AKSA 55, 100 - Listening impressions

[jacco vermeulen]

To me it seems that the AKSA has a proven record, and Mr Dean is praised for his after sales skills.
No one is forced to buy anything, and the starting price of the 55 kit isn't that exclusive.
It's a free market, make a better one and compete Mr Dean out of business.

Problem with good beers is that it takes 30 pints before i start barfing, i'm a southern carnival boy.

 

[Tube_Dude]

Not at all. It could even make things worse since Vce of the CCS transistor will be lower, and Cob will then be higher due to the non-linearity.

Adding a resistor to a CCS output , always increase the output impedance of the CCS specially at the high frequencies .

Furthermore, you now have two noise sources instead of one

Not at all. As the bases of the LTP transistors are referenced to ground and the CCS are referenced to the power rail, the resistor will help to isolate the supply noise especially in the high frequencies , that can pass trough the parasitcs capacitances of the CCS transistor (as Forr have already mentioned).

 

[destroyer X]

Be carefull friends, not to broke rules...Pinkmouse is around, and he already moved

and closed a wonderfull AKSA thread.

So...he is serious related his job and obligations.

I personally feel afraid of him, as i asked, under knees, to return the aksa thread alive, and i received no answer from him.

So...my bonds shake when he is around.

He may be a nice guy....but even nice guys can hurt us when doing their job.... example....Dentist!...ahahahaha!

I will take care of my words.

I suggest you to do the same...do not even comment my words, as i can be easy accused as Moutiny...and again....booom....thread closed....please...read and do not comment.

There are more two Aksa users, that are so happy that will open threads, to explain his feelings and subjective analisis about Aksa or will enter and publish here...if they had time enougth...they are busy...you know.....

regards,

Carlos

 

[forr]

Hi Hugh Dean,

Thanks for your reply. Your philosophy is very clear to me.
A comparison can be made with photographic lenses : the sharpest lens is not necessarily the one which gives the nicest pictures. I agree... However if you start with a very sharp lens, there are some ways to slightly degrade things to get niceness. An example : cinematographer Geoffrey Unsworth for the film "Murder on the Orient Express" by Sidney Lumet took the best lenses he could find and then added some special degrading filters wich gave a wonderful overall hazy picture. Among other amplifiers based on the philosophy parallel to yours, I would like to mention one I would have won : the oustanding Nytech, available twenty years ago.

I know Bailey amps and Self writings as well as many, many, other authors, the most important being Self, Baxandall and Perrot. I wonder why you write towards Self with some acidity :
"From what I have researched, Doug Self has made no significant creative contribution beyond Bailey and the public domain. In some areas he is plain wrong, such as his love of the CFP. But he does write well, and his math is very good."

It seems to be a conflict between two persons and/or two conceptions rather than a misunderstood fundamental of electronics from either.
Sure, Self has designed with different goals than yours. His contribution is giant because he has gone as nobody before him in the minute details of every part of the circuit. We can credit him for the invention of a new class A biasing. This is creative. In the contrary, sorry to say, from what I know from your amps is not more innovative than his. I would add as his major contributions : his insistance about the importance of the bias circuit and its physical implementation, as well as recommandations for feedback, ground, and power supply connexions for which Linsley-Hood was completly wrong.

What so with the CFP ? I read Rod Elliott advocates coupound pairs. Your negative comment is subjectively biased, and then debattable. No problem when used in class A, and slightly more efficient than followers. In class B, it requires a very precise bias control for the quiescent current and its distorsion have quite a lot of high harmonic components. A sophisticated way to minimise them would need an additionnal power supply of a few volts, clamped above the main one, and the connection to it of the collector resistor of the driver wich biases the output power device.
My listening experience with CFP pairs or even triples as used by Quad, Crimson, and 1978 Linsley-Hood bipolar is that they are very good for bass, under, say, 1 kHz, because of the high gain local feedback loop. At high frequencies, I never found them offensive to the ear but I may prefer output stages with smoother transitions in the crossvover region. I obtain quite good overall results with the same Linsley Hood architecture but using hybrid power stages (bipolar driver / mosfet output)

Otherwise the general topology used by Self is nothing else, in a robust form with power, than a standard op-amp. An often seen criticism is its use of a Miller compensation cap but yours uses it as well. My experience, which has to be confirmed, just says it's less prone to parasitics than the advance phase one.

