Audio by Van Alstine ABX Comparator Review

[Spkrdctr]

@amirm thanks for testing.

It clears things up... a little.
A little analysis.

The added distortion (-75dB) is in the estimated ballpark range of what was expected. Fortunately on both channels in the same way and have to assume this is not attenuation level dependent as in all cases the added distortion will be caused by the 'switch selector' in the volume control IC.
The AD5262 datasheet states worst case is 0.014% and the measurement is a little worse 0.017% so seems correct.
Assuming this is the same for 90 setting as well there is a measurable amount of distortion added and is also relatively high for the more audible higher harmonics.
0.02% THD is something I have never been able to detect using music (because of all the masking going on) so is not alarmingly high but ... undesirable.
Distortion is the same on both channels so the difference (which is what counts) is much, much smaller and this is what's important in an ABX.... the difference.

Some missing info is the used load capacitance and load resistance.
The roll-off will depend on the load capacitance which most likely will be determined by the cable and when it is going into an amplifier might also be influenced by the input capacitance. It thus can only get worse.

An interesting measurement is the attenuation vs frequency response.
At 99 setting (max volume) and, I assume, 100k load of the APX555 we get 1.56dB attenuation and with the used RCA cable a -0.2dB roll-off at 20kHz.

This would calculate as the box having an output resistance of 20k (assuming 100k load, when another load is used that doesn't fly anymore).
This would be an unusual high output resistance as the chip itself would be around 60ohm at 99 volume setting.
It is not unthinkable that, because the 20k IC was used the designer added 20k to the 'wiper' so input and output R would be the same.
This would not have been a good decision if that were the case.

When the box was loaded with 600ohm the output R would be 118ohm which would be closer to the minimum 60ohm of the used volume control IC.
When the box was loaded with 10k (your average power amp) the output R would calculate as 2k.

When the second DAC were to have a different output level that resistance value would increase a little resulting in a lower cut-off frequency.
This is what is measured as well. Seeing what Josh wrote that the different DACs were only different a little and the setting was somewhere between 99 and 90 the 90 setting measurement is handy.
At 90 setting an extra 1.2dB attenuation is present and at 20kHz the roll-off is -0.4dB so -0.2dB with the used load extra.
The roll-off (and subsequent phase rotation) difference will have been max. 0.2dB at 20kHz (and less at lower frequencies).
No idea if such a small difference at 20kHz (and much smaller at 10kHz) is audible and could be a reason.

At setting 80 (2.5dB difference which could happen if DAC A is 2V out and DAC B is 1.5V or 2.7V the roll-off at 20kHz (for the used RCA cable) is -0.6dB so a 0.4dB difference which may well be audible for a young tester.

Missing info is the load resistance and capacitance.

I'm afraid the results aren't conclusive as only the difference between the inputs is what is ABX'ed and those differences are negligible for distortion despite the decrease in fidelity of the box itself (SINAD 75, distortion dominated).

The questionable part is either a difference in 'mute' time (which most likely is not present or too small to detect and is not measured) or a barely audible difference in HF roll-off (at 20kHz) and/or phase difference which would depend on the used RCA cable and load impedance.

I'm fairly certain (given my experience with blind testing in the past) that I would not be able to pass the test based on the measured and expected differences.
There is a considerable measured change brought by the box when using the volume control of the box.

Anyway interesting results.

Jeepers! Solderdude goes all in. I think I can sum up that long and interesting post. "The box doesn't work for its purpose". Thanks for a great post solderdude!

 

[Janfi67]

Thanks to @amirm and everyone who contributed to the acquisition of this wonderful relay box.

@Josh83 is a true marketing genius.

With this dissonant relay box that allows identifying the selected source, he writes a review on a controversial topic, with results that should astonish anyone who relies more on their intellect than their emotions. It goes viral on ASR, attracts tons of people to his website who would have never gone otherwise, and on top of that, he pockets $1,300.

Talk about a masterclass.

