Help GoldenSound / GoldenOne Get An Audio Precision Analyzer

[Confused]

13 hours ago, GoldenOne said:

Yep paul miller certainly has some excellent stuff. Though I do wish the measurement suite was a little more thorough.

Which actually I suppose is a good question, what are some measurements that people wish were taken of more products?

 

One measurement that interests me is amplifier output impedance.  The reason for this interest comes from a time I listened to three amplifiers on the same day, with the same speakers.  The amps were at slightly different price points, with a variation in specs as you might expect.  One thing that I did pick up was a noticeable difference in the presentation of the bass.  One amp seemed to have better bass depth and accuracy, versus another that had maybe slightly more prominent upper bass, but lacked the depth and accuracy of the other.

 

Later, when I checked the specs, I noticed that one amp had a lower output impedance versus the other.  I have always suspected that this was a least a contributing factor to what I observed.  Later amp demo's seemed to follow a trend, that in terms of bass performance, my subjective preference went to amps with the lowest output impedance.  OK - This hardly meets the requirements for double blind testing or any kind of statistical significance, but it does seem to me that output impedance is given very little prominence in measurements, in comparison to power output, signal to noise ratios and similar.

 

I could also answer your question in an entirely different way.  It seems to me that the big issue with measurements, and the objective vs subjective debate is that there is very little correlation between what is observed and what is measured.

 

To take an example.  One controversial area at the moment is "audiophile" ethernet switches.  I have seen measurements of switches that show differences at the DAC, but these differences are so small that they could not possibly relate to anything auditable.  Yet there remains many reports of clear audible differences with switches.  Indeed, some people are buying switches costing $1k, $2k and even $5k.

 

So am I suggesting that you should measure switches to find out what is going on?  Not really.  OK, you could do this if you wanted to, and who knows, you might find something, but I don not think this is the right approach.  Step 1 should be establish that there is an audible difference, and this should be done with a properly controlled blind test.  Once a clear situation is established where X demonstrably sounds different to Y, then the job of the measurements would be to try to establish the exact mechanism causing the audible difference.  From here we would learn.

 

The problem as I see it is that we have a vast body of subjective observations claiming all sorts of things, but proper controlled blind tests are almost non-existent.  I know that there are some that state that blind tests do not work for whatever reason.  OK, fair enough, but I would happily take part in a blind test, and if we are to establish a scientific link between the measured and the observed, I cannot see any other way (sensibly) of doing this other than starting with an observable difference established via blind testing in a controlled way.

 

I did have a slightly crazy idea of persuading a dealer to run blind test events.  I do know of one dealer that likes to have lots of events.  Of course, conventional wisdom would dictate that this is not is the said dealers interests, he want to have sighted tests where potential customers easily "hear" the difference and then happily buy the expensive thing tried.  This is probably how it will forever work, although I would add the contrary view that I might listen to two products, think I hear a difference, but then convince myself it is probably just imagined, placebo or expectation bias, thus keeping hold of my money.  If however I could reliably hear a difference between X and Y in blind conditions, this would give me the confidence to spend.  A crazy thought I know, so no need for anyone to explain why it will never happen.

 

So this is the key I think, somehow we need to organise more independent blind testing.  From this, interesting things to try and measure could follow, and if measurable differences were found, we could then try to correlate this to what is already established as audibly different.

 

Without such blind testing, I am not sure if more measuring will get us anywhere?

[opus101]

The measurement closely related to output impedance is 'damping factor'. That's the ratio of a typical speaker impedance (normally 8ohm) to the output impedance. Its often specified but I think it would be better replaced by output impedance, at a given frequency.

[GoldenOne]

15 hours ago, idiot_savant said:

something to bear in mind is that there *may* be a bit of shenanigans going on for products dynamically optimising themselves for the AP test suite, feel free to ping me if you want a bit of devious thinking here

Oh I'm sure, in fact we've seen situations where this has occurred before like the SMSL M500  https://www.audio “science” review/forum/index.php?threads/strange-smsl-m500-high-3rd-harmonic.22710/

 

 

14 hours ago, idiot_savant said:

I’ve reread this, and sound like a right idiot…

 

I’m trying to applaud your efforts yet point out a bit of subtlety could be how you differentiate yourself from other measurement websites - it’s easy ( but not cheap! ) to just buy an AP and run the test suite…

 

your friendly neighbourhood idiot 

Not in the slightest! Don't worry I took it as constructive :)
I agree, pressing the "APx go brrr" button is pretty easy. But that's why I'm keen to get one, to do the 'other' stuff. To do additional tests and look deeper into things. And unlike 'certain other sites', not just to do everything solo. I'm fully planning to work with and get advice and input from knowledgeable and relevant people wherever possible because that's how stuff moves forward and it'll make for much better and more interesting content too.
 

