren't doing any different with your Auralic Aries Mini in place of a CAPS. Just a different type of low power computer built towards audio. You can still use the smart phone/tablet to operate the media player on a CAPS.

The real question is how much isolation downstream the streamer/CAPS can negate the effect on USB from streamer/CAPS? Enough to where it doesn't matter? as much?

 

Maybe a Rasberry like gstew is the way to go? or some other itx low power board that could be operated by a galvanically isolated power supply? Many different ways to go at this isolation.

 

ElviaCaprice,

 

I would argue that a purposed designed Ethernet to USB renderer like Austinpop's Aries Mini (or the rest of the Auralic line or the uRendu or the SMS200) SHOULD clearly outperform something like a CAPS which is made with carefully selected, but still general-purpose HW that has no intention or pretensions of being designed for audio playback like this other units. Having no un-needed HW on board, carefully designed power and signal paths, and adding things like the 'improved Regen' incorporated at the output of the uRendu (and I suspect similar care, if not identical circuits were used at the outputs of the other units I listed) will all tip the SQ needle in favor of the purpose-designed/build audio endpoint.

 

AND I don't expect that USB improvements can negate the need for any Ethernet input optimizations. If you read through the posts I referenced in my long note a few pages back and the recent one from John Swenson in my post just above, similar mechanisms are in play, which means (to me at least) that related mitigations are in order (but optimized for Ethernet instead of USB).

 

As far as the R-Pi being the way to go, as a generalized audio endpoint, it is not a good choice IMHO. This is largely because of the Ethernet AND USB processing both being handled by the same chip, so electrical noise produced by one protocol DIRECTLY impacts the quality of the processing of the other. Where the R-Pi works fairly well is when it is used with an I2S-connected DAC, but this limits one to fairly low-end DAC HATS that require a bit of DIY to make into really good-sounding units OR a concerned DIY effort to build a fairly good I2S input DAC (I use them in both ways). AND then only in concert with an I2S reclocker like Ian's FIFO (available via Group Buy on DIYAudio) or the newly released Allo Kali (which I expect is close to the quality of Ian's FIFO if powered well, but I've not compared them yet). This is due to limitations in the on-board clocking and GPIO of the Pi that result in a fairly jittery I2S output which these devices correct.

 

I will say that with good power supplies, a tweaked out low-end DAC HAT or a good high-ish-end one, and with appropriate network tweaks (all linear supplies, an FMC segment, and multiple Baaske filters), I've gotten the R-Pi as a player to sound very good in my systems, besting my also tweaked-out Sony HAP Z1-ES, not a shabby unit even in stock form (though far from SOTA even tweaked-out).

 

And it is worthwhile noting that the uRendu (the purpose-built audio endpoint for which the most details have been published, AFAIK) is not far from a more powerful R-Pi done right as an audio device... with a good USB output. (And I keep poking various manufacturers to do a purpose-built audio R-Pi... one with the exact same active HW complement as the Pi so that it will work with all the existing distros/players AND DAC HAT hardware, but done with audio quality in mind... I suspect that would be another good killer product!).

 

But no, IMHO, a CAPS of any sort nor an R-Pi or any of its relatives such as the BeagleBoneBlack or Cubox will beat a good purpose-built audio endpoint such as the various Aries, uRendu, or SMS200 as a generalized audio-quality USB out device.

 

Greg in Mississippi

[gstew]

I have observed that devices that feature internal switching regulators can have a wide DC input voltage range. When fed a higher DC voltage, the current consumption is lower than when fed a lower DC voltage. <SNIP>

 

I think I know why my MC110CS has a higher minimum DC input voltage than MC100CM, MC200CM & MC210CS. The MC110CS has an extra bridge rectifier (made with 4 discrete 1N4004 diodes) in line with DC input jack. This causes the input DC voltage to be lowered by more than 1V. With this bridge rectifier, it looks like this MC110CS can work with reversed polarity DC or even AC input! I'm not going to try though.

 

These TP-Link FMCs all appear to feature one internal switching regulator, plus one or two LDO linear regulators downstream from the switcher to generate lower voltage rails. The internal switcher explains the DC input voltage latitude.

