Transparent amplification

[sergeauckland]

Hmm. I may be a total rube here. But, from what little I know, the test amp would only be able to operate at the extreme low power end of its output so as not to overdrive the following reference amp. Yet, most all solid state amps exhibit increasing noise and distortion as a % of signal as output level is reduced.

Do they instead run the test amp at higher power, but externally lower the output? If so, is the padding circuit itself transparent? It is not in the circuit when using just the reference amp. @sergeauckland may have answered this.

So, these tests, while interesting, require the test amp to commit an unnatural act, an extreme case, unlike their normal operation at normal output power directly into a speaker. So, what do these tests really mean?

I agree with Keith. I would rather trust the direct amp measurements, even into dummy loads.

A Straight-wire Bypass test only tests for transparency. Can you hear it when it's in circuit? Yes/No. It can be done as an ABX, or my preferred alternative, AA, AB,BB, BA, and one has to choose between same/different.

The output of the first amplifier has to be attenuated so it doesn't overload the second amp, but I think anyone with half an idea of how this stuff works will accept that a simple resistive attenuator is transparent. I know there are those who think even resistors have a sound, but then there are those who believe the earth is flat, that there are fairies at the bottom of their gardens, and that Father Christmas comes down the chimney every year. I think we can discount those people, as they're the same people that won't take a blind test as the stress of the situation affects their hearing, and would much rather change the cables for tonal correction than ever sully their pristine systems with DSP.

S.

 

[Thomas savage]

Not sure about fairies, I do still hold out hope of bumping into a woodland nymph or two... three...

 

[sergeauckland]

Not sure about fairies, I do still hold out hope of bumping into a woodland nymph or two... three...

They're very different. One's a childish fantasy, the other's a grown-up's fantasy!

S.

 

[oivavoi]

Having said all that, I'm feeling a bit uneasy arguing that amplifiers may not be transparent, as I don't think any tiny lack of transparency in modern well-designed amplifiers will be noted when listening to music.

For actual listening, I suspect that the S/N-ratio might be the most important factor, plus providing sufficient power for dynamic peaks. When listening to very dynamic recordings with loudspeakers that are able to go loud in a quiet room - which I will do when I move to a new apartment in February - the noise floor of the electronics becomes one of the limiting factors. I'd like a S/N ratio of at least 100 db for the first watt. So my requirement is essentially that the amp should provide lots of power, and also be very quiet. A difficult combination for many amps. Modern class d might be the best bet for achieving that, even though those Swedes don't think it's a technology which is completely transparent at this point.

 

[sergeauckland]

Having said all that, I'm feeling a bit uneasy arguing that amplifiers may not be transparent, as I don't think any tiny lack of transparency in modern well-designed amplifiers will be noted when listening to music.

For actual listening, I suspect that the S/N-ratio might be the most important factor, plus providing sufficient power for dynamic peaks. When listening to very dynamic recordings with loudspeakers that are able to go loud in a quiet room - which I will do when I move to a new apartment in February - the noise floor of the electronics becomes one of the limiting factors. I'd like a S/N ratio of at least 100 db for the first watt. So my requirement is essentially that the amp should provide lots of power, and also be very quiet. A difficult combination for many amps. Beyond that, I think amplifiers should just do their job and amplify.

Completely agree with that last sentence. As to the noise floor of the electronics, I'm a lot less bothered by that, given that with LPs, the noise floor of the LP is so high, and even with CD, the noise floor of most live recordings is pretty high, and studio recordings can be anything. With low dynamic range rock recordings, there's never any silence for the noise floor to be apparent, and even with higher dynamic range Jazz or chamber works, the noise floor on the recording is audible, well above the noise floor of the electronics themselves.

S.

 

[oivavoi]

Completely agree with that last sentence. As to the noise floor of the electronics, I'm a lot less bothered by that, given that with LPs, the noise floor of the LP is so high, and even with CD, the noise floor of most live recordings is pretty high, and studio recordings can be anything. With low dynamic range rock recordings, there's never any silence for the noise floor to be apparent, and even with higher dynamic range Jazz or chamber works, the noise floor on the recording is audible, well above the noise floor of the electronics themselves.

S.

Agree! Sorry, I edited my comment while you wrote yours, and the sentence you highlighted mysteriously disappeared. But that remains my approach to amps though. They have a simple job to do and mostly they do it well enough.

EDIT: And I only listen to digital, and mostly classical/jazz, and I try to find as good recordings as possible, so dynamic range and noise floor is an issue for me.

 

[Bjorn]

Members of LTS, the Swedish AES, have claimed on Swedish forums that they have done before/after blind tests on various well regarded class d amps, and that they still haven't found any d amps they deemed transparent enough to write about. They refuse to provide any other details though. I think that they want to do it thorouhly if they put a product down.

This surprised me, as objectivists in most other places seem to praise ncore etc. But I have no reason to doubt them. They ususally avoid audiophoolery. Would be interesting if others with the technical skills to conduct before/after tests could put the Ncore under the bench: is it transparent or not?

