Welcome to Solid State Guitar Amp Forum | DIY Guitar Amplifiers. Please login or sign up.

March 29, 2024, 08:38:16 AM

Login with username, password and session length

Recent Posts

 

Discrete poweramp build

Started by phatt, January 03, 2014, 03:06:55 AM

Previous topic - Next topic

phatt

Something I whipped up over the Crissy break.
Well I decided to have a crack at a discrete power amplifier circuit and ESP p3 looked like a simple straight forward build so lets see how long it takes to let the magic smoke out. xP

Use this page for any reference;
http://sound.westhost.com/project03.htm
(Figure 1a is what I've used as reference)

OK I built it on a bread board for testing and after much double checking I fired it up with 40 watt limiter lamp and all seemed ok for a while but when I pushed a big volume of audio signal the lamp lit up and big DC volts on the output. arr-huh? not as simple as it seems.
Now I'm not the sharpest tool in this place but I've had quite a few attempts at discrete power amps long ago and I know that even with the best advice I'm always a little nervous about claims of stability.
Without the tecky knowledge background I can't say why but my breadboard does have rather long wires and that may not help stability.

I do remember ~20 odd years back (while learning that I still had a lot to learn)  about DC coupled pwramps that low values of R4 can cause all sorts of wacky things to go wrong. So I changed this from 560R to 1k which fixed the situation. I assume the latchup might also be due to the lamp limiter but was not brave enough to take the chance that it would work without.

I decided after an A/B test with another power amplifier that I needed a little more output so I changed R5 down to 360 Ohm and this made a big improvement. Of course my concern was would this upset something else but to my surprise it seemed stable and nothing got excessively hot.
To my surprise, all the 5Watt resistors seem to run rather cool, the hottest (make that warmest) part at idle seems to be Q3 and output devices only get warm with high signals for extended periods.
R7/R8 do get warm but again nothing concerning.

The circuit is now working with no limiter lamp so running full rails of 33-0-33VDC and has done so now for a many hours. So just wanted to run this past better minds to see if I've missed something.

Can this be improved by say use of MJ802/MJ4502 (or better) or am I just wasting money for little improvement.
Maybe this is capable of more power but I don't wish to run it on the bleeding edge just to bragg that I can fry eggs on the heat sinks.

Oh yep,, The PTC short protection seems to be working as I did my short circuit test by bridging the output socket and nothing has blown,,, Yet.

Thanks chaps, Phil.

J M Fahey

 :dbtu: :dbtu: :dbtu: :dbtu:
Congratulations.
Now all you need is a PCB  ;)

Roly

Some thorts;

Voffsets:
Try putting an identical resistor in the collector of Q2 (to fully balance the diff pair).

Q1 and Q2 should be matched at the operating current, and should be intimately thermally coupled, face to face with some thermal goo, in an ali clip for example.

I've taken to putting my thermal sensing on the main heatsink, which is after all what it is supposed to be monitoring.

Output offset can be trimmed out with a pot strung between B+ and B-, and fed from the wiper via a large resistor, to end up producing +/-100mV at the Base of Q1, trimmed for Vout = 0.00V

Protoboards are an instrument of the devil.  You can draw this up on laminate with a waterproof felt tip pen, etch, drill, stuff and test in less than a day.

{Groovy test rig BTW}
If you say theory and practice don't agree you haven't applied enough theory.

phatt

Hi Roly,
          Thanks for the input, Yes I will try a 1k on Q2 collector, I've noted a few circuits with that in place.

Seems there is some contradicting views in this area, some saying they don't do the same job so no need for perfect balance but I did go to the trouble of checking HFE of Q1,2 reading within one point of each other (301 302) the rest of the devices read between 100~130.

Well I belted it with guitar signal for an hour today and it was 42 Deg Celsius today (cricky,,close to record i think) so any biasing issues would have shown up surely.
Output stayed rock solid around 80mV.
The bias transistor on the heat sink will likely depend upon the PCB layout which I have not considered much at this point.

