Theoretical "optimium" bias for a class AB amplifier?

The Lord Flashheart

Hi all,

Is there a specific "optimum" bias for a class AB amplifier, independent of the emitter resistors in the output pair?

I have an amplifier that uses a uPC1270H with 28V rails and 0.1 Ohm emitter resistors.

The amp has 4 channels and one appears to have been much more heavily (ab)used than the other; dried up small value caps, drifted carbon resistors, etc.

It lacks bias pots and the bias is set with fixed value resistors. The bias current across the emitter resistors in the "good" channels measure consistently at around 30mA (3.1mV across the 0.1 emitter resistor) whereas the "bad" channels differ from each other and are higher at 40-50mA.

I want to fit pots and re-bias the amp but have no figure to set the bias to.

Is there any rule of thumb or calculation applicable in these circumstances?

rayma

See page 11 here:

Attachments

• 1971-02.pdf
3.3 MB · Views: 109
1 users

1 user

The Lord Flashheart

See page 11 here:
Optimum as in lowest distortion means choosing both optimum emitter resistors and optimum bias current, and these depend on the load resistance.

A wing-spread diagram is the best way I've found of visualizing cross-over issues (gain against input voltage - though the gain of an output stage is a slight attenution), there are some examples in this thread:

The Lord Flashheart

Isn't the 'Oliver' criterium 26mV across Re?
https://hifisonix.com/power-amplifiers/cross-over-distortion/
Should also be in the above linked HP journal.

Jan
Thank you, will have a read. Had not come across the term "Oliver Criterion" before but did recall seeing the figure of 26mv across the emitter resistor. In this case that seems far too much, Is that figure meant to between both emitter resistors (negative and positive half of the output pair) or just one?

jan.didden

Paid Member
Actuallu Oliver said 'between 14 mV and 26mV', and Russell found that with much lower value it worked fine.
I guess you really need to read the links and see what they say. The hifisonix is clearest.

Jan

wg_ski

26 mV (which can result in 100mA or more bias current) may be the optimum, but very useable results are obtained down to about 10% of that, especially if EF3 is used. Usually they cut it in half to a third and call it good.

1 user

BesPav

There are too much talk about EF1 optimum bias point which straightly trasfered to EF3 optimum bias.

1 user

WhiteDragon

30 to 45ma per device is fine
if with differential depending on performance
feedback takes care of the rest
or depending on compensation additional bias
wont change the numbers much

With a amplifier that can actually show
or measure any significance with additional bias.
80ma to 100ma per device can be closer to ideal

wg_ski

There are too much talk about EF1 optimum bias point which straightly trasfered to EF3 optimum bias.