Objective proof the UpTone Regen ISO can improve a DAC's output(*)

[Superdad]

39 minutes ago, plissken said:

 

No, the only leakage loop spurie is with YOUR SMPS. The USB bus power and his lab power supply don't exhibit this.

 

Do I really need to explain this again?  His Audio Precision analyzer uses an SMPS.  So the leakage of it and that of our supplied Mean Well SMPS interact to produce the extra 10-20dB of harmonics (still very low down at -120 -- -130dB)

 

And AC spurie are still seen with his "lab" supply--because it too has a little leakage and interacts with the PS of his AP.

[plissken]

Just now, Superdad said:

 

Do I really need to explain this again?  His Audio Precision analyzer uses an SMPS.  So the leakage of it and that of our supplied Mean Well SMPS interact to produce the extra 10-20dB of harmonics (still very low down at -120 -- -130dB)

 

And AC spurie are still seen with his "lab" supply--because it too has a little leakage and interacts with the PS of his AP.

 

It's not lost on me that the AP has a SMPS. What's lost on you is that your SMPS made things the worse out three options: Your SMPS, his lab supply, or vanilla USB bus power. 

 

My money is still on someone falling flat on their face with DAC that keeps it's shit together on USB bus power when tested for it blind.

 

You want to do this at RMAF in October? 

[pkane2001]

35 minutes ago, scan80269 said:

Regarding jitter, Amir (and/or his followers) have adopted an assumption regarding the total correlation between incoming USB signal jitter and DAC analog output jitter.  Has it occurred to you that some DACs may be capable of complete de-correlation between the two two jitters, i.e. high incoming USB jitter doesn't necessarily translate to high analog output jitter?  Better DAC designs should be able to take a boat load of bad incoming USB jitter and signal integrity (waveform, etc.) and still output excellent analog with no hints of the USB crap coming through.  With such DACs I would not expect the ISO-Regen to yield much tangible SQ benefit, if at all.

 

 

 

Scan, I'm not clear as to what point you're making here. Are you saying that uncorrelated jitter might not manifest itself at the DAC outputs and yet, can affect SQ? Or that some DACs are just not subject to the incoming USB crap and therefore the output will not exhibit jitter, regardless? If former, I'd like to understand how that's possible that an audible effect is not measurable in the analog domain.

[scan80269]

4 minutes ago, pkane2001 said:

 

Scan, I'm not clear as to what point you're making here. Are you saying that uncorrelated jitter might not manifest itself at the DAC outputs and yet, can affect SQ? Or that some DACs are just not subject to the incoming USB crap and therefore the output will not exhibit jitter, regardless? If former, I'd like to understand how that's possible that an audible effect is not measurable in the analog domain.

It's the latter.  USB DACs all have to re-clock the incoming data from USB clock domain to digital audio domain (based on 44.1KHz or 48KHz multiples depending on sampling rate carried by the digital stream).  There's a FIFO implementation in some form in every USB DAC.  A well designed reclocking circuit in a DAC can prevent the USB side jitter from contributing to the jitter on the digital audio clock side.

 

For your last question, my viewpoint is that measurements that we know how to do collectively have yet to be able to reflect/demonstrate what our ears can hear.  I strongly believe it is a major fallacy to claim that audibility must correspond to measurability, and the opposite being a fallacy as well: no measurement differences mean there cannot be audible differences.  I just don't believe the two need to be 100% correlated with each other.

 

 

 

[plissken]

11 minutes ago, scan80269 said:

It's the latter.  USB DACs all have to re-clock the incoming data from USB clock domain to digital audio domain (based on 44.1KHz or 48KHz multiples depending on sampling rate carried by the digital stream).  There's a FIFO implementation in some form in every USB DAC.  A well designed reclocking circuit in a DAC can prevent the USB side jitter from contributing to the jitter on the digital audio clock side.