My version of Self's amp is the first one published with emitter followers : it adds nothing and lacks nothing bar a bit of own "flavour" which may be the character of too perfect things, therefore amps with no disto. Anybody, commercially uninvolved, heard the Halcro apparatus ?

~~~~~~ Forr

§§§

 

[lineup]

A resistor from differential pair emitters to set current
is in fact not a constant current source.

At +1 volt input the voltage across resistor is increased.

Say the +input transistor needs +-0.1 mA to produce voltage swing in VAS.
And this at +-1.0 Volt input.

Say we use a common emitter resistor 5 kOhm
+- 1.0V/5 kOhm = +-0.2 mA

This means the non-inverting input transistor will use +-0.1 mA
and the 'left-over' will also be +-0.1 mA in inverting input transistor.

Means at +1.0V in, both transistors will have +0.1 mA.

( Using another resistor value, can give +0.1 mA in one, and +0.2 mA or more in the other transistor.
And if current needed to feed VAS-stage is more than the current increase in resistor,
the other transistor will have to go down in current. )


In a constant current source,
if non-inverting increase 0.1 mA, the other will decrease 0.1 mA, to balance.

So when one transistor has +0.1 mA, the other will have -0.1 mA.


This is clearly a difference between using a resistor to feed the input pair.
The sum of current in input pair is changing, because of input voltage.

 

[AKSA]

Hi Forr,

Thank you for your post.

I accept your criticisms in the spirit in which they are intended, but my opinions are my opinions, and I really don't see a need to defend them here. I design amps, but I do not attach ridiculous importance to my opinions. Everyone has them, and they are all different. I'm not an aggressive salesman, so I have nothing to be ashamed about.

You see Doug one way, I see him another. Have you heard this amusing anecdote about two academics arguing? 'The fighting was bitter, and personal, and vindictive, because the stakes were so low'. I'm always reminded of this when I see fights about opinions. There are so many other things in life which are more important........

Cheers,

Hugh

 

[hifryer]

I think the differences between Hugh and Forr / Greg are that the former insist on total objectivity based on the measurements they choose to use, whereas Hugh is objective and does measure but is driven by what he thinks gives the best sound even if this measures less well. He still keeps his distortion measurements below those consided audible by many.

Let's look at some goals. What are we trying to achieve here?

We are not trying to build objective test equipment or medical equipment.

I, for one, aim to to reproduce as much as I can of the subjective musical experience that I get from live performances. This can only be a subjective / emotional thing. I make changes based on this.

It is clear to me that harmonic distortion needs to be reduced below a certain level and that below that lower measurements achieve little and can reduce the fidelity if their achievement causes other effects.

With lots of negative feedback ( as in AKSA and SKA ) a low enough level is achieved by both. In terms of this factor in isolation it doesn' matter if achieved with aid of sources and mirrors or not. So the argument that these techniques MUST be better is invalid in my eyes.

Others have pointed out that there is long chain of imperfect devices between the recording microphones and our ears. We will never totally faithfully reproduce the source experience. What we hear is imperfect in a host of unknowable ways.

We choose the audio components that get us a close as we can to the original experience. These are all flawed.

So people like Hugh and myself listen and experiment with ideas to try to get closer. Sometimes this results in greater simplicity rather than added complexity - so be it. Sometimes some objective measurements deteriorate - so be it. If it works and we dont yet understand why or how - so be it.

Yes, this is totally subjective. But not necessarily a retrograde step. The"improvement " must pass critical evaluation in many system setups and rooms and favorable comment from many listeners.

This is true of the AKSA base design and all offered upgrades. There have been MANY failed changes that held promise but were rejected.

Yes, I am an AKSA fan and recognise that pride of building can colour judgement - but I design and build my own amplifiers which I always hope and wish will blow AKSA away!!! So I am not "one eyed"!!
Unfortunately I have found to date that only my much more expensive efforts can compete. Others may do better.

Hugh Dean has told me that I am his most carping, critical and demanding customer! Tough luck Hugh! I only want to push you to greater heights.

My last comment!! BG's are a PIA!!!!

cheers

 

[forr]

Hi Christer,

Thanks for your analysis of CCS. I've never found a good analysis of CCSs in the technical literrature. Douglas Self does not make any comment about the resistor in series with it in its amps.

I built some real circuits using insertion of a cascode BJT at the current output of a CCS. They were very prone to oscillations. One of the best and reliable results I obtained was exactly the same I saw on a valve circuit designed by Morgan Jones.

The real current source transistor is biased with a LED with no parallel cap (no need, I think, the LED has a low dynamic impedance at all frequencis).

Biasing the second common base transistor (which makes the whole circuit being a cascode) with a second LED, biased in series with the first, oscillates.

Instead, I used a 1 kOhm resistor. If the current through the series circuit (LED + 1 kOhm) is obtained from a stable voltage refered to the power supply, the CCS is very insensitive to external conditions. This is the Morgan Jones's circuit.

I obtained better results, even at high frequencies, with a unique BJT (no cascode) biased by a simple TL431 controling the voltage across the emitter resistor. I attended some misbehaviour at high frequencies but couldn't see it on the scope, which does not mean that this circuit has no problem.

