JLH - Pass - Krell clone - Cresendo what sounds best

[Upupa Epops]

More philosofical words, but only little actions

Gentlemen Kees, Wingfeather and everybody who can - if you can seriously discuss about amplifier problems, write on my email, I try to help you, when I will know answer.

 

[Jan Dupont]

This thread in general

So fare from reading all the replies to this thread (which could have been very interesting) originally started by Kess, I can conclude as follows:

Personall oppinions (like those from Upupa Epops, Wingfeather, PMA and a few others - me included), that sound is determent by many other things than "magic caps and transistors" are not very welcome on diyaudio.

When I go back and compare to earlier threads, it clearly shows that it is the same members that (time after time) uses up the
space to chritise (read: ) other members ideas and inputs without contributing with anything positive.....

We all have our personal believes (thanks for that), and I find it interesting to read the posts from the (sadly, but obviously) very few really experienced people in here. I may/may not agree with them, but I do not just post a reply without having a quastion or a constructive input.

Now you (the majority) can put me on your "hate list" togehter with the others, who tries to create a educational and interesting debate.

My input to this thread is to support the oppinions from Upupa Epops and Wingfeather among others, that "good sound" is individual, and that many factors (symmetri and PCB-designs) has to be taken into consideration.

After this "input, I better go

 

[DrG]

I'm not sure you read exactly what Upuppa and Wingfeather said. In summary their contentious points were:

Upuppa: can predict from circuit diagram how an amplifier will sound...

Wingfeather: amps that measure the same (THD etc) will sound the same...

Didn't have anything to do with individual perceptions of what sounds "good". And this was part of the advice they were offering to Kees.

 

[Jan Dupont]

I read exaly what they read

As Upupa I feel I can predict the "sound" from the schematics.

Wingfeather also wrote:

For example, from looking at a design it is fully possible to tell whether there will be excess second harmonics. Second harmonics have a very distinct sound, synonymous with Valve Sound.

I agree to this.......

More from anyone one we are at it????

 

[Peter Daniel]

More from anyone one we are at it????

Mr. Dudek wrote:

Crescendo - objective much more clear trebles and sujective less bass - it is ideal amp for jazz, folk and classical music. Krell - subjective more bass and not so clear trebles like Cresc. It is amp for rock and pop. JHL : similar like Krell on bass, treble more distorted (quasicomplementary configuratoin is not ideal for low distortion ) - using the same like by Krell. Pass : here is many fans of Mr. Pass, becouse its easy to build this amplifiers and they maybe will be object, but this amplifiers have big "sound coloration", especialy types for DIY

He only commented on tonal balance of the amps and didn't mentioned anything about other qualities like soundstaging, resolution, immediacy, liveliness, pace and rythm, airyness. While tonal balance can be easily changed by substituting different components, the other qualities I mentioned, are more difficult to control and depend not only on a given circuit, but also on chassis design, materials, layout, wiring, physical size and so on. While I agree that certain aspects of the sound can be predicted by looking at the schematic, I would still argue that not everything.

 

[DrG]

Well in that case, my next question would be "How do you explain how the same circuit/boards can sound different with different types/makes of parts?" To which you *must* answer: "They sound no different at all"... if what you say is true.

So, perhaps you are the most valuable man in audio design in the entire world. You seem to have "golden eyes".

Hell I don't know... but I wish I had that gift, as I'm sure do hundreds of other amateurs and professional audio engineers.

I'll believe my ears. You can believe your eyes or whatever you please. I really don't care...

 

[Jan Dupont]

Peter Daniel:

While I agree that certain aspects of the sound can be predicted by looking at the schematic, but not everything.

Agree But you can tell if the amp will be best suited for
Jazz, rock or classics, and you will also in some degree could tell
the "sonic" result of the amp.......

If you study the schematics for different amps, and compare to the prefered use for each amp design, you will see that almost all amps for a specific use share a lot of common circuitry......

Some designs are better than others, but you are able to classify them........

