Blind Listening Tests & Amplifiers

Status
Not open for further replies.
SY said:

nw, there are deviations that aren't worth bothering with. Consider two amps with identical and vanishingly low distortion. They null nearly perfectly. Now, put a perfect 1 ms delay at the input of one of the amps.
Well, again I agree, but in my real-world experience, uniform group delay hasn't been a big problem. Phase shift at the highest frequencies is, and that's why you bandwidth limit the results and use a 44khz sampled CD to drive the amp.

SY said:
And another question in the case of HD and IM is what to do (if anything) to minimize them even after they're below the threshold of audibility. If I have a transparent amp (transparent in that input and output cannot be distinguished by a skilled listener) with 0.5% THD and a 0.1 ohm source Z and want to make that null better, am I "hurting anything" if I increase the parts count, making the amp more expensive and less reliable? Am I "hurting anything" if I increase the O/P stage idle current, thus increasing heat, lowering reliability, and increasing the customers' electricity bills?
The above are valid concerns, but they go directly back to my statement of "How good of a null is good enough?". My solution to the above was to do a reasonable (but hardly exhaustive) amount of correlation between null results and blind listening tests. As I said, I'm pretty confident that even the golden ear types here couldn't detect differences below -60db in their own systems.

So I would answer your question by proposing you make the amplifier only as complex as necessary to achieve a difference signal that's 60db down in the audio band. That 60db number, however, is obviously open to debate. Your mileage may vary. As I said, even 50db seems to be plenty. But this still doesn't invalidate the test. It simply means you have to do a sufficiently well run test to correlate difference data with what people can hear in a blind test. Sounds like a good research project for one of those academic types huh?

Remember, I'm not proposing that null difference testing be the ONLY tool in the designer's arsenal. But it's a very good one, don't you agree?
 
Re: Re: Re: Re: Topological acuity....

mikek said:


...and the notion that AES papers are written by 'academics with a particular agenda, and little or no experiance designing audio', is as ill-informed, as it is vacuous......The overwhelming bulk of audio technology as known today, is due to such pioneering members of the AES as T. Holman, (THX fame), Dolby, D. Self (currently at audiolab), G. Stanley, (at Crown), E. Benjamin, Sondermeyer, Dennis bohn of Rane,...etc...etc......People who posess more experiance more audio design experiance in each of their finger nails than you're ever likely to accumulate in your lifetime....dear...dear...

I will not deny there are not excellent audio engineers in the list.

Let me give you another list.

John Curl, Nelson Pass, William Z. Johnson, Keith Johnston, Erno
Borbely, Ben Duncan...... all of whom think listening is an important part of designing audio equipment the last time I heard. A few of these guys even think passive components sound different.

I wasn't familiar with Jack Sondermeyer but he has a number of patents from the following:
http://www.audioannals.com/05675656-Sondermeyer-etal-txt.htm

"BACKGROUND OF THE INVENTION

The invention relates to a solid state power amplifier which emulates tube distortion. In particular, the invention relates to a solid state power amplifier having a tube distortion circuit in which the power level of the amplifier may be continually varied while maintaining an appropriate tube distortion sound."
I wonder how that would do on the null test?

I think you must be correct about the "audio design experiance in each of their finger nails" part in at least one case. The last Crown amplifier I heard sounded just like fingernails on a black board. Doesn't Crown also design car stereo amps now?
 
nw, sure, it's a tool, but there are other ways of skinning that cat. Plain vanilla THD measurements, for example. MLS looks like a more interesting way to do it, but one could argue that it's a null test of its own, since you're comparing input and output. In my own testing, done on a shoestring, I've used an instrumentation amp to compare input and output; the IA's noise and distortion set a floor, but it can be a pretty low floor, certainly below your assumed 60 dB criterion. (Personally, I'd argue that, fer sher fer sher, one should set the floor at about 98 dB below max output level, i.e., 16 bits)

When you talk about null tests, what sort of excitation signal do you have in mind?
 
diyAudio Senior Member
Joined 2002
TRANSPARENT.

Hi,

If I have a transparent amp (transparent in that input and output cannot be distinguished by a skilled listener) with 0.5% THD and a 0.1 ohm source Z and want to make that null better, am I "hurting anything" if I increase the parts count, making the amp more expensive and less reliable? Am I "hurting anything" if I increase the O/P stage idle current, thus increasing heat, lowering reliability, and increasing the customers' electricity bills?