~~~~~~~~~~ Forr

§§§

 

[AKSA]

Forr,

That's a very good idea, thank you! I have used the TL431 in the cathode circuit of a tube. You can use a voltage divider to the adj terminal from either the B+ (as a power regulator) or the plate (as a servo to control tube operating Vak).

Recently I pulled apart a faulty 300W twitch mode supply for my PC. I found no less than four TL431s. These are very good ICs, and ubiquitous.

Cheers,

Hugh

 

[thalis]

"""A comparison can be made with photographic lenses : the sharpest lens is not necessarily the one which gives the nicest pictures. I agree... However if you start with a very sharp lens, there are some ways to slightly degrade things to get niceness."""

Dear Forr,
As I can understand it, Hugh is doing exactly the same thing. By choosing a resistor instead of a constant current source in the differential input ;-))

You say this: Resistor will end up to greater distortion (true), therefore the amplifier is worse, therefore there is a flaw in circuit design.
You also believe that if someone denies these consequences, he believes that having more distortion means better sound, in general.

(I will speak generally about these matters, not just for you, but because I read about these subjects all over the forum)
I think that there is a flaw in this logic, and that the consequences are not true.
Adding this resistor maybe alters another quality for which we haven’t got yet scientific proof and can’t quantify it and measure it.
We can’t deny the possibility of existence of such qualities, for two reasons.
First, because everything refers to our ear and our perception of the musical event. If our ear and brain believe that something sounds better, we have no right to say that it doesn’t know what it hears and thinks, but we have to research further to find why this event occurs. Various distortion figures and various theories about the structure of harmonics etc. have been introduced, just to explain facts, which were considered as facts because of our ears preferences! So, the first thing that someone should do, would be to built this specific amplifier with a resistor in its dif.input, and a second amplifier with a constant source, mirror etc., to see what sounds better. I presume that Aksa has done the test, since it gives no significant rise in cost and since it is something common to work with current sources, and that he found that he prefers his tiny resistor.
BTW, in some other circuit, I have also preferred such a solution, and, in general, I have front-ends and loudspeakers of very considerable cost and quality, and a good experience in music since someday I used to be a concert piano player as well (classical). I agree that perhaps I’m wrong (and Aksa and everybody else with similar experience) and perhaps the circumstances and the rest of my stereo chain where randomly in such a form in which they favoured something poorer which would in other circumstances prove to be such, but I don’t agree that you could be so sure for these other circumstances and so confident about the ‘overall better solution’ for a given circuit, as a rule.
The second reason is this: If your certainties where true, listening to a line stage consisted of 40 pieces AD-825 Op amps in series, well constructed with good pcb work, supply, decoupling caps, well defined input impedances etc. etc., and some appropriate dividing stages so to not operate the amps in unity gain, wouldn’t be very different from listening to only one Op amp. In our simulations and in our oscilloscope and sp. analyser it would be perhaps (correct me if I’m wrong) even better from one (1) TL71, for instance. The truth is it isn’t - in fact, it is a disaster I have built one such stage with just 15 (if I remember correctly) and just couldn’t bear it, though I could bear one TL. Btw, I recommend this test to all.