 

This is another area where Ethernet based DAC's are going to win out. The non-realtime, high bandwidth, low latency, deep cache and much better implemented clock domain boundaries. 

 

Just give me an appliance with  an 1/8th, 1/4th, or 1/2 GB of buffer and kiss the rest goodbye. 

 

I like my DC-1 but my next DAC won't have a USB port on it. 

[esldude]

1 hour ago, Superdad said:

 

Yes, plissken is "Jinjuku" at ASR (he is more polite here than there).  Dennis (esldude here) is Blumlein88 there.

 

Yes, I am Blumlein 88 at ASR.  Nothing nefarious.  After using the same moniker for 20 years thought I would try something different.

 

 

[esldude]

42 minutes ago, scan80269 said:

 

For your last question, my viewpoint is that measurements that we know how to do collectively have yet to be able to reflect/demonstrate what our ears can hear.  I strongly believe it is a major fallacy to claim that audibility must correspond to measurability, and the opposite being a fallacy as well: no measurement differences mean there cannot be audible differences.  I just don't believe the two need to be 100% correlated with each other.

 

 

 

Audibility must correspond to measurability in principle.  It may not be possible to yet do or interpret those measurements, but if something sounds different due to a change in soundwaves it must be a difference that can potentially be measured. 

 

That measurable differences must always be audible is of course a fallacy, but one rarely made by people who measure.   For instance here some have acted as if some 60 hz related hum was being offered up as proof of an audible problem.  It was merely offered up as something measurably different and slightly worse when the device was in use.  It was not claimed to be an audible issue.  I don't even know the higher jitter levels exhibited by the Modi 2 without a USB cleaner would be audible.  It does clearly indicate that device is more effected by what is on the USB buss than several others tested.  As well as indicating there are DACs that can benefit from a device like the Regen measurably.  The potential that the ISO Regen improves sound with that DAC is at least possible as we can see it measurably improves the result. 

[esldude]

40 minutes ago, plissken said:

 

This is another area where Ethernet based DAC's are going to win out. The non-realtime, high bandwidth, low latency, deep cache and much better implemented clock domain boundaries. 

 

Just give me an appliance with  an 1/8th, 1/4th, or 1/2 GB of buffer and kiss the rest goodbye. 

 

I like my DC-1 but my next DAC won't have a USB port on it. 

I haven't had problems with USB DACs, but I sure hope the Dante network or something like it becomes common and affordable. 

[esldude]

1 hour ago, kumakuma said:

 

Got it. I guess "plissken" wasn't available as a handle when he joined Amir's site.  

 

Maybe plissken is the hidden special secret identity. I have run across a Jinjuku on some other audio sites.  I am sure I have seen one at AVS which I think might be the same person.  Not sure, nor do I think it a big deal.  It was easy to figure out if you read both sites without him saying anything about it.  So I don't think he was hiding anything. 

 

Maybe I need a second secret identity here.  I might choose spirograph.  That way I have an abundance of fitting avatars to choose from like:

 

 

 

[scan80269]

2 hours ago, esldude said:

Audibility must correspond to measurability in principle.  It may not be possible to yet do or interpret those measurements, but if something sounds different due to a change in soundwaves it must be a difference that can potentially be measured. 

I agree in principle.  The challenge here is how to have measurements adequately illustrate what our ear/brain are capable of discerning.

 

We have come far over the decades in creating (discovering?) tech attributes like THD+N, intermodulation, TIM, S/N, dynamic range, etc. and techniques like FFT to help us characterize the performance of audio gear, but I'd like to think that our ear/brain is by far the most sophisticated audio instrument, and these attributes and techniques fall far short of being able to characterize what we can actually perceive.  In other words, the capability of our ear/brain is far advanced beyond our current knowledge in math & science related to audio & sound.  Just because no graph can currently illustrate a difference we heard doesn't mean that difference cannot exist.  We are either giving our ear/brain much too little credit for its abilities or attempting to distill the perception down to a bunch of fairly rigid and narrow attributes that are insufficiently representative.

