Standardized Low to Medium Power Testing

[miero]

Take the THD percentage, divide by 100, take the inverse (1/x), log it and multiply by 20 (as it is voltage we are talking about)...

How about using a lookup table? ;-)

Source: http://www.sengpielaudio.com/calculator-thd.htm

 

[restorer-john]

or if you are really lazy like me go here
http://www.sengpielaudio.com/calculator-thd.htm

Haha, I was just looking at that exact site but figured the engineers would have their calculators (HP of course) and a de-rigueur Pentel P205 on their desks...

 

[restorer-john]

How about using a lookup table? ;-)

A lookup what? Only 3 decimal places? What's wrong with you?

 

[miero]

@amirm I'm not sure if such graphs would be useful for comparing a sound of amplifiers. Maybe separate odd/even harmonics and/or also low/high order harmonics would help too.

 

[sergeauckland]

Firstly, I'd like to say that I'm pleased that a standard set of tests is being developed for amplifiers. I think it will be of great benefit to the reputation of ASR.

I don't know whether the AP test set can do it, but a screen shot of the distortion residual at 1W would be most useful to see. It will show whether the distortion is largely noise based, has any crossover artefacts, or is mostly second or third harmonic. It's easier to see this than an FFT if that's quite busy.

One further 'nice to have' is a screen shot of the waveform into clipping as this will show up instability or just clean clipping and recovery. A square wave into a capacitative load is also a nice to have.

My interest is mostly with vintage equipment, and these tests were done in the past, before FFTs became possible, and have been lost over the years.

Anyway, thanks for taking on-board more extensive amplifier testing.

 

[LF78]

How about displaying a distortion number in dB instead of %?

I agree, dB is more readable at a glance.

Slightly off topic, but still on the topic of measurements: it would be very interesting for DACs to perform a test on jitter rejection. Something like: connect a well known jittery digital source (Chromecast Audio?) and check if the DAC is able to reject jitter or not.

 

[restorer-john]

I agree, dB is more readable at a glance.

'At a glance' is not good enough for me. dBs tend to get rounded up or down to whole numbers. Percentages, especially when into the three or four digits below 0 are generally presented to their confidence level. Much more useful IMO.

 

[FrantzM]

Hi

Qualify the following any way you wish... you can even call it flattery ..

Epitome of dedication. It is very rare to have a person with that level of knowledge and dedication at the helm of a hobby website. It is even more rare to see the individual actively pushing the envelope. Amir is doing it. I believe, Siegfried Linkwitz in a quite different fashion tried that too. There may be others I haven't visited their websites yet. It remains however rare
I am slowly trying to ameliorate the performance of my HT. After several years (50) in the hobby and having owned serious HT and 2-ch gears (by serious I mean expensive ) along the years, I thought I had an handle on things... I am discovering that I don't, that I knew close to nothing... It is complicated. Tuning audio or video systems to the maximize their performance is not easy. Now , for a few seconds please try to compare this with testing a product at that kind of level .. In the absence of clear standards and trying to relate it to audibility ... That is Quixotic, yet needed. Sorely needed. We may not fully appreciate while it is being done since aside from a few contributions, it cost us nothing... There is no doubt the site is striking fear in some hearts and minds.. Some with the knowledge and heart ( I am talking to you too DonH56!!!! and you Blumlein and You too Ray and Ron And Don Hills, Solderdude, Restorer, even you Cosmix and do not think for a second that I have forgotten you, JJ !! although you don't post enough !!!!!! ) are showing that the emperor has no cloth and that there are better ways to reach our nirvana however subjective it might be at the end ...

I could write more about this but I'll stop there.

Congrats Amir. Congrats people! Keep it going. To the people that I have not mentioned but who have contributed immensely to the ASR cause, Our Cause, i apologize...
This is great! Long Live ASR and the King!

P.S. Presenting results in an ergonomic fashion is a complex area .. To have an idea read this http://www.openculture.com/2018/04/the-genius-of-harry-becks-1933-london-tube-map.html

 

[flipflop]

The channel imbalance vs volume position charts show that more than one unit can be displayed on the y-axis. Just leave % to the left and add dB to the right. Problem solved, everyone's happy.

 

[Blumlein 88]

The 8 and 4 ohm testing is incredibly useful for determining the quality of the power supply and yes, distortion varies considerably at high powers into 4 and 8 ohms.

I agree. But if I've not misunderstood Amir plans to test at 4 ohms and to 50 watts. Testing a 150 wpc into 8 ohms amp at 50 watts in both 4 and 8 ohms won't tell us much about the power supply current capability.