Our stereos play only some part of the real event. Much of it is lost in our recordings, much of it in our machines as well. In fact, there is a huge gap between what you will hear in a concert hall with 100 musicians, and what your stereo will reproduce. This gap is filled up in our brain.
Though we have in our knowledge specific quantities of which the real event is consisted and could be reconstructed, we can be sure that there are other quantities that we don’t know (if we knew, our reconstruction would be much better). Since we don’t know these, we cannot be sure that amplifier X is better from Z because of what we know of them in technical terms. If this was the case, everyman’s line stage would be an AD-825 per channel or something like that (perhaps 20 of them .

Now I’ll suggest a different way to think about a Stereo system: Its mission is not to present the absolute reality based on our existent knowledge of the various portions of which reality is consisted, since we don’t know them all, but to draw a painting of reality.
A good painter who would have only a red, a yellow and a black paint wouldn’t produce only the red as red, the orange as orange, but the green, brown, grey as something else, but would find ways to paint every colour different from the original but the whole scene still good for us to dream about it. In the same way the engineer that produces good paintings is not the engineer who knows to produce circuits of zero distortions, but the engineer who knows ways (topologies, layouts, etc., working schemes) that produce acceptable listening impressions. If some of them produce some distortion, it doesn’t matter.
I’m sure you are such an engineer as well, but you haven’t realized it yet ;-)

Lastly, I disagree with the way this thread defines the terms ‘objectivist’ and ‘subjectivist’. I could say that subjectivist is he who has such belief in correlation of distortion measurements and quality of reproduction, that he denies that a product that doesn’t perform so well in these measurements can sound better from another that does measure well, since he denies the real objective, which is the ears pleasure - he denies the object of his efforts.



P.S I haven’t heard an Aksa amplifier neither have I seen its schematics

 

[mastertech]

Tube-Dude-

Hi Hugh

And don't you think that making the voicing of the amplifier (listening tuning ) ,the outcome will be only valid for the speakers and the source ( including the recording themself ) used in the evaluation.

And please don't interpret this as a personal attack .

Cheers
--------------------------------------------------------------
Rubbish, there is no such thing as perfect/standard by which
you fine tune an amp, an amp is fine tuned by its designer
using the knowledge he has, and i am not prepared to disclose
knowledge only to be told that all depends on sources and trans
ducers and nothing on amp

Have you ever really listened to a class-a wideband high-speed amp
obviously not!

cheers

 

[AKSA]

MTech,

I'm not quite sure what you mean here, but I'm not sure I agree.

I designed a wideband Class A hybrid amp back in the mid-nineties, and it remains my gold standard.

Class A, like anything else, has its good and bad points.

You need to be careful with categoric statements. Invariably someone proves you wrong!

Cheers,

Hugh

 

[Tube_Dude]

Tube-Dude-

Have you ever really listened to a class-a wideband high-speed amp
obviously not!

In the last 35 years I can't count the number of amps that I have designed and listened , of the various kinds.

-Bipolar germanium class A/B

-Bipolar silicon class A and class A/B

-Mosfets class A and class A/B

-Hybrids

-Tubes , PP and SE

But now , I ask you , have ever really listened to a amp , that in a null test , the residual is only thermal noise?
Is better that you don't listen to it , because it can became addictive.

Cheers.

 

[mastertech]

Aksa-
Class A, like anything else, has its good and bad points.
You need to be careful with categoric statements. Invariably someone proves you wrong!
------------------------------------------------
not a problem...enjoy the music!

 

[Christer]

Adding a resistor to a CCS output , always increase the output impedance of the CCS specially at the high frequencies .

Yes, but as long as we are talking small signal BJTs and currents on the order of a few mA the output impedance for LF is already very high (1/hoe). Since the value of the resistor is limited by how much voltage you can afford to drop, you migh perhaps be able to double the output impedance in the best case. It is a different case for HF, though, but that will typicaly not kick in until you are up in the MHz range.

