newbie de-lurks

[comegetsome]

Hello everybody at diyaudio,

I have finally decided to de-lurk after a few years of viewing the threads in diyaudio and have learnt alot from the gurus.

My primary interest is in building amplifiers, with my first in my teens (I am in my 50's now) being the 4-transistor Motorola design via their application note, and my last being the 200W mosfet amplifier published in the Sep 1999 Popular Electronics. I really never know if the designs were really any good, as I don't have golden ears. I have recently invested in an oscilloscope and a signal generator, and I thought that perhaps I can build something better, using an op-amp and a triplet output design with gain.

Well, I managed to blow up the power transistors and the 100uF Vcc filtering caps (I was driving the amp at 10KHz sine with 8 ohm + 33mH in series at clipping point). I did not see any thing bad, no ringing or artifacts when I drove it at 10KHz square wave with 8ohm+33mH at half power, so I really don't know what happened.

So I thought I ought to start from scratch and ask if any gurus out there can help me and give me some comment on the triplet OPS with gain. I have 2 fairly similar designs, and all comments/criticisms will be appreciated. The front end will be similar to the Amos design, but perhaps I should get the OPS design right first.

Vcc +/- 65
predriver 2sb649/2sd669
driver MJE15033/15032
power MJ21195/21196 (will be paralleled)

Thanks in advance

[wg_ski]

That OPS looks a lot like the Tiger. It takes a certain combination of intuition, mathematical genius, and luck to get one working properly. As you can see, high gain OPS'es don't always clip cleanly. And 33mH is an awfully high impedance reactive load at any audio frequency - start by driving resistive loads or lightly reactive until everything is under control.

Why you would blow up 100uf bypass caps is a mystery - unless either you didn't use big reservoir caps and just used those 100 uf'ers directly off the bridge rectifier, or put 'em in backwards.

[comegetsome]

The 100uF Vcc caps got quite hot, perhaps they weren't big enough physically (100v 100uf 85deg C). Possibly some cheap Chinese made caps.

The chasis is an old Phase Linear 200 power amp, I was using the psu (beefed up to 15,000 uF per rail) so it can't be the psu.

I was reading somewhere that oscillation can bounce between the rails so maybe that was what happened. Even if there was power sag (the amp was clipping at 50V pk-to-pk, no load at 65V) I dont suppose 10kHz ripple at 15V could blow the caps.

I am not sure if the transistors blew first or the caps, but I assume by logical deduction that the caps blew first, and the lack of bypass on the rails caused oscillation and then the transistors blew

Any comments anyone, especially on the basic design of the triplet OPS?

Thanks everyone for reading

[wg_ski]

If they were in parallel with 15kuF, there's no way you put enough ripple into them to hurt them. If the caps got hot, they were either in backwards of bad to begin with. If they were old caps, they may have gone bad just sitting.

If you lost the bypass caps, I guarantee you that OPS will oscillate. It is possible to get them to work - some of the more modern QSC amps (the PLX series) with switch mode supplies are using something like this. But the last two in the are configured as darlington in common emitter in that implementation.