Quality Power - Page 21 - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #601 of 637 Old 03-25-2014, 06:21 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 13,875
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 435 Post(s)
Liked: 1057
Quote:
Originally Posted by amirm View Post

Quote:
Originally Posted by arnyk View Post

Sure:




However the above in my opinion represents a fool's mission because it can't be heard and its spectral content was not asked for.

In this case crest factor was calculated from Average RMS power and Peak Amplitude. The display and analysis tool was Adobe Audition. Note that the actual crest factor is more than 10 dB, which is what I meant by "10 dB or more".

I expect major hypercritical nit-picking to follow! ;-)
I apologize for causing you angst Arny. I was genuine in my question. I have seen the argument you are making that music has a lower crest factor (difference between peak and average) than sine wave used for amplifier power testing but I have yet to find one that says how they computed that "10 db" number. A few say that they used the DR Meter which is inappropriate for this type of analysis.

Anyway, looking at your sample data, I see that the background is white. That tells me that you have selected the entire song.

That would be false. The timeline is at the bottom of the graphic. The time range selected is approximately 2.625 to 3.080 minutes into the song. It is the crescendo at the end of the song minus most of the natural build up and fade out.
Quote:
I hope this didn't come across as "major hypercritical nit-picking." smile.gif The analysis behind this argument is key to its validity.

How about "Just plain wrong"?

The numbers calculated by Audition 1.0 are simple arithmetic averages and peak values - essentially the properties of the waveform as perceived by a piece of electronic equipment.

It appears that the software you used calculated these numbers in accordance with ITU Recomendation BS 1170 which anybody can find online. These calculations indicate perceptual weighting - the properties of the waveform as perceived by human ears.

Since we are talking about how an amplifier perceives the waveform the use of weighting based on human perception is also just plain wrong.
arnyk is online now  
Sponsored Links
Advertisement
 
post #602 of 637 Old 03-25-2014, 06:30 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 13,875
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 435 Post(s)
Liked: 1057
Quote:
Originally Posted by Frank Derks View Post

Sinewave testing is kind of a worst case scenario. It can be done with short burst too if cooling is not engineered for continuous sine wave loads.
For eaxemples 7 amps sharing a PSY and 'intentional' limited heatsink capability in av amps is designed with a reasonable crest faktor chosen by the designer.

Point is that with a sinewave test and a 8Ohm load were dealing with two knowns as opposed to dragging an undefined speaker load and crest factor into the mix.
From this simple test we get a distortion level and the actual amount of power at a given distortion level.

Simple and repeatable.

Then there is the minor matter of irrelevancy. ;-)
arnyk is online now  
post #603 of 637 Old 03-25-2014, 06:33 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 13,875
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 435 Post(s)
Liked: 1057
Quote:
Originally Posted by Frank Derks View Post

A low bass note sustained for a couple of seconds at full power is what the power supply needs to be able to cope with.
Amplifiers don't discriminate so testing with a 1kHz or 10kHz sine wave at the same power levels makes no difference for an amplifier.

I had exactly that in mind when I selected the musical selection for my analysis.
arnyk is online now  
post #604 of 637 Old 03-25-2014, 06:38 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 13,875
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 435 Post(s)
Liked: 1057
Quote:
Originally Posted by mcnarus View Post

Quote:
Point is that with a sinewave test and a 8Ohm load were dealing with two knowns as opposed to dragging an undefined speaker load and crest factor into the mix.
From this simple test we get a distortion level and the actual amount of power at a given distortion level.

Simple and repeatable.
Exactly. A spec is just a spec, and a measurement is just a measurement. Neither is a universal truth. They're just indications of an amp's capabilities. And as an indicator, they work well enough.

I interviewed some of the people, for example the chief engineer of the electronics department at Consumer Reports, who were influencing specs at the time that the IHFM's procedures for measuring power were being decided. There were also articles that contained statements by them in the audio magazine of the day. I owuld generalize their statements as being along the lines that they were a lot more interested in having a repeatable test than a test that was optimally relevant. The test equipment that was available to them and that they used was fairly simplistic and not easy to use to measure crest factor like we do today.
arnyk is online now  
post #605 of 637 Old 03-25-2014, 10:32 PM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 17,797
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 524 Post(s)
Liked: 336
Quote:
Originally Posted by arnyk View Post

How about "Just plain wrong"?

The numbers calculated by Audition 1.0 are simple arithmetic averages and peak values - essentially the properties of the waveform as perceived by a piece of electronic equipment.
Audition 1.0? Why are you using such old software Arny? When did 1.0 come out? 2003? I am running Audition 6.0 ("CC") which is the latest version.

And how sure are you of those numbers? Did you run a set of controls through them to make sure they are computing what you think they are displaying? I have Audition 3.0 still which has similar stats as yours. Here is a control test that I ran through it:

i-dkZDw85-XL.png

Does it look right to you? The "peak value" is -15.45 db. How can that be? The top of the waveform is almost at 0 db. I happen to know the real value is -0.20 db. That is nowhere in the neighborhood of -15.45 db. The Average RMS is 0 db. Do you think that is right?

