Audio by Van Alstine ABX Comparator Review

[solderdude]

Wow. Ad hominem aside, do you think he should be expected to dismantle or re-wire his expensive Klippel, potentially impacting calibration, to satisfy your curiosity.

No, not at all, and also not just to satisfy my curiosity but to add tangible proof !

Just place the box on the speaker stand and/or record (using the connected the APX555 or whatever is connected).

No need for any impact on any device at all. No need for even unplugging anything nor any dismantling nor rewiring nor losing calibration or whatnot.
In fact the recorded sound level would even be calibrated !
Again I did not 'demand' this just asked if Amir could post a recording of the clicks. If he doesn't ... that's fine too.

However, it might be simpler and about as effective to just make a YT video and put the mic he uses for that close to the box ?
Possibly Amir shoots the videos on his phone using built-in mic though and having it directly aside from him should be enough to make it audible.

It would be extremely illuminating and remove any doubt some may have and disprove the remarks Josh made not being able to hear a difference in relay sound. Hence my request to record and simply post the relay sound. No analysis is even needed (like the one from Pavel). It would just be more evidence of some aspect that has not been 'measured' and was likely the most 'telling' tell of all possible tells.

 

[pma]

Background noise.

 

[pjug]

Background noise.

Sure, but if the recording is anything like your ABX recordings of your own comparator box relays, it is clear enough to anyone who is being reasonable. FWIW my ABX results below. I think anyone would pass your ABX.

foo_abx 2.0.6d report
foobar2000 v1.6.6
2026-10-04 09:19:59

File A: off.wav
SHA1: c6fcdd59754c7a41359513c21392c37ab023b718
File B: on.wav
SHA1: a9fb51fc314910e927d504730d15e07d8af2f1e0

Output:
Default : Primary Sound Driver
Crossfading: NO

09:19:59 : Test started.
09:20:28 : 01/01
09:20:43 : 02/02
09:21:03 : 03/03
09:21:14 : 04/04
09:21:24 : 05/05
09:21:31 : 06/06
09:21:37 : 07/07
09:21:44 : 08/08
09:21:52 : 09/09
09:22:03 : 10/10
09:22:15 : 11/11
09:22:21 : 12/12
09:22:26 : 13/13
09:22:36 : 14/14
09:22:44 : 15/15
09:22:51 : 16/16
09:22:51 : Test finished.

----------
Total: 16/16
p-value: 0 (0%)

-- signature --
dd66e1ce04ec6e7e1078dab1b2533a6227581926

 

[David_M]

I don't understand why the main switching circuitry in this box needs to be active, aside from the DC voltages needed to switch the relays and any possible remote-control receiver circuitry. It's supposed to be a pass-through box, a straight wire with zero gain. Any gain adjustments would be done upstream by the AB sources. A passive box would provide much better specs, imho.

 

[pjug]

I really hope that this is not the end of the evaluation of the box. I understand if @amirm does not want to do more, but maybe @NTTY or someone else would be willing do do more. Such as:

- Acoustic recording of the relay clicks (maybe include switching from X too, not just between A and B)
- recording of electrical artifacts injected into signal by the switching, to see if it is uniform between switching between A, B, and X
- Video an ABX of the same DAC going to both A and B. If this can be passed it proves the box has some tell.

Edit: silly me, there is no additional relay for X

 

[solderdude]

I don't understand why the main switching circuitry in this box needs to be active, aside from the DC voltages needed to switch the relays and any possible remote-control receiver circuitry. It's supposed to be a pass-through box, a straight wire with zero gain. Any gain adjustments would be done upstream by the AB sources. A passive box would provide much better specs, imho.

The amp-speakers selectors are simple relay contacts. And even those can (and do) add a little distortion.

The audio path is 'passive' (can only attenuate) but there are semiconductors in the signal path. These are the 'switches' inside the volume control IC where the signal has to pass though and its these 'electronic switches' that add the distortion. Using the chip was the designers choice for whatever reason he had.

The fun part, and also its downfall, is the possibility to attenuate different RCA input levels in small steps (0.1dB) so one can level match RCA output level devices so one can ABX gear using this device.
The idea the designer had was to not use any active electronics in its path and use a 'passive attenuator'.
This probably because people are afraid active electronics would impart a sound.
Sadly the distortion the volume control chip used in this device is magnitudes higher than even the worst op-amps around and would impart more 'sound signature' than any op-amp possibly could.

