Armstrong 621 power transistors

[peter31416]

I am attempting to resurrect an Armstrong 621 stereo amp. The unit blows the fuse in one channel when switched on, even without speakers connected. I think I have located the problem to one of the 40636 output transistors which has almost zero emitter to collector resistance.
Has anyone found a suitable alternative to the RCA 40636 ? Looking at the list of alternatives it is not that simple as just matching parameters it seems.

[snowman_al]

Hi Peter
2N3055 or 2N5882. Actually quite a few 'will do'.
Check the driver transistors before making any replacements...
Alan

[paulsherwin]

The 2N3055 is one of the most widely faked transistors, so you should be careful when buying them. Use an established supplier with a good reputation, even if it means paying more.

[Marconi_MPT4]

RCA transistor type 40636 was renamed RCA1B01 in their 1978 power devices data book. VCE/VCB maximum rating is 95V and the example circuit has an 84V supply, Armstrong 621 schematic shows 82V used. 2N5882 has a maximum rating of 80V and no safety margin. 2N3055 appears to be 60V max. There are higher voltage types available such MJ15015 and others.

Both 40636 and RCA1B01 are hometaxial devices with a transition frequency of 0.8MHz, while modern replacements have a much wider bandwidth typically 4MHz or more. In most cases these will perform but there is always a possibility of parasitic oscillation without modification of the forward or feedback paths.

Rich

[Diabolical Artificer]

I have some old 2N3055's and maybe some 40636's, they are functional but have rusty top cases. They came off vintage power supply boards, am pretty certain they are the old hometaxial devices. If you want them you can have them FOC, just pay the postage. Drop me a PM if interested, Andy.

[Trevor]

The 2N3055 might not be quite correct they have to be at the high end of the voltage range of that device the 40636 will be great as would 2n3773 etc
Some 2n3055 can be used but these were selected to a higher than standard voltage device
I know these amplifiers well and used to sell a lot of them back in the day
Trev

[peter31416]

Many thanks to all for your info. It is still not that straight forward but I will try the 2N3773 and hope it does not result in RF oscillation. I will let you know how it goes. BTW my expertise is in mechanical engineering, not electronic but I can use a soldering iron.

[peter31416]

I have finally got back to this project. I have replaced both output transistors in the dud channel with 2N3773 type and the amplifier works, sort of.
At low volume levels the amplifier works on both channels. At around 50% volume it produces a continuous ultrasonic oscillation on both channels even with all inputs and the speakers disconnected. At lower volume levels although not continuously oscillating, when playing music, oscillation spikes are produced on the peaks of the signal.
I have been unable to determine the frequency but it is above my hearing range.
It has been suggested it could be produced by positive feedback but with all input and outputs removed it still oscillates. I have tried putting an additional 15 Ohm resister in series with the speakers but this had no effect other than reducing the sound output slightly.
It looks like the problem is internal to the amplifier. Any suggestions would be welcome.

[Diabolical Artificer]

This is a bit of a bodge but you could try a 220p - 470p from the junction of the emitter resistors to ground. If you can find out what frequency the oscillation is at you can craft a better filter using 1/(2 * pi * f * c) to find a resistor value. That means something like a 1k or 2k2 and a small cap. For instance to stop HF oscillation on some of the big amps I've designed I've used a 2200p & 1k resistor, the cap goes in front of the cap (at an anode or in your case the output), cap to ground.

This isn't doing the job properly but might be a quick fix. Hope that helps.

[Craig Sawyers]

The various iterations of the Quad 405 sprouted inductors and capacitors as transistors became obsolete, and then obsolete again. and they had to use different ones with different fT.

Might be worth looking at the 405 service manual (freely available on the the web; just search). Might give some hints what might work in stopping modern high fT devices from oscillating.

Craig

[RichardGM]

Might something as simple as capacitors between collector and base simulate lower fT?

[Radio Wrangler]

Originally Posted by RichardGM

Might something as simple as capacitors between collector and base simulate lower fT?

Simulate, in some ways. What it does is add an extra pole in most circuits, which is not quite the same as moving the existing pole.

In the Quad Triplet case, the issue is having its poles clustered too close together in a circuit needing one to be dominant. To make it work, it needs a pole moving, and adding an extra one just makes things a lot worse. Trying a pole-zero cancellation would get nasty on tolerancing, if the effect on current gain and swing didn't already eliminate the approach.

The Armstrong circuit, if I remember rightly, isn't quite so intolerant.

David