 

[Peter Daniel]

I pretty much agree with that too

 

[Jan Dupont]

In order for everybody to participate, I have started a new thread:
Controversial topic - Components vs. Design

Happy fighting

 

[Kees]

Dear All,

I posted my design in the hope to get some feedback from people who tried similar topologies:
1- The output stage; triplet design instead of darlington
2- Differential stages with double transistors
3- Current mirrors with 3 transistors

Please do NOT mind the transistor part numbers. Remember that's a result of what models are availble. I just have to start somewhere.

The design is focused on more local linearity, therefore less loop gain is needed (less negeative feedback).

Are there any people who tried the same?
More important: How does it sound?

 

[Wingfeather]

I know I'm still treading on thin ice here, but I want to make it clear that I do not think that if two amps have 0.1% THD they will sound identical. Such an idea is just like saying "well both of these puppies weighs 12.5kg exactly, so I got no idea how they can sound different"
A spec like THD is pretty much useless by itself, though I believe if it's very low then the amplifier can be considered distortionless.
Other factors exist with this as well. The THD specification is given for a single set of circumstances, such as an 8Ohm resistive load at 1kHz. The THD spec also lists total distortion, but not the spectral distribution of it. It's widely agreed that high-order distortion is much more upsetting to the listener.
If two amps have been measured at exactly 0.1%-second-harmonic-and-nothing-else putting 1W into an 8R load, 1kHz then I believe that yes, the two amps will sound identical if they're playing a 1W, 1kHz sine wave into 8R. However these two amps will most likely show different figures if you load them capacitatively, or if you alter the frequency, or both, or anything else.
My belief is though, if a complete set of numbers for two amplifiers matches into a given load over the entire audio spectrum, then into that load the two amps will sound indistinguishable. The same goes if overall THD is below the 0.3% 'threshold of discernability', regardless of the distortion's distribution.
Perhaps I'm being too scientific about it, but it's just way I think.

Thanks to ACD for some backup!

 

[sss]

hi kees
i tryed the triple configuration and it gives lower output impedance and distortion (if desined correctly) but its harder to stabilize it

 

[DrG]

My belief is though, if a complete set of numbers for two amplifiers matches into a given load over the entire audio spectrum, then into that load the two amps will sound indistinguishable.

Okay... which numbers? THD? 2nd harmonic content? Even vs odd harmonic contributions? IMD? Damping factor? OL gain? Feedback factor?

Do you see the problem? Assuming 2 amps measure absolutely identical in terms of every last conceivable parameter then yes, I agree... except to say that the only way this is possible is if you're dealing with 2 identical amps!

The same goes if overall THD is below the 0.3% 'threshold of discernability', regardless of the distortion's distribution.

Now this is not true, unfortunately. I can recall doing demos in the days of the MF A1 and cheaper B1 models (I had an A100X in those days...). THD of both was less than 0.3%. In the same room, same source, same speakers the A1 was ALWAYS preferred. Same with the MF P180 and Rotel power amps (forgot model no.). MF was preferred hands down, every time.

 

[PMA]

I know I'm still treading on thin ice here, but I want to make it clear that I do not think that if two amps have 0.1% THD they will sound identical.

A spec like THD is pretty much useless by itself, though I believe if it's very low then the amplifier can be considered distortionless.

The THD spec also lists total distortion, but not the spectral distribution of it. It's widely agreed that high-order distortion is much more upsetting to the listener.

If two amps have been measured at exactly 0.1%-second-harmonic-and-nothing-else putting 1W into an 8R load, 1kHz then I believe that yes, the two amps will sound identical if they're playing a 1W, 1kHz sine wave into 8R. However these two amps will most likely show different figures if you load them capacitatively, or if you alter the frequency, or both, or anything else.
My belief is though, if a complete set of numbers for two amplifiers matches into a given load over the entire audio spectrum, then into that load the two amps will sound indistinguishable. The same goes if overall THD is below the 0.3% 'threshold of discernability', regardless of the distortion's distribution.
Perhaps I'm being too scientific about it, but it's just way I think.

Thanks to ACD for some backup!

Wingfeather,

IMHO most of your assumptions are correct, I can not agree with your conclusion. After having measured (see few spectras on my www) and listened to many many amplifiers I must say that many other factors come into the game. Some of these factors seem quite incredible in a focus of contemporary engineering approach. To conclude - I think that we fail when we measure static and quasi-static distortions. What we should do is to measure dynamic distortions with natural (real) musical signal, preferably directly at speaker's terminals.

