Sound Quality Vs. Measurements - Page 929 - diyAudio
Go Back   Home > Forums > Member Areas > The Lounge

The Lounge A place to talk about almost anything but politics and religion.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 29th May 2013, 12:43 PM   #9281
godfrey is offline godfrey  South Africa
diyAudio Member
 
Join Date: Nov 2009
Location: Cape Town
Quote:
Originally Posted by dvv View Post
Godfrey, again - if we knew all there was to be known to be able to make a shall we say credible amplifier, many would be doing it.

Yet, they are not. Most sound generally artifical, implyong but not really conveying the musical message.

Now, why would that be? Poor designer education?

Nicely sidestepped.

The question was: Why do you find the claim that "when two amps measure the same, they must sound the same" preposterous, when your own experience tells you that amplifiers with audible differences also have measurable differences?
  Reply With Quote
Old 29th May 2013, 12:53 PM   #9282
diyAudio Member
 
Join Date: Mar 2007
Location: Oxfordshire
I think the point people miss is we can just about make two amplifiers that have almost zero distortion that sound different . The almost zero is perhaps easier than going for two that are lets say 0.1 % .

The thing to get excited about is they will be different . 10 years from now we know why . RFi , who knows .

I have been out with one of identical twins . Joe was nicer than Caroline . Guess which one I choose ?
  Reply With Quote
Old 29th May 2013, 01:12 PM   #9283
diyAudio Member
 
Join Date: Feb 2009
Location: UK
Quote:
Originally Posted by fas42 View Post
I've mentioned variations on the following general testing methodology many times, but it's always ignored or discounted as being too similar to current techniques. Input a mix of several high frequency audio sine waves at very low levels, say -40db from max - do a spectrum measurement of the output. Then, stress the whole system in various ways so that theoretically the distortion components resulting from that original mix are not directly impacted, look at the spectrum content again; my belief is that a 'good sounding' system will show much less variation of levels of the original spectral frequencies than a 'bad' one.
But why sine waves? If transients are where we think the system is falling down, why not zoom in directly in the time domain?

The frequency domain is used for historical reasons (analogue equipment only) and because it allows us to ignore phase and frequency response, I think.

But take a mystery amplifier which adds a 'blip' at the zero-crossing of the input waveform but only if it's steep enough, or something like that, and the frequency domain is a very indirect method indeed of trying to see what's happening. Maybe it results in a 0.001% average increase in THD+N or something if you're lucky enough to trigger it. But zoom right in on the output in the time domain, driving it with arbitrary waveforms, and suddenly you can isolate and see the 'blip' directly, so you know what's going on and maybe how to fix it.

Edit: You know, this sort of thing might be fun to try in practice, but it might even be possible to simulate it in software.

1. Design the test amplifier in SPICE, including mechanism for introducing blip/wobble/instability, and a realistic load.
2. Write a program to probe the circuit with test signals, harnessing SPICE to generate the output. Temperature can be included as a variable, as could power supply variation etc.
3. Compare frequency domain results to the results from the proposed time domain 'null test'.
4. See which is the more useful or informative.

Edit 2: I see I'm talking into a vacuum again.

Last edited by CopperTop; 29th May 2013 at 01:30 PM.
  Reply With Quote
Old 29th May 2013, 01:24 PM   #9284
fas42 is online now fas42  Australia
diyAudio Member
 
fas42's Avatar
 
Join Date: Jun 2012
Location: NSW, Australia
Blog Entries: 11
Quote:
Originally Posted by nigel pearson View Post
One test that has been suggested is to take a recording of a piece of music then convert the amplifier output back into digital and see what differences there are . Some say it can be dramatic . Why not standard null tests also ?
If only it were as simple as this ..

Unfortunately, a tiny variation, which has zero audible impact, can result in a large difference file, where the meaningful distinguishing features, from the point of view of your hearing, will be swamped. It's not trivial to separate the features that matter from the ones that are far less important acoustically - otherwise it would been done a long time ago ...
__________________
Frank the truth is, I just like a bit of ASMR ...
  Reply With Quote
Old 29th May 2013, 01:33 PM   #9285
diyAudio Member
 
Join Date: Feb 2009
Location: UK
Quote:
Originally Posted by fas42 View Post
If only it were as simple as this ..

