Bias transistor location

[Mick Bailey]

Looking at the Session PM-120 module and I was wondering about thermal stability, as TR3 (BC548) is not located on the output transistor heatsink. Would relocating it be an improvement, or would it introduce crossover distortion when hot? I've read conflicting arguments with Darlington outputs.

[Enzo]

What problem are we trying to solve? has the amp shown itself to be unstable or thermally troubled?

[Alan0354]

This is just my understanding and opinion:

Thew BC548 is set up as a Vbe ( diode drop of about 0.7V per Vbe) multiplier. The resistor across B and E is 270K, then the resistor across B and C is 680K X 2 in parallel with an unknown resistor to get 4Vbe.

But to your question, I would put the BC548 onto heatsink. The temp co of a silicon diode is about -2mV/deg C. and the multiplier should track as 4 X (-2mV) per deg C. This compensates the 4 X (-2mV) decrease of the two Dalington.

But again, it's your amp, I am just talking out loud!!!

[Mick Bailey]

More of a theoretical comparison really with a later amp I had in (the revised late-90s version) rather than fixing a fault. This had the transistor in contact with the heatsink and smeared with thermal compound, though I've seen other amps from the same period that don't. Just interested to get a 'take' on the arrangement. Maybe the manufacturer tried both methods and decided the amp was better with the transistor located off the heatsink. Either that or someone had modified the amp, maybe to correct a thermal runaway fault.

The new PM-120 modules currently sold as a replacement for older failed modules are effectively the power amp bit sliced off the last production run of Session amps, but the transistor is not in contact with the heatsink. Maybe it's not strictly necessary, though the idle current does increase with temperature rise which suggests to be that the biasing doesn't optimally track temperature change and maybe could be better.

Rod Elliott suggests that with Darlington output stages the transistor is best not in thermal contact with the output devices as the bias shift is too much in the opposite direction - that is, an amp correctly biased cold will possibly not be sufficiently biased when hot and introduce crossover distortion as a result.

[Alan0354]

This is just the calculation of the ideal case, I cannot speak for the designer:

If the BC548 is not in contact with the heat sink, the temperature does not rise up much. So it hold the Vbe multiplier setting. But the darlingtons do heat up. And the temp co for each BE junction is about -2mV/deg C. So the 4 BE junction of the two darlington is -8mV/deg C and the idle current increase. Maybe that's the way the designer want it to be. But you can get into thermal run away if the power transistor heat up too much.

From the diode equation, the amount of Vbe increase to double the current is V=VT ln(2)=.025V X 0.693=17.33mV. This mean for 4 Vbe drop in the two darlington, the idle current doubles in every 4 X 17.33=70mV increase of the votage drop across the two base of the darlington.So if the BC548 is not onthe heat sink, then every 70/8 = 10 deg C rise of the heatsink temperature will double the idle current. The design does have emitter degenerate resistor R16 and R17, I don't understand what's written on the value, but that help in mitigating the problem. But the idle current definitely increase as temperature goes up.

[Mick Bailey]

Alan,

Thanks for the calculations. This explains more fully what I'm seeing. The voltage drop across the darlington emitter resistors (0.33 Ohms) at idle increases from 3mv cold to 17mv when even just moderately warm - I guess when fully hot after a session of playing it would continue to rise, but likely still within safe limits as plenty of these amps run fine so I don't consider there to be an inherent thermal problem, but just from an efficiency (or academic) point of view holding the idle current steady regardless of temperature would seem to be a better proposition.

As an experiment I've made up a larger heatsink and clip to see what happens with the transistor mounted on flying leads in close contact with the heatsink. Tomorrow I'll wire it up and see what happens. Nearly time for bed over here so a fresh look in the morning.

[Mick Bailey]

Here's an update; after a few minutes at full output, the drop across the emitter resistors rose to 31mV with the stock circuit. I decided to leave off experimenting with that particular amp and instead built a PM-120 circuit to replace the ILP module in my own amp, with a longer heatsink to locate to the original module's footprint.

I built this as-is, with the vbe multiplier on-board and produced the same results - changing bias with temperature, though my figures were slightly different; cold quiescent emitter drop of 6mV and hot of 28mV.

After removing TR3 and attaching it with flying leads the bias remained rock-solid throughout. Now I've made up a clip from a 1/4" spade connector and permanently located the transistor on the heatsink, with a smear of thermal paste. The result is the amp runs cooler after sustained playing, so the relocation is beneficial.