DAC burn-in/break-in? My Gustard X16 measurements

[Pdxwayne]

Because ruling it out in such persistence is actually a good indicator of a bias.

If I didn't do measurements between my KTB and e30 last year, I would have accepted that persistent ASR claim.

Sometimes something really simple can explain the difference heard is not imagination.

See https://www.audiosciencereview.com/...topping-e30-listening-tests.17988/post-665826

Could a certain components on a board require more than 24hrs to reach full potential? Why can't I get an answer here?

 

[pma]

The problem is that the difference you showed before/after some time is not in correspondence with known minimum audible differences so your explanation is not and will not be accepted as a proof that there is an audible difference after burn-in.

 

[Pdxwayne]

The problem is that the difference you showed before/after some time is not in correspondence with known minimum audible differences so your explanation is and will not be accepted as a proof that there is an audible difference after burn-in.

I agree with most people here about such small voltage differences likely not audible.

But the question here is about larger than expected voltage changes for x16. Up to 0.004v from 2khz and up vs always around 0.001v for E30 and KTB.

What could cause it? What components on the board are causing it? Could it cause other issues that made me dislike x16 initially?

Can you give me any tips based on your experience? On a modern board like X16, are there any components that may need more than 24 hours to reach full potential?

 

[Frio]

I had e30 and the x16 didn't hear a difference. They are both transparent dacs. They are by design not intended to alter the tone.

But when is see now how every mV is interpreted to match the impression I have my doubts if anything could convince me of those 'sparkles' or 'punches' or whatever.

 

[Pdxwayne]

I had e30 and the x16 didn't hear a difference. They are both transparent dacs. They are by design not intended to alter the tone.

But when is see now how every mV is interpreted to match the impression I have my doubts if anything could convince me of those 'sparkles' or 'punches' or whatever.

Yeah, maybe there are differences in board components quality. So far I have heard one person returned x16 because it sounded like blanket over speakers. The other return it because other than BT, he gets no sound.

May be my board contains components that not quite functioning optimally without a few warm up and cold down? Any clue would be appreciated. But so far what I got is "attack of the multimeter".
; )

 

[Frio]

Those cases are actually quite normal name one mass product that doesn't have a single return. Now to assume that it has to do with some boards or a suboptimal chip without going deeper and measurements of those units just from a forum impression is just another proof of a bias that needs to be fulfilled.

 

[solderdude]

There are no parts in there that need to have had 3 weeks of being 'switched on' to reach 'full potential'.
If that were the case building measurement equipment would be a real pain.
There are no 'special' parts in there that 'require' several weeks of 'forming' that are not used in other DACs as far as I can see.

What is possible is that there is a broken component in there. For instance a ceramic capacitor in the post filter.
The problem however is that such a part would need to fail in a way that is extremely unlike to fail in that manner.
That capacitor would have to have had a higher capacitance (failed) and later on started to work properly.
IF that were possible at all the FR would not change in this particular way.

Firmware and software in chips are not known to fail in the way you describe and do not need 'burn-in'
The majority of signal in a DAC are digital.

Another possible thing could be something was up in the power supply and a ripple was present at a higher frequency.
However, capacitors do not need a few weeks to 'form' and do not suddenly start to function 'better' after weeks.

Not many people will use a multimeter and submit that as proof. You have been told this before. Certainly not when measuring such small voltages and then not even considering the last digit is always +/-1.

When you measure FR you need a good ADC (and even these are tricky) not a measurement device that is almost 5dB 'off' by itself and only measure up to 10kHz. In 1kHz increments no less.

It looks like you measured the X16 twice only. One time shortly after you got it and heard (needs to be substantial to hear) it sounded off and the second time 3 weeks later.
It looks like you are 'looking' for something to blame because you heard something. Differences you thought you heard as you have been going around with audible differences between KTB and E30 before.

Several people with electronics knowledge have indulged you and gave pointers on how to measure and whenever you discovered something have replied with patience.

I know you want to validate your hearing with tools you are not very familiar with and are either not up to the task or used incorrectly.
If what you said is really an issue people that have measured DACs/ADC's etc. surely would have found this.
In fact Amir did show 'warming up' effects but these do not suddenly disappear after a few weeks. They happen every time you switch them on while not being on operational temperatures.

So the problem is only 2 measurements and possible a perception issue and maybe a defective device that somehow 'fixed' itself within a few weeks. And only by 0.1dB max and different (and not logically) at various frequencies at that in ways not known to folks having experience with electronics for decades.

So far I have heard one person returned x16 because it sounded like blanket over speakers

For this to be audible (he didn't measure it so can be anecdotal only) you would need several dB's roll-off in the highs, Not 0.01 to 0.09dB differing in the 2kHz to 10kHz range only.