 

11 hours ago, opus101 said:

have they increased that top limit on their newer models?

I believe the oscillator is still up to 200khz

[PeterSt]

55 minutes ago, GoldenOne said:

I believe the oscillator is still up to 200khz

 

There has to be a typo somewhere. 🤪

[GoldenOne]

11 minutes ago, PeterSt said:

 

There has to be a typo somewhere. 🤪

DAC is 0.001hz to 80khz
sine oscillator is 5hz to 204khz

ADCs are 192khz bandwidth for the ultra-high performance stuff and then a little lower on the 1.2mhz bandwidth

So you can still test switching amps etc, but I don't think there's going to be a need to actually generate a signal at >200khz so it should be fine :P

[PeterSt]

5 minutes ago, GoldenOne said:

so it should be fine :P

 

Of course it is fine. It is only that no oscillator of 200KHz or 192KHz etc. will be in there.

 

sine oscillator is 5hz to 204khz

 

Maybe they call it like that. But if so, that would be a (signal) generator. Thus something else again.

Oscillators in this realm will be in the xxMHz range.

 

And the LF oscillator in a synth should be able to go as low as 0.05Hz (preferably much lower). Haha.

 

The gist is clear: these (24 bit) machines still don't go beyond a useful range of 96KHz. Strange actually; Already 10 years or so back I asked "so what about DAC's being able to do 786KHz ? isn't it time that you dunk in an ADC doing that ?". ... The ADC's did not exist of course (not for 24 bits). This has changed now. ... I suppose AKM is not good enough for them ?

 

So Yes, I have been contemplating how my RME could be supplemental (I still did not unpack it).

[GoldenOne]

2 minutes ago, PeterSt said:

 

Of course it is fine. It is only that no oscillator of 200KHz or 192KHz etc. will be in there.

 

sine oscillator is 5hz to 204khz

 

Maybe they call it like that. But if so, that would be a (signal) generator. Thus something else again.

Oscillators in this realm will be in the xxMHz range.

 

And the LF oscillator in a synth should be able to go as low as 0.05Hz (preferably much lower). Haha.

 

The gist is clear: these (24 bit) machines still don't go beyond a useful range of 96KHz. Strange actually; Already 10 years or so back I asked "so what about DAC's being able to do 786KHz ? isn't it time that you dunk in an ADC doing that ?". ... The ADC's did not exist of course (not for 24 bits). This has changed now. ... I suppose AKM is not good enough for them ?

 

So Yes, I have been contemplating how my RME could be supplemental (I still did not unpack it).

To be specific I was referring to the analog sine generator. These are often referred to as oscillators such as here https://quantasylum.com/collections/frontpage/products/qa480 or general "Wien bridge oscillator" designs (though I don't know what specific design the one in the AP is.) 

I wasn't meaning crystal oscillators etc.

[The Computer Audiophile]

Hi @GoldenOne Here's something I've been thinking about because the only way I know how to test it is kind of crazy. If there was a test to discover an issue like this, it would be really neat. Otherwise, only reviewers using equipment for periods of time like normal users, will discover this type of stuff. 

 

Issue: I had a component here that would stop playback and require a reboot when I flipped the switch for my bathroom fan. It turns out this only happened in the winter when the humidity in my house is really low. The component, I believe, built up static in a chip, that eventually caused the issue. I also believe there was an issue with pops and ticks at strange intervals when the air was extremely dry in my house. Like the static was being discharged only partially, rather than "fully" by the fan switch. 

 

Question: Can you think of a way to test for this?

[GoldenOne]

2 minutes ago, The Computer Audiophile said:

Hi @GoldenOne Here's something I've been thinking about because the only way I know how to test it is kind of crazy. If there was a test to discover an issue like this, it would be really neat. Otherwise, only reviewers using equipment for periods of time like normal users, will discover this type of stuff. 

 

Issue: I had a component here that would stop playback and require a reboot when I flipped the switch for my bathroom fan. It turns out this only happened in the winter when the humidity in my house is really low. The component, I believe, built up static in a chip, that eventually caused the issue. I also believe there was an issue with pops and ticks at strange intervals when the air was extremely dry in my house. Like the static was being discharged only partially, rather than "fully" by the fan switch. 