Unfortunately, a tiny variation, which has zero audible impact, can result in a large difference file, where the meaningful distinguishing features, from the point of view of your hearing, will be swamped. It's not trivial to separate the features that matter from the ones that are far less important acoustically - otherwise it would been done a long time ago ...
What are those tiny variations with zero audible impact? As I mentioned earlier, the key is for the test setup to be periodically making a linear model of the circuit under test (gain, delay, phase, frequency response). What's left is not linear, and so may not be dismissed as having zero audible impact.
  Reply With Quote
Old 29th May 2013, 01:44 PM   #9286
fas42 is online now fas42  Australia
diyAudio Member
 
fas42's Avatar
 
Join Date: Jun 2012
Location: NSW, Australia
Blog Entries: 11
Quote:
Originally Posted by CopperTop View Post
But why sine waves? If transients are where we think the system is falling down, why not zoom in directly in the time domain?

...

But take a mystery amplifier which adds a 'blip' at the zero-crossing of the input waveform but only if it's steep enough, or something like that, and the frequency domain is a very indirect method indeed of trying to see what's happening. Maybe it results in a 0.001% average increase in THD+N or something if you're lucky enough to trigger it. But zoom right in on the output in the time domain, driving it with arbitrary waveforms, and suddenly you can isolate and see the 'blip' directly, so you know what's going on and maybe how to fix it.
Yes, that's perfectly reasonable too, to catch other types of misbehaviour. An industry identity, I think it was Tom Danley - my apologies if it wasn't - mentioned monitoring some reputable amplifiers with a transient capture 'scope, and, seeing intermittent instability at the output. So there's room for more exploration there also ...

Probably most important is trying to create a groundswell of interest and desire of audio people in general to get better measurements happening. Lone individuals doing their thing always seems to fizzle out, fade away to nothing in this game ...
__________________
Frank the truth is, I just like a bit of ASMR ...
  Reply With Quote
Old 29th May 2013, 01:50 PM   #9287
dvv is offline dvv  Serbia
diyAudio Member
 
Join Date: Sep 2004
Location: Belgrade, Serbia
Quote:
Originally Posted by nigel pearson View Post
My ideal amp would be > 100 W 8 R ( 1000 W 1 R transient ) . < 0.1 % THD with suppression bridge reducing harmonics . Output impedance < 0.5R including cables . No dramatic change in distortion spectrum with level . 5 Hz to 20 kHz -0.1 dB . Noise > - 100 dB . Stable into real loads .

My valve amp sounds more detailed and open than most . If the bass was tight I would not bother with better .

Global feedback is the best there is if it suits the amplifier . That's the problem . In most cases it doesn't . With my valve amp feedback could be made to work . If I did it would be nothing to do with distortion . Trends I saw with the feedback version valve amp also are seen with all feedback amps I have played with . I suspect an amp has to be very fast indeed to be an " ideal " feedback amp ? I was shocked how unhappy valve amps are with feedback . Logically that would be true . It can be made to work . In the process one sees the whole problem . If I built an OTL valve amp the first thing to try is gallons of feedback . With the right valve it has to be fast enough . I have heard OTL .They are so dam fast and open . A tad bright I would say . Without an analyzer on hand I suspect a lot of third harmonic and perhaps some fifth ? To be honest I have never heard anything like them . So dam fast .

Some will say I would prefer my 0.1% distortion ideal amp over one that is " more perfect " because I like distortion . Nonsense , what speakers could prove that ? If it is true it will be mechanisms in the machine the feedback loop is hiding that are different . As Dvv says , what is open loop like . Dvv your amp is almost ridiculously good open loop . If you added just enough feedback to meet my output impedance ideal and pot down the input I bet your amp would meet a good spec ? Try it sometime . I think you might be shocked how you like it . It will sound very open and fast is my guess ? Perhaps a bit grainy at times like a good photograph .

One thing is . An amp might have no virtue except the ability to accept feedback . Like an inherently unstable aircraft with the right mechanisms it might be OK . One designer told me that linearity in amps is totally unimportant if there are no dramatic jumps in gain at any level ( open loop ) . Given enough speed that's all that mattered to get the feedback to work .
Nige, my amp is still in the process of birth. After that, there's the development stage, and that will take some time, I have an unholy list of things I want to try out.

But yes, I am rather proud of how it works with no global NFB. It's possible to get even more mileage from it, but just as I'm afraid of falling short, so I'm afraid of overdoing it.

With a THD of 0.3%, 20-20,000 Hz, 100WRMS/8 Ohms, but much more importantly with a harmonic decay rate ALMOST as good as I want it to be (more work is needed), I hope I am home and dry, with reasonably more development work.

Also, very important, even so it is pretty tolerant of not so nice loads with no global NFB. At least, so says my simulator, which has never before lied to me. Nevertheless, I don't completely believe in it all until I have a test unit running on my table and am starting to photograph 'scope screens.

According to my calc, it should be capable of working steady statedown to 3 Ohms; it should also be capable of around 560 Watts into 2 Ohm in impulses. IEC says impulses should be 20 mS, but to me, no less than 45 mS is what I want.