Not as transparent as the cables I designed, I hope?

Different colours to different people, I reckon.

The validity of the nulling test is questionable here, to say the least.

As for the Wondercaps, the test method suggested is not realistic, sounds more like grasping at straws to me, the easiest way to do is just replace the caps with a more recent, well respected part and have a listen.
If you can measure a difference, than fine, although I'd doubt it.
If you don't, but still hear an audible difference than we're getting somewhere.

I very much doubt a nulling test would throw up any difference, which proves our point in the first place...if you don't hear any difference than the only conclusion I can come up with is that you must be one of those fortunate individuals that can actually live with whatever gear they have and enjoy it...

Me, I always want to know what's going on behind the scenes,:)
 
Re: Re: Re: Re: Re: Topological acuity....

Fred Dieckmann said:
The last Crown amplifier I heard sounded just like fingernails on a black board. Doesn't Crown also design car stereo amps now?
The last one I heard (a Crown K1) sounded amazingly good considering it's chopping the signal into width modulated high current pulses. BCA amplifiers, like the K1, are genuinely groundbreaking in terms of minimizing some of the negative qualities normally associated with digital amplifiers. In blind testing the K1 fares better than any other digital amplifier we've tested in our project.

It's rather unfortunate that Harmon International holds patents on it because I expect we'd be seeing a lot more BCA amps otherwise. You have to give credit to the designers for thinking outside the box and advancing the state-of-the-art with respect to digital power amps. 2500 continuous RMS watts in a 3.5" high chassis with no fan is an amazing feat (referring to the K1's bigger brother, the K2)--especially when you consider it sounds good!

And Crown helped JBL (another Harmon brand) design the world's most powerful car stereo amp using BCA technology (which as far as I know is still just a prototype). It's rated at a staggering 6000 watts RMS. When you consider who spends the big bucks on car stereo, and why, it's probably a smart marketing move to have the biggest car amp in the world--regardless of how practical it is.
 
nw_avphile said:
I'm not trying to be cranky, but can we please try for a little better signal-to-noise ratio in this thread? On the previous page, out of 15 posts, 5 have something to do with the thread topic. Some previous pages are even worse.

Sure.

But seeing as your post has nothing to do with the topic either, isn't it just adding to the noise? :)

se
 
Re: TRANSPARENT.


If I have a transparent amp (transparent in that input and output cannot be distinguished by a
skilled listener) with 0.5% THD and a 0.1 ohm source Z and want to make that null better, am I
"hurting anything" if I increase the parts count, making the amp more expensive and less reliable?
Am I "hurting anything" if I increase the O/P stage idle current, thus increasing heat, lowering
reliability, and increasing the customers' electricity bills?

Let's say we test this amp in a nulling test and it checks very well. We are not doing any change to the amp itself, except for placing it on spikes or some sort of feet that it makes audible difference. We check the amp again. Would it measure different this time in a null test or not?

PS: Depending on how I place my amps, the difference is so audible that even my 6 year old daughter asks: Daddy what's wrong with Aqua?
 
Re: TRANSPARENT.

SY said:
I've used an instrumentation amp to compare input and output; the IA's noise and distortion set a floor, but it can be a pretty low floor, certainly below your assumed 60 dB criterion. (Personally, I'd argue that, fer sher fer sher, one should set the floor at about 98 dB below max output level, i.e., 16 bits)
Well that brings up another good point about null testing... and that's what level you test at. I think Tube_Dude says he uses 2 volts RMS (I assume he means into the speaker) which is less than 1 watt into 8 ohms. I find with some amps, the noise floor of the amp starts to dominate the difference signal if you go too low. But when I say -60db, I mean -60db below the level of the signal, not below the maximum output level.

SY said:
When you talk about null tests, what sort of excitation signal do you have in mind?
I use real music! In fact, for the null/blind comparisons, I used the same music for each. I've tried different types of music and the more high frequency energy a given recording has, the worse the null result usually is (as would be expected from what we know about amplifier distortion and given the HF phase shift).

I also do null testing with sine waves, but music is the more difficult test as it addresses the real-world slew rate, T.I.M and other distortion the amplifier may produce with a transient or non-repetitive signal. I also do the testing driving real speakers and it's much nicer to listen to music while running the test than sine waves :)

fdegrove said:
As for the Wondercaps, the test method suggested is not realistic, sounds more like grasping at straws to me, the easiest way to do is just replace the caps with a more recent, well respected part and have a listen.
Why is it not realistic? If that's a BLIND listen you're suggesting (between two identical amps using the different caps), we're in total agreement. If you know what you're listening to, the psychological bias (already discussed at length elsewhere) is going to dominate what you hear.


fdegrove said:
If you can measure a difference, than fine, although I'd doubt it. If you don't, but still hear an audible difference than we're getting somewhere.