 

And I think you are right! Thanks for catching that.

 

I'd assumed that most if not all of the various TP-Link FMCs had the same or at least very similar power setups... it looks as though that is a bad assumption on my part. Also notice that the versions of the MC200CM and MC210CS shown have provisions for those diodes, but they aren't populated.

 

Looks like we'll have to open up the units to tell what we can really do with them instead of assuming that anything with that initial switching regulator will take the same wide range of voltage inputs.

 

Greg in Mississipi

[gstew]

I have observed that devices that feature internal switching regulators can have a wide DC input voltage range. When fed a higher DC voltage, the current consumption is lower than when fed a lower DC voltage.

 

An interesting question is given a specific device (e.g. microRendu), is the sound better with high V / low I or low V / high I? I can imagine the answer being on a device-by-device basis or perhaps even system-by-system basis, requiring listening to compare.

 

Until recently I also thought that lower input voltage should translate to better sound, but I got hit with a couple of counterexamples:

 

(1) replaced Auralic LPS 16V with Vinnie Rossi MINI LPS 12V for Aries Femto

 

(2) replaced stock Netgear FS105v3 switch powered by Teradak 12V LPS with modified FS105v3 (low-noise linear regulator inside) power by Teradak 5V LPS.

 

In each case there was an audible SQ degradation, and my friend and I went back and forth several times to ensure what we heard was not a fluke.

 

One hypothesis here is that higher input current level (w/ lower input voltage) may translate to more board-level electrical noise. Things may not be quite that simple, however.

 

This is consistent with my observations here, especially the FMCs.

 

I think this is an area worth more investigation.

 

When I started DIY'ing back in the early '80s, one hot tip was to power the then latest-and-greatest LM317/LM337 regulators with a pretty high input voltage as they were supposed to sound better with a lot of drop and running hot. With modern linear regulators and LDO regulators, I think this is less of an issue. BUT in my 'test-bed' setup, I have 2 6.5v DC rails and 2 13v DC rails. Devices I use in that setup (all with linear regulators) that require only 1 or 2 rails and regulate down to 5v generally sound better (and never any worse) with the 13v rail(s) than the 6.5v ones, LDO regulators or not.

 

OTOH, there are the devices like the FMCs with switching regulators. In general, I'd tend to power them with a higher rail to lower the current consumed (and the current pulse produced by the devices). OTOH, my TP-Link MC100CM produces better SQ with a 5v TeckNet battery alone than with either a good DIY 7.5v linear or a Jameco 7.5v linear adapter.

 

Then there were some tests by SCAN80269 comparing various power options on several different pieces of EQ:

 

http://www.computeraudiophile.com/f22-networking-networked-audio-and-streaming/auralic-aries-and-sonore-microrendu-listening-impressions-29351/index5.html#post613478

 

and

 

http://www.computeraudiophile.com/f22-networking-networked-audio-and-streaming/auralic-aries-and-sonore-microrendu-listening-impressions-29351/index5.html#post615876

 

His results suggest that units with switching regulators DO work better in audio with input voltages on the higher side of their allowable ranges.

 

AND I've seen other posts suggesting that similar devices sound better with higher input voltages, but did not spend time to hunt them all down... sorry!

 

I don't yet see a clear path forward here.

 

Then there is this:

 

This is something I noticed too when I first got my PS Audio P3 AC regenerator. I am not an electrical engineer so I am not sure if my experience is analogous to what @scan80269 posted but here goes. My P3 allows me to adjust the output voltage to any of the devices connected to it. A few years ago when I first got the P3 I messed around with the output voltage and found that the closer I was to 120VAC, the music sounded more clear and energetic. I am not saying it was a huge difference but I could definitely hear it. At the end, I settled on 116VAC out.

 

Does my experience make sense for AC versus DC?

 

Makes sense to me!