Still have confidence in the statistics you used in that ancient copy of Audition? "Simple arithmetic averages and peak values?" I hope not smile.gif. Clearly you are relying on incorrect data output from Audition.
Quote:
It appears that the software you used calculated these numbers in accordance with ITU Recomendation BS 1170 which anybody can find online. These calculations indicate perceptual weighting - the properties of the waveform as perceived by human ears. Since we are talking about how an amplifier perceives the waveform the use of weighting based on human perception is also just plain wrong.
You are mistaken Arny. I did not use ITU loudness rating at all. That is an additional piece of information the latest version of Audition shows at the bottom of that window. I made no reference to that field in my answer. I simply used the same fields you mentioned: Peak Amplitude and Average RMS. The only difference was that I used a different clip and most up to date version of Audition. That you didn't recognize the software I was using and further thought I was relying on ITU loudness metric is quite surprising.

So what do you think now? My data is valid and yours is suspect. Right?

Amir
Founder, Madrona Digital
"Insist on Quality Engineering"

amirm is offline  
post #606 of 637 Old 03-26-2014, 04:10 AM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 13,875
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 435 Post(s)
Liked: 1057
Quote:
Originally Posted by amirm View Post

Quote:
Originally Posted by arnyk View Post

How about "Just plain wrong"?

The numbers calculated by Audition 1.0 are simple arithmetic averages and peak values - essentially the properties of the waveform as perceived by a piece of electronic equipment.
Audition 1.0? Why are you using such old software Arny? When did 1.0 come out? 2003? I am running Audition 6.0 ("CC") which is the latest version.

Holy nit picking, Batman!

Didn't I predict this almost precisely?

Proverbs 26:11-12

"Like a dog that returns to its vomit Is a fool who repeats his folly. Do you see a man wise in his own eyes? There is more hope for a fool than for him."
arnyk is online now  
post #607 of 637 Old 03-27-2014, 09:32 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 17,797
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 524 Post(s)
Liked: 336
Quote:
Originally Posted by arnyk View Post

Holy nit picking, Batman!

Didn't I predict this almost precisely?

Proverbs 26:11-12

"Like a dog that returns to its vomit Is a fool who repeats his folly. Do you see a man wise in his own eyes? There is more hope for a fool than for him."
Darn tootin'! I can't believe I committed that sin again! biggrin.gif

Between us chickens, that proverb counts on the common man who thinks dogs are like people instead of animals. The reality is different of course. Dogs vomit back what they think is dangerous to keep digesting. By throwing it up and chewing it further into smaller pieces, they avoid that potentially deadly blockage. Among dog owners is a saying that dogs get to enjoy their food twice while we once. smile.gif We feed our dogs raw food with bones and I get to watch this scenario all the time. They swallow a big piece of bone, bring it back up and chew and eat it again.

Anyway, enough religious discussions. Back to the topic at hand, there are three things that usually don't mix: someone who writes great software, knows signal processing, and math. When I first joined Microsoft I was testing our audio encoder and kept hearing this artifact that didn't sound like compression artifacts. The source files were all 44.1 but at the bit rates I was testing the sample rate was lower so I knew there was sample rate conversion going on. So I used my audio software and downsampled using that before encoding and the artifact vanished! I went to the head of our codec group and asked him what he was using for resampling and he said it was the Windows service. I told him about the problem and they built a high fidelity resampler in the encoder and problem was fixed. Similarly, I discovered something that had cost me a lot of time. The fact that changing the volume control affected sound! Again, poor design causing quantization errors. Don't know who had written these but no doubt it was a good software engineering not understanding the math and signal processing. I hired JJ to shoot the Windows audio stack in the head and fix all of these problems and the result was the much improved one that has shipped from Windows Vista on.

By the same token, I have found the stats in Audition suspect. The clues are right there in the pop up stats. Here is the same analysis that I did with latest version of Audition but now, with the version of the stats you have:

i-WwzLJtv.png

Immediately we see lack of knowledge on behalf of the person who programmed that dialog. It says "Average RMS Power." We can have Average power but what the heck is Average RMS power? There is no such thing as RMS Power. From the bible of the Internet, the wiki biggrin.gif, we have this:

7c1c446258305f4c3c58251e5122d9ab.png

So right away we see that there is no such thing as RMS Power. Only Average Power.

The formula calls for knowing both voltage and current. The waveform gives us the voltage. But where on earth do we get the current? Computing that requires knowing the load resistance which we do not have.

There is another problem in that dialog that I leave as an exercise for the reader to find.

Now let's look at the stats in my updated version of Audition:

i-Pwxnd5V-XL.png

We see an immediate improvement as you see in the line I highlighted values. The word "power" has been expunged from the stats and replaced with Amplitude. As I explained that is all the data we have in the audio file so nothing other than that can be computed. We still have a problem because they are using the term "Average RMS." RMS is actually an average so the term Average is redundant. Let's review what RMS is. Here is a simple sine wave:

400px-Sine_wave_voltages.svg.png

If we just summed all the values we get zero since the positive and negative cycles cancel each other out. If we feed that sine wave to an amplifier, it would not generate zero power. Rather, it will generate power in both cycles, and the direction of current will simply be different from the positive to negative side. The concept of RMS came about to solve this problem. It creates an average that is insensitive to polarity by squaring the values, summing them and then dividing by the number of samples. We then take the square root and we find our RMS value:

014a453df92ed77eca97b228f0624d9f.png

So why still call this an Average? The key is that innocent looking "100 msec" field in the first graph above. The default for this is 50 msec as you can see in Arny's post. When you ask Audition to give you the statistics, it computes a piece-wise RMS value that is limited to that interval. If you give it a much longer interval than 50 msec, it would compute RMS values for each segment and then simply average all of those. In other words, it is the average number of a N number of RMS computations.