I very much doubt a nulling test would throw up any difference, which proves our point in the first place...if
Well if the listening isn't blind, it only proves the point that non-blind listening is really flawed. We ARE getting somewhere, however, if you doubt a person could measure a difference between the caps you hate and ones you like.
 
Re: Re: TRANSPARENT.

Peter Daniel said:


Let's say we test this amp in a nulling test and it checks very well. We are not doing any change to the amp itself, except for placing it on spikes or some sort of feet that it makes audible difference. We check the amp again. Would it measure different this time in a null test or not?

A good question!

Because you're driving real speakers with the null test, you could easily perform the test with the amplifier sitting on top of the speaker if you'd like. Any vibration/resonance/sound waves/electromagnetic interference/RFI/gamma rays from another galaxy or anything else that influences the sound would show up in the difference signal.

So yes, as an experiment, it would be interesting to conduct the test with the amp say on an isolation platform in a different room from the speakers, and again with it sitting on top of the speaker cabinet with no isolation with the speaker being driven fairly hard in both cases. If you got the same results, you can be quite confident that your amp is immune to vibration and doesn't need any special treatment or isolation.

If you get a different result, you could use the null testing to narrow down just how much isolation you needed to avoid the problem, or dribble some goop on the circuit until it went away, etc. That's an example of how null testing can be very useful.
 
diyAudio Senior Member
Joined 2002
REALISM.

Hi,

Why is it not realistic?

Because chances are against you to have Peter and myself in the room doing blind tests are rather slim...not that I'd mind.

Other than that, I'm pretty much convinced both Peter and myself would bust statistics in a hurry...in other words, choose better guinea bigs like most other stasticians do.:mafioso:

Relax and listen, ;)
 
The null...that tell the story!!!

Well that brings up another good point about null testing... and that's what level you test at. I think Tube_Dude says he uses 2 volts RMS (I assume he means into the speaker) which is less than 1 watt into 8 ohms. I find with some amps, the noise floor of the amp starts to dominate the difference signal if you go too low. But when I say -60db, I mean -60db below the level of the signal, not below the maximum output level.

I have used 2volt input and 2 volts output in my test(inspired in the hafler but more simple) only for comparatif pourposes...and if you see mi post #69 in this thread
you will see that a tweked onkyo can have a null of 1mV....more than 60dB under the 2 volts output...

One more point i felt that anytime i decrease the null the sound quality improve...but we can't forget that this improvement depends of the rest of the audio system....specialy loudspeakers!!
 
nw_avphile said:


A good question!

Because you're driving real speakers with the null test, you could easily perform the test with the amplifier sitting on top of the speaker if you'd like. Any vibration/resonance/sound waves/electromagnetic interference/RFI/gamma rays from another galaxy or anything else that influences the sound would show up in the difference signal.

So yes, as an experiment, it would be interesting to conduct the test with the amp say on an isolation platform in a different room from the speakers, and again with it sitting on top of the speaker cabinet with no isolation with the speaker being driven fairly hard in both cases. If you got the same results, you can be quite confident that your amp is immune to vibration and doesn't need any special treatment or isolation.

If you get a different result, you could use the null testing to narrow down just how much isolation you needed to avoid the problem, or dribble some goop on the circuit until it went away, etc. That's an example of how null testing can be very useful.

This is a fine answer, yet slightly oversimplified. So far I didn't find a piece of SS gear that wouldn't be influenced by resonances and effectively it is observed sonically.

I did some experiments recently and just placing a strip of 1/4 thick wood or acrylic under amp's feet changes its tonal character. Acrylic sharpens high frequencies, wood tones everything down, rubber makes for muddy sound with amp staying in the same spot.

I notice it both with my 50kg Alephs and with 1 kg Gainclones.
 
Re: REALISM.

fdegrove said:
Because chances are against you to have Peter and myself in the room doing blind tests are rather slim...not that I'd mind.

Other than that, I'm pretty much convinced both Peter and myself would bust statistics in a hurry...in other words, choose better guinea bigs like most other stasticians do
It was a hypothetical question, but only one of you need be there to prove the point and it doesn't have to be a gainclone or Peter's $75 capacitors. It could be any amp you respect that we can come up with two identical versions of and your favorite cap against the Wondercap.