 

When I got my first P10, it was powering a system with a really tweaked-out cMP/cPlay setup... super-low underclocking/undervolting BIOS settings, a super-tweaked WinXP that with player was <15mB, full linear supplies including ATX-24 (13 linear supplies total!), SSD with minimized music file load, an ultra-modified Juli@ sound card digital section with an attached I2S input DAC card, basically the works for that setup. AND it sounded best with my P10 set to 115vAC.

 

Fast forward several years and I modified the heck out of a Sony HAP Z1-ES that was a definite, but not huge step up. That unit is basically an all-in-one player with a lower-power CPU (same line as used in the uRendu), a DSP for upsampling PCM to DSD (I prefer the unit with playback in native format), and a fairly sophisticated DAC/output stage setup all fed from 8 linear supplies. With it in the system replacing the cMP/cPlay setup, 120vAC sounded best.

 

This is with the same amps.

 

So a number of different data points with some seemingly conflicting results. I don't have an answer here yet, just looking for more data and patterns.

 

Greg in Mississippi

 

P.S. Note that the Sony HAP Z1-ES stock uses 3 switching regulators on the digital side powering the screen/processor/DSP, the harddrive (which I replaced with an SSD), and the cooling fan (which I promptly removed). Replacing these 3 regulators with good (but not great) linear regulators was the single largest improvement of any of the mods I did to that unit (and I did a LOT!). That makes me think our FMCs could be improved with a little DIY tweaking.

[gstew]

Fair to say you're Baasking in the glow of great sound, Greg.

 

LOL!

 

I got 4 thinking I would use 2 in my main system and move 2 downstairs to my smaller system as I set it up to work in networked mode. BUT I still need to listen to the effect of all of them in the main system... I don't want to regret pulling 2 of them.

 

I did confirm that 1 between the downstream FMC and the R-Pi was a positive change. The rest just got slotted in when I received them with no comparisons. I DON"T think they are degrading the sound, but don't know if they are helping either.

 

Thanks for the laff.

 

Greg in Mississippi

[gstew]

I have it on my list to try MW.

 

Right now I am using Sine wave.

@gstew had some suggestions - did you see them up-thread?

 

LOL Rajiv, I think I sent those to you in an email and never posted them.

 

In any case, here they are... do get them as 'What I've tried that works'. This is definitely an area where YMMV... for example, for some people Multiwave is the cat's meow. It has always made things sound a bit thick and less tuneful in my setups.

 

My PS Audio P5/P10 setup tweaks... not sure how many of these are not available on the P3:

 

1. Feet. I use Herbies Iso-Cups with his SuperSonic Hardball. I am sure there are much better ones out there, but this is a good high bang-for-buck solution.

 

2. Voltage Selection. 115v worked best when I was running my super-tweaked out motherboard-based cMP/cPlay setup with all linear supplies. 120v works better now for my modified Sony HAP-Z1-ES setup. Go figure....

 

3. Phase Tune. I currently have +2 on mine. I adjust this to get the most even waveform on the 'Difference' display screen, with the least amount of difference from neutral, which seems to work best. I have done it by ear in the past, but realized that I ended up with the same setting as above, so I now stick with the easy-to-do adjust to the waveform.

 

4. High Regulation. The manual says to use the 'Low Distortion' setting for SS gear, I have always found the High Regulation to sound better for my setups.

 

5. SW version. See this thread from a few years back: Variable Multiwave Update For P5/P10 - Interesting Results | Power products | ForumsPS Audio Variable Multiwave Update For P5/P10 - Interesting Results | Power products | ForumsPS Audio . The OP, DarqueKnight, liked V37.. So do I.

 

6. Pre-filtering. The early manual I have suggests using PS's Soloist AC in-wall receptical/filter. I prefer it straight into the wall, but with 2 Hammond choke filters in parallel on the other outlet.

 

7. Aftermarket fuses. I use AMR fuses because they were a high bang-for-buck option, there are better ones, but it does make a difference.

 

No it does not. Some filtering, yes, but not isolation.

 

I can't cite a source, but if you look on PS Audio forums, this has been asked and answered by Paul McGowan, founder and CEO.

 

He's an opinionated fellow, and has his own strong views on isolation. Not all here would agree.