 

Also, I listen to music, not test tones.  Using an FFT plot to illustrate differences in jitter, noise level, etc. with a single test tone may be straightforward, but how do you do that for a piece of complex music, say a polyphonic Mahler symphony with a whole bunch of instruments playing concurrently?  I don't find it surprising that people get offended when told they are delusional in claiming to hear a sonic difference when the FFT graph clearly shows no difference.  Again, being unable to illustrate a measured difference does not automatically mean no difference can possibly exist.  It is a fallacy to claim so.  In my book, the openness to accept the possibility that current measurements can easily fall short of representing what we can actually hear is a form of humility that can help us get farther.  To admit we have yet to learn everything there is to learn about sound and perception should motivate us to continue working to better characterize what our ear/brain does so effortlessly.

 

Also, good sound requires a balance of art & science/engineering.  It's never been just one or the other.  The greatest audio engineers in the world designing by tech specs alone may not yield the best sounding gear.  There are many counter examples.  Tube amplifiers often have poorer specs (S/N, THD, noise floor, etc.) vs. solid state amps, but many audiophiles prefer the tube sound.  The takeaway?  S/N, THD, etc. by themselves are inadequate to characterize or predict what the humans prefer.  Even engineers in Intel Corporation today have not fully grasped some of the basics of good sound engineering.  Great tech specs alone do not necessarily translate to great sound, and in real life the opposite often occurs.

 

It is with these realizations that lead me to believe that listening with ears is essential during the development of any audio related product.  For example, I know that UpTone's final decision on the USB hub chip brand/model for their ISO Regen was heavily based on listening feedback from a few highly respected individuals.  This tells me that UpTone knows that tech spec superiority and audio superiority are not necessarily strongly correlated.  It's another way of saying it's virtually impossible to engineer audio excellence without any listening involved.

 

I'm not belittling the importance or value of our technical/scientific knowledge of audio.  For example, I'm thoroughly convinced of the usefulness of FFT.  It's just that I believe we still have a long way to go to fully comprehend and characterize the human perception we call hearing.

 

[esldude]

16 minutes ago, scan80269 said:

I agree in principle.  The challenge here is how to have measurements adequately illustrate what our ear/brain are capable of discerning.

 

We have come far over the decades in creating (discovering?) tech attributes like THD+N, intermodulation, TIM, S/N, dynamic range, etc. and techniques like FFT to help us characterize the performance of audio gear, but I'd like to think that our ear/brain is by far the most sophisticated audio instrument, and these attributes and techniques fall far short of being able to characterize what we can actually perceive.  In other words, the capability of our ear/brain is far advanced beyond our current knowledge in math & science related to audio & sound.  Just because no graph can currently illustrate a difference we heard doesn't mean that difference cannot exist.  We are either giving our ear/brain much too little credit for its abilities or attempting to distill the perception down to a bunch of fairly rigid and narrow attributes that are insufficiently representative.

 

Also, I listen to music, not test tones.  Using an FFT plot to illustrate differences in jitter, noise level, etc. with a single test tone may be straightforward, but how do you do that for a piece of complex music, say a polyphonic Mahler symphony with a whole bunch of instruments playing concurrently?  I don't find it surprising that people get offended when told they are delusional in claiming to hear a sonic difference when the FFT graph clearly shows no difference.  Again, being unable to illustrate a measured difference does not automatically mean no difference can possibly exist.  It is a fallacy to claim so.  In my book, the openness to accept the possibility that current measurements can easily fall short of representing what we can actually hear is a form of humility that can help us get farther.  To admit we have yet to learn everything there is to learn about sound and perception should motivate us to continue working to better characterize what our ear/brain does so effortlessly.

