3/4 + 3/4 the strange A1

Hello MOrten,
Please reserve 2 boards for me.
I made some suggestions and I didn't tell right. I mean that You put perhaps 2kOhms from the 24v to the bases of Q3 and Q4 to give Q5 more Current than the small basecurrents of Q3 and Q4.

I Think You can take the two 1yf as I suggested but You need the fixation of the voltage between the bases of
Q9 and Q10 perpaps with 220 yf as in the original mf a1.
This should be an advantage because the feedbacksignal goes through Filmcaps instead of elcos.
Sorry , I have still no simulationprogram
With regards
 
Dear @StevieK
this one is the original, the real Musical Fidelity A1:

1737471673364.png



If you have any further questions about your future 3/4 + 3/4 Amplifer, don't hesitate to send me a PM - I'll answer them.

Good luck,
HBt.
 
Hello MOrten,
Please reserve 2 boards for me.
I made some suggestions and I didn't tell right. I mean that You put perhaps 2kOhms from the 24v to the bases of Q3 and Q4 to give Q5 more Current than the small basecurrents of Q3 and Q4.

I Think You can take the two 1yf as I suggested but You need the fixation of the voltage between the bases of
Q9 and Q10 perpaps with 220 yf as in the original mf a1.
This should be an advantage because the feedbacksignal goes through Filmcaps instead of elcos.
Sorry , I have still no simulationprogram
With regards
Send me a PM for boards
 
Yes Morten, there is an ulterior motive in this selection - and if possible you should select and mate exactly this type.

Although all AF transistors of this Siemens / Philips series are similar and it could perhaps also work safely with BC556B /546B at this point - but whether the unusual type of frequency response compensation (acting on the input capacitance and with it) will only become apparent at the end.


But now you're going the whole hog Morten!

😎

greetings,
HBt.
 
  • Like
Reactions: m0rten
Is there any advantage of using the old 3055/2955 instead of newer ones?

If we use the TIP xxxx, we make sure that the simulated frequency response compensation (i.e. the stability/ oscillation tendency) works.

I have plenty of 2SA/2SC types with better specs, but are there disadvantages with using those?

What types are these exactly? The only (critical) point is that we form a classic darlington with the drivers (the BD xxx)!

The credo here is:
try it out, but I don't expect an improvement in overall performance.


The situation is completely different with the competing idea from ZenMod (with the likeness of Pippi Longstocking), which benefits enormously from modern power transistors, the 3/4 + 3/4 amp doesn't care about that. If it does not cause any damage (instabilities) and is thermally up to the task, then "fire away".


#
I still have to ask for a lot of patience, it will be some time before I get to the construction.

:umbrella:
greetings,
HBt.
 
  • Like
Reactions: m0rten
Might be it is too late as you have already got the board that looks so beautiful.

I want to input another idea.
There is a topology called "Transnova". It is equivalent to common emitter output stage that A1 has, but with conventional bias setup. One of the benefits with a conventional Vbe multiplier is that it can work into deep class B region without THD spike. Thus, you don't have to bias it heavily at the first place. The drawback is you need a floating power supply. I am not sure if you can overcome this with a simple bootstrap.

Here, I believe, is the original paper.
https://forums.melaudia.net/attachment.php?aid=43792
 
I cannot directly understand the TRANSNOVA insertion. I do not yet recognize a connection with the MF-A1 and its derivatives.

#
The current model (version 2023) uses the inexpensive and robust TIP35 & TIP36 pair. In this model, it is mainly the completely undersized mains transformer that is cooking!
 

Attachments

6, 5, 4, 3

I'm actually missing some resistor values, it's always the same - you think you have everything in your own stock, but there's always a component missing and so the stock keeps growing.

#
In any case, the tension is rising ...
 

Attachments

  • 2.png
    2.png
    1.4 MB · Views: 23
  • 1.png
    1.png
    3.6 MB · Views: 26
The first stage, the differential amplifier with its current mirror, is precisely selected, paired and symmetrical. You can estimate 60 minutes for this work step!

Reference points
BC3x7
HFE 343

BC5x0
HFE 442


BC327/BC337 are the least noisy of that bunch. Better than the always returning BC550/BC560.
Cob 12pF (8pF) & Cib 60pF, ft 200MHz (170MHz), F -> ?
vs.
Cob 3pF (4pF) & Cib 8pF (10pF), ft 200MHz (250MHz), F -> 2dB

I cannot find a noise figure for the BC3x7 type in the old Siemens data books. Whether there will be noticeable noise with a 500µA IC, I doubt it. I have more cuffs before the frequency response compensation, you never know beforehand whether everything is stable, especially not with this concept ... we'll see!


HBt.

tbc