 

All I can say is the P3/P5/P10 regenerators are da bomb! Whatever the secret sauce is, I'll take it!

 

Rajiv, thanks for the great links. AND I totally agree! My first P10 made a very significant difference in my first system, so much that I stretched my budget and got another when I setup my 2nd (main) system. Basically it was the turning point in tuning a pair of DIY amps I was using such that I decided I could live with them... and still have them today!

 

The topic of isolation is a very interesting one... it really means a LOT of things. In the context of an LPS-1, what it means is that the actual circuit producing the output into the powered device IS NOT connected through to the AC line at the time it is powering. They do this with 2 alternating Ultracap banks where one is charged as the other is used... and the one being used is not connected to the charging circuit or anything else BUT the output to the powered device. Vinnie Rossi's supplies do the same.

 

There's a good background note from John on the problems solved by the LPS-1s on the Uptone Audio website:

 

John Swenson's Tech Corner – UpTone Audio

 

In the context of an 'isolation' transformer, the output is 'isolated' from the input in that the input (primary) and output (secondary) windings are not directly electrically connected. Put a DC voltage across the primary windings (small so you won't burn it up) and you won't read anything across the secondary windings. BUT put 60hz AC voltage across the primary and you do get a corresponding voltage on the secondary. AND the ones recommended by John & Alex don't conduct much at higher frequencies, so they act as a filter for a lot of the noise that's on our modern AC lines.

 

Then there are 2 types of isolation inherent in a PS Audio P3/P5/P10. One is that provided by the power transformer. Since the units are basically high-powered amplifiers putting out a clean 60hz AC sine wave, they start with a large (VERY LARGE in the case of a P10) power transformer the is part of the basic power supply. This transformer does exhibit some of the same aspects of isolation as the isolation transformers above, but likely does not do as good of a job of filtering out high frequency noise. Still, it is much better than none.

 

Then there is what one could call isolation in that each piece of equipment plugged into a P3/P5/P10 has a lower negative impact on the voltage seen by all of the other pieces of equipment sharing the unit than if they were plugged into the wall socket together. This is because just as a speaker-driving amplifier is designed to have a low output impedance to provide 'damping factor' or control over the drivers, the P3/P5/P10 outputs also have a low output impedance and within their current limits, will produce MUCH less voltage variation for all of the gear plugged into the outputs. The wall socket has a much higher output impedance, though still sufficiently low that there is not much sag for most of our systems and circumstances.

 

John Swenson made a post a few months back going over many of the different types of what is termed 'isolation' and comparing their pluses and minuses... it's worth search for if you've not seen it.

 

OTOH, I am very curious about the current recommendation from John Swenson/Alex Crespi to use isolation transformers AND closely-coupled AC plugs in coordination with at least one LPS-1. I DO wonder if a hybrid solution, an appropriately-sized isolation transformer ahead of a P3/P5/P10 and for the DIY'er with an expired warranty like myself, modifying the P3/P5/P10 by removing the AC input surge protection devices (whose function is now served by the isolation transformer) PLUS possibly bypassing the filtering on the AC outlets of the P3/P5/P10 (if it is in series with the output) MIGHT be a better solution than either by themselves.

 

Someday when I have another $1000 to spend on isolation transformers!

 

Greg in Mississippi

[gstew]

Thanks Greg, makes sense what you say and for the most part I would agree, but I can't help think that there is not a way to isolate those 0's and 1's from the PC via USB to rid of all electrical issues. If so then it shouldn't matter what one does on the streamer side of the issue. But for now it would appear that the networking Ethernet option has a leg up, with it's own dirty issues. I'll wait and hold down the fort here for a USB only option. Till then, I am placing my future upgrade on an isolation transformer to see what it can bring to the table, especially for the amp side of the equation. Getting that heavy pig here to Costa Rica should be fun. You guys keep firing away on the testing mods.

 

Mark

 

Mark,

 

I think the isolation transformer should be a good step up. Looking forward to your report.

 

OTOH, I think you underestimate how much electrical noise consumer-level computer gear makes compared to the purpose-built units and how much that noise impacts the SQ in our setups, even the much lower level of noise produced by the purpose-built gear.

