Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: 'Differenztonfaktor' and tube amp superiority (Read 8546 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

'Differenztonfaktor' and tube amp superiority

There has been a project at the technical university of Berlin, where they have tried to build the best tube amp they could think of in an engineering project. The signal path is "all tube" and the necessary voltage regulation discrete. It measures quite well and they have asked themselves why everybody they showed it to was so excited about its sound. Discrete amps would also often measure well but not cause such enthusiasm.

As engineers they couldn't stop until they thought that they had found some measure capable of explaining the difference. The claim that their tube amp has got an Differenztonfaktor of 0,002% while already common discrete pre(!) amps would have over 180 times as much (0,366%). Their theory is that amps are too often built around single wave test singals (like 1kHz sweeps), but that their amp excels at real life multitude wave content and that the Differenztonfaktor is the appropriate metric to describe this.

What do you all think of this?

It looks a lot like intermodulation distortion to me. The effect of the latter is generally not described as important as they claim Differenztonfaktor to be.

Links:

1. Black Cat 2 Project Page (German)
2. Definition of Differenztonfaktor at Wikipedia (German)

English robot translation of 1.
English robot translation of 2.

'Differenztonfaktor' and tube amp superiority

Reply #1
Quote
Im Rahmen der angesprochenen Diplomarbeit wurden auch vergleichende Blind-Hörtests mit anderen, kommerziellen, Verstärkern durchgeführt, bei denen die "BLACK-CAT"-Verstärker sehr gut abschnitten.

They did blindlistening tests but don't mention the conditions and the exact findings, only that it "did prety well".

Sounds interesting to say the least, but on some parts it reads like the typical audiophoole magazine.

'Differenztonfaktor' and tube amp superiority

Reply #2
What do you all think of this?

It looks a lot like intermodulation distortion to me.

1. Black Cat 2 Project Page (German)
2. Definition of Differenztonfaktor at Wikipedia (German)

English robot translation of 1.
English robot translation of 2.


Yes, I hadn't heard of "differenztonfaktor" before but after reading the link I'd say it's basically just a more specific IMD measurement. You see when people quote IMD figures they often express the results as the total harmonic power over the fundamental power (of the two sine waves combined). I know for a fact that this is how RMAA (rightmark audio analyser) does it. The problem with this is that it's really bundling the normal harmonic distortion and the intermodulation products harmonics togther, so if there's harmonic distortion it also appears as a component of the IMD results. I assume they just do it this way because it's easy, you only have to measure the total RMS output and the RMS output at the two signal frequencies and you can deduce the rest mathematically.

To me it looks like with "differenztonfaktor" they're just being more specific as in measuring the ratio of only the intermodulation product harmonics power to the total power.

'Differenztonfaktor' and tube amp superiority

Reply #3
But aren't harmonic distortion and intermodulation distortion both basically the result of the same thing, non-linear response of the amplifier, just measured differently? Or am I missing something?

'Differenztonfaktor' and tube amp superiority

Reply #4
I did have a longer response, but I think most of it was covered in the time I was writing it and getting sidetracked away from my PC, so I deleted it for brevity.

I did think of a couple of things:

1) The difference in sound might just as well be down to the interaction of the output stage of each amplifier with the crossover and speaker's complex impedance (reactance and resistance). It doesn't appear they've eliminated such effects. Measurements made across a 10 ohm resistor aren't necessarily representative of driving a real speaker load.

2) To isolate and test the perceived quality of intermodulation distortion magnitude couldn't they use a digital source with DAC, say at 24-bits, and apply very simple DSP to the test signal. For example for each input sample, I (normalized to the -1...+1 scale), create an output sample O, where:
O = ½(I + a.I²)

The half is there to prevent clipping. I guess dividing by (1+a) or something would be better for level matching.

Vary a and audition the results against a=0 (no mixing) through a very clean amplifier. Possibly use good headphones instead of a speaker with crossover to help eliminate reactance effects. Being digital it's possible to pre-process the same music samples two ways, level match, then run ABC/HR or similar. That way one could approximately establish the audibility threshold for factor a and by measurement, the equivalent "differenztonfaktor"

This method acts like using a mixer (non-linear element that acts as a multiplier) to generate sum and difference frequencies, then attenuating them to limit the effect. You can find out more about this using google. It's widely used in superheterodyne radio receiver designs where it's desired. In audio amplifiers, it's considered undesirable distortion, at least at levels where it's audible.
Dynamic – the artist formerly known as DickD

'Differenztonfaktor' and tube amp superiority

Reply #5
Would it be possible to take a fairly low-performance amplifier, add a DSP that "learns" the amplifier's transfer function, complete with the effect of the speaker's complex load impedance, and then compensate for it?

