What's a common source of transistor overheating in SS amps?

[liddokun]

So here's the deal.
I finished building a solid state 12w stereo amp, and I plugged it in and it worked fine. But after about 5 minutes or so of having it on, two of my transistors (TIP31 and TIP32) started to smoke and sizzle, and release this kind of fluid which I don't know of. Does anyone here know what a common mistake is in wiring these that would cause this to happen about 5 minutes into working properly?

[R.G.]

Was this a kit?
Or your own concoction, or something from the net?
Were there any heat sinks on the transistors?
Do you have any means to (a) set the output stage bias or (b) tell if it's oscillating?

[liddokun]

Hey R.G.,
This is something from a school assignment for my electronics engineering course.
The schematic is here:

No, there were no heatsinks on the transistors, I built one recently (a mono one) and it did not require heatsinks.
Also, I have no means to set the output stage bias or tell if it's oscillating.
I know it has to be a wiring problem somewhere, I'm almost certain, because I built it as two channels, each side being identical, and only the right channel had this problem. Meaning I wired the left channel properly.

[R.G.]

OK. That being the case, your most likely problem is a bias issue.

D5, 6, and 7 bias the output stage almost but not quite into being on, to help eliminate crossover distortion. The voltage between Q3 base and Q4 base is what sets the output stage bias (and an important thing for you to learn from this assignment!). If you short Q3 base to Q4 base the amplifier will still work, but should sound noticeably ugly. That's the crossover distortion.

If you remove the short and measure the voltage between Q3 base and Q4 base, it should be no more than three diode drops, between 1.5 and 2.0V. If it's more than that, your wiring error is in the biasing arrangement in D5-7. It is also possible that there is something wrong with Q2 or R4/speaker. This particular design relies on the speaker being there to hold the correct bias on Q4 - it's a poor design in that sense. The idea was to use the output capacitor and DC resistance of the speaker to provide a bootstrap voltage for driving Q4, saving one resistor and one capacitor from the classic bootstrap driver circuit. I personally think that was poor economy, but then they didn't pay me to write up the example for the school assignment. 

Measure the DC voltage with respect to ground to Q3 base and Q4 base. These should be within the D5-7 diode drops of centered in the power supply; if not, there is something wrong with Q2/R4/C3 or LS1.

[Gus]

liddokun

I think I saw schematic this in another thread.  People posted about the 1K emitter resistors at R6 and R7.

I would think no heatsinks is your problem.  You are most likely just lucky the mono one has not melted down yet.

The output bootstrapping is an issue like R.G. posted, what is the voltage across the speaker with the volume pot at 0?

If you can find some older transistor data books some have simple transistor power amp circuits in them as examples of how to use the transistors.

[liddokun]

Gus and R.G.,
Thanks for the suggestions! Yea, I posted a question about this design in a different thread, when I was building the mono version. The schematic is indeed incorrect, there are two 1k resistors (R6 and 7 I believe) that should be 1R instead. Also, the cap on the speaker should have polarity the other way around. I was just absent the day my teacher told this to the class.

I'm going to look at it and try to make it work, but as it stands, it was actually for a demo for the younger grade class.  My teacher and I are looking at some monolithic chips from digikey, some of which put out up to 300 watts per side (though with the help of darlingtons).

[PerroGrande]

Hey R.G...

What would the down-side be to tying the lower side of R4 to ground rather than involving the speaker in a bootstrap configuration?

[R.G.]

What would the down-side be to tying the lower side of R4 to ground rather than involving the speaker in a bootstrap configuration?

If you look at the schematic and imagine a resistor from R4 base to ground:
- when Q2 is trying to drive the output low, it is turning off, so its collector voltage goes lower
- the lower voltage on Q4 base pulls Q4 emitter down
- Q4 base current has to come through the new resistor to ground
- as Q4 emitter goes down, it reduces the voltage at Q4 base to ground, and starves Q4 base for base current

The result is that Q4 is only quite softly driven on as it gets toward ground, and so the output stage cannot pull down as well as it can pull up. Feedback tries to correct this, but with only the current from a resistor to ground, it can't completely. So the undistorted output range is restricted by Q4's current drive needs and the value of the resistor from base of Q4 to ground. You can make the resistor smaller, but this runs the current in Q2 up.

A bootstrap connection drives the voltage available at base of Q4 about 1/2Vcc negative through the output cap (or a separate bootstrap cap, which is how most designs do it) so there is plenty of voltage available to drive Q4 fully to the negative supply (ground in this case).