 

Question: Can you think of a way to test for this?

That's an interesting one, and would definitely be a design fault.
I'd imagine it'd probably need some probing on the board itself rather than something that could be properly tested at the output, but the fact that it was so responsive to humidity is definitely odd.
Was this device grounded or floating?

To be honest I imagine there are some situations where buildup of charge or other similar things may impact a device. One that I've been curious about for a while is the schiit yggdrasil. That is notorious for having a DAYS Long "warmup" time. But It wouldn't really make sense for it to be a thermal issue. As that would sort itself out within a couple hours at most. It'd make more sense if it was some sort of buildup or stabilization of some other electrical property.

It'd certainly be interesting to look at. To create a proper test would probably need to know more about the specific component though.
The fact that it stopped playback when you turned on a mechanical component suggests that it could also be that it was extremely sensitive to DC on the mains. Fans, compressors, fridges etc are usually the main culprit for that. Any kind of reactive load.

[The Computer Audiophile]

2 minutes ago, GoldenOne said:

That's an interesting one, and would definitely be a design fault.
I'd imagine it'd probably need some probing on the board itself rather than something that could be properly tested at the output, but the fact that it was so responsive to humidity is definitely odd.
Was this device grounded or floating?

To be honest I imagine there are some situations where buildup of charge or other similar things may impact a device. One that I've been curious about for a while is the schiit yggdrasil. That is notorious for having a DAYS Long "warmup" time. But It wouldn't really make sense for it to be a thermal issue. As that would sort itself out within a couple hours at most. It'd make more sense if it was some sort of buildup or stabilization of some other electrical property.

It'd certainly be interesting to look at. To create a proper test would probably need to know more about the specific component though.
The fact that it stopped playback when you turned on a mechanical component suggests that it could also be that it was extremely sensitive to DC on the mains. Fans, compressors, fridges etc are usually the main culprit for that. Any kind of reactive load.

 

It took a little while to figure this one out, but I think the larger items of importance would be some way to test for things like this, even if not humidity related, there are likely many other ways products can exhibit strange behavior. 

[GoldenOne]

Just now, The Computer Audiophile said:

 

It took a little while to figure this one out, but I think the larger items of importance would be some way to test for things like this, even if not humidity related, there are likely many other ways products can exhibit strange behavior. 

I'm sure a device could be made in order to simulate mains DC offset and therefore test a devices resilience to it. Though would need to be careful as to the level of offset tested to avoid damaging something.

Normal ESD testing should be done by the manufacturer, and is part of the requirement to get the CE mark (though I'm sure many manufacturers put the CE mark on their products regardless of if they actually qualify for it or have had their product tested).
But that would only be protection from external ESD. 
As to buildup of static charge in the device itself or components in it, this would need to be done with a static meter https://fraser-antistatic.com/products/715-static-meter/
 

I think unfortunately even with the tools to do the above, it'd be tricky to get much insight as to how it affects the product and might just need to be something that feedback is given on through normal use. Because without having deeper knowledge of the design or working with the manufacturer it's not necessarily possible to know what acceptable or normal levels are and how abnormal levels might affect things.

Proper grounding should alleviate most issues though I'd expect
 

[fas42]

IME controlling static influences are a very large part of obtaining optimum sound - which means that there is no such thing as measuring the performance of some apparatus at a point in time, and therefore "having the full story" - if you are really measuring something that is truly relevant to the subjective perception, then this will highly likely change over a period of time, depending upon all sorts of factors - and therefore will end up as a complex graph, with many axes.

[GoldenOne]

So I'm going to have an APx555 here for a couple weeks (not mine, not keeping it, just here temporarily). This likely won't be long enough to do some of the more in-depth stuff but I'm going to measure anything I can get my hands on just cause why not, but also will probably do a brief video explaining a little about why the APx555 is different to just using a good ADC etc.

If anyone has anything specific that they think I should do whilst it's here lmk!

[GoldenOne]

Friend of mine sent a small present to keep the analyzer company. 

 

 

[Confused]

On 6/14/2021 at 2:51 PM, GoldenOne said:

So I'm going to have an APx555 here for a couple weeks (not mine, not keeping it, just here temporarily). This likely won't be long enough to do some of the more in-depth stuff but I'm going to measure anything I can get my hands on just cause why not, but also will probably do a brief video explaining a little about why the APx555 is different to just using a good ADC etc.

If anyone has anything specific that they think I should do whilst it's here lmk!

I am a little surprised there was not more response to this one.  Anyway, reading through some recent threads here on AS gave me an idea of something that could be tested.