 

[Spkrdctr]

Well, as you all know i have passed the torch to Amir in debunking all this silliness in supposed "tests" that are not really tests and so on. But from my history I immediately thought the best test is to have a switch, like a $1 toggle switch on a box with wires going to nowhere. The box will look like it is all hooked up and everything is fine but in reality it is a hollow box with an unconnected to anything switch on it. Then run tests. Of course the satisfying "click" will be heard every time the switch is toggled. Then see the results. It should be that the person never hears ANY difference at all as it is impossible if nothing is hooked up to anything. The results will surprise the hell out of everyone here. I would bet out of a hundred people tested they would all hear differences that are not there. The test results of course would be all over the place and show that people can't really hear what they think they can. I hate to be the guy bursting bubbles all the time but it is what it is. This is why it is so important to go by Amir's tests on products. Many on ASR though think they have God level ears but I have stopped bringing that up as no one cares! Great job Amir! Once again Amir brings reality and clarity to a confused mess.

 

[pma]

My point is that any Golden Ear should immediately have concluded that the device is inadequate by just listening to it

I have had quite enough personal experience with self-appointed “Golden Ears”. Some of them did not realize when I disconnected the tweeter of a 3-way speaker, some of them did not notice phase inversion in one of the stereo channels.

 

[mlsstl]

I just breezed through this thread so may have missed something, but one item I don't think I saw mentioned was that the "level matching" appears to be done in 10 dB intervals. If so, that is hardly a level match -- things could still be off by 5 dB. Or, is the level adjustable by 1 dB increments?

 

[voodooless]

the "level matching" appears to be done in 10 dB intervals.

It’s not.

 

[gzost]

It's in 0.1 dB increments, which should be good enough.

 

[Kremmen]

It's in 0.1 dB increments, which should be good enough.

Absolutely, especially when one considers that our bodies have their own internal 'noise floor.' The sound of our heartbeat, the rushing of blood through the carotid arteries in our neck, and the microscopic twitching of the hair cells inside our cochlea all create internal acoustic activity.

Mind you, there will always be some 'golden-eared' BS artist who will claim that despite this fact they possess superhuman hearing and can detect sound level differences etc. beyond what mere mortals can. Rob Watts is the perhaps most notorious BS artist with regard to such ludicrous exaggerations. So very sad, especially for such an obviously intelligent man.

 

[Sir Sanders Zingmore]

I tried recording the sound using my phone but the mic is not good enough to record the full spectrum.

Surely you of all people have access to a proper microphone…

 

[welwynnick]

Surely you of all people have access to a proper microphone…

He also has a huge backlog of other gear to test.

 

[welwynnick]

All the further discussion and turmoil here is pointless.

Agreed.
Amir has answered the question.
And yet the discussion goes on.

EDIT : I think what you're doing in the "DAC listening test files" thread is much more valuable.

DAC listening test files - updated with new filters files

UPDATED - NEW TEST FILES, September 21 / 2026 Hello all, after a short break that was needed to prepare and test new files, there is a link to new set of test files. https://drive.google.com/file/d/15aR4OTuDIHystUynvuWIpBiVoj0p2zEe/view?usp=sharing The test files were prepared by...

www.audiosciencereview.com

For those who weren't aware, the ASR endeavour to make a proper ABX comparator is ongoing, perhaps now with renewed interest? :

Reference ABX hardware design and implementation

Nice. I see two diodes for each mosfet pair. Is it zenerers or TVR's? Have you done any calculations on the spike energy from the speakers? Hey @Armand - One is a 15V gate-protection Zener, and the other is in the gate-drive circuit to open a lower resistance path to speed up discharge. The...

www.audiosciencereview.com

 

[Kremmen]

Agreed.
Amir has answered the question.
And yet the discussion goes on.

Sadly, it's human nature!

 

[pma]

Others that would have liked to see measurements simply continue to think the result is a fluke.

Let me paraphrase. Hypothetically, I was given a task to measure my mains voltage with 1mV precision. Like 230.013V +/-0.001V. I have had an instruments calibrated to measure with 0.0001V precision, but now it is lent. The other instrument I have measures with only 1V resolution and shows 230V. My mains voltage has changed of 0.01V and I am declaring I was able to measure it with an instrument with 1V resolution, in a good faith. Would you believe me? That is the analogy with Van Alstine box used to differentiate between DACs with linearity better than 120dB. It is similar as guys telling differences of DAC sound from youtube videos with sound recorded by a microphone, not level matched, and then listened to heavily lossy compressed sound.

 

[shuppatsu]

Do you not trust his result?