The problem with this is that if you select a large segment of your audio file or the entire track, the Average RMS value gets pulled down by areas of the track that are quiet or silent. You can see such a fade out in Arny's graph. If you are after some general stats about the clip, this is fine. And that is what Pros use it for. Trying to use it to compute amplifier power/Crest factor however, causes a problem. The amplifier does not get to average out an entire clip and amplify that. It has to amplify what it is told to amplify at any one instance. Its capacitor power reservoir can only handle milliseconds worth of power drain, not seconds and minutes. So what we want to see is short segments of the track analyzed as I showed in my examples, not long segments. Using long segments will artificially lower the average and hence "crest factor."

Anyway I better stop here before confusing everyone more smile.gif. The net is that the statistical analysis in Audion cannot be used superficially to make an argument about crest factor and amplifier power. While useful data can be extracted from it, it is not designed for this use and can easily produce garbage as I showed in my question to Arny. I have yet to see anyone making this amplifier crest factor argument correctly walk through proper analysis of the waveform and this discussion is another version of it. The famous expression, "Lies, damned lies, and statistics"[/i} most definitely applies here.

Amir
Founder, Madrona Digital
"Insist on Quality Engineering"

amirm is offline  
post #608 of 637 Old 03-27-2014, 11:26 AM
AVS Special Member
 
DonH50's Avatar
 
Join Date: Feb 2010
Location: Monument CO
Posts: 5,928
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 97 Post(s)
Liked: 234
Typo, Amir, in first paragraph under the picture: "Immediately we see lack of knowledge on behalf of the person who programmed that dialog. It says "Average RMS Power." We can have Average power but what the heck is Average RMS power? There is such thing as RMS Power. " I think you meant to say NO such thing as RMS power...

Actually, I went up and looked up Audition, glad I know not to buy it now, not cheap...

"After silence, that which best expresses the inexpressible, is music" - Aldous Huxley
DonH50 is offline  
post #609 of 637 Old 03-27-2014, 11:51 AM
AVS Special Member
 
Frank Derks's Avatar
 
Join Date: Jul 2002
Location: Region A,B,C
Posts: 1,889
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 96 Post(s)
Liked: 87
Amir the formula you give for the RMS Is for a continuous-time waveform.

For a function over time it is:


"The RMS value of a continuous function or signal can be approximated by taking the RMS of a series of equally spaced samples"

I think this is what audition is calculating over the selected wave form.
Frank Derks is online now  
post #610 of 637 Old 03-27-2014, 02:07 PM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 17,797
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 524 Post(s)
Liked: 336
Quote:
Originally Posted by Frank Derks View Post

Amir the formula you give for the RMS Is for a continuous-time waveform.

For a function over time it is:


"The RMS value of a continuous function or signal can be approximated by taking the RMS of a series of equally spaced samples"
I am sorry but I don't follow your argument. Anything with an integral is for continuous signals. The discrete domain version of it is a Sigma which is what I post.
Quote:
I think this is what audition is calculating over the selected wave form.
It can't compute the integral because that assumes a continuously varying value. What you quote in the italics is the way it is doing it which as I said, is a sum of squares function.

Amir
Founder, Madrona Digital
"Insist on Quality Engineering"

amirm is offline  
post #611 of 637 Old 03-27-2014, 02:10 PM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 17,797
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 524 Post(s)
Liked: 336
Quote:
Originally Posted by DonH50 View Post

Typo, Amir, in first paragraph under the picture: "Immediately we see lack of knowledge on behalf of the person who programmed that dialog. It says "Average RMS Power." We can have Average power but what the heck is Average RMS power? There is such thing as RMS Power. " I think you meant to say NO such thing as RMS power...
Thanks Don. The dogs were literally barking was I was trying to finish the post and didn't get a chance to proof read it smile.gif. Made the correction in the post.
Quote:
Actually, I went up and looked up Audition, glad I know not to buy it now, not cheap...
It is not. I am stuck also needing Photoshop, Premier, and Illustrator which pushes the bill into crazy numbers. I used to upgrade every other version but even that was too expensive. So I finally bit the bullet and subscribed to the whole suite. I think it is like $50/month. Feels good to be up to date finally but it is costly to be sure.

Amir
Founder, Madrona Digital
"Insist on Quality Engineering"

amirm is offline  
post #612 of 637 Old 03-27-2014, 08:15 PM
AVS Special Member
 
DonH50's Avatar
 
Join Date: Feb 2010
Location: Monument CO
Posts: 5,928
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 97 Post(s)
Liked: 234
I bought Photoshop ages ago, upgraded a few times, but never really used it. Just too much for an amateur hack and I did not have time, though I have several Kelby books. Kept buying them when I upgraded thinking "this time, I will read the book!" -- never happened. tongue.gif The main reason I bought it was for raw import capability, but now that is in the "lite" version that is much more user-friendly (or, less user-hostile wink.gif ) so I quit upgrading around CS3 or 4.

My dog barks now and then, as do the neighbor dogs, but dogs barking now makes me think of my neighbor's friend who was awakened last summer by her dogs barking to see the deck going up in flames. She got out with the dogs in one car as the house and the other car burned. I still get shivers and hearing a dog bark makes me look sharply around. She raced out and met her husband outside the evac line -- he was frantic but they wouldn't let him pass, with good reason!