Not at all. As the bases of the LTP transistors are referenced to ground and the CCS are referenced to the power rail, the resistor will help to isolate the supply noise especially in the high frequencies , that can pass trough the parasitcs capacitances of the CCS transistor (as Forr have already mentioned).

Oh yes. The CCS is one noise source and the resistor is one noise source. 1+1=2, as I am sure they teach in Portugal too (just joking).
I do realize now, however, that you seem not to talk about the noise generated in the tail circuit, but rather about supply rejection.
There I agree a resistor could help somewhat, realistically some 6 dB or so maximum for LF. It will, however, also shift the HF pole, but once again, that will be effects far above the audio band. Incidentally, due to the reciprocity theorem the analysis is exactly the same as for CCS current modulation by varying tail voltage.

So yes, you could improve PSRR by a few dB by using a resistor, but you could improve it much more by using a cascode transistor, or why not just use an RC filter between the rail and the CCS? It only cost a capacitor extra and could give more improvement than the tail resistor.

 

[Christer]

Thanks for your analysis of CCS. I've never found a good analysis of CCSs in the technical literrature. Douglas Self does not make any comment about the resistor in series with it in its amps.

Thanks, but I guess many authors consider it just routine to do a small-signal analysis. It is also easy to play with Spice and see what happens.

I built some real circuits using insertion of a cascode BJT at the current output of a CCS. They were very prone to oscillations. One of the best and reliable results I obtained was exactly the same I saw on a valve circuit designed by Morgan Jones.

Yes, cascodes are known to oscillate sometimes. Did you try a base stopper?

The real current source transistor is biased with a LED with no parallel cap (no need, I think, the LED has a low dynamic impedance at all frequencis).

Yes, the main reason a capacitor is sometimes used is probably to reduce noise from the LED (or whatever). That is moderately effective due to the low dynamic impedance of the LED, so a better alternative might be an RC filter.

 

[Christer]

Hi Hugh Dean,

Thanks for your reply. Your philosophy is very clear to me.
A comparison can be made with photographic lenses : the sharpest lens is not necessarily the one which gives the nicest pictures. I agree... However if you start with a very sharp lens, there are some ways to slightly degrade things to get niceness. An example : cinematographer Geoffrey Unsworth for the film "Murder on the Orient Express" by Sidney Lumet took the best lenses he could find and then added some special degrading filters wich gave a wonderful overall hazy picture.

Sorry, but that is not a good analogy. The photographer is an artist, just like the musician. Just like a violinist can prefer to use a Stradivarius of a Guernari, or a pianist can put chains or other strange objects on the strings to alter the sound, the photographer chan choose to blur pictures, play with depth of field, etc. Those are artistic decisions by the performer.

If you look at a slide photo using a projector, you are reproducing the artwork of the photographer, and then you typically don't want to blur or otherwise distort the picture. You want to see it as the photographer intended it. In the same way a sound reproduction system should (at least in my opinion) achieve an objective acccount and not add to the artistic decisions of the musician. You can, of course, choose to deliberatly modify the reproduction of the photo, or the music, but then you are infringing on a work of art. It would be like buying a painting and then start to repaint parts of it, or to buy a novel an rewrite the parts you don't like.

 

[mastertech]

Tube_dude
But now , I ask you , have ever really listened to a amp , that in a null test , the residual is only thermal noise?
Is better that you don't listen to it , because it can became addictive.
-------------------------------------------------------
enjoy the music dude!!! not null tests

 

[Christer]

Tube_dude
But now , I ask you , have ever really listened to a amp , that in a null test , the residual is only thermal noise?
Is better that you don't listen to it , because it can became addictive.
-------------------------------------------------------
enjoy the music dude!!! not null tests

With they type of "music" most people listen to nowadays, I wish they did listen to null tests.

 

[Tube_Dude]

I do realize now, however, that you seem not to talk about the noise generated in the tail circuit, but rather about supply rejection.
There I agree a resistor could help somewhat, realistically some 6 dB or so maximum for LF.

I was talking about, the performance of the addiction of a resistor to a CCS , in the tail of a LTP in an amplifier and not of the analysis of a perfomance of a CCS per si.

And in a amp, the rail is the greatest source of noise , with many harmonics , from the swithing of the output devices and from the rectifiers .

But in the simulations the rails are pure DC.