implimenting a baker clamp

[mjona]

Your amplifier has no output coil and lacks source resistors which would introduce a modicum of inductance. You need to consider that the likely result will be overshoots in the wave form. You can quantify the distortion from the scale on an oscilloscope - this could be in double figures before a decimal point in percentage.

 

[26mVpdc]

The most obvious place for compensation here would be Q9b - Q9e to place a dominant pole in the loop response. Shunting the Q3-Q4 emitters may improve phase margin but you loose the advantage of Miller compensation smoothly trading gain for linearity. May avoid slew rate issues as well.

 

[ctrlx]

The most obvious place for compensation here would be Q9b - Q9e to place a dominant pole in the loop response.

do this with or without C4 ?

 

[mjona]

do this with or without C4 ?

The method of compensation is a matter of preference. C4 is already there - it is simply a matter of increasing the value to create a dominant pole.

The impedance of C4 is in parallel with the feedback resistor which the feedback resistor. The effect is the same as a lead capacitor in a feedback loop.

 

[26mVpdc]

I wouldn't bother changing C4. Just add a miller cap in the b-c of Q9. The designer was attempting to do something with C4 but compensation in the feedback loop may not be the best solution.

 

[mjona]

I wouldn't bother changing C4. Just add a miller cap in the b-c of Q9. The designer was attempting to do something with C4 but compensation in the feedback loop may not be the best solution.

If you take this advice the added capacitor will need to be charged by the LTP. The amount of available current from this has a limiting effect on slew rate capability - Looking at one of Self's examples he uses twice the amount of current you are using for his LTP CCS.

See Bob Cordells web pages for two articles about TIM distortion to see how to quantify and deal with the foregoing.