 

Also, good sound requires a balance of art & science/engineering.  It's never been just one or the other.  The greatest audio engineers in the world designing by tech specs alone may not yield the best sounding gear.  There are many counter examples.  Tube amplifiers often have poorer specs (S/N, THD, noise floor, etc.) vs. solid state amps, but many audiophiles prefer the tube sound.  The takeaway?  S/N, THD, etc. by themselves are inadequate to characterize or predict what the humans prefer.  Even engineers in Intel Corporation today have not fully grasped some of the basics of good sound engineering.  Great tech specs alone do not necessarily translate to great sound, and in real life the opposite often occurs.

 

It is with these realizations that lead me to believe that listening with ears is essential during the development of any audio related product.  For example, I know that UpTone's final decision on the USB hub chip brand/model for their ISO Regen was heavily based on listening feedback from a few highly respected individuals.  This tells me that UpTone knows that tech spec superiority and audio superiority are not necessarily strongly correlated.  It's another way of saying it's virtually impossible to engineer audio excellence without any listening involved.

 

I'm not belittling the importance or value of our technical/scientific knowledge of audio.  For example, I'm thoroughly convinced of the usefulness of FFT.  It's just that I believe we still have a long way to go to fully comprehend and characterize the human perception we call hearing.

 

Null testing can do the trick. 

 

As for preference, well preference and fidelity are two very different things.  Yes many prefer tube amps, many prefer compressed musical recordings, neither has anything to do with fidelity.  That tube amp signature can be fully replicated with solid state gear of good design.  So don't confuse highly respected listening for preference with fidelity. 

[Superdad]

2 hours ago, scan80269 said:

I agree in principle.  The challenge here is how to have measurements adequately illustrate what our ear/brain are capable of discerning.

 

We have come far over the decades in creating (discovering?) tech attributes like THD+N, intermodulation, TIM, S/N, dynamic range, etc. and techniques like FFT to help us characterize the performance of audio gear, but I'd like to think that our ear/brain is by far the most sophisticated audio instrument, and these attributes and techniques fall far short of being able to characterize what we can actually perceive.  In other words, the capability of our ear/brain is far advanced beyond our current knowledge in math & science related to audio & sound.  Just because no graph can currently illustrate a difference we heard doesn't mean that difference cannot exist.  We are either giving our ear/brain much too little credit for its abilities or attempting to distill the perception down to a bunch of fairly rigid and narrow attributes that are insufficiently representative.

 

Also, I listen to music, not test tones.  Using an FFT plot to illustrate differences in jitter, noise level, etc. with a single test tone may be straightforward, but how do you do that for a piece of complex music, say a polyphonic Mahler symphony with a whole bunch of instruments playing concurrently?  I don't find it surprising that people get offended when told they are delusional in claiming to hear a sonic difference when the FFT graph clearly shows no difference.  Again, being unable to illustrate a measured difference does not automatically mean no difference can possibly exist.  It is a fallacy to claim so.  In my book, the openness to accept the possibility that current measurements can easily fall short of representing what we can actually hear is a form of humility that can help us get farther.  To admit we have yet to learn everything there is to learn about sound and perception should motivate us to continue working to better characterize what our ear/brain does so effortlessly.

 

Wonderful post Sam.  Really cuts to the heart of what has become a dichotomy but should not be.

 

 

Quote

It is with these realizations that lead me to believe that listening with ears is essential during the development of any audio related product.  For example, I know that UpTone's final decision on the USB hub chip brand/model for their ISO Regen was heavily based on listening feedback from a few highly respected individuals.  This tells me that UpTone knows that tech spec superiority and audio superiority are not necessarily strongly correlated.  It's another way of saying it's virtually impossible to engineer audio excellence without any listening involved.

 

Just to be clear about this: We chose the new USB3.1 hub chip for both measurable technical reasons AND (presently) unmeasurable sonic reasons.  But we knew which one sounded clearly better weeks before I separately sent Jud and Larry samples of enhanced REGEN boards--identical except for the 2.0 versus 3.1 hub chip--and marked only with the letters "G" and "M."  