I would suggest to return the X16 and have it replaced. Then on reception of a new one put it on a proper FR measurement from the start.
Then listen to it to hear if this one is off too.
Then reconnect it to the measurement chain (= NOT a multimeter) and take plots at increasing time intervals over the course of weeks.

Then you can research and prove 'burn-in' of the X16. Not the way you handled it so far.

 

[pma]

This thread again and again confirms that any debates with those with naive amateur reasoning are pointless and nothing but wasting time. This is a big problem of open platforms, anyone without any kind of knowledge and education in the field feels he could argue with knowledgeable and educated people. There is no way to explain and to teach (because of so different background of the parties) so all the effort is useless and contra productive. The only result are higher barriers.

 

[chris719]

This thread again and again confirms that any debates with those with naive amateur reasoning are pointless and nothing but wasting time. This is a big problem of open platforms, anyone without any kind of knowledge and education in the field feels he could argue with knowledgeable and educated people. There is no way to explain and to teach (because of so different background of the parties) so all the effort is useless and contra productive. The only result are higher barriers.

Exactly. If he had any knowledge at all he probably wouldn't be performing this test. Even if he did, he wouldn't be looking at burn-in as a root cause because there's nothing to age in two weeks that would affect gain drift.

 

[solderdude]

The differences OP 'measured', by lack of measurements below 1kHz and above 10kHz, are not even consistent (in dB) over the measured range.

Trying to 'prove' burn-in in the X16 would not be admissible in court based on 2 basic 'measurements'. I can understand OP thinking it 'confirms' his subjective findings or at least serves as 1 of the indicators.

One could see this attempt as learning to drive a car and after a few days of non guided attempts think you can drive cars in a rally successfully.
No electronics engineer would ever feel the measurements done were anywhere near the quality and rigor to base any conclusions on.

 

[JSmith]

So measurements error out of questions.

Oh really, margin of error out of the question is it? Your equipment is 100% perfect, infallible and the best possible type of tool?



JSmith

 

[voodooless]

One could see this attempt as learning to drive a car and after a few days of non guided attempts think you can drive cars in a rally successfully.
No electronics engineer would ever feel the measurements done were anywhere near the quality and rigor to base any conclusions on.

Yeah, said as much here already: https://www.audiosciencereview.com/...echo-trails-length-illusion.20618/post-683705 .. and now we're yet again at 7 more pages

 

[Jimbob54]

I think the observations about optical vs coax are interesting too. The OP seems to find audible differences with every new set up/ component he tries, in scenarios where received wisdom is there should be none or minimal at best.

 

[mansr]

Was the DAC thoroughly warmed up for all "measurements"? Here's a graph of the output voltage from a random DAC for the first two hours after power on.

As we can see, it takes a full hour for the voltage to stabilise, at which time it has increased by about 4.5 mV from the initial value. And that's not all. Continuing to measure for 12 hours gives the graph below.

Even after the initial warm-up, the output voltage fluctuates by nearly 1 mV.

The DAC in question is DC coupled, so for better accuracy, I simply fed it a constant value. The meter is a Keithley DMM6500, warmed up for more than the specified 30 minutes. The calibration is expired, but it should still be good enough for this measurement.

 

[Pdxwayne]

Was the DAC thoroughly warmed up for all "measurements"? Here's a graph of the output voltage from a random DAC for the first two hours after power on.

View attachment 114804

As we can see, it takes a full hour for the voltage to stabilise, at which time it has increased by about 4.5 mV from the initial value. And that's not all. Continuing to measure for 12 hours gives the graph below.

View attachment 114803

Even after the initial warm-up, the output voltage fluctuates by nearly 1 mV.

The DAC in question is DC coupled, so for better accuracy, I simply fed it a constant value. The meter is a Keithley DMM6500, warmed up for more than the specified 30 minutes. The calibration is expired, but it should still be good enough for this measurement.

First instant it was powered on for 8+ hours continuously.

Second was on for over 24 hours.

I do know about fluctuations on initial power on. I even started a thread about it.

: )

 

[Pdxwayne]

Sigh....

I am trying to present a possible clue....

In various home use scenarios, once E30 warm up to over 30 minutes, once I voltage matched at 1KHz, all subsequent measurements at 2khz up to 10 kHz are stable.

It means, if I measure voltage output for 2khz, I get same exact voltage that I got when I measured 1 month ago and the same exact voltage I got 2 months ago. Majority of the kHz measured, I got EXACT matching voltage, with a few deviate by 0.001V. This is with E30 in 3 different rooms, in multiple different audio setups. Always EXACTLY matching voltages for majority of kHz measured, with just a few deviate by 0.001V.