We could first verify the two amps sound the same in a blind test using the same caps, and then pop the Wondercaps into one of them and repeat the test. I'm assuming you're saying, if you get to choose these hypothetical amplifiers and that we do this in your own system, you could pick out the one with the Wondercaps if you didn't know which was which?
 
diyAudio Senior Member
Joined 2002
VIBERATIONSSSS

Hi,

Any vibration/resonance/sound waves/electromagnetic interference/RFI/gamma rays from another galaxy or anything else that influences the sound would show up in the difference signal.

If your test does not show that up than it only adds to my believe that is is seriously flawed, not that I didn't know it already.

And you keep on avoiding the subject of component signature, as it is often refered to...colouration in my book, which will NOT show up in nulling tests as you know them.

Why you have such a hard time accepting that the test method you advocate is flawed is beyond me.
Naturally labelling people as subject opposed to objective measurements is easy, rest assured neither Fred, Peter, any other so called subjectivists are and neither am I.
I only feel it takes some experience to interpret what the scope and other analyser is showing you...no it is not in any book, that much I can tel you.

Pffft,;)

/Singing for the moment.
 
diyAudio Senior Member
Joined 2002
TO CAP IT.

Hi,

I'm assuming you're saying, if you get to choose these hypothetical amplifiers and that we do this in your own system, you could pick out the one with the Wondercaps if you didn't know which was which?

Yes, I am rather confident that I'd pass such a test with flying colours.
I always have in the past, not getting any younger though.;)

Oh, and in case you wonder about my resume and credentials too..I could send you those too...:clown:

Cheers,;)
 
Re: VIBERATIONSSSS

fdegrove said:
If your test does not show that up than it only adds to my believe that is is seriously flawed, not that I didn't know it already.
I never said it didn't? I've never run the null test with and without vibration present. That's why I said it would be an interesting test.

fdegrove said:
And you keep on avoiding the subject of component signature, as it is often refered to...colouration in my book, which will NOT show up in nulling tests as you know them.
I'm not sure you understand the test? Coloration was what SY was talking about in point #1 that he made on the previous page and I answered at the top of this page. It does show up. ANY amplifier colorations, intentional or otherwise, will show up in a null test.

fdegrove said:
Why you have such a hard time accepting that the test method you advocate is flawed is beyond me.
Please explain to me how it's flawed and we can go from there?
 
Hey Tube_Dude,

I not finished reading the whole thread yet, but I have to say about the quote below


So it emerges that are 2 scholls of thougt here

1-only with two known menbers me and Nw_phile that belive that audio is a aplication of cience and that good design and some tests can can help in acessing the fidelity of the amp...

2-the other, the vast mayority of members that belive that listening is all...that more important that the nuling test, is the opinion of some one, that from other part of planet claim that capacitor A sound better that capacitor B witout any cientifical proof...and that a amplifier can make a role of Mozart and make better the original sound...

So my friend Nw_phile...now we are two!!!

Alone but happy!


We are three...

Saudações
 
diyAudio Senior Member
Joined 2002
TURNING THE TABLES.

Hi,

Please explain to me how it's flawed and we can go from there?

Has it occured to you how you try to avoid the real issues at hand?
You're the one bringing this method of testing, comparing gear to the forum, I feel the onus is on you not me.

Why is it so hard to follow some suggestions made and report back?

Cheers,;)
 
Re: TO CAP IT.

fdegrove said:
Yes, I am rather confident that I'd pass such a test with flying colours.
I always have in the past, not getting any younger though.;)
Hey! We're getting closer... now what we need are two members here who actually live halfway close to each other to make this sort of challenge a reality. Two members from each side of the subjectivist/objectivist fence would be ideal.

It doesn't have to be over Wondercaps. It can be the ability to tell say a $15,000 Mark Levinson from a $300 Onkyo or Yamaha mainstream integrated amp. That actually should be a much easier challenge, right? After all, the cheap Japanese amp will have not only have cheap ten cent capacitors, but cheap plain copper wiring, cheap resistors, cheap connectors, a cheap volume control, no (obvious) vibration/resonance supression, a wimpy power supply, etc.

So any takers on this one? A cheap Asian made integrated up against (hopefully your) high-end four or five figure power amp (one with good specs) in your own system blind?
 
Status
Not open for further replies.