I said nothing to them about how they differered or which one I liked, but like me, they both chose "G" within a minute of comparing them.

 

Of course stuff like that--and the choice to use the Crystek 575--were big obviously audible selections.  Much harder, but cumulatively as important sonically, are all the smaller details, such as elements of the power network for the hub and clock.  Selection of bypass caps for multiple VDD pins of the hub were important, and while John was very careful with low inductance planes for the power domains, I was not truly happy with the SQ until we had optimized the PS bypass caps (some of which were difficult to squeeze into the ideal spots right next to the hub chip power pins;  thank goodness for X7R MLCCs).

 

Unless one has been involved in the design and parts selection--with step-by-step listening comparisons--of many products (I don't pretend to be the engineer, I'm just another skilled pair of ears), it is hard for many to understand how real and critical this part of the design process is.

 

But ask any reputable audio engineer today--be they from Ayre, PS Audio, Conrad-Johnson, Chord, MBL, or Berkeley Audio Design--and they will all confirm that hundreds of critical choices get made both on the bench and in the listening room.  And they will all admit that a great many of the choices they make--based on listening--are not things they can even think of how to measure.

(I realize I mostly named electronics firms, but speaker designers fall under this too; Good luck trying to measure the benefit of pricey film capacitors used in crossovers, but good luck getting any high-end speaker manufacturer to give up their preferred brands.)

 

Still, I readily admit that it seems a bit nuts to be hearing the effects of tiny (1/4 the size of a grain of rice) capacitors bypassing the power pins of a USB hub chip.  Yet sweating this and dozens of other details is what results in a winning performance.  It's the audio olympics and the competition is tough.  And as any athlete will tell you, only the winners are remembered.  (Sorry to be a bit over-the-top here; but I just read that iFi announced their delayed-since-April iGalvanic will at last be released at the end of August.  Will it surpass the ISO REGEN?  Have they been studying our product and sound?  They can certainly out–market us.  We--and the audience--shall see/hear.)

 

Goodnght all,

--Alex C.

 

P.S.  I am back home with our sons from our weekend trip visiting my 88 year old father in North Carolina.  It meant the world to my dad--and our sons are at an age to appreciate knowing him and hearing the history of the Crespi clan.

Biut we now really hate Denver airport.  The connecting gate was well over a mile away--and we had to run almost the whole way to make it!

 

P.P.S.  Dennis I dug your reference to Spirograph.  When I was a kid I had what must have been the largest set they ever made.  Something like 65 pieces.  Played with it for weeks on end.

[Quadman]

16 hours ago, plissken said:

Maybe it isn't but it's not there with his lab supply. His lab supply does take care of a 180Hz bump that was their with USB bus power.

 

Why not tell Amir to repeat the test but this time put the lab PSU on the AP analyzer and put the SMPS supply back on the IR, if that cleans up the result like the lab supply did when put on the IR then you have to give credibility to Alex's explanation, and amir would have to admit such error in his measurements and reporting of.

[Jud]

12 hours ago, esldude said:

So don't confuse highly respected listening for preference with fidelity.

 

I don't claim to have golden ears, so I think we can just say "a coupla guys" rather than "highly respected listeners." 

 

(Always loved the Garth Brooks song, "I've Got Friends in Low Places.")

[pkane2001]

1 hour ago, Quadman said:

 

Why not tell Amir to repeat the test but this time put the lab PSU on the AP analyzer and put the SMPS supply back on the IR, if that cleans up the result like the lab supply did when put on the IR then you have to give credibility to Alex's explanation, and amir would have to admit such error in his measurements and reporting of.

 

I could be wrong, but I don't think AP analyzers allow for an external DC power source. Otherwise, this would be a good test.

