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

[Pdxwayne]

Claiming significance at the <0.1% level with a cheap multimeter, no temperature control, and measurement far outside the design range is egregiously wrong.

Did I tell you about the time I measured a bacterium with a yardstick? It was very consistent. And the hell with you stuffy scientists with your insistence on things like gauge R&R and error bars.

Regarding temperature control,

Error bars and R&R. Stop avoiding the basics.

What error bar and R&R are you talking about? Examples please?

My home is temperature controlled. I have learned that before measurements, the DAC need at ~30 minutes warm up to stabilized (else multi meter won't get a stable reading).

I have measured my E30 at 3 different locations at 3 different times (Dec, Jan, Feb). All 3 times after the initial warm up, the measurements not varied by 0.001V.

 

[Koeitje]

Just follow an online introductory course to statistics and you are good to go.

Not meant as flame or anything, its just that humans have very bad intuitions when it comes to what is a significant change and what is not.

 

[solderdude]

E30 was also remeasured at the same time and same location when I first measure x16. It was to double check my previous results for E30. Same as today. Both times, DAC were/are in same location. Ambient temperature should be consistent both days.

Did you change the filters for both measurements ?

 

[Pdxwayne]

Just follow an online introductory course to statistics and you are good to go.

Unfortunately I can't turn back time on my X16.

Like I mentioned, I have measured KTB and E30 multiple times at multiple dates (totally over 100 times). No errors bigger than 0.001v. That was why I said my multi meter have consistency.

Changes of 0.004V of X16 was never seen before on E30 nor KTB..

 

[SIY]

You have no idea what your errors are.

 

[Pdxwayne]

Did you change the filters for both measurements ?

For X16, my initial test was with filter 1. My measurements yesterday were also filter 1 for apple to apple comparison.

 

[Pdxwayne]

You have no idea what your errors are.

Even without careful matching of temperatures, measured over 100 times and I got consistentcy of 0.001V.

 

[pma]

Measurement errors. DACs transfer function makes no difference even after years.

 

[cjm2077]

Unfortunately I can turn back time on my X16.

Like I mentioned, I have measured KTB and E30 multiple times at multiple dates (totally over 100 times). No errors bigger than 0.001v. That was why I said my multi meter have consistency.

Changes of 0.004V of X16 was never seen before on E30 nor KTB..

From the manual for the DMM: Temperature Coefficient: 0.1*(specified accuracy) / °C

You're also well beyond the rated AC voltage measurement range of 50-60Hz.

It's just not the right tool for the job. You're right that over a short period of time it may be precise but not accurate. But over longer periods of time various drifts means that it will be neither.

 

[Pdxwayne]

Measurement errors. DACs transfer function makes no difference even after years.

For my initial measurements, I specifically measured x16 2khz multiple times because it was odd as compared to E30 and KTB. I also measured E30 again to check 2Khz number. So measurements error out of questions.

 

[Pdxwayne]

From the manual for the DMM: Temperature Coefficient: 0.1*(specified accuracy) / °C

You're also well beyond the rated AC voltage measurement range of 50-60Hz.

It's just not the right tool for the job. You're right that over a short period of time it may be precise but not accurate. But over longer periods of time various drifts means that it will be neither.

Like I said, my home is temperature controlled. I have measured my E30 in December, January, and February. Consistent results within 0.001V.

 

[SIY]

Even without careful matching of temperatures, measured over 100 times and I got consistentcy of 0.001V.

Uh huh.

 

[Pdxwayne]

Uh huh.

Want a video?

 

[solderdude]

Here's the thing:
You are talking about 0.02dB difference at 4kHz and 0.1dB difference at 7kHz.

The question is how reliable your measurements are anyway.
For example, the E30 measures 0.486V @ 1kHz and 0.288V @ 10kHz.
This means that acc. to your measurements 10kHz is -4.5dB which it isn't at all, in fact it is way less than -0.1dB off.

How can you be sure your measurements are correct when every MM is always +/- 1 last digit per definition ?

Maybe, just maybe 0.1dB is barely audible in blind tests (I have never been able to detect any audible differences below 0.2dB amplitude differences with tones) but certainly will not be an audible subjective difference you described hearing.

 

[Koeitje]

You need to setup a more controlled test environment, use better testing equipment and use statistics to determine if the difference is significant.

 

[Pdxwayne]

Here's the thing:
You are talking about 0.02dB difference at 4kHz and 0.1dB difference at 7kHz.

The question is how reliable your measurements are anyway.
For example, the E30 measures 0.486V @ 1kHz and 0.288V @ 10kHz.
This means that acc. to your measurements 10kHz is -4.5dB which it isn't at all, in fact it is way less than -0.1dB off.

How can you be sure your measurements are off when every MM is always +/- 1 last digit per definition.

Yeah, as mentioned multiple times, meter not accurate enough to show real voltages.

But, it is consistent. It can show trend of changes when you have multiple filters measurements to compare. For example, my meter can show a deviation for e30 filter 5.

See https://www.audiosciencereview.com/...topping-e30-listening-tests.17988/post-665605.

It is also consistent when I measured E30 in Dec, Jan, Feb. Never deviate more than 0.001V for various freq measured.
Same for KTB. Measured multiple times, no deviation larger than 0.001V.

X16 on the other hand, shows up to 0.004V deviations from 2Khz on.

 

[Pdxwayne]

You need to setup a more controlled test environment, use better testing equipment and use statistics to determine if the difference is significant.

Unfortunately that is all I have. There is only one change to measure a brand new X16. I can't turn back time for X16.

 

[pma]

Here's the thing:
You are talking about 0.02dB difference at 4kHz and 0.1dB difference at 7kHz.

You know, it is a hard work to measure psycho-bias. This place turns weird.

 

[Tatteredmidnight]

Unfortunately that is all I have. There is only one change to measure a brand new X16. I can't turn back time for X16.

Unfortunately it may just be insufficient. In measurement labs all equipment is calibrated and the behavior is properly classified within the test parameters. Without knowing these factors, you cannot confidently say you are properly controlling for outside variables and you can’t say your results are outside your standard deviation. Too many unknowns for any level of certainty here.

 

[Pdxwayne]

Unfortunately it may just be insufficient. In measurement labs all equipment is calibrated and the behavior is properly classified within the test parameters. Without knowing these factors, you cannot confidently say you are properly controlling for outside variables and you can’t say your results are outside your standard deviation. Too many unknowns for any level of certainty here.

Of cause I don't have the correct setup for optimum correctness. I already stated standard deviation is 0.001V. Anyone here with proper equipments want to buy x16 and do same experiments? That would be awesome.