Also, could you apply this DSP to a tube amp, have it learn that amplifier's characteristics and then replicate them on a SS amp?


'Differenztonfaktor' and tube amp superiority

Reply #7
My point was that it hardly seems necessary to try to measure just the IM distortion and not the harmonic distortion if they both just basically measure the non-linearity of the ssytem.

'Differenztonfaktor' and tube amp superiority

Reply #8
But aren't harmonic distortion and intermodulation distortion both basically the result of the same thing, non-linear response of the amplifier, just measured differently? Or am I missing something?


Right, THD and IM are both the consequences of nonlinear distoriton.

However, there are many variations in how one measures THD and IM.

The problem I have with THD measurements is that they are questionable at high frequencies in band-limited analog systems. The harmonics fall out-of-band and can easily be lost. Then you get an optimistic view of nonlinear distortion at high frequencies.

I personally look the hardest at 2-tone IM for swept tones when they are available. Someone mentioned RMAA and yes RMAA produces this measurement. At times it has been questionable for those situations where the tones are close to each other.

To review, the Black Cat people seem to be looking at two-tone IM, for two tones that are generally pretty close to each other.

On quick review, the amp itself seems to be a farily ordinary design with lots of open loop gain, loop feedback,  and some lag compensation, perhaps to maintain stability at high frequencies given that there is an output transformer in the circuit.

The pot in the feedback loop makes the distortion performance of the amp dependent on how that pot is adjusted.



'Differenztonfaktor' and tube amp superiority

Reply #9
My point was that it hardly seems necessary to try to measure just the IM distortion and not the harmonic distortion if they both just basically measure the non-linearity of the ssytem.


In the days of tubes, there used to be lengthy discussions about the relationships between THD and IM measurements.

Note that there are two common IM "single number" specs - SMPTE and CCIF.  SMPTE were motion picture engineers and they had a curse called optical sound that was high on their agenda. Their test had a LF tone and a HF tone.

If you are interested in SQ, you care most about whether or not the equipment has low nonlinear distoriton at all frequencies of interest, which neither a single IM nor a single THD measurement will give you.

A plot of swept IM with the test frequencies close to each other seems to me to be the most preferred. It would probably be just a compuational chore to use that to get a good estimate of whatever THD number you were interested in.

CCIF was a measurement  of high frequency two-tone nonlinear distoriton, which is probably not the only thing to look at if there were transformers in the signal path because transformers can have problems with NL distortion at low frequencies (or not).

'Differenztonfaktor' and tube amp superiority

Reply #10
BTW, the Black Cat people are no tube fundamentalists. In a 2008 presentation (PDF on the linked page), Henry Westphal, one of their developers told the audience that low difference tone factors could also be achieved with good SS designs. They consider it much harder to accomplish with SS, though, and an intrinsic feature of proper tube based designs (even simple ones).

'Differenztonfaktor' and tube amp superiority

Reply #11
BTW, the Black Cat people are no tube fundamentalists. In a 2008 presentation (PDF on the linked page), Henry Westphal, one of their developers told the audience that low difference tone factors could also be achieved with good SS designs. They consider it much harder to accomplish with SS, though, and an intrinsic feature of proper tube based designs (even simple ones).


I'm willing to stipulate the fact you presented, but your analysis of it is something else.

I wouldn't call people who begrudgly lower themselves to come anywheres near agreeing with the well-known fact that SS amps generally have far less distortion than tubed amps, "no tube fundamentalists".

IMO, it takes a tube fundamentalist to seriously consider using tubes in audio in 2009, except as marketing tools.

'Differenztonfaktor' and tube amp superiority

Reply #12
IMO, it takes a tube fundamentalist to seriously consider using tubes in audio in 2009, except as marketing tools.