 

[Quadman]

22 minutes ago, pkane2001 said:

but I don't think AP analyzers allow for an external DC power source

 

If that is so, darn that would have been revealing.  I'll have to research that a bit.

[esldude]

On 8/6/2017 at 8:28 AM, Superdad said:

 

Because it is not a ground loop he is forming (though if he puts the ISO REGEN switch in the defeat position, his setup--using single-end connection to the AP and having the same computer connected to both the DAC and the analyzer--would surely form one of those too!).  No, he is forming an AC leakage current loop.

 

That is very different.  It is a significant, widely overlooked issue in audio (and even with measurement equipment), and it very many exists, as exemplified by Amir's graph--which he completely misinterpreted as being the ISO REGEN putting mains noise on the 5VBUS line into the DAC.

 

John has written extensively about leakage/loops.  He has explained:

• what they are (AC traveling over every sort of connection including DC connections);
• where they come from (virtually every power supply; linears have leakage though not as much as switchers, and batteries have none);
• why they form (must be two PSUs--it is the interaction between two or more);
• and how best to mitigate them (reduce impedance between gear by plugging all system mains power cords into a heavy power stirip with zero filtering elements--use a low interwinding-capacitance isolation transformer for protection of gear)

By the way, one of the hallmarks of our own UltraCap LPS-1, its raison d'etre, is that by being "battery-like" (without the performance shortcomings of batteries), it blocks the path of leakage currents.  Or I should more correctly say that it keeps the device being powered from contributing any PS leakage the system's leakage loops.  EVERY audio system has leakage loops, but the combination of the ISO REGEN and the LPS-1 is what creates a complete "moat" between the computer (with its very high and typically quite nasty and "bursty" leakage) and the DAC.

 

Remember, galvanic isolation (just one of the functions of the ISO REGEN) is the blockage of DC not AC.

 

You can read Amir's reply here:

 

http://audiosciencereview.com/forum/index.php?threads/uptone-iso-regen-review-and-measurements.1829/page-8#post-46534

[Speed Racer]

Alex,

 

You really could help yourself out if you would take the time to do some real world testing and post the results on your website. Showing a nice looking USB eye pattern here doesn't really tell us anything about what is coming out the analog side of the DAC. 

[lmitche]

It is highly likely that powering either the Meanwell, AP or MacIntosh with an AC isolation transformer with a floating neutral such as this:

 

https://www.amazon.com/Precision-1604A-Single-Isolation-Transformer/dp/B000LDLF3M/ref=sr_1_4?s=industrial&ie=UTF8&qid=1502155827&sr=1-4&keywords=120v+isolation+transformer

 

will break the AC ground loop in Amir's testbed.

 

Powering the laptop with its internal battery disconnected from it's power supply may accomplish the same thing. If not the isolation transformer should be added as well.

 

Jud had a similar problem with the prototype ISO Regen and MacBook. The isolation transformer fixed it.

[plissken]

On 8/6/2017 at 9:28 AM, Superdad said:

 

Because it is not a ground loop he is forming (though if he puts the ISO REGEN switch in the defeat position, his setup--using single-end connection to the AP and having the same computer connected to both the DAC and the analyzer--would surely form one of those too!).  No, he is forming an AC leakage current loop.

 

 

Amir just responded very well to what I and other were pointing out. No matter the straw man you want to parade about, the ISO regen with power supply made things worse. End of story. 

 

Quoted in full:

 

OK, back from the trip and had a bit of time to answer this. I am going to provide two answers. One high level that is easy to understand and should completely do away with this argument. The second answer, will be a deep dive into power supplies, emissions control, safety, etc. that explains why the problem is created with ISO regen and not otherwise.

First, here is the argument from the first link above:


In a nutshell, he is saying that because ISO regen uses a switchmode power supply and so does the Audio Precision (how he knows the latter is beyond me seeing how he doesn't have one), and that switchmode power supplies have a "required Y capacitor" that what I am observing with respect to increased mains frequency and harmonics in the output of the DAC as measured by my Audio Precision Analyzer, is unavoidable. He is also pointing to an iFi post which was measurements of their power supply (and not the DAC) as proof point of this.