So.. it should not be a straight wire with 0 gain but rather a wire with finely adjustable gain.
It is not remotely a wire nor an attenuated wire (despite of being 'passive') in fact it is bandwidth limited and adds of distortion (below audible thresholds with music IMO).

 

[pma]

The amp-speakers selectors are simple relay contacts. And even those can (and do) add a little distortion

Depends strongly on contact material + geometry. I may add some measurements, later.

 

[pma]

I think anyone would pass your ABX.

Such is the goal.

 

[pma]

Sure, but if the recording is anything like your ABX recordings of your own comparator box relays

And believe me it is not that trivial to get a clear and quite clean result. In a discipline like this a human ear is better than a microphone. I assume Amir had had a reason not to post his cell phone record of relay clicks.

 

[pjug]

And believe me it is not that trivial to get a clear and quite clean result. In a discipline like this a human ear is better than a microphone. I assume Amir had had a reason not to post his cell phone record of relay clicks.

Just record at night in winter after the wind has calmed down following a nice snowfall. Easy conditions to create, ha ha.

 

[pma]

That is not the case, in fact.

 

[amirm]

FYI I started to do some listening tests. The chain is highly susceptible to ground loops/noise. So much so that with my IEMs at higher gains I can easily hear my computer operating! Close a browser tab and you clearly hear its effect in the IEM! It has been decades since I have heard this kind of noise.

Background noise is also strong, variable and even muting doesn't stop it. This is all with one DAC driving the switcher and then a headphone amp. I will hook up another DAC and see what happens. But for now, what the AP measurements showed was the best case scenario.

 

[pma]

FYI I started to do some listening tests. The chain is highly susceptible to ground loops/noise

It is a horrible design. Horrible and expensive! I wonder why so many audio designers have been that incompetent.

 

[amirm]

I also put the unit in its ABX mode. Boy, it does a lot less than I thought. You run the test. But then it is your responsibility to record your selection, and then to turn on the switcher LEDs and walk through to see which ones you got right! In other words, there is no score keeping at all. I could very well note a trial wrong during blind and/or then reveal cycle or make up the entire results!

I thought it would have a voting button for each trial that it would then report the results at the end. But there isn't. In this regard, there is no way to show evidence that you passed a test. Hence, it is just for one's personal satisfaction than proving an outcome externally. You are completely at the mercy of the person reporting the results as to accuracy and honesty.

 

[donjoe0]

But this is just not the correct assumption. In statistical terms, since Amir's testing shows the testing setup is inadequate

Inadequate in some ways but not in others, see comments by Solderdude.

This is what everyone in scientific research does.

Scientific research is done in may ways for different purposes. In this case, if you come at it with conservative intentions like you're a DAC manufacturer who would really hate having their existing designs called into question after being marketed for decades as "audibly transparent (to all humans in all situations)", you will try to use only existing evidence to estimate(!) that Josh probably has typical or close to typical hearing, if there's any evidence whatsoever that points to one confounder he could've fallen for, and leave it at that, not explore any other possibilities that may be still open. Saves you from having to design anything better, saves the reputation of your existing products, mission accomplished, case closed.

However if you come at it with pure curiosity and a desire to expand human knowledge wherever the opportunity seems genuine, you dig deeper, investigate every possibility not yet ruled out, try to *really* prove if the seeming outlier is an outlier (yes, still human, like the ones with the verified outlier results I just pointed you to) or not.

He's saying that the variable FR depending on attenuation levels could conceivably enable Josh to identify 'X' as 'A' or 'B,' even if 'A' and 'B' measured identically apart form voltage.

That's one of the possibilities, but of course the more interesting one to me is that he heard actual FR differences between the DACs themselves, which maybe do not measure identically. Or IOW I assume we don't have very good data on how close they measure to absolute transparency, because no measurements-reviewers I've seen in audiophiledom so far have paid proper attention to the actual limits of FR deviation audibility, nor measured and presented DAC FRs at appropriate resolutions to support such categorizations as "transparent".

I couldn't open the AES paper he linked. Have JNF levels under 0.4dB@20KHz been demonstrated?

Yeah, me neither, but @RandomEar might still have a copy, or at least had access for a while and left us some of the relevant parts here: https://www.audiosciencereview.com/...n’t-even-measure-”.72288/page-15#post-2651009
(It's paper [D], mislabeled "Scott et. al".)

 

[Plcamp]

FYI I started to do some listening tests. The chain is highly susceptible to ground loops/noise.

In my experience, such susceptibility can greatly vary with small changes in I/O cable placement and orientation. Will be interesting to see if you detect recognizable differences between two attached devices simply from this effect.