Sorry for the off-topic posts, back to debating RMS or whatever is up next. - Don

"After silence, that which best expresses the inexpressible, is music" - Aldous Huxley
DonH50 is offline  
post #613 of 637 Old 03-28-2014, 04:48 AM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 13,875
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 435 Post(s)
Liked: 1057
Quote:
Originally Posted by amirm View Post


By the same token, I have found the stats in Audition suspect.

If I could teach some people basic science, their lives and by extension my life as their lives impinge on it, could be simplified.

If you suspect a measurement tool don't psychoanalyze it or speculate. Just give it some varied problems to solve that you already know reliable answers for by other means.
arnyk is online now  
post #614 of 637 Old 03-28-2014, 05:08 AM
FMW
AVS Special Member
 
FMW's Avatar
 
Join Date: Jun 2007
Posts: 4,558
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 134 Post(s)
Liked: 641
Quote:
Originally Posted by DonH50 View Post

I bought Photoshop ages ago, upgraded a few times, but never really used it. Just too much for an amateur hack and I did not have time, though I have several Kelby books. Kept buying them when I upgraded thinking "this time, I will read the book!" -- never happened. tongue.gif The main reason I bought it was for raw import capability, but now that is in the "lite" version that is much more user-friendly (or, less user-hostile wink.gif ) so I quit upgrading around CS3 or 4.

My dog barks now and then, as do the neighbor dogs, but dogs barking now makes me think of my neighbor's friend who was awakened last summer by her dogs barking to see the deck going up in flames. She got out with the dogs in one car as the house and the other car burned. I still get shivers and hearing a dog bark makes me look sharply around. She raced out and met her husband outside the evac line -- he was frantic but they wouldn't let him pass, with good reason!

Sorry for the off-topic posts, back to debating RMS or whatever is up next. - Don

Don, I stopped at CS2. It is old enough now that, not only does Adobe no longer support it but it is now free. I can now put on as many computers as I like. They gave me a universal activation code.
FMW is online now  
post #615 of 637 Old 03-28-2014, 08:20 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 17,797
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 524 Post(s)
Liked: 336
Quote:
Originally Posted by arnyk View Post

If I could teach some people basic science, their lives and by extension my life as their lives impinge on it, could be simplified.
Your fountain of wisdom, experience and knowledge is a gift to us all Arny. I don't know what we would do without you.
Quote:
If you suspect a measurement tool don't psychoanalyze it or speculate. Just give it some varied problems to solve that you already know reliable answers for by other means.
Ah. Thank you! I have been feeding self-help files to Audition 3.0 hoping it would fix its bugs but had not worked. Now I know why!!!

This is the type of problem I am having. Here is a simple 7 KHz sine wave where I have (roughly) selected the positive side of the waveform and asked the older Audition 3.0 to show its statistics:

i-SXxRhtd.png

The horizontal white line shows the 0 dB line which matches the way I created the sine wave with a peak of 0 db. The dialog however is telling us its peak value is 4.56 db! I think the software inside Audition 3.0 has lost its mind. How can we be looking at a 0 db peak in the waveform yet have this dialog say it is actually 4.56?

Then there is the bit about Average RMS Power. It is saying it is 0 db. How can the average of the upper part of that waveform be 0? Only the top value is 0. The rest are lower than 0 so their average should not be zero.

Please help me Arny. My next move was going to be a lobotomy or shock therapy to my computer but someone told me that could screw up my system. I hope that is not true! I really need this to work to win an online argument.

Much thanks in advance.

Amir
Founder, Madrona Digital
"Insist on Quality Engineering"

amirm is offline  
post #616 of 637 Old 03-28-2014, 09:42 AM
AVS Addicted Member
 
Ratman's Avatar
 
Join Date: May 2002
Location: Collingswood, N.J.
Posts: 14,297
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 91 Post(s)
Liked: 251
You two can actually make "learning" a joke with your constant competitive bickering.
Ratman is offline  
post #617 of 637 Old 03-28-2014, 01:13 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 13,875
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 435 Post(s)
Liked: 1057
Quote:
Originally Posted by amirm View Post

Quote:
Originally Posted by arnyk View Post

If I could teach some people basic science, their lives and by extension my life as their lives impinge on it, could be simplified.
Your fountain of wisdom, experience and knowledge is a gift to us all Arny. I don't know what we would do without you.
Quote:
If you suspect a measurement tool don't psychoanalyze it or speculate. Just give it some varied problems to solve that you already know reliable answers for by other means.
Ah. Thank you! I have been feeding self-help files to Audition 3.0 hoping it would fix its bugs but had not worked. Now I know why!!!

This is the type of problem I am having. Here is a simple 7 KHz sine wave where I have (roughly) selected the positive side of the waveform and asked the older Audition 3.0 to show its statistics:

i-SXxRhtd.png

The horizontal white line shows the 0 dB line which matches the way I created the sine wave with a peak of 0 db. The dialog however is telling us its peak value is 4.56 db! I think the software inside Audition 3.0 has lost its mind. How can we be looking at a 0 db peak in the waveform yet have this dialog say it is actually 4.56?

Then there is the bit about Average RMS Power. It is saying it is 0 db. How can the average of the upper part of that waveform be 0? Only the top value is 0. The rest are lower than 0 so their average should not be zero.

Please help me Arny. My next move was going to be a lobotomy or shock therapy to my computer but someone told me that could screw up my system. I hope that is not true! I really need this to work to win an online argument.

Much thanks in advance.

I think the following is a more useful test:



I selected a larger but convenient number of cycles of a 1 KHz FS sine wave.