Yes J M Fahey will just have to wait a while longer for a PCB,,,as I'm not as fast as Him,,,, YET. :P

The pot across the output sounds interesting, Never heard of that trick but is rather obvious now that you mentioned it.
I assume a 1k per volt rule will suffice for that?
Got plenty of 100k pots.
Yeah I know you hate these breadboards but convenient.
It got to damned hot to do much more today so inside the house is Aircon. :dbtu:

Phil.

Roly

You've measured +50mV at R2, 22k, ( -> -80mVout) which implies a Q1 Base current of;

I = E/R
50/22 = 2.27272727uA

0.05V/22000r = 0.00000227A * 1000000 = +2.27uA

Now,
Vsup = 33V

R for -2.27uA from 33V

R = E/I
33V/2.27uA = 1.4537444Meg

Simple, fixed;

470k from -33Vsup to a *small* bypass cap (short TC <100ms), 1M to the input

t = CR

C = t/R
0.1/.470 = 0.21276596uF

...so, a 0.22uF to ground.


And your output offset is reduced from -80mV to ... ?


{or just stick in a 1.5Meg resistor and see what happens.}
If you say theory and practice don't agree you haven't applied enough theory.

phatt

Ta mate,
Hey I must be improving as I did get most of the maths. :)

So we are simply trying to pull that +50mV (pos offset) at Base of Q1 back down to zero and this in turn will leed to a zero DC offset at output by bleeding some neg voltage back to the input thus we bleed it from the neg rail.

Makes sense as it's all Direct coupled.

As the R values are quite large it will  have no effect on the AC signal.

Am I right in assuming that this idea is what some of those *DC servo control* amp circuits (old hifi amps) did years back?
My old Sansui had an opamp setup dedicated to just keeping it all close to zero.

All on hold for now ,, to darn hot to do much of anything but the page is in my folder under the bench for later in the week.

Car needs attention this week so that might chew up time and money.
Darn thing won't stay locked,, just keeps unlocking itself. :grr
Phil.

Roly

Quote from: phattSo we are simply trying to pull that +50mV (pos offset) at Base of Q1 back down to zero

Good.  :dbtu:   It's all just Ohm's Law and the occasional capacitive reactance.  Relax.   8|

Yes!
  :dbtu:  'tho the gotcha is that the sum of the offsets at the moment relates +50mV in as -80mV out, but due to these offset voltages (the difference between Q1 and Q2 mainly), 0mV out may not occur at 0mV in.

There may be a small residual input voltage (current sink actually) required to bring the output to exactly 0mV, and it may be thermally dependent too.  This is why I mentioned, matching the input pair, balancing their collector circuits, and thermally coupling them.

Quote from: Roly{or just stick in a 1.5Meg resistor and see what happens.}
:afro:

{I'm only mostly crazy;  :duh  the rest of the time I'm, like, it's just hard to tell the difference.   :lmao:   }


Quote from: phattAm I right in assuming that this idea is what some of those *DC servo control* amp circuits (old hifi amps) did years back?
My old Sansui had an opamp setup dedicated to just keeping it all close to zero.

Yeah, essentially there is a DC VLF loop that rebalances the output slowly, on average, at a frequency much lower then the bass, seconds to tens of seconds time constants.

Interestingly the best studio and recorder AGC/limiters's of Ye Olden Days has release time constants around 30 minutes!
If you say theory and practice don't agree you haven't applied enough theory.

phatt

Thanks Roly,
                 Been kinda busy so no play time.
I will get back to this eventually but everything has gone on hold.
SWMBO,,, wants to move house again. Arrrh @$&* :-X

Really do appreciate your amazing ability with the maths. :tu:
Will keep pondering some capacitive reluctance while I pack up all my junk. :lmao:

Hey looks like We might end up down Bairnsdale way, got some good friends down Lakes Entrance way.
Phil.