Emitter resistors in multiple BJT output stage

[bocka]

Is there any limitation in the lowest possible/practical value
for the emitter resistor of a push-pull BJT output stage? I want
to parallel multiple 2SA1943/2SC5200 transistors. My simulation
shows that decreasing the value of the emitter resistor value
significally reduces the distortion generated in a triple EF. This
is because gm is higher, more linear and also spread over a wider
voltage band which redueces crossover distortion in a class AB
output stage. Because real transistors are not equal they have
to be selected. My transistors used (all the same datecode and bought
from Arrow) are selected to about 2mV Ube. Out of a lot of 25 I
have differencies in Vbc of nearly 40mV.

Currently I'm ended with a value of 0.12ohm, with about 70ma for
each transistor. This results in a voltage drop of about 9mV across
the emitter resistors. When mounting these multiple transistor bank
to a heatsink (0.5 K/W) in two banks of 3 at the upper and 3 the lower
side of the heatsink, I expect a thermal difference between these
two banks from maybe 5K (not measured yet). This would result in a
current difference between this two banks of about 30%.

The question is: May the value of the resistors lowered further
or will this result in thermal instability? (I have some amps seen
blown by this effect!)

[mikeks]

Originally posted by bocka
Is there any limitation in the lowest possible/practical value
for the emitter resistor of a push-pull BJT output stage?

In general not less than 0R1, and no greater than 0R22.

[Damon Hill]

Lowering the value of the output emitter resistors much more may get you into layout effects; the individual traces' inherant resistance may start to become significant.

MOSFETs without the equivalent emitter resistors aren't
uncommon, but I'll bet you'll need at least .1 ohm to help
make a bipolar design reasonably bullet-proof.

Carefully matching gain and thermal conditions would
possibly help. How many transistors can you afford to
blow up testing your hypothesis?

How much lower is distortion in your simulation with lower
emitter values? I'm using .27 ohm in my Leach amp simply
because I had a supply of those parts, but I also have .22
and .1. I wanted to try banks of four and five pairs with
higher rail voltages, but my supply of output transistors is
very limited, and so is my budget!

[john curl]

In general you should have 12-22mV across each emitter resistor. Going too low will cause problems, not only in thermal instability, BUT with the transistion between class A and class AB. Going too high will also cause added distortion in the transition between class A and class AB.

[sss]

imho adding more output pairs will give u the same effect as lowering the resistors ,but will be "safer"

[sam9]

Rather than simularing different values, I tried it once some time ago. 0.1R is a stated above the lowest. (Actually it is hard to find ordinary sandcast resistors below this value.) However, using a themocouple to measure how hot the ouput devices became I concluded that 0.1R was too small to dependably prevent "current hogging". Sometimes it is Ok and sometimes not. Closely matching output transistors may well make the lower value sufficient.

On the other hand 0.22R never showed any sign of "current hogging" (as reflected by temperature). BTW, I used the temperatur of the output devices as as indicator because the DMM I had at the time wasn't real accurate with low R values and voltages the calculated current was a little ambiguous.

In addition to lower distortion using the lowest RE values you get a small bonus of a couple more watts!

[mikeks]

Originally posted by john curl
In general you should have 12-22mV across each emitter resistor.............

This is misleading, as it is as dependant on the value of quiescent current as the size of the resistors themselves.

[john curl]

Mikek, you don't know what you are talking about.

[thespeakerguy]

Build it up, debug it, then do this test:

Hit one output device with a heatgun, and monitor it's current. The resistor needs to be large enough to prevent thermal runaway. Don't let the amp sit powered up without supervision until you're sure it won't go into thermal runaway.

[ppl]

Hi John Thanks for input on this tyopic i asume that the 12-20Mv be the target voltage and select your actuial resistor value to obtain a 12-20 mV reading across each emitter resistor at your operating current and so the actual resistor values should be selected to acomidate these conditions is this correct?