The +/- 32767 peaks seem right.

The peak amplitude of -0.05 dB makes no sense but it is close enough to the expected 0.00 dB

The 3 clipped samples - same answer

The DC offset of 0.062 % is probably due to not selecting a precise integer number of waves. I tried to improve that, and to no avail.

The Average RMS Power: -3.01 dB is exactly what I expected

The Total RMS Power: -3.02 dB again exactly what I expected

Hope this helps.
arnyk is online now  
post #618 of 637 Old 03-29-2014, 08:37 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 17,797
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 524 Post(s)
Liked: 336
Quote:
Originally Posted by arnyk View Post

I think the following is a more useful test:



I selected a larger but convenient number of cycles of a 1 KHz FS sine wave.

The +/- 32767 peaks seem right.

The peak amplitude of -0.05 dB makes no sense but it is close enough to the expected 0.00 dB

The 3 clipped samples - same answer

The DC offset of 0.062 % is probably due to not selecting a precise integer number of waves. I tried to improve that, and to no avail.

The Average RMS Power: -3.01 dB is exactly what I expected

The Total RMS Power: -3.02 dB again exactly what I expected

Hope this helps.
It doesn't Arny. Knowing the answer, i.e. 3 dB, you worked backward until that answer matched that. With arbitrary piece of music, you wouldn't know what the answer is supposed to be to play around this way.

Even in this instance you have random/buggy data. The minimum RMS Power is displayed as 996.99 dB! That is some minimum. The Tunguska Meteor is supposed to be the loudest sound produced on earth and it clocked at ~300 dB. The minimum in your sine wave is supposed to be 600 db more??? And the maximum is negative infinity? The Average RMS Power is supposed to be the average all the numbers including these max and minimum values so the fact that it shows -3 db is actually a problem.

As to Average RMS Power being right, it actually isn't. You have selected 0 db = FS Square Wave. You are analyzing a sine wave and making crest factor estimations relative to a sine wave. As such, you need to select sine wave, not square wave. sine wave and square wave differ by the ratio of square root of two. Therefore selecting sine wave as your reference, will modify Average RMS Power. Repeating that with my copy of Audition 3.0 I get 0 db and peak of "-19 db!" So another set of garbage numbers.

Net, net, what I said originally is true. You cannot use this feature of Audition to find peaks in very short segments. Its use is limited to analyzing very large segments and even there, you need to use the new edition to get reasonable numbers in some but not all cases.

Bottom line on the crest factor argument is that no one has provided any data behind music crest factor being "8 to 10 db" and lower than a sine wave. The sample test that I showed with MPEG-2 test clip of French Horn, actually looks very close to a sine wave:

i-k6N48NW-XL.png

It is impossible to make a case that the RMS power of this waveform is lower than a sine wave by 5 to 7 dB. Or that this waveform is wildly different than a sine wave. As I said, music is whatever it wants to be. We don't get to define it as something that has 5 yo 7 dB lower average.

Even if we had correct stats, we have no idea how much capacitive reserves we have in any one amplifier. Designers can put in 10,000 microfarad or 100,000 microfarad. Worse yet, we don't know how much power is being pumped into those capacitors. The less, the faster the capacitor drains. The argument backfires on us anyway as high-end gear as a rule has much more capacitor reserves than mass market gear. Large electrolytic capacitors used for this purpose are expensive and take up a lot of space so the tendency in cost sensitive mass market products to skimp on them. Same is true as far as larger transformers versus smaller.

So let's move on and not make arguments like this that quickly get us in trouble.

Amir
Founder, Madrona Digital
"Insist on Quality Engineering"

amirm is offline  
post #619 of 637 Old 03-29-2014, 09:01 AM
AVS Special Member
 
DonH50's Avatar
 
Join Date: Feb 2010
Location: Monument CO
Posts: 5,928
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 97 Post(s)
Liked: 234
So, what is the pp, average, and RMS value of the French horn's tone?

Years ago a number of studies were done showing the peak-to-average ratio for music was 17 dB. The results were published in the AES Journal. I am no longer an AES member but the results may be openly available. IIRC compressed pop was lower, closer to 10 dB (some might even be worse), and some small group and classical larger, but overall the 17 dB ws pretyy consistent. I know the studies I did were pretty much in line with that. This was back in the 80's when the dynamic range needed for CDs was in question (some things never change). Of course this is not quite the same thing under debate here, I know.

Brass instruments tend to have the most overtones, then some winds and strings, while the flute has a tone nearest a sine wave. At least IIRC. One of the most frustrating things these days is "knowing" things I find very difficult to prove by providing references and all the usual stuff people expect.

And so,

IMO - Don

"After silence, that which best expresses the inexpressible, is music" - Aldous Huxley
DonH50 is offline  
post #620 of 637 Old 03-29-2014, 10:07 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 17,797
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 524 Post(s)
Liked: 336
Quote:
Originally Posted by DonH50 View Post

So, what is the pp, average, and RMS value of the French horn's tone?
I think the ratio is pretty close to sine wave as I showed so ~3db. The peak can be seen from the waveform.
Quote:
Years ago a number of studies were done showing the peak-to-average ratio for music was 17 dB. The results were published in the AES Journal. I am no longer an AES member but the results may be openly available. IIRC compressed pop was lower, closer to 10 dB (some might even be worse), and some small group and classical larger, but overall the 17 dB ws pretyy consistent. I know the studies I did were pretty much in line with that. This was back in the 80's when the dynamic range needed for CDs was in question (some things never change). Of course this is not quite the same thing under debate here, I know.

