1
\$\begingroup\$

(Context: I don't have a background in e.e., but getting better at basic bench work.) I'm repairing/recapping an early 70s German integrated stereo. The power amp output section has four large heatsinks for its transistors, two each corresponding to the left and right stereo channels. After a very short period of time, even a couple minutes, the left-channel's sinks become extremely hot to the touch while the right channel remains cold.

The audio output is correct and expected, a typical radio or aux music signal amplified in stereo. Sound on both channels is totally fine coming out of the speakers. I only noticed the heating asymmetry by accident.

I don't know that this is a problem that needs solving, exactly, but the heat in that area seems likely to reduce the life of the new left channel caps somewhat.

I'd like to understand if (1) this indicates a problem and (2) what might explain it? Could those power transistors be deteriorating unevenly with age? Could something else that feeds them be behaving strangely? (I've replaced all the electrolytics and a couple questionable looking film caps on this board, for what that's worth.)

Any insight or theories/pointers/assistance very welcome. I have a bench oscilloscope and can probe signals in an entry level kind of way if that's useful but not sure what I'd be looking for in this case since output seems fine.

Worth noting that the differential in heat remains over long durations-- the right channel does eventually (an hour?) become perceivably a bit warm to the touch, but the left channel becomes scorching very quickly and stays that way. (This is also true if the left speaker is not even plugged in, fwiw)

Photo of the board and image of the schematic for this board below. The left channel is the left half of the board, and the two left heat sink sections.

enter image description here

Schematic. Input from preamp at left, output to speakers at right. Left channel is top half of the diagram. T708 (and maybe T709? maybe T705?) appear to be the very hot transistors.

enter image description here

\$\endgroup\$
7
  • 1
    \$\begingroup\$ Measure the voltage across R619 and R719 for the other channel. That tells you the quiescent current in the output stage : the usual culprit. Hot channel will have a higher voltage. R621/721 likewise. Adjust the appropriate one of R615,R715 until R619/R719 match. (Given the age of those trimmers, I would switch off, wait, adjust, switch on, test rather than turning them when switched on, for the safety of your transistors). \$\endgroup\$
    – user_1818839
    May 18, 2020 at 22:34
  • \$\begingroup\$ Your schematic conveniently shows DC voltages at various points, measured from ground (which is the heavy horizontal line down the middle of the schematic). I see +20V at the bottom of R719 for example. There are a handful of others. Measure these with no audible signals. \$\endgroup\$
    – glen_geek
    May 18, 2020 at 22:40
  • \$\begingroup\$ @BrianDrummond Thank you. Do you mean measure the current (vs voltage) across R619/R719? Or voltage to ground from each side of the resistor, or? (Sorry if dumb question). Is there a "correct" value for that to have? Ie, understood that they should match but how to know what the right matched value should be? \$\endgroup\$
    – Ben Zotto
    May 19, 2020 at 1:44
  • 2
    \$\begingroup\$ Measure the voltage across the resistor : divide by R to find the current. The manufacturer would have specified a quiescent current : 20mA would be typical; you might choose slightly less to be on the safe side and save power, with a slight increase in distortion. \$\endgroup\$
    – user_1818839
    May 19, 2020 at 11:27
  • 2
    \$\begingroup\$ I like the hats to attach the to92 transistors to the heatsink. \$\endgroup\$
    – Colin
    May 19, 2020 at 14:29

3 Answers 3

Reset to default
2
\$\begingroup\$

Digging down way deep in my knowledge base, but...

Looks like what you have is a push-pull (totem pole) output stage, with the two transistors in the top (T706, T708) and bottom (T707, T709) legs of the totem forming a darlington pair.

These darlingtons should be biased so that with no signal, both the top and bottom pairs of transistors are conducting, just a bit. This is done to reduce the distortion (crossover distortion) that would occur if the output transistors were operated in a pure B mode. This distortion occurs because with pure B type operation, both sets of transistors would be OFF near the 0 V point of the waveform, creating a couple of volt dead zone in the output.

By biasing the outputs transistor pairs to be slight ON at near 0 V, this distortion is rediced/eliminated. The downside is that because both transistor pairs conduct near 0 V, there is higher dissipation in the output transistor due to the current that flows from the top to the bottom pair.

The resistors Brian Drummond mentioned (R715, R615), along with the transistor, establish this quiescent operating point, the amount of current that flows through the top and bottom pairs, and so impacts the power dissipation. If this quiescent point is not set correctly, you have excessive current flow (sort of a shoot-thru affect) that increases dissipation.

\$\endgroup\$
2
  • \$\begingroup\$ Thanks. I checked your answer here because you gave me the notions of the theory of what the circuit was supposed to be doing and what could lead to the overheating. The culprit turned out to be a damaged trace that effectively disconnected the "top" (in schematic) of R714. I don't know enough to reason through why that would result in normal sounding output but massive quiescent current. But jumpering the connection allowed me to normalize with the pots, overheating vanished. Thank you. \$\endgroup\$
    – Ben Zotto
    May 20, 2020 at 19:08
  • 1
    \$\begingroup\$ Glad I/we were able to help. \$\endgroup\$
    – SteveSh
    May 20, 2020 at 19:10
5
\$\begingroup\$

See the resistors I've marked with red boxes: -

enter image description here

With no signal at the inputs, use a DVM (on DC millivolts) and measure the voltage across each of R721 and R621 - they should be about the same - maybe 5 mV to 20 mV. If the hot channel has a significantly larger voltage then try adjusting its associated bias current trimpot (the other red boxes) to reduce it to the "good" channel.

I think R721 and R621 might be these 0,24 Ω resistors marked in green but double check the PCB legends: -

enter image description here

Also up for consideration - do the transistors get hot when the speakers are disconnected, connected, or both? If they only get hot when the speakers are connected then it might be that C709 (the output DC block capacitor) has gone low resistance.

\$\endgroup\$
1
  • \$\begingroup\$ Thank you for taking the time to mark up the images, clarifying the practical side of the comments on the question and other answers. Your response and others led me to discover that the voltage (current) in the left channel was not just a little bigger but orders of magnitude larger than the other channel, not fixable with the trim pot. The culprit turned out to be a damaged trace that effectively disconnected the "top" (in schematic) of R714. Jumpering the connection allowed me to normalize with the pots, overheating vanished. Thank you. \$\endgroup\$
    – Ben Zotto
    May 20, 2020 at 19:02
2
\$\begingroup\$

Trimpots R715 and R615 are set differently but they do the same thing for each channel. Try adjusting R715 to look like the setting of R615. Adjusting to match the same voltage across R721 and R621 would be best.

amplifier heating

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.