I'm quite sure they haven't tested a Ncore amp. If they did and the amp wasn't built in a way to add distortion, you can almost be certain it would pass their pass. I think even UcD amps with HG and HxR would. Some have found the best UcD amps to sound cleaner and more transparent than Bryston amps, subjectively that is......

While I appreciate the tests LTS does, I personally find their acoustics of their listening room a major weakness. Sitting close to the rear wall with a Skyline diffuser behind them doesn't lay the best foundation for hearing smaller details.

 

[oivavoi]

I'm quite sure they haven't tested a Ncore amp. If they did and the amp wasn't built in a way to add distortion, you can almost be certain it would pass their pass. I think even UcD amps with HG and HxR would. Some have found the best UcD amps to sound cleaner and more transparent than Bryston amps, subjectively that is......

While I appreciate the tests LTS does, I personally find their acoustics of their listening room a major weakness. Sitting close to the rear wall with a Skyline diffuser behind them doesn't lay the best foundation for hearing smaller details.

Thanks Bjørn, didn't know about their listening room. If anything, that could lead to non-transparent devices passing their tests, I guess.

Concerning Ncore: As I said, they don't want to reveal which amps they have looked at. But I have a strong suspicion they have listened to Ncore, since they have stated that they have listened to D-amps "which people were raving about and which were said to be completely transparent", and apparently they've been doing this until recently also. This seems to me to a good description of the reactions to Ncore (and possibly Anaview and/or Icepower some years ago). Some of the engineers who are involved in LTS claim that ordinary measurement methods fail to capture some of the distortion that even the best class d may introduce. I have no way of knowing whether this is true or not. Beyond my paygrade

As to subjective listening of ncore vs others; it is what it is. I don't think there's any reason to think that objectivists are less prone to listening bias than subjectivists. If an objectivist knows that amp A "is completely transparent", it might bias his listening in the same way as a subjectivist gets biased when listening to amp B after reading that it "provides rich and deep layers of musicality".

 

[Blumlein 88]

Firstly, don't let the perfect be the enemy of the good. If someone has the totally foolproof, unassailable dead bang certain amplifier test method I am all ears.

So what I call a series amplifier test is more or less the same method as the Swedish AES is using. You could call it a straight wire bypass comparison test.

You place a second amplifier between the source and the power amp which is powering speakers. At the output you load it so it can be played at its normal power level. Ideally with a group of components that accurately simulate your speaker load. You tap the result off that load, using a resistor divider network, you lower the level so that the output of this is exactly unity gain with the input. It is not difficult to do this with a few hundred ohm resistor load which will not effect the amplifier output, and yet will have a very low output impedance not to effect frequency response. You can switch between interconnect for source and power amp or source, amplifier under test, and power amp. A perfect amplifier would sound identical either way.

I can think of a a half dozen possible fringe issues that haven't been mentioned by anyone. Of course the main one is whether or not the amplifier powering the speakers is itself transparent. The way I initially dealt with that was to take two amps and reverse position of them. Compare the sound this way. I usually though not always found one amp would convey the same sound regardless of which position it was in. This was your less transparent amp. It had a coloration or character which the opposite amp was revealing as the opposite amp had lesser character or coloration. The opposite amp might not be perfect, but whatever color it had was well below the amp signature that could be heard.

Now sometimes two amps would result in neither being transparent to the other each having some area that had its own sound. In such a case neither amp is transparent. I don't know how the Swedish AES has dealt with this issue.

The other big issue would be that an amp can be transparent with speaker A, but might not be with speaker B and vice versa. Presumably a completely unassailable amp would be transparent with any speaker connection.

It has been some years since I did any of this. My speakers at the time were Quad ESL63's. I found only one amp which was completely transparent. Now I didn't test dozens of amps. More like one dozen. I found a Spectral DMA50 to be straight wire with gain good. You could put it in the middle position and it sounded in no way different than interconnect. You put it powering the speakers and hear every other amp had a sound vs straight wire. I didn't have a pair of those to series, but I did feed one channel into the other and listened in mono. I could not hear that it had any sound.

Initially I did this test with a simple group of power resistors not trying to mimic the loudspeaker load. Surprisingly even this way most amps had a signature. The Swedish AES does this test with simulated loudspeaker load sighted, and then blind. Unless there are new results, they had one large Audio Research solid state amp pass the sighted portion as being transparent. It however was detectable in blind testing. They had a large Bryston come close, and they figured out why it wasn't quite transparent. They suggested to Bryston a change to the input circuit. Bryston made the change and that amp became the first to pass their sighted and blind test as not having a sound signature. I have heard those amps in a very high quality system, and they are indeed superlative amplifiers.

I like this test because firstly it is very clearly very discriminating. If it wasn't an effective test you wouldn't get these results showing differences. Secondly it is a listening test. Some people will never be convinced by any bench test results. This is a listening test.

It is my opinion this could be simplified by tapping the signal at speaker binding posts, reducing the voltage level with resistors until it is safe to feed it into a quality ADC. Then record the result and use that for switching between amp and no amp on the reference system. I haven't confirmed with actual listening this would be transparent itself though I believe it would.