Pavel

 

[MBK]

What we should do is to measure dynamic distortions with natural (real) musical signal, preferably directly at speaker's terminals.

Pavel

I would go further: we should measure the actual speaker output. After all we want undistorted sound, not just undistorted electrical signal. Of course at the beginning of the design we want to isolate the tasks into subsystems: source - connection - preamp - connection - amp - connection - speaker - room - listener. Ultimately though we have to integrate the system and here the complexity comes in.

Regarding the string "amp-connection-speaker" as a larger subsystem:

The approach of building the amp standalone to work with "any" speaker / X-O / and driver, will lead to compromises. I think the ideal approach would consider the speaker as a part of the amplifier's feedback system. That means of course active speakers.

In this thread for instance we discuss the interaction of amp distortion with speaker distortion and possible distortion cancellation:

http://www.diyaudio.com/forums/showthread.php?s=&threadid=19030&pagenumber=3

In this thread we discuss feedback of speaker movement parameters to the amp:

http://www.diyaudio.com/forums/showthread.php?s=&threadid=20037

None of this new of course but I think of higher merit than simply discussing amps as if they were enough.

PMA - I know you are reading these above threads so I'm not telling you anything new

 

[PMA]

I would go further: we should measure the actual speaker output.

In case that I understand well, you would like to measure acoustic pressure, not the electric value at the speaker.

 

[MBK]

yes... and I would like the amp's feedback to incorporate the nonlinearities in the SPL generated, not just in the U or I delivered to the terminals. Then you have a chance to get rid of the (usually dominating) driver distortion.

 

[Kees]

SSS,

You say the triplet is harder to stabilize. I assume you mean thermal stability. How did you ensure stability? Just the standard Vbe multiplier, as in my circuit diagram ?

About the sound, can you hear the difference in comparrison to a darlington configuration?

 

[Nelson Pass]

I would like the amp's feedback to incorporate the nonlinearities in the SPL generated, not just in the U or I delivered to the terminals. Then you have a chance to get rid of
the (usually dominating) driver distortion.

And I would like to be declared a national treaure and made
exempt from taxes or death...

You say the triplet is harder to stabilize. I assume you mean thermal stability. How did you ensure stability? Just the standard Vbe multiplier, as in my circuit diagram ?

The triplet (by which I assume your mean Triple Darlington
follower, is a nice circuit which is occasionally required but usually
overused, a choice to which I have to plead guilty in my ancient
past.

The problem of thermal stability is fairly easily addressed by a
number of methods, generous heat sinking being a favorite.

The frequency stability issues are often more complex, but nothing
that experiments with Base resistances and a little CE capacitance
can't map out, assuming some common sense and the willingness
to put in the time.

 

[MBK]

And I would like to be declared a national treaure and made
exempt from taxes or death...

That would be nice indeed...

But it may be easier to achieve power amp feedback that includes the driven device's nonlinearities. And maybe this has all been tried and found impractical or counterproductive. But maybe not so i keep on imagining.... Or please refer me to a source that I don't know of.

I imagine to monitor actual *power* output to the speaker driver, to compare this power output with the *power* that the amp would generate over a resisitive load as standard, and to use the error as feedback in the power amp's loop.

In this way we'd become independent of the load's impedance variations (not just in the frequency domain but most importantly in the nonlinear excursion areas) and would therefore deliver linear power. We're still left with efficiency variations in converting electric power to SPL, and the above makes sense only for a naked driver in an active system. But at least we'd incorporate some useful information the driver can give us about its condition.

Example:

near excursion limit

- the voice coil enters lower magnetic flux areas. So, we get
- reduced impedance through reduced magnetic flux.
- Voltage drive-> power amp maintains U, reduces I.
- Current drive-> power amp maintains I, reduces U.

In both cases power delivery is cut by the power amp's feedback. We don't want that do we? Power feedback the way I imagine it would raise both U and I to restore the power delivered according to the amp's ower gain factor. In other words, not voltage gain, not current gain, but power gain. And power feedback.

Impossible?