Isn't that position itself fundamentalistic? The part costs of that amp project are much lower than many commercial SS amps. It looks hot and measures extremely well while even being relatively simple in its circuit design. If their claims about 'Differenztonfaktor' were true, as written backed by DBT, and one would need quite some effort to accomplish the same with SS, it doesn't sound too fundamentalistic to me. (I'm still trying to find their researcher's diploma thesis).

This thread has neither seen any substantial statement for nor against the 'Differenztonfaktor' claim, yet.

'Differenztonfaktor' and tube amp superiority

Reply #13
Also, could you apply this DSP to a tube amp, have it learn that amplifier's characteristics and then replicate them on a SS amp?


Of course.  Theres nothing special about any particular kind of distortion.

'Differenztonfaktor' and tube amp superiority

Reply #14
You could 'fix' an SS amp's THD signature like the following (right), for example, by digitally adding harmonic distortion at k2 (left):



But a form of inverse distortion, if your amp has audible flaws, would be very hard to accomplish, if at all possible.

'Differenztonfaktor' and tube amp superiority

Reply #15
IMO, it takes a tube fundamentalist to seriously consider using tubes in audio in 2009, except as marketing tools.


Isn't that position itself fundamentalistic?


No, it is a demonstrable fact.

Quote
The part costs of that amp project are much lower than many commercial SS amps.


Cheaper to build than SS Amps with comparable power output and bandwidth?  I don't think so.

Let's see your Bill Of Materials.  Usually, one can just go out and buy a superior SS amp assembled and tested for less than the parts cost of a tubed amp.

Quote
It looks hot and measures extremely well while even being relatively simple in its circuit design. If their claims about 'Differenztonfaktor' were true, as written backed by DBT, and one would need quite some effort to accomplish the same with SS, it doesn't sound too fundamentalistic to me. (I'm still trying to find their researcher's diploma thesis).


Differenztonfaktor is a measurable parameter. If it is the determining factor, that measurements of  it would be all that we need to compare.

Ever do a DBT on a good SS amp, comparing it to the proverbial straight wire (plus a simple attenuator)?  It's about as much fun as watching paint dry.

Quote
This thread has neither seen any substantial statement for nor against the 'Differenztonfaktor' claim, yet.


Then let me speak for it. If it is as I discern swept two tone IM with the two tones relatively close in frequency, then it is a great addition to a measurement suite for power amplifiers. Not the be-all and end-all, but a great parameter to use to search out possible  substandard performance.

'Differenztonfaktor' and tube amp superiority

Reply #16
You could 'fix' an SS amp's THD signature like the following (right), for example, by digitally adding harmonic distortion at k2 (left):



But a form of inverse distortion, if your amp has audible flaws, would be very hard to accomplish, if at all possible.


In principle a wide range of deterministic distortions, both linear and nonlinear, can be compensated for using an appropriate compensator. In practice, you can't thoroughly cancel the distoriton out because real-word compensators aren't perfect.

Of course we use linear compensators to correct frequency response all the time - its called pre-emphasis and de-emphasis in the case of LPs.

If you can accurately simulate the non linear distortion of an amp, which you would have to do to add it in to a clean amp, then you can invert it and use it cancel it out of a dirtry one with pretty good effectiveness.

I've actually done this in the design of an audio generator with residual distortion in the 0.003% range. I ended up with mostly second order distortion due to the stability control device that I used. I created a simple circuit that created broadband second order distortion that was out of phase. When installed, the nonlinear distortion compensator cancelled out about 3/4 of the generator's distortiion, taking it down to about 0.001%.

'Differenztonfaktor' and tube amp superiority

Reply #17
Quote
The part costs of that amp project are much lower than many commercial SS amps.


Cheaper to build than SS Amps with comparable power output and bandwidth?  I don't think so.


If you arbitrarily limit the issue to only two boundary conditions for the sake of your argument, yes. If this amp really excelled at (hypothetically) perceivable qualities, that only very sophisticated SS designs are able to deliver, then maybe no.

Differenztonfaktor is a measurable parameter. If it is the determining factor, that measurements of  it would be all that we need to compare.


So have you actually done exactly these measurements? It could be that a discrete design isn't cheap when you want to optimize that specific parameter. Without having had our hands on it, we don't know. However, somehow you act like you already do. The mere fact that this is a measurable parameter should have been out of question.