 

I believe you have an "NAA" type device?  (SOtM sMS-200?)  In other threads there has been debate about computer / PC "noise" that can find its way to the DAC, so how you play back a bit perfect file can influence the sound, depending on the configuration of the PC, software, noise etc. 

 

As a subset of such discussions, it has been suggested by some that "network attached end point" type devices should effectively be immune to whatever might be going on in the associated PC, as long as the said machine is delivering the bits, nothing else matters.  Others dispute this notion.

 

Which gives me an idea for a test.  You could measure the output of the NAA (or perhaps the output of a DAC connected to the NAA) using different PC's to feed the NAA, or maybe the same PC is different states, first just running simple playback software, then maybe have the computer mutli-task, running the playback software and at the same time with the PC running near to the processor limit running another task.  There are very many ways this could be done, but the key would be to have the feeding computer in definitively different states, or maybe use a simple LPS powered NUC versus a big noisy SMPS ridden high powered PC.

 

Just a thought, maybe it is a bad idea for some reason, maybe a more optimum test could be devised?

 

Might be interesting though?

[Confused]

2 hours ago, GoldenOne said:

Friend of mine sent a small present to keep the analyzer company. 

 

The Panther looks a little disgruntled, and maybe looks as if it is actually "leaving a small present" on your analyzer in response.

[GoldenOne]

16 minutes ago, Confused said:

I am a little surprised there was not more response to this one.  Anyway, reading through some recent threads here on AS gave me an idea of something that could be tested.

 

I believe you have an "NAA" type device?  (SOtM sMS-200?)  In other threads there has been debate about computer / PC "noise" that can find its way to the DAC, so how you play back a bit perfect file can influence the sound, depending on the configuration of the PC, software, noise etc. 

 

As a subset of such discussions, it has been suggested by some that "network attached end point" type devices should effectively be immune to whatever might be going on in the associated PC, as long as the said machine is delivering the bits, nothing else matters.  Others dispute this notion.

 

Which gives me an idea for a test.  You could measure the output of the NAA (or perhaps the output of a DAC connected to the NAA) using different PC's to feed the NAA, or maybe the same PC is different states, first just running simple playback software, then maybe have the computer mutli-task, running the playback software and at the same time with the PC running near to the processor limit running another task.  There are very many ways this could be done, but the key would be to have the feeding computer in definitively different states, or maybe use a simple LPS powered NUC versus a big noisy SMPS ridden high powered PC.

 

Just a thought, maybe it is a bad idea for some reason, maybe a more optimum test could be devised?

 

Might be interesting though?

I did actually touch on this briefly in my DDC video.

Using a network streamer can be a big benefit especially if you were previously using a beefy (and very electrically noisy) PC.
And dacs can and do indeed show different measurements in both things such as outright THD+N and Jitter when connected via a more ideal source such as a streamer vs a PC.
How much this change is will depend on the specific DAC, with some dacs showing very obvious changes, something like an apple dongle being a rather extreme example. And some dacs which have proper internal galvanic isolation like a Yggdrasil or May should have no change at all.

In terms of USB vs an I2S/AES/SPDIF streamer though, the latter will affect performance more due to the fact that the source device/streamer is the master clock, not the DAC itself.

[Confused]

12 minutes ago, GoldenOne said:

I did actually touch on this briefly in my DDC video.

Using a network streamer can be a big benefit especially if you were previously using a beefy (and very electrically noisy) PC.
And dacs can and do indeed show different measurements in both things such as outright THD+N and Jitter when connected via a more ideal source such as a streamer vs a PC.
How much this change is will depend on the specific DAC, with some dacs showing very obvious changes, something like an apple dongle being a rather extreme example. And some dacs which have proper internal galvanic isolation like a Yggdrasil or May should have no change at all.

In terms of USB vs an I2S/AES/SPDIF streamer though, the latter will affect performance more due to the fact that the source device/streamer is the master clock, not the DAC itself.

Yes, I enjoyed watching your DCC video, excellent stuff.  (Although I was left wondering how a Mutec MC3+USB might have faired, but you can't test everything I guess)

 

That said, what I am proposing is slightly different.  Not a PC direct versus NAA test or similar, but a fixed NAA / DAC system, with the only variable being the PC.  The question being, does the Ethernet / NAA route reduce susceptibility to noise from a big noisy PC, or does it eliminate it completely?  Maybe you have already established this?