You must be thinking of Peter Chapman's paper: "Programme Material Analysis" from B&O company:

i-Q8jZK4S.png

The 17 dB comes from the symphonic material as you noted. Alas, it is important to understand that the purpose of the paper was to determine how hard a speaker is driven as to be able to predict its lifetime. To that end, the above number is an average across many tracks and many CDs. That data does not apply in the amplifier argument because again, the amp is not trying to amplify your entire collection as an average. It sees instantaneous voltages at its input and if a couple of those arrive back to back it will be running on empty with respect to its power supply reserves and clip.
Quote:
One of the most frustrating things these days is "knowing" things I find very difficult to prove by providing references and all the usual stuff people expect.
I hear you smile.gif. Every time I get the bill for my AES and ASA membership there is a moment where I consider saving money and not paying. But then immediately have a need to look up something and I will go ahead and pay. It seems like a never ending tuition for audio knowledge smile.gif.

Amir
Founder, Madrona Digital
"Insist on Quality Engineering"

amirm is offline  
post #621 of 637 Old 03-29-2014, 10:49 AM
AVS Addicted Member
 
Ratman's Avatar
 
Join Date: May 2002
Location: Collingswood, N.J.
Posts: 14,297
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 91 Post(s)
Liked: 251
Quote:
Originally Posted by amirm View Post

Every time I get the bill for my AES and ASA membership there is a moment where I consider saving money and not paying. But then immediately have a need to look up something and I will go ahead and pay. It seems like a never ending tuition for audio knowledge.
You can afford it. biggrin.gif
Ratman is offline  
post #622 of 637 Old 03-29-2014, 11:58 AM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 13,875
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 435 Post(s)
Liked: 1057
Quote:
Originally Posted by amirm View Post

It doesn't Arny. Knowing the answer, i.e. 3 dB, you worked backward until that answer matched that. With arbitrary piece of music, you wouldn't know what the answer is supposed to be to play around this way.

I'm not sure what that means, but the methodology I used was to consult a standard reference, such as this one

http://en.wikipedia.org/wiki/Crest_factor

Which includes this data:



And see what the software (in this case CEP 2.1) says for test files of the same kind of waves as close as I could generate them in a few minutes:

Sine Wave: Book Answer: -3.01 dB


Min Sample Value: -32767
Max Sample Value: 32767
Peak Amplitude: 0 dB
Possibly Clipped: 83
DC Offset: .017
Minimum RMS Power: -3.02 dB
Maximum RMS Power: -3.02 dB
Average RMS Power: -3.02 dB
Total RMS Power: -3.02 dB

Full Wave Rectified: Book Answer: -3.01 dB (This test file was a little approximate)

Min Sample Value: 0
Max Sample Value: 32767
Peak Amplitude: 0 dB
DC Offset: 62.341
Minimum RMS Power: -3.11 dB
Maximum RMS Power: -3.11 dB
Average RMS Power: -3.11 dB
Total RMS Power: -3.11 dB

Triangle Wave: Book Answer: -4.77 dB

Min Sample Value: -32767
Max Sample Value: 32767
Peak Amplitude: 0 dB
DC Offset: .004
Minimum RMS Power: -4.78 dB
Maximum RMS Power: -4.78 dB
Average RMS Power: -4.78 dB
Total RMS Power: -4.78 dB

Square Wave: Book Answer: -0.0 dB

Min Sample Value: -32767
Max Sample Value: 32767
Peak Amplitude: 0 dB
DC Offset: -.213
Minimum RMS Power: -.01 dB
Maximum RMS Power: -.01 dB
Average RMS Power: -.01 dB
Total RMS Power: -.01 dB

Let the nit picking begin!

And finally, a french horn:

http://www.beginband.com/sndclips.shtml




Crest factor approximately 6 dB, about as low as I've ever seen it with music. The example is questionable because it is a solo instrument that does not play all that loudly, and would only reach peak levels as part of an orchestra with many other instruments playing. Nevertheless, 6 dB supports the point I have been making all along.

And finally an enlargement of the french horn note with approximately 6 dB crest factor:

arnyk is online now  
post #623 of 637 Old 03-29-2014, 01:16 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 13,875
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 435 Post(s)
Liked: 1057
Quote:
Originally Posted by amirm View Post


Even if we had correct stats, we have no idea how much capacitive reserves we have in any one amplifier.

Hmm, Service manuals are banned in Washington state? ;-)
Quote:
Designers can put in 10,000 microfarad or 100,000 microfarad.

The first number is typical, +/- 20%.
Quote:
Worse yet, we don't know how much power is being pumped into those capacitors.

Voltmeters are banned in Washington state? ;-)
Quote:
The less, the faster the capacitor drains.

Wrong again. Actually, if anything the reverse is true. The general answer is that the capacitor drains at the same rate when it is charged to any voltage that is sufficient to keep it from clipping. That is unless Coulomb's Law is banned in Washington State!
Quote:
The argument backfires on us anyway as high-end gear as a rule has much more capacitor reserves than mass market gear.

After reading a lot of service manuals, not so much.
Quote:
Large electrolytic capacitors used for this purpose are expensive and take up a lot of space so the tendency in cost sensitive mass market products to skimp on them.

Maybe a month ago I illustrated how that is not exactly the truth. The size and cost of the caps that one puts into power supplies has been shinking pretty dramatically since the days when I built my Dyna 400.
Quote:
Same is true as far as larger transformers versus smaller.