Ever do a DBT on a good SS amp, comparing it to the proverbial straight wire (plus a simple attenuator)?  It's about as much fun as watching paint dry.


In theory, if DTF really was a relevant measure and many amps failed at it, then the headphone/power amp, that you have used to amplify the test signal coming out of the wire respectively test unit, could have masked that the test unit also has had this flaw.

So straight wire tests, which you have conducted in the past, may have been affected by an unmeasured parameter (DTF), that you didn't know back then.

I really don't know what to think about the Black Cat claims, yet. Maybe they really are right and comparable SS designs (including DTF specs) are not cheaper than their tube based. Or they totally overrate this parameter (or its true status among common SS amps) and it is the same kind of snake oil many from that scene have tried to sell for years. I just wanted to ask for some opinions or experiences.

'Differenztonfaktor' and tube amp superiority

Reply #18
Quote
The part costs of that amp project are much lower than many commercial SS amps.


Cheaper to build than SS Amps with comparable power output and bandwidth?  I don't think so.


If you arbitrarily limit the issue to only two boundary conditions for the sake of your argument, yes. If this amp really excelled at (hypothetically) perceivable qualities, that only very sophisticated SS designs are able to deliver, then maybe no.


Sophisticated SS designs are not necessary for sonic transparency. 

Quote
Differenztonfaktor is a measurable parameter. If it is the determining factor, that measurements of  it would be all that we need to compare.


So have you actually done exactly these measurements?


Of course not, but one need not be that exactly close to have seen representative results.

Quote
It could be that a discrete design isn't cheap when you want to optimize that specific parameter. Without having had our hands on it, we don't know. However, somehow you act like you already do. The mere fact that this is a measurable parameter should have been out of question.


What I know is that once power amps get sonically transparent they their SQ can't be meaningfully improved upon. Sonic transparency is much more difficult and much more costly in a tubed design.

Quote
Ever do a DBT on a good SS amp, comparing it to the proverbial straight wire (plus a simple attenuator)?  It's about as much fun as watching paint dry.


In theory, if DTF really was a relevant measure and many amps failed at it, then the headphone/power amp, that you have used to amplify the test signal coming out of the wire respectively test unit, could have masked that the test unit also has had this flaw.


Your argument fails because it presumes that amplifiers that are sonically transparent are somehow difficult to obtain. It fails again because it presumes that there must be an active stage after the bypass switch.

Quote
So straight wire tests, which you have conducted in the past, may have been affected by an unmeasured parameter (DTF), that you didn't know back then.


STF is not a new parameter, its very closely related to twin-tone IM tests where the tones were relatively close. The familiar form of the CCIF IM test spaces the tones at 1 or 2 KHz. DTF appears to space them at 700 Hz. 300 Hz isn't going to change the world. The wider spacing makes it a tad easier to obtain greater sensitivity. There's nothing magic about 700 Hz spacing versus 1,000 Hz spacing.

Quote
Or they totally overrate this parameter (or its true status among common SS amps) and it is the same kind of snake oil many from that scene have tried to sell for years. I just wanted to ask for some opinions or experiences.


For example, when I use the cost argument on people who actually build tube amps, there is little controversy. They pay the bills, they know.

For example, a Hammond OPT for an amp like this is about $130 the cheapest retail source I cab find online.  SS amps simply don't need OPTs or anything like them, so the OPTs are 100% overhead. I can buy a whole assembled and tested 120 wpc SS amp for less than the cost of just the OPTs for a 60 wpc tubed amp. The power transformer for the SS amp is also cheaper because the amp is more efficient.

There's nothing magic about discrete output stage SS power amps. The SS receiver in my main system is >100 wpc @ 8 ohms, all-discrete output stages, and cost less than $100.

'Differenztonfaktor' and tube amp superiority

Reply #19
<duplicate post deleted by author>

'Differenztonfaktor' and tube amp superiority

Reply #20
Sophisticated SS designs are not necessary for sonic transparency.


Transparency against what?


In theory, if DTF really was a relevant measure and many amps failed at it, then the headphone/power amp, that you have used to amplify the test signal coming out of the wire respectively test unit, could have masked that the test unit also has had this flaw.

Your argument fails because it presumes that amplifiers that are sonically transparent are somehow difficult to obtain. It fails again because it presumes that there must be an active stage after the bypass switch.