 

[Bjorn]

Thanks Bjørn, didn't know about their listening room. If anything, that could lead to non-transparent devices passing their tests, I guess.

Concerning Ncore: As I said, they don't want to reveal which amps they have looked at. But I have a strong suspicion they have listened to Ncore, since they have stated that they have listened to D-amps "which people were raving about and which were said to be completely transparent", and apparently they've been doing this until recently also. This seems to me to a good description of the reactions to Ncore (and possibly Anaview and/or Icepower some years ago). Some of the engineers who are involved in LTS claim that ordinary measurement methods fail to capture some of the distortion that even the best class d may introduce. I have no way of knowing whether this is true or not. Beyond my paygrade

Well, hard to say but I would pass a judgement on simply thinking they have tested it.

I've been a subscriber to the magazine for years and have never seen any reviews of Ncore amps. I mentioned Ncore to them, can't remember exactly when but it might have been two years ago, and asked if they could test it. They hadn't heard about it before, thanked me for the idea but I never saw any tests. And why wouldn't they write about it if they have tested it?

As to subjective listening of ncore vs others; it is what it is. I don't think there's any reason to think that objectivists are less prone to listening bias than subjectivists. If an objectivist knows that amp A "is completely transparent", it might bias his listening in the same way as a subjectivist gets biased when listening to amp B after reading that it "provides rich and deep layers of musicality".

It's based on measurements, not subjectivist listening tests. I would say that's very different than typically listening bias.

If LTS has done the before/anfter test and it failed and they can point to what measurements are revealing it, I'm all ears and would be very interested to hear it. But until then, it becomes only thoughts.

Personally and after seeing a lot of measurements of Ncore, I highly doubt it could fail their test. But I could see Icepower, Pascal and several Anaview amps not pass it.

 

[amirm]

Would be interesting if others with the technical skills to conduct before/after tests could put the Ncore under the bench: is it transparent or not?

It is on my TODO list to create a setup like theirs to perform such tests. It requires a speaker emulator load and then a voltage divider.

 

[sergeauckland]

It is on my TODO list to create a setup like theirs to perform such tests. It requires a speaker emulator load and then a voltage divider.

Amir,
If you have any circuits for a loudspeaker emulator, I'd love to have one. It's fairly easy to devise an RCL network for the 'typical' loudspeaker, but one that generates back-emf has been beyond me so far.

S.

 

[amirm]

Here is the LTS setup by the way. https://audiosciencereview.com/foru...first-watt-abx-facepalm.766/page-3#post-37815

I am trying to find the link to Ing's page where he described the dummy speaker load....

 

[amirm]

Amir,
If you have any circuits for a loudspeaker emulator, I'd love to have one. It's fairly easy to devise an RCL network for the 'typical' loudspeaker, but one that generates back-emf has been beyond me so far.

S.

Our posts crossed. I have been searching but so far, no luck. There was a nice write-up for it in English even. But I can't remember where. I will keep searching.

 

[amirm]

Wow, that took a lot of work but i found it: http://www.sonicdesign.se/amptest.htm

 

[oivavoi]

It is on my TODO list to create a setup like theirs to perform such tests. It requires a speaker emulator load and then a voltage divider.

From me to you Amir:

 

[Thomas savage]

From me to you Amir:

You both smell of fish, I’m not hugging either of you here...

 

[oivavoi]

It's based on measurements, not subjectivist listening tests. I would say that's very different than typically listening bias.

I think it's the same thing, really. Sean Olive showed in his listening tests that the Harman employees always preferred their own high end speakers over the speakers of the competitors - but this preference was much stronger when the comparison was done sighted. Unsighted, the mean preference for the Harman speakers was subsantially diminished. Similarly, I would assume that the preference for Ncore amplifiers among objectivists to be substantially stronger than it would have been absent the knowledge about how it measures. It may be that people would have preferred Ncore anyway. But I'm pretty sure that the knowledge that it measures better than most other amps conditions people to like it even more.

Also: Assume that the LTS engineers are right that there is something in class d amps that don't show up in measurements (because of the filter that gets applied to measurement gear in order to be able to measure class d, as I've understood it). Assume that this becomes known in systematic tests. I am very sure that people then will start to "hear" these things, which they didn't hear before.

This is all speculation, of course. But I think we should be honest enough with ourselves to acknowledge that listening bias goes in all directions. It's not only about the flowery language of the audiophools, listening bias can also arise from knowledge about measurements and objective performance.

 

[sergeauckland]

Wow, that took a lot of work but i found it: http://www.sonicdesign.se/amptest.htm

That's interesting, thanks. I wonder how much difference the lack of a back-emf generator would be in practice. I really don't know. I did wonder if taking a 'normal' difficult 'speaker crossover and substituting fixed resistors for the drivers as the circuit you linked to would be good enough an emulation.

S

 

[oivavoi]

@amirm , you might contact this guy and try and convince him to give you some tips for setting up the gear: http://ki.se/en/people/svagra

I believe he has been in charge of their testing procedures for the last years.