I agree with @pjug that it isn’t about trust. I trust Amir, just as the A/S people trust Josh. But if we’re the side that represents science and rationality than we have to do better than “trust me bro.”

For the most part both sides agree that extraordinary claims require extraordinary evidence. But we have markedly different ideas of what claims are extraordinary. When Josh presented his findings we demanded a bunch more information. Recordings, access to the equipment chain, etc. Now that the shoe’s on the other foot we should at least be able to provide a recording of the relay switch sounds.

 

[PjV]

The positive news is that now this box has great value for manufacturers.
You can prove in ABX that cables sound different or that a rack makes the silence more black, or whatever...

 

[markk]

This falls into the "extraordinary claims require extraordinary proof." Here's a guy who is claiming to distinguish between DAC's at a rate of ~98%. Given the high standard of the modern DAC, with flat FR, very high SN/vanishingly low distortion, this would be a most unusual result.

Not surprising to see that the AB testing methodology and equipment lacks rigor and is not truly blind in its operation. Additionally, there is no independent verification of Josh's results.

While it's impossible to see inside someone's heart to ascertain their motives (i.e. if they were lying), if someone can reliably pick a particular unit, it's not unreasonable to conclude it's on the basis of something other than sonic differences in DAC's. In this case, sonic differences between relays.

If that is the case though, It would be surprising if Josh was completely unaware of what the actual difference was. To be clear, the sound of the relay...

Ultimately, nothing really to see here. Another extraordinary claim without extraordinary proof and a cloud of suspicion regarding motives.

Move along...

 

[pjug]

I agree with @pjug that it isn’t about trust. I trust Amir, just as the A/S people trust Josh. But if we’re the side that represents science and rationality than we have to do better than “trust me bro.”

For the most part both sides agree that extraordinary claims require extraordinary evidence. But we have markedly different ideas of what claims are extraordinary. When Josh presented his findings we demanded a bunch more information. Recordings, access to the equipment chain, etc. Now that the shoe’s on the other foot we should at least be able to provide a recording of the relay switch sounds.

@Josh83 said the clicks are uniform (I don't see him ever saying the IEMs or anything else prevented him from hearing the clicks). Amir said the clicks are clearly different. A recording will show everybody one way or another. If the clicks are significantly different then Amir has a pretty good summation of all of this when he says:

I also have to wonder how someone who says he can tell the difference between DACs can't hear differences between relay clicks!

 

[shuppatsu]

@Josh83 said the clicks are uniform (I don't see him ever saying the IEMs or anything else prevented him from hearing the clicks). Amir said the clicks are clearly different. A recording will show everybody one way or another. If the clicks are significantly different then Amir has a pretty good summation of all of this when he says:

I also have to wonder how someone who says he can tell the difference between DACs can't hear differences between relay clicks!

I can believe that very easily. There are headphone aficionados who have confessed to accidentally listening in mono for months at a time. Many people here have admitted to not noticing less dramatic inadvertent changes to their systems. If the click changes are too subtle to be picked up by the iPhone microphone, it's probably pretty subtle. If Josh wasn't trying to differentiate the sound by relay clicks or stress test the comparator, he may have never noticed the difference and later written them up as "uniform."

But of course he doesn't need to have consciously noticed the difference in relay clicks in order to have it influence his testing.

 

[JeffS7444]

Telltale relay noise aside, I'd think the speaker-switching portion would be okay..?

But seems to me that a purely passive line-level section was always going to be problematic, and that an active buffer such as @pma's would be a must in order to minimize the effects of cable LCR.

 

[pjug]

I can believe that very easily. There are headphone aficionados who have confessed to accidentally listening in mono for months at a time. Many people here have admitted to not noticing less dramatic inadvertent changes to their systems. If the click changes are too subtle to be picked up by the iPhone microphone, it's probably pretty subtle. If Josh wasn't trying to differentiate the sound by relay clicks or stress test the comparator, he may have never noticed the difference and later written them up as "uniform."

But of course he doesn't need to have consciously noticed the difference in relay clicks in order to have it influence his testing.

And also of course it would be impossible to block the clicks from his conciousness if he could hear them at all clearly. Like most people do in difficult ABX, Josh was switching back and forth between small sections of the tracks. So he would be exposed to a whole lot of clicking, and over and over again same clicking differences no matter what DACs or music selection.