Here is the diagram for how I measured the Behringer UMC204HD as is:



Right away we see the flaw with the statement Alex is making. There are plenty of switchmode power supplies in my playback and measurement chain. The Behringer was powered by the USB power of my desktop workstation. This is a custom PC I built with a switchmode power supply. I don't pretend to know what power supply configuration is used in the Audio Precision Analyzer. I do have a laptop connected to it that controls the analyzer over another USB bus and the laptop was connected through yet another switchmode power supply to mains.

So if switchmode power supplies have "required Y capacitors" and anything upstream of the analyzer would create mains harmonic noise, it should have been there all along. But it was not at nearly a significant level until I changed to this configuration:



In this configuration, yet another power supply made by a half-decent Chinese/Taiwanese company called Meanwell is added in the middle of the chain. This is powering the ISO Regen which in turn generates a new USB bus power. In this configuration we managed to increase the mains related harmonics as I showed before:



By now it should be obvious that the mere fact there are switchmode power supplies upstream of the DAC/USB bus does not at all say that the measurement scheme is faulty. The only change here was insertion of the ISO regen and it created the heightened noise products in the output of the DAC. Clearly what Alex is saying regarding design and operation of switchmode power supplies is incorrect (I will dig into this in the later detailed post).

But let's say for the moment that he is right. Nothing about that means that we can't measure the output of the system with the Audio Precision Analyzer. The AP acts just like any other audio device. It has an analog pre-amp, power supply for the same, and other bits to capture that stream. There is no law against such an audio having a switchmode power supply. Class-d amps for example mostly have switchmode supply. Is Alex saying that in such configuration ISO Regen does screw up the fidelity of the system? If so, why is that not indicated in the specification for the device?

Last but not least, I suspect the last thing anyone would think of when reading about a device that says it provides isolation, that it actually creates current leakage of its own!

Summary
There is nothing wrong with the testing as performed. Had UpTone bothered to measure the performance of their system with DACs, they would have likely found this issue and hopefully would have re-looked at their system architecture and chose a different design. AC mains current leakage, ground loops, etc. are part and parcel of unbalanced, RCA connections. Fact that adding yet another power supply to what is already a complex configuration causes more mains related distortions should not be surprising. Simplicity sometimes is our friend and ISO regen moves away from that.

 

Amir
Founder, Audio Science Review
Founder, Madrona Digital
Contributing Editor, Widescreen Review Magazine

 

 

[lmitche]

6 minutes ago, plissken said:

 

Amir just responded very well to what I and other were pointing out. No matter the straw man you want to parade about, the ISO regen with power supply made things worse. End of story. 

 

Quoted in full:

 

OK, back from the trip and had a bit of time to answer this. I am going to provide two answers. One high level that is easy to understand and should completely do away with this argument. The second answer, will be a deep dive into power supplies, emissions control, safety, etc. that explains why the problem is created with ISO regen and not otherwise.

First, here is the argument from the first link above:


In a nutshell, he is saying that because ISO regen uses a switchmode power supply and so does the Audio Precision (how he knows the latter is beyond me seeing how he doesn't have one), and that switchmode power supplies have a "required Y capacitor" that what I am observing with respect to increased mains frequency and harmonics in the output of the DAC as measured by my Audio Precision Analyzer, is unavoidable. He is also pointing to an iFi post which was measurements of their power supply (and not the DAC) as proof point of this.

Here is the diagram for how I measured the Behringer UMC204HD as is:



Right away we see the flaw with the statement Alex is making. There are plenty of switchmode power supplies in my playback and measurement chain. The Behringer was powered by the USB power of my desktop workstation. This is a custom PC I built with a switchmode power supply. I don't pretend to know what power supply configuration is used in the Audio Precision Analyzer. I do have a laptop connected to it that controls the analyzer over another USB bus and the laptop was connected through yet another switchmode power supply to mains.