You have to have an amp somewhere in your test setup that is not the tested amp, wether integrated into your source, before or after the bypass switch doesn't matter. Else you wouldn't be able to generate the energy to hear anything from your straight wire bypass. And if that amp is flawed with a defect not known at that time, bypass and tested amp may sound identical, since both reference and tested amp may share the kind of flaw not known at that time.

Sonic transparency is never only a statement about amp vs. reality (or vs. a plain wire alone). At best it is a statement amp vs. best available amp at the time of testing (implied to power the bypass wire).

I really do appreciate the value of ABX testing for engineering and evaluation. However, sometimes one can get the impression you stretch it a little. You present sonic transparency as an universal fact, what it isn't (the claim is constrained by the equipment and knowledge available at the time of testing), but mean your experience, which I am definitively interested to hear.

'Differenztonfaktor' and tube amp superiority

Reply #21
Sophisticated SS designs are not necessary for sonic transparency.


Transparency against what?


In theory, if DTF really was a relevant measure and many amps failed at it, then the headphone/power amp, that you have used to amplify the test signal coming out of the wire respectively test unit, could have masked that the test unit also has had this flaw.

Your argument fails because it presumes that amplifiers that are sonically transparent are somehow difficult to obtain. It fails again because it presumes that there must be an active stage after the bypass switch.


You have to have an amp somewhere in your test setup that is not the tested amp, wether integrated into your source, before or after the bypass switch doesn't matter.


Exactly. A practical approach is to make your signal source an audio production grade audio interface that has as good performance as it gets, and can also drive high quality headphones. In my case that was a LynxTwo.

Quote
Else you wouldn't be able to generate the energy to hear anything from your straight wire bypass. And if that amp is flawed with a defect not known at that time, bypass and tested amp may sound identical, since both reference and tested amp may share the kind of flaw not known at that time.


You can play the "and if that was flawed" game until the cows come home... ;-)

How about this? Play and re-record the same audio signal through a computer audio interface through as many generations as it takes to float your boat, like 20. Then ABX the first generation with the last. If the audio interface masks or distorts enough to matter in one or two generations, after 20 or however many iterations of that distoriton adding up and multiplying, hearing it should pretty easy. 

If your first generation versus umpty-umth generation recordings are indistinguishable in an ABX test, it seems to me that you can treat that audio interface like a trusted tool within the bounds of your tests.

I've done this with a number of audio interfaces, and Ethan has done it with a low cost consumer interface "Sound Blaster". His test files are on his site, while mine were on the now-departed www.pcabx.com web site.

Quote
Sonic transparency is never only a statement about amp vs. reality (or vs. a plain wire alone). At best it is a statement amp vs. best available amp at the time of testing (implied to power the bypass wire).


That's the ideal. Reality is that all of the recordings that we listen to have been through a fairly complex recording process that is unknown to most. Recordings are like sausages, idealists should never watch them being made. ;-)

Quote
I really do appreciate the value of ABX testing for engineering and evaluation. However, sometimes one can get the impression you stretch it a little.


True, I'm not an idealist. However, a lot of fear of stretching gets dispelled by familiarity. Drive long enough over enough roads and you know which corners to cut, and which to avoid.

Quote
You present sonic transparency as an universal fact, what it isn't (the claim is constrained by the equipment and knowledge available at the time of testing), but mean your experience, which I am definitively interested to hear.


I don't think that sonic transparency is universal. Iterative testing allows quantifying transparency, even when a component is transparent over one generation. I think you will find that ABXing 5 iterations through a component that you don't want to trust for one to be quite enlightening.  Humans are naturally risk adverse, which is why Insurance is a business.

Don't trust my experiences, by all means get some of your own. For example, Ethan's little experiment with iterated looping of a cheap audio interface is something that anybody with a modern computer can do with a $5 budget.


'Differenztonfaktor' and tube amp superiority

Reply #23
Which way do you want to go?


Well, honestly outside and soon!  Spring is finally locking in after some ups and downs here. I have spend too much time here on HA over the last few weeks waiting for that. It was quite entertaining now and then and sometimes also very insightful. Thanks for your comments. Hold the fort!

rpp3po

 
SimplePortal 1.0.0 RC1 © 2008-2021