 

Then there is the issue of SQ versus flexibility and convenience. There IS currently a way "to isolate those 0's and 1's from the PC" and get them to a DAC with very little electrical noise. It is called 'sneaker-net'.

 

I have an SDTrans384 SD Card player that I use as my digital reference source. It plays back WAV-only music off SD Cards no larger than 32Gb. No formats other than WAV, no networking, no upsampling, no fast-forward/reverse, no remote-control (though a rudimentary one is available), very limited output connections (I2S, S/PDIF, TOSLINK, PS Audio compliant I2S over HDMI cable only). Only plays multiple tracks (if that function is enabled) in the order they were written to the SD Card. Only commands available are select repeat mode, select directory, skip to next directory or file, play, stop.

 

VERY hair shirt!!! Heck, my turntable at least lets me pickup the needle and place it down later in a record for a 'fast forward' function!

 

Plus it is only available as a DIY board.

 

Because it does so little, it is VERY electrically quiet compared to any computer-processor based setup (it uses a very dumb, very low power FPGA, part of why it does so little). It is definitely my best digital source. A friend of mine has one in a tweaked out setup and says it bests his analog setup (which is a super-tweaked out setup too!).

 

This won't work for most people.

 

To get remote control, you add circuitry and functionality, which means more noise. To get additional connectivity options (such as USB out or in from a network-connected NAA), you add circuitry and functionality, which means more noise. To handle different music formats, you add circuitry and functionality, which means more noise. And on and on.

 

The purpose-built units along with an Ethernet input and a USB output hit the sweet spots in functionality AND do it at a level of performance unavailable using consumer-level HW. And as better chips and circuits become available for consumer-level HW, they will of course be adopted for purpose-built HW, keeping them ahead of consumer-level HW.

 

AND the people who really know this design space (people like John Swenson and Ted Smith, the PS Audio Directsteam DAC designer, among others) now are clearly saying that absolute isolation is not available and that all isolation introduces some problems of it's own. Isolation is a part of the solution, but quiet, purpose-designed HW is also going to be needed for the best SQ in a setup that most people will want to buy.

 

See my post a few above referencing a post by John Swenson on different types of isolation. Here is that post:

 

http://www.computeraudiophile.com/f27-uptone-audio-sponsored/mystery-revealed-uptone-audio-ultracap%99-linear-power-supply-1-a-28609/index8.html#post560434

 

And while I was looking for it, I found this one that is also VERY relevant to the topic at hand:

 

http://www.computeraudiophile.com/f26-sonore-sponsored/sonore-microrendu-power-supply-unit-observations-considerations-and-commentary-28480/index7.html#post578499

 

I'm not sure there is any way around this.

 

Greg in Mississippi

[gstew]

<SNIP>

I have sold the EMO, the MFCs and the intona.

I confirm this is definitely better to get the lowest possible noise at the begining of the chain than to add noise trying to remove it later with added components.

B.

 

I think FOB69 has written one of the most profound and fundimental truths of this realm... Once you have noise polluting your silence, you will never get it out. The best you can do is reduce it, but you can NEVER get it out!

 

Keep that in mind!!!

 

Greg

[gstew]

Batteries can be great but also tricky to work with. Supercaps have become dirt cheap. What is needed is a simple and inexpensive dual bank controller that auto switches between two banks of either batteries or supercaps and keeps the mains supply disconnected from the battery's or supercaps which are discharging. This project may be of interest: Dual Bank Floating Supercap Supply - diyAudio

 

Jabbr,

 

Thanks for the design and schematics over at DIYAudio. I have several places I could use this in my setups,, mostly for battery packs and less-critial gear

 

Is this a tested design and are boards available? Any pictures of finished projects using it.

 

Look for a good starting point on this.

 

THANKS!

 

Greg in Mississippi

[gstew]

Watching with interest, looking forward to seeing actual built HW and experiences with how well it works.

 

No question that floating (not-grid-connected) power sources have benefits in certain situations. I'm hoping this project makes implementing them easier for the DIYer.

 

Greg in Mississippi