 

[LF78]

dBs tend to get rounded up or down to whole numbers. Percentages, especially when into the three or four digits below 0 are generally presented to their confidence level.

0.000215% = -113 dB
0.000134% = -117 dB

I believe dB scale is much more readable, and being logarithmic it also better reflects the perceptual effect of the measurement. But maybe it's just me

 

[RayDunzl]

I am graphic five power levels in between 1 and 50 watts.

Exponential power increase/levels?

6dB jumps might be about right.

 

[levimax]

Here is the same graph as post 1 but with 90 kHz bandwidth:

View attachment 20389

It is much less cluttered and easier to understand even though I am showing all 10 graphs now. So maybe you are talking me into running it this way.

Here is the same graph as post 1 but with 90 kHz bandwidth:

View attachment 20389

It is much less cluttered and easier to understand even though I am showing all 10 graphs now. So maybe you are talking me into running it this way.

This looks good to me.... I understand about audibility of 20 KHz and harmonics but it is nice to see what it looks like "unfiltered" especially as a basis for comparison.

 

[amirm]

The channel imbalance vs volume position charts show that more than one unit can be displayed on the y-axis. Just leave % to the left and add dB to the right. Problem solved, everyone's happy.

I have done that in some of my measurements:

Problem is that the AP graphing software then creates dual instances for each measurements. For a simple stereo you now have 4 graphs. As soon as you make a few measurements, this gets out of control. Worse yet is that the graphing system is buggy that way when it comes to cursors. I struggle to get it to attach the cursor to the right graph. Half the time I have to restart the app just to get it to put the cursor on the right one to measure power vs distortion.

What they need is a simple option to show more than one value on an axis or put it on both sides but not treat it as a new measurement.

 

[amirm]

You aren't requiring much current. If you wish to use 4 ohms, I'd like to see the test to 100 watts. 20 volts.

The issue here is that these tests take fair amount of time to run. Each line is probably a minute's worth. Running a small amp at 100 watts for that long can be dangerous. I know, I have already blown one that way (OK, it died after two minutes but you get the picture).

As I have noted, with loaner units I like to keep the stress level low so that there is no risk of damage. That is why I labeled this test as low to medium power measurement.

I will continue to have the THD+N versus power at full amplitude including clipping.

 

[amirm]

@amirm I'm not sure if such graphs would be useful for comparing a sound of amplifiers. Maybe separate odd/even harmonics and/or also low/high order harmonics would help too.

Graphs make comparisons easier. But yes, they lack the detail. I am developing this test because it was requested in review threads. I am open to testing otherwise.

 

[amirm]

Exponential power increase/levels?

6dB jumps might be about right.

That was the default I was using (logarithmic). Can also do linear.

 

[flipflop]

I have done that in some of my measurements:

Problem is that the AP graphing software then creates dual instances for each measurements. For a simple stereo you now have 4 graphs. As soon as you make a few measurements, this gets out of control. Worse yet is that the graphing system is buggy that way when it comes to cursors. I struggle to get it to attach the cursor to the right graph. Half the time I have to restart the app just to get it to put the cursor on the right one to measure power vs distortion.

What they need is a simple option to show more than one value on an axis or put it on both sides but not treat it as a new measurement.

Have you contacted Audio Precision about these issues? Quite poor performance for a $28k product.

 

[amirm]

Have you contacted Audio Precision about these issues? Quite poor performance for a $28k product.

Yes. They say they have licensed the graphing section from a third-party and they can't do anything about it. Their suggestion is to export the data and graph it using Excel and such!

Very few people if any are doing what I do with their instrument. Most people are running it in automated way producing graphs and pass/fail for manufacturing.

Still I give them feedback hoping they will address them some day. Alas, with $1000/year for software updates now, even if they update it I am not sure I want to keep paying that way to get these incremental features.

 

[sergeauckland]

Yes. They say they have licensed the graphing section from a third-party and they can't do anything about it. Their suggestion is to export the data and graph it using Excel and such!

Very few people if any are doing what I do with their instrument. Most people are running it in automated way producing graphs and pass/fail for manufacturing.

Still I give them feedback hoping they will address them some day. Alas, with $1000/year for software updates now, even if they update it I am not sure I want to keep paying that way to get these incremental features.

I think that's the main use for sophisticated test equipment these days. Few people/manufacturers operate test facilities these days in development, even fewer for technical reviews for magazines or web sites like ASR.

If you want real in-depth detail, it pretty much has to be done manually, which of course is hugely time consuming.

S.