So if switchmode power supplies have "required Y capacitors" and anything upstream of the analyzer would create mains harmonic noise, it should have been there all along. But it was not at nearly a significant level until I changed to this configuration:



In this configuration, yet another power supply made by a half-decent Chinese/Taiwanese company called Meanwell is added in the middle of the chain. This is powering the ISO Regen which in turn generates a new USB bus power. In this configuration we managed to increase the mains related harmonics as I showed before:



By now it should be obvious that the mere fact there are switchmode power supplies upstream of the DAC/USB bus does not at all say that the measurement scheme is faulty. The only change here was insertion of the ISO regen and it created the heightened noise products in the output of the DAC. Clearly what Alex is saying regarding design and operation of switchmode power supplies is incorrect (I will dig into this in the later detailed post).

But let's say for the moment that he is right. Nothing about that means that we can't measure the output of the system with the Audio Precision Analyzer. The AP acts just like any other audio device. It has an analog pre-amp, power supply for the same, and other bits to capture that stream. There is no law against such an audio having a switchmode power supply. Class-d amps for example mostly have switchmode supply. Is Alex saying that in such configuration ISO Regen does screw up the fidelity of the system? If so, why is that not indicated in the specification for the device?

Last but not least, I suspect the last thing anyone would think of when reading about a device that says it provides isolation, that it actually creates current leakage of its own!

Summary
There is nothing wrong with the testing as performed. Had UpTone bothered to measure the performance of their system with DACs, they would have likely found this issue and hopefully would have re-looked at their system architecture and chose a different design. AC mains current leakage, ground loops, etc. are part and parcel of unbalanced, RCA connections. Fact that adding yet another power supply to what is already a complex configuration causes more mains related distortions should not be surprising. Simplicity sometimes is our friend and ISO regen moves away from that.

 

Amir
Founder, Audio Science Review
Founder, Madrona Digital
Contributing Editor, Widescreen Review Magazine

 

 

Yes, I read Amir's response before making my post above that contains a recommendation that will likely fix Amir's testbed.

[scan80269]

Just now, lmitche said:

Yes, I read that before making my post above that contains a recommendation that will likely fix Amir's testbed.

And I suggested yesterday that Amir replace the Mean Well SMPS with a battery to power the ISO REGEN and re-measure.  A purely scientific control experiment with only one component changed.

 

UpTone's LPS-1 power supply should also do the trick, but given Amir's obvious anti-UpTone stance, I thought a battery would at least be "neutral territory".  

 

[esldude]

So if one had a switch mode amp one needs to spend $96 more to isolate the ISO Regen to isolate the USB?

[lmitche]

14 minutes ago, scan80269 said:

And I suggested yesterday that Amir replace the Mean Well SMPS with a battery to power the ISO REGEN and re-measure.  A purely scientific control experiment with only one component changed.

 

UpTone's LPS-1 power supply should also do the trick, but given Amir's obvious anti-UpTone stance, I thought a battery would at least be "neutral territory".  

 

When we had this same conversation about the Amber Regen I was told that no one has an isolation transformer in their audio systems. My response was that no one has an AP in their audio systems. Clearly care must be taken to ensure that the testbed doesn't impact the results especially when those results are posted in a public forum.

[lmitche]

18 minutes ago, esldude said:

So if one had a switch mode amp one needs to spend $96 more to isolate the ISO Regen to isolate the USB?

Again, it was the MacBook SMPS that created the problem in Jud's system, not the amps.

 

I have no experience with switched mode amps and the ISO Regen, so have no answer to your question.

 

I would be curious to know if the 60hz is audible in Amir's testbed. It was definitely so in Jud's system. If Amir owns an amp and speakers this test would be helpful.