 

[respice finem]

Just a thought: If a device should be used to compare other devices, it should be at least not inferior compared to them. If it's inferior, it's not "fit for the job".

 

[tmtomh]

Inadequate in some ways but not in others, see comments by Solderdude.


Scientific research is done in may ways for different purposes. In this case, if you come at it with conservative intentions like you're a DAC manufacturer who would really hate having their existing designs called into question after being marketed for decades as "audibly transparent (to all humans in all situations)", you will try to use only existing evidence to estimate(!) that Josh probably has typical or close to typical hearing, if there's any evidence whatsoever that points to one confounder he could've fallen for, and leave it at that, not explore any other possibilities tbat may be still open. Saves you from having to design anything better, saves the reputation of your existing products, mission accomplished, case closed.

However if you come at it with pure curiosity and a desire to expand human knowledge wherever the opportunity seems genuine, you dig deeper, investigate every possibility not yet ruled out, try to *really* prove if the seeming outlier is an outlier (yes, still human, like the ones with the verified outlier results I just pointed you to) or not.


That's one of the possibilities, but of course the more interesting one to me is that he heard actual FR differences between the DACs themselves, which maybe do not measure identically. Or IOW I assume we don't have very good data on how close they measure to absolute transparency, becuse no measurments-reviewers I've seen in audiophiledom so far have paid proper attention to the actual limits of FR deviation audibility, nor measured and presented DAC FRs at appropriate resolutions to support such categorizations as "transparent".


Yeah, me neither, but @RandomEar might still have a copy, or at least had access for a while and left us some of the relevant parts here: https://www.audiosciencereview.com/...n’t-even-measure-”.72288/page-15#post-2651009
(It's paper [D], mislabeled "Scott et. al".)

The core of your position here is that Josh heard frequency response deviations and that we don't know what the limits of human hearing are with regard to such deviations, and so the DACs might very well have FR deviations from each other that are audible, and DAC manufacturers don't want to hear that (no pun intended) because it would force them to design better products.

This all sounds reasonable, but it employs one of the most common misleading moves in audiophile discussion: making key questions sound more unknown than they actually are.

For your point to be persuasive, at least in my view, the DACs in question would have to have measurable FR nonlinearities, or else we would have to be lacking in any measurements of them at all (in which case we wouldn't know if they have such nonlinearities). But the FR measurements of DACs are easily accessible, and anyone can examine them to see if they're ruler-flat or not.

For your point to be persuasive, we also would have to have no clear sense of what kinds and magnitudes of FR deviations are likely to be audible. Here again, we can't just say "X dB is definitely audible or inaudible" as a single number - but we do have access to studies that give us a clearer sense than you suggest. For example, differences of about 0.1dB, perhaps slightly less, can conceivably be audible with signals like sine waves, in the right circumstances. With musical signals the differences increase, if memory serves to 0.2dB (but even if I'm wrong about the magnitude, the point is that the question has been studied and tested).

And for your point to be persuasive, any small high-frequency response differences between the DACs have to be compared to the small high-frequency response differences of the ABX box itself at different volume levels. Josh stated that to volume-match the DACs he had to adjust the ABX box's volume level for each DAC. The adjustments were small and the resulting FR differences of the box itself were therefore small - but they were not necessarily smaller than the FR differences between the DACs themselves. So for example a 0.13dB ABX box-induced drop in response with a DAC whose top-octave response was already, say, 0.09dB down compared to another DAC, could potentially produce the kind of tiny effect that Josh says he found audible. Why should a DAC manufacturer or designer bother to investigate a 0.09dB drop between 10 and 20kHz if the drop is not actually audible and was only audible to one particular listener when it was increased (in this example) to 0.22dB because of the ABX box?

There is also the as-yet untested question of the impact of impedance on the FR of the ABX box.

And finally, your entire claim that DAC manufacturers don't want to have to improve their products is undermined by Josh himself, who has repeatedly characterized the differences he heard as very small and not significant in terms of determining which one would sound "better."

So in terms of the position you are asserting here, you are way, way out in front of your skis.

 

[amirm]

We have claims from Josh that what he heard was soundstage differences, not frequency response differences. Maybe he can't tell the difference but assuming he does, then frequency response variation doesn't explain his claims.

 

[RandomEar]

Yeah, me neither, but @RandomEar might still have a copy, or at least had access for a while and left us some of the relevant parts here: https://www.audiosciencereview.com/...n’t-even-measure-”.72288/page-15#post-2651009
(It's paper [D], mislabeled "Scott et. al".)

See attachment.