diyAudio

diyAudio (https://www.diyaudio.com/forums/index.php)
-   Solid State (https://www.diyaudio.com/forums/solid-state/)
-   -   "vintage" quasi complementary power amp (https://www.diyaudio.com/forums/solid-state/323668-vintage-quasi-complementary-power-amp.html)

catd 9th June 2018 12:25 PM

"vintage" quasi complementary power amp
 
4 Attachment(s)
Hi all!

In the late 1970s I bought an output stage called "dynax sc160" from a german electronic dealer "Bühler", - some of the older germans here will remember -, it came completely assembled on a big heatsink along with a very poor "power supply", containing only a rectifier B80C5000 and two caps 2200uf. But the transformer (M-Core) had enough power.
The stage wasn't very reliable and I got more then once smoke signs at greater loading.
There was a built-in overload protection, which did however not function properly and instead produced clipping already at medium loads.
https://www.diyaudio.com/forums/atta...1&d=1528545960

I completly omitted the protection circuit.
(Instead of it I built an over current protection circuit in the power supply and so a voltage control resulted by itself.)
A pair of additional output Transistors were fitted to the output stage and some foil capacitors added (or changed from electrolytics).
https://www.diyaudio.com/forums/atta...1&d=1528546181
In this form I used it since then.
http://www.diyaudio.com/forums/solid...ml#post5036286


Now I'm thinking about changing the old electrolytic caps and by the way I could do some further changes to the circuit.
So I made some simulations. (I don't have model for BD249, so I used TIP35-model. I know, there are better amps out there, but I want to stay with this kind, because I have a bunch of the used transistor types and I like simple approaches. Also I have to keep the PCB size.)
https://www.diyaudio.com/forums/atta...1&d=1528546398


What do you think about the changes of the circuit?

anti 9th June 2018 04:24 PM

I would delete the entire DC offset voltage bias network; and make the R26 the exact same value as R8 = 18K (?) and connected directly to GND.
The DC offset could be instead made trimmable by making the R2 a trim-pot.

Then do you really need the R21 there? Imho you could delete it. Also see if you can alter the R3 value - around 100R-ish maybe?

catd 9th June 2018 06:09 PM

2 Attachment(s)
Hello anti,
thank you for your suggestions.
Done!
This reduces parts number and THD a bit.
https://www.diyaudio.com/forums/atta...1&d=1528567677

EssB 9th June 2018 07:02 PM

Quote:

Originally Posted by anti (https://www.diyaudio.com/forums/solid-state/323668-vintage-quasi-complementary-power-amp-post5457513.html#post5457513)
I would delete the entire DC offset voltage bias network; and make the R26 the exact same value as R8 = 18K (?) and connected directly to GND.
The DC offset could be instead made trimmable by making the R2 a trim-pot.

Then do you really need the R21 there? Imho you could delete it. Also see if you can alter the R3 value - around 100R-ish maybe?

.. and R5 can also be deleted Edit, OK just seen you already have in your reply

anti 10th June 2018 10:36 AM

Quote:

Originally Posted by catd (https://www.diyaudio.com/forums/solid-state/323668-vintage-quasi-complementary-power-amp-post5457601.html#post5457601)
...
Done!
...

Zehr gut...

But you still have that 470R resistor there (R3? R5? can't see clear on my screen) that goes to the base of the bottom CFP PNP (BD136_138_140).

Don't delete it, but lower the value incrementally (my gut says it will end up somewhere around 100R region). Slam the input so the amp clips and set the value of that resistor so that the amp will clip as symmetrically as possible.

Also run square-wave-input analysis and check if you have any oscillation or overshoot in the CFP (quasi "pair"). That C7 at 100pF is supposed to kill the oscillation if present; so tweak it to get rid of any oscillation AND to get the "prettiest" square-wave response AFTER you determined the best value for that "470R" resistor ...

catd 10th June 2018 03:13 PM

2 Attachment(s)
Hi...

I had already done a square wave simulation.
No overshots etc. 1 khz is quite OK.
At 10 khz one can recognize a slight rounding of the corners. But this is due to the used devices, I think. Playing around with values 22p -100p for capacitor C7 at Q3 base doesn' t affect this.
It also makes no real difference in simulation, if I chose R9 at Q3 base as 470 or 100 Ohm. Only THD can be decreased by 1/1000. No effect on clipping symmetry. So I took 220 Ohm as compromise.
I think this fine tuning is more an issue for the real building later.

I made a further change to the LTP.
In the simulation this halved THD, in difference to previous version.
It remains to be seen if I could mount all of this on the given PCB-space.
https://www.diyaudio.com/forums/atta...1&d=1528643426

chriquet 11th June 2018 04:24 AM

I'd be happy to see a choke on the o/p

voltwide 11th June 2018 06:01 AM

I doubt the loop compensation is really stable. There are three caps: C5,C6, C7. Best stability should be obtained with one dominant pole, so omit C6, C7 and optimize C5.

steveu 11th June 2018 10:26 PM

The original purpose of R9 was to limit the current drive from Q4 so that the over-current protection would not fight with Q4. I would remove R9 or reduce it to maybe 22 Ohms. The only remaining reason for it is driver snubbers.
I would not use Q11 (~diode) unless it was to cross couple R14, but there is no good way to increase the shoot through prevention off-drive of R18, but I would reduce both R14 and R18 to about 47 Ohms. Simulate the ~TIP35 current with a 10KHz drive near clipping to see the effect. This was a common failure mode for these old amps, when a ground connection broke or shorted causing oscillations. See "Miller effect".
C6 should have a small resistor in series, say 100 Ohms to prevent RF rectification form the speaker cable into the diff amp. I assume the real amp has a build-out choke+resistor in series after R23. This is required to prevent the speaker and cable from shorting the feedback at high frequencies and cause oscillations.
Fast diodes around the VA and drivers are often used to solve over-drive recovery delays by preventing saturation of Q4, Q12, Q10, Q3 . Schottky diode are the simplest but most Schottky do not have a high enough reverse voltage. Usually 1n4148 help enough to be worth while.

steveu 11th June 2018 10:31 PM

Adding Q14, Q15 is a great idea but the increased gain may require a larger cap at C5.

The current limit protection should have worked without causing problems so that bothers me. Do you have some kind of passive crossover with a series resonance that is ~shorting the amp?

TIP36C are not very expensive so I would scrap half the BD249s and change the output to a cross-coupled complimentary design.


All times are GMT. The time now is 09:56 AM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio

Wiki