RC constants & power amps

[AndrewT]

Hi all,
Q1. Is there a relationship between the Hi pass filters in a power amplifier a) the DC block at the input and b) at the DC block (gain =1) on the NFB lower leg?
Q2. Should the RC time constants be different by a set amount?
Q3. If different which should have the larger time constant?
Q4. Can the RC be the same?
Q5. Are any other roll off points related to these two filters?
hope to hear from you all
regards Andrew T.

[Jan Dupont]

I will try to give it a shot on some of your questions

Q1 to Q4: The input Hi Pass should have a higher freq than the NFB, otherwise you will have a gab not covered by the NFB resulting in unwanted low freq amplification.
However many uses DC NFB that solves the filter problems

[AndrewT]

Hi ACD,
do you mean DC NFB = DC servo? or something else. Please explain.
regards Andrew T.

[Jan Dupont]

By DC NFB I mean no cap in the feedback to Gnd!
In this way, small DC offsets will be corrected by the NFB.
DC-servo is a complete other thing

[AndrewT]

Thanks ACD, that's clear now. Feedback frequency = 0Hz and input frequency is much higher, fits with your recommendation.
regards Andrew T.

[Jan Dupont]

High pass on input at app. 5 to 10 Hz should be ok, depending on what you want from the amp

[djk]

Power supply caps and load form a pole, the NFB cap must have its pole higher than this.

The input pole must be higher than the NFB pole.

NAD uses small supply caps that would normally force the input pole to be quite high, so they make the input pole a two pole filter with a Q=1 (cute).

[AndrewT]

Hi DJK,
how do you turn the PSU/load into a pole?
Use V/I to generate the load R?
What current(I) do you use for the test condition? max power into nominal load or max current of output stage or idling current?
regards Andrew T.

[Graham Maynard]

Hi djk,

Yes, and these interacting poles can sometimes make bass notes sound completely different as you turn the volume up. I've heard it as a 'pumping' or 'pouf' sound as a bass driver causes phase shift.

I have for some time now simulated my circuits by applying zero signal input and moving the signal generator to the power rails. This can show up surprising results below 1kHz, not least illustrating a need for high series input capacitor value, if not DC coupling.


Cheers ........ Graham.

[djk]

I figure minimum load impedance seen by the power supply and the total filter capacitance of either the positive or negative supply.

Example

Hafler DH220 has a pair of 10,000µF and is driving a stereo 4 ohm load.

I use 10,000µF and 2R to calculate the pole of 8hz.