[GoldenOne]

12 minutes ago, Confused said:

Yes, I enjoyed watching your DCC video, excellent stuff.  (Although I was left wondering how a Mutec MC3+USB might have faired, but you can't test everything I guess)

 

That said, what I am proposing is slightly different.  Not a PC direct versus NAA test or similar, but a fixed NAA / DAC system, with the only variable being the PC.  The question being, does the Ethernet / NAA route reduce susceptibility to noise from a big noisy PC, or does it eliminate it completely?  Maybe you have already established this?

Ethernet is by spec transformer coupled and so noise will not be an issue. (It would be a nightmare trying to overcome the levels of noise in something like a datacentre if it wasn't I imagine)

The only thing that would realistically ever change would be jitter, but due to how ethernet and TCP/UDP works this simply isn't an issue.
SPDIF/AES/I2S sources can and do make a difference, as do USB sources to a lesser degree (key component with USB is the USB implementation in your DAC), but with ethernet the simple answer is it just doesn't work that way and any audiophile Ethernet tweak I personally cannot see how it could ever make a difference.

I've tried a few different networking setups and was never able to hear an audible difference nor measure any change whatsoever.
 

I'd be more than happy to test as extensively as I could both subjectively and objectively if someone were to lend me an 'audiophile' switch/router. I'm always open minded, but at the moment there is no reason and no evidence that networking tweaks would make an improvement.

[firedog]

On 5/28/2021 at 2:12 AM, GoldenOne said:

I couldn't agree more.
I disagree with John on MQA but the work he's done over the years is exceptional and he's provided immense insight both into the objective performance of various products as well as the reasoning for and explanations of his own and other testing methodology and the impacts of what is being shown.

His work is a hugely valuable resource.

I think JA's tests are great, and his comments/interpretations add a lot.  He is also willing to say sometimes that he doesn't understand the intersection of the subjective sound and the objective measurements.

 

The fact that Stereophile does a subjective review and then has JA test the unit is also a great setup. The reader often gets to see that the Stereophile reviewers have noticed something that then shows up in the measurements - or the opposite. That makes for a good read. 

[GoldenOne]

2 minutes ago, firedog said:

I think JA's tests are great, and his comments/interpretations add a lot.  He is also willing to say sometimes that he doesn't understand the intersection of the subjective sound and the objective measurements.

 

The fact that Stereophile does a subjective review and then has JA test the unit is also a great setup. The reader often gets to see that the Stereophile reviewers have noticed something that then shows up in the measurements - or the opposite. That makes for a good read. 

Absolutely
Subjective impressions and objective performance are both important, and often they either don't align, or one provides insight to the other.

It's useful to do things in the way that stereophile is doing, providing subjective feedback and then independent objective testing.
I'll aim to do the same. I'll include objective tests in my reviews where I personally feel it is appropriate, but I'll always do the objective testing AFTER forming my subjective opinions to avoid having the measurements create a bias.

The full measurements will then be posted separately and without much in the way of opinion/comment. To allow the two approaches to stand on their own.

 

[sphinxsix]

10 minutes ago, firedog said:

He is also willing to say sometimes that he doesn't understand the intersection of the subjective sound and the objective measurements.

In other words he admits sometimes there is no correlation between measurements and our perception, something which most objectivists would strongly disagree with. Well.. they know better..

 

[Superdad]

3 hours ago, GoldenOne said:

...but with ethernet the simple answer is it just doesn't work that way and any audiophile Ethernet tweak I personally cannot see how it could ever make a difference.

 

Well 2,500 EtherREGEN owners--plus thousands more using and discussing their audible experiences with other switches, clocks, filters, SFP transceivers, and cables--will disagree with you there... 

[firedog]

1 hour ago, sphinxsix said:

In other words he admits sometimes there is no correlation between measurements and our perception, something which most objectivists would strongly disagree with. Well.. they know better..

 

Or just something he can't explain. Not the same thing.

[fas42]

What I would be interested in, is to test the tester, 🙂. An example would be, hook the APx555 to some extremely low distortion setup, with close to undetectable distortion, and frequency artifacts ... see how consistent the readings are, over some hours, say. Then, subject the APx555 to severe operating conditions; say plug in some nasty mains noise generator next to the socket that the AP unit is plugged into; also, place some bad RF noise generator right next to the APx555 box ... obviously, the aim is to see how robust the AP unit is in its readings, when in a noisy electrical environment.

 

The tricky bit is to make sure that the setup hooked up to the APx555 is not that which is possibly reacting to the extra noise - some effort would be needed to properly isolate the former.