But the fact that large transformers aren't necessary is the point for which supporting evidence continues to pile up. Crest factor > 0 dB means that power power transformers for amps that pump out music and drama need to be only a fraction of the size of what is required to pump out sine waves on the test bench.
arnyk is online now  
post #624 of 637 Old 03-29-2014, 03:03 PM
AVS Special Member
 
DonH50's Avatar
 
Join Date: Feb 2010
Location: Monument CO
Posts: 5,928
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 97 Post(s)
Liked: 234
@Amir: Thanks for the name, I thought it was an AES paper. Or I might have seen it an an AES event, or it may have been quoted in another paper, anything to stave off the feeling of forgetfulness from aging... smile.gif Still, even as a coarse average guide, it seems to support relatively high crest factors for a lot of music even if we don't know exactly how long the peaks are in a given tune.

Capacitors can ride out short peaks and recharge reasonably quickly (especially fir switch-mode supplies). Longer loud passages or sustained loud notes tend to be more strenuous, but of course this all assumes the amp is near its power limits, a debatable premise.

I keep my IEEE and various other memberships active, let the AES lapse because for a long time I wasn't "doing" audio. I keep thinking I need to up again. I gave them my info a couple of years ago but they did not find my old membership from a couple of decades ago, not that it would help.

@Arny: One quibble: For a conventional power supply the capacitor is recharged at ~120 Hz, so a large sustained pulse above that frequency will discharge the caps faster than they can be recharged. Fortunately there is more than ample headroom for most of us, and a brief clipping event is usually unnoticed. I have observed movies seem to have greater dynamic range than music, but then again clipping on an explosion is not likely to be noticed either... smile.gif Reality is always much more complex.

All: IEEE Standard 1241 deals with ADCs and defines procedures for generating test signals that do not require windowing (that is, are continuous at the ends, no glitches).

FWIWFM - Don

"After silence, that which best expresses the inexpressible, is music" - Aldous Huxley
DonH50 is offline  
post #625 of 637 Old 03-29-2014, 06:26 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 13,875
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 435 Post(s)
Liked: 1057
Quote:
Originally Posted by DonH50 View Post


@Arny: One quibble: For a conventional power supply the capacitor is recharged at ~120 Hz, so a large sustained pulse above that frequency will discharge the caps faster than they can be recharged.

That seems to disagree with what one observes on the test bench. The power supply tends to progressively lose more and more of its ability to provide a steady voltage source when the amp is amplifiing signals further and further below 120 Hz. It is far more solid and consistent above 120 Hz.
Quote:
Fortunately there is more than ample headroom for most of us, and a brief clipping event is usually unnoticed. I have observed movies seem to have greater dynamic range than music, but then again clipping on an explosion is not likely to be noticed either... smile.gif Reality is always much more complex.

Agreed that certain minimal and infrequent amounts of clipping seem to be tolerated by the ear.
arnyk is online now  
post #626 of 637 Old 03-29-2014, 06:57 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 13,875
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 435 Post(s)
Liked: 1057
Quote:
Originally Posted by amirm View Post


i-Q8jZK4S.png

The paper is online in its entirety here:

http://www.mountain-environment.com/AES_paper_1996_4277.pdf

"The peak-to-r.m.s, ratios for the programme material groups are much higher than the ratio used
in the existing power tests. Even the more compressed hip-hop material has a ratio of 8.2 dB
which compared to the 3 dB currentlyused is an increaseof 3.3 times.Hence, using the 3 dB
peak-to-r.m.s, ratio does not representprogramme material."
arnyk is online now  
post #627 of 637 Old 03-29-2014, 08:47 PM
AVS Special Member
 
DonH50's Avatar
 
Join Date: Feb 2010
Location: Monument CO
Posts: 5,928
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 97 Post(s)
Liked: 234
Quote:
Originally Posted by arnyk View Post

Quote:
Originally Posted by DonH50 View Post


@Arny: One quibble: For a conventional power supply the capacitor is recharged at ~120 Hz, so a large sustained pulse above that frequency will discharge the caps faster than they can be recharged.

That seems to disagree with what one observes on the test bench. The power supply tends to progressively lose more and more of its ability to provide a steady voltage source when the amp is amplifiing signals further and further below 120 Hz. It is far more solid and consistent above 120 Hz
.

Guessing other factors at play, like more sustained current, higher thermal load, etc. I have seen issues @ HF and LF during testing but doubt I have as much experience as you measuring various amplifiers (almost certainly not). Interesting observation, something I need to think about.

Thanks for the link to the paper! At least I didn't completely forget everything, it was an AES paper. Gonna' toddle off and download it. - Don

"After silence, that which best expresses the inexpressible, is music" - Aldous Huxley
DonH50 is offline  
post #628 of 637 Old 03-30-2014, 05:21 AM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 13,875
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 435 Post(s)
Liked: 1057
Quote:
Originally Posted by DonH50 View Post

Quote:
Originally Posted by arnyk View Post

Quote:
Originally Posted by DonH50 View Post


@Arny: One quibble: For a conventional power supply the capacitor is recharged at ~120 Hz, so a large sustained pulse above that frequency will discharge the caps faster than they can be recharged.

That seems to disagree with what one observes on the test bench. The power supply tends to progressively lose more and more of its ability to provide a steady voltage source when the amp is amplifiing signals further and further below 120 Hz. It is far more solid and consistent above 120 Hz
.