Same for KTB I measured multiple times in December. Always EXACTLY matching voltages for majority of kHz measured.

Now with X16, ALL measurements from 2Khz up changed. Some up to 0.004V.

This doesn't make most of you go "huh?". No curiosity in anyway?

 

[BDWoody]

This thread again and again confirms that any debates with those with naive amateur reasoning are pointless and nothing but wasting time. This is a big problem of open platforms, anyone without any kind of knowledge and education in the field feels he could argue with knowledgeable and educated people. There is no way to explain and to teach (because of so different background of the parties) so all the effort is useless and contra productive. The only result are higher barriers.

I would argue a little bit, to the extent that I have seen several members allow themselves to be guided through the process to get results that there is no way that they could have gotten to on their own.

Unfortunately, in this case, we have someone who falls into the second part of your post, where he seems only interested in supporting what he wants to support, not learn how to take proper measurements. It isn't as easy as you all (who know what you're doing) make it look.

@Pdxwayne , I suggest you start trying to learn from the handful of quite qualified experts who are still, remarkably, replying with a mountain of patience. For you to reply to them with sighs and general dismissal really isn't going to help you learn anything.

 

[Pdxwayne]

I would argue a little bit, to the extent that I have seen several members allow themselves to be guided through the process to get results that there is no way that could have gotten to on their own.

Unfortunately, in this case, we have someone who falls into the second part of your post, where he is only interested in supporting what he wants to support, not learn how to take proper measurements. It isn't as easy as you all (who know what you're doing) make it look.

@Pdxwayne , I suggest you start trying to learn from a the handful of quite qualified experts who are still, remarkably, replying with a mountain of patience. For you to reply to them with sighs and general dismissal really isn't going to help you learn anything.

I really appreciate all the feedbacks here. Thanks all!

I also know I don't have the appropriate equipment. I use what I have to the best of its limitations.

I especially appreciate @solderdude patience through out the many threads I started. Excellent patience! Thanks!

 

[solderdude]

First instant it was powered on for 8+ hours continuously.

Second was on for over 24 hours.

I do know about fluctuations on initial power on. I even started a thread about it.

: )

Consider that mansr measured DC output (or overall amplitude) variations of 0.01dB. He did not check if the frequency response changed.

Now with X16, ALL measurements from 2Khz up changed. Some up to 0.004V.

This doesn't make most of you go "huh?". No curiosity in anyway?

Of course it does but the variations are not logical (not consistent in frequency either which would be more logical) and there is no 'mechanism' for it.
I have never met a component that alters its properties on specific frequencies in a different way (other than a notch filter) but magically stops doing that after a certain time (permanently settled in a positive way). And I have been active in electronics, analog audio and measuring for over 40 years now.

A multimeter is the wrong tool, certainly when measuring mV differences on a 300mV signal.

The ONLY way this can be proven is by ordering a new one (and if it is indeed a defect the new one may not show it) and measure FR correctly over a period of weeks. Not with a multimeter but with a good sound card or measurement device.

1kHz = ref
2kHz = + 0.06dB
3kHz = + 0.04dB
4kHz = + 0.02dB
5kHz = + 0.06dB
6kHz = + 0.09dB
7kHz = + 0.1dB
8kHz = + 0.08dB
9kHz = + 0.08dB
10kHz = + 0.09dB

This looks more like measurement errors than 1 or more components needing a special weeks taking burn-in process.

 

[Pdxwayne]

Consider that mansr measured DC output (or overall amplitude) variations of 0.01dB. He did not check if the frequency response changed.



Of course it does but the variations are not logical (not consistent in frequency either which would be more logical) and there is no 'mechanism' for it.
I have never met a component that alters its properties on specific frequencies in a different way (other than a notch filter) but magically stops doing that after a certain time (permanently settled in a positive way). And I have been active in electronics, analog audio and measuring for over 40 years now.

A multimeter is the wrong tool, certainly when measuring mV differences on a 300mV signal.

The ONLY way this can be proven is by ordering a new one (and if it is indeed a defect the new one may not show it) and measure FR correctly over a period of weeks. Not with a multimeter but with a good soundcard or measurement device.

Again, appreciate your patience!

Yup, not logical voltage changes for x16, thus seems like a potential clue.

I also share your concern about defective unit vs getting a normal new unit as replacement...

I do know my tool is the main issue here.

Any good, reasonable priced, measurements equipment you would recommend?

For soundcard, I do have Focusrite Scarlett Solo (3rd Gen) as ADC. Good enough for loopback check?