I nearly pulled it off with my last one some years ago, but even then, I somehow knew that wasn't the real McCoy and that I'd be back at it sooner or later.

Since most people here probably have no idea of what we are on about, here's the incriminated device:
Attached Images
File Type: jpg CENTURION-PCB-PIC-1-by-Alex.jpg (389.2 KB, 69 views)
Attached Files
File Type: pdf Economy 100100 SE AC Amp Full SCH.pdf (79.2 KB, 26 views)
__________________
Per Aspera Ad Astra.
  Reply With Quote
Old 29th May 2013, 01:55 PM   #9288
dvv is offline dvv  Serbia
diyAudio Member
 
Join Date: Sep 2004
Location: Belgrade, Serbia
Just a quick note: some part designations are wrong, and the note lists the wrong transistor to be mounted on the heat sink.

And the second board, one to be piggybacked, is still to come any day now.

The end result should be a complete power module, with everything but the large filter caps on it. So someone may want to install say 3 in one case.

Also in the works is the basically same amp adapted to half the nominal power, 50/100 W into 8/4 Ohms, but REALLY pushed into pure class A.

Then I would have enough to upgrade my loudspeakers into fully active.
__________________
Per Aspera Ad Astra.
  Reply With Quote
Old 29th May 2013, 01:58 PM   #9289
fas42 is online now fas42  Australia
diyAudio Member
 
fas42's Avatar
 
Join Date: Jun 2012
Location: NSW, Australia
Blog Entries: 11
Quote:
Originally Posted by CopperTop View Post
What are those tiny variations with zero audible impact? As I mentioned earlier, the key is for the test setup to be periodically making a linear model of the circuit under test (gain, delay, phase, frequency response). What's left is not linear, and so may not be dismissed as having zero audible impact.
Nigel was talking of the Diffmaker type of thing, which apparently struggles with variations of timing, phase and frequency in certain situations, even though that is what it's designed to deal with.

Yes, a sophisticated enough mechanism will do the job - just requires someone to be motivated enough to make it happen ...
__________________
Frank the truth is, I just like a bit of ASMR ...
  Reply With Quote
Old 29th May 2013, 02:16 PM   #9290
diyAudio Member
 
Join Date: Jan 2008
Quote:
Originally Posted by dvv View Post
Godfrey, again - if we knew all there was to be known to be able to make a shall we say credible amplifier, many would be doing it.

Yet, they are not. Most sound generally artifical, implyong but not really conveying the musical message.

Now, why would that be? Poor designer education?
Musical message may be conveyed perfectly played even as wax cylinder. Very little realism is required for this, only ability to listen to the music, and let it inside the mind.

As per credible amplifier, once non-linearity and noise are controlled, differences between specimens playing same speaker reveal only differences in transfer function of how load couples with amplifier as system. Quotient of transfer function of system using amplifier A, with transfer function of system using amplifier B may be used to both express the difference between the two amplifiers, and to null the perceived differences between the two systems.

Amplifiers with no effective load, transfer no real information. Gross distortion and gross noise are obvious defects that are easily avoided.

This is a little more difficult with electro-acoustic transducers, but is readily achieved to levels of statistical significance over large population of listeners, using wide range of source signals. This is in ballpark of 0.1%-1% THD and encompasses IMD as well, and fits my personal experience.

Once noise and gross distortion are accounted for, information is neither created or destroyed by transmission system, only rearranged in completely reversible manner.

To wit; any recording is discrete in time may be convolved with MLS pattern that renders recording as noise to human perception. Convolution with reverse MLS returns recording to original state.

Likewise a discrete signal may be digitized. When this is played on analog system is sounds like noise, but original waveform is perfectly reconstructed by transfer function of D/A converter.

Capturing sound on wax cylinder was based on observation that sound causes physical vibration, and that physical vibration causes sound. Sound quality made little advancement until quantifiable, repeatable measurements were made possible.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

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


Similar Threads
Thread Thread Starter Forum Replies Last Post
Quality Control differences = variations in sound quality? KT Class D 3 4th June 2014 12:02 AM
Sound Card for Measurements Marik Solid State 2 2nd January 2012 08:59 PM
Sound Card Recommendations (For Audio Measurements) dchisholm Equipment & Tools 5 16th July 2011 09:40 AM
How to protect sound card during amp measurements? okapi Everything Else 13 2nd September 2008 03:06 PM
Sound cards - test and measurements jackinnj Everything Else 2 5th July 2003 03:02 PM


New To Site? Need Help?

All times are GMT. The time now is 10:22 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2