Guessing other factors at play, like more sustained current, higher thermal load, etc. I have seen issues @ HF and LF during testing but doubt I have as much experience as you measuring various amplifiers (almost certainly not). Interesting observation, something I need to think about.

Thanks for the link to the paper! At least I didn't completely forget everything, it was an AES paper. Gonna' toddle off and download it. - Don

The thrust of the paper seems to be similar to what has been somewhat controversial here - the existing testing methodologies put more stress on the UUTs (in this case speakers) than actual use. Because of the intimate connection between speakers and amplifiers, the same test signals might be appropriate for both.
arnyk is online now  
post #629 of 637 Old 03-30-2014, 07:36 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 17,797
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 524 Post(s)
Liked: 336
Quote:
Originally Posted by arnyk View Post

Hmm, Service manuals are banned in Washington state? ;-)
They are. They were being used to wrap pot which we outlawed in the last referendum. biggrin.gif

Seriously the schematic will not give you the answer at all. It is like looking at a car and knowing its specs and telling me you know at any one instant what gas mileage I am getting.

In this case, we are feeding the amplifier a dynamic and unpredictable signal. That power drain which is a function of the efficiency of the amplifier impacts the capacitor in the power supply. The cap in turn is getting charged by the upstream transformer and rectifiers. There is no way in you head you can compute the level of drain in the capacitor by looking at the schematic.
Quote:
The first number [10,000] is typical, +/- 20%.
Typical in mass market gear. Not so in high-end. Here is Mark Levinson No 53 monoblock: http://www.marklevinson.com/downloads/products/prod_22_634473655136955941_ML%20No53%20Technology%20Background%20V5%2004032010_5.17.10.pdf

"In addition to 188,000µF of local capacitance in the main power supply section, the №53 power amplifier boards feature an additional 105,600µF of local capacitance, for a total of 293,600µF capacitance (146,800µF per voltage rail), for a combined storage capacity of approximately 680 Joules. "

Since this is one channel, comparing to a stereo amp it would have double that which would put it a whopping ~600,000 µF or 60X more than your typical number.

Here is a the Parasound JC 1 monoblock amp: http://www.parasound.com/halo/jc1.php
"18,000 µF filters for driver stage; 132,000 µF Nichicon Gold Tune filters for output stage"

132,000 + 18,000 = 150,000. Multiplying this by two to get the stereo level we arrive at 300,000 or 30X more than your number.

Those are for stereo numbers. Here is my trusty Proceed Amp 5:

CES99-Proceed%20Five%20Channel%20Amplifier%20inside%20chassis.jpg

Too lazy to pull out my monster 125 pound unit out of the rack to see the capacitor values smile.gif. Suffice it to say it is more than 10,000 uf.

As I mentioned, this whole argument can backfire since high-end and high-power amps have more capacitor reserve as a rule.
Quote:
Wrong again. Actually, if anything the reverse is true. The general answer is that the capacitor drains at the same rate when it is charged to any voltage that is sufficient to keep it from clipping. That is unless Coulomb's Law is banned in Washington State!
I don't know what this means seeing how you are mixing voltage and energy in the same breath. That aside, what I said was that the larger your capacitor bank, the more reserve capacity you have and vice versa. I don't know how you would think the reverse is true. It is like saying a 100 gallon water tank holds less water than a 50 gallon one. Coulomb's law is just name dropping and of no consequence in what I said.

Amir
Founder, Madrona Digital
"Insist on Quality Engineering"

amirm is offline  
post #630 of 637 Old 03-30-2014, 08:15 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 17,797
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 524 Post(s)
Liked: 336
Quote:
Originally Posted by arnyk View Post

I'm not sure what that means, but the methodology I used was to consult a standard reference, such as this one

http://en.wikipedia.org/wiki/Crest_factor
Thanks for running those stats Arny. Alas it gets us no closer to the answer. With respect to known waveforms and RMS values, you can mess with the parameters and waveform until you get your answer. As I mentioned, you can't do that with music as you don't know the actual value.

Recall that I started this discussion by showing a waveform and selecting one of its peaks -- the very thing we are discussing as to the strain on the power supply -- and showed that the computed value by Audition were wrong. I later said that I had left one of the important fixes as an exercise for the reader. Let me show the effect of that in these two demos. Here is a song with a segment of the waveform selected and stats for it:

i-67xkWDB-XL.png

I purposely placed the selection to the vertical scale on the right so that we can visually do the math. We see that the "average" value is computed as -23 dB. If we look at the scale on the right we see that none of the selected waveform dip below 9 db. So clearly that statistic is wrong. Remember that "RMS" value is supposed to be the average. In this case the waveform is all positive so the RMS value is simply the average of all of those numbers.

Now look at what you get when you correctly use the program:

i-L56DNC6-XL.png

Ah, our average is now -5.31 db which is much more reasonable. The peak is 3.63. So the crest factor is 1.68 db! That is even smaller than a sine wave which makes sense since our selected waveform is spending its time at high amplitudes instead of gracefully dipping to zero as sine wave does.

BTW, I can select a different segment and have it spit out basically the same value for both!

Net, net, the tool requires expertise to use. You cannot ignore such things as -infinity as the maximum signal level and still think you are getting reasonable output by cherry picking the number that matches your argument.

Amir
Founder, Madrona Digital
"Insist on Quality Engineering"

amirm is offline  
Reply Audio theory, Setup and Chat

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off