DC Servo loop tunning

[Ted205]

Hi guys

I've taken this layout from the lm3886 datasheet.

schematic

I want to use in on an opamp with a gain of 2.

I've been playing around with it and though it does work, its very slow. If I reduce R2, to say 2.2kohms, it becomes alot quicker.

However i'm not entirely sure how it works so i dont want to do anything thats going to alter things with out me even realising as after all its a fairly important part of my system (blocking input DC).

Will R2 at 2.2k be ok, or maybe even lower ?



cheers

ted

[pacificblue]

The DC servo counteracts all signals that are below its operating frequency by feeding them back to the inverting input. It is a low-pass filter. If you make it faster, it will start to counteract at higher frequencies. Make it too fast and it cancels out your music signal below its corner frequency.

If you change R2, change R3 likewise. Changing them by a factor of 1000 is way too much.

[peranders]

The servo has possibly too low adjustment window and it's too slow also. The opamp runs always at the gain of one!

Each input of the opamp must always have the same impedance. Equal resistors and equal capacitors.

[Ted205]

i've found an error with my schematic. R1 should be 10k. As it should be 10 times the value of Rf (1kohms)

[AndrewT]

now split R1 into two resistors and form a T filter with the resistor junction going via a cap to audio ground. Try 5k1 for both sides of the T filter and 4.7uF or 6.8uF to ground.

Now go and read the servo thread, particularly Tom's contribution.

[peranders]

Andrew and everybody else, the opamp itself isn't too slow. The servo is too slow for my taste. I think around 1 Hz is good.

The gain of the integrator is +1 in closed loop and DC in is which means that 1 mV offset will result in 1 mV at the output. Since the integrator is non-inverting it's essential it's important to have identical resistor and cap values seen from the servo opamp's inputs.

[peranders]

Originally posted by AndrewT
now split R1 into two resistors and form a T filter with the resistor junction going via a cap to audio ground. Try 5k1 for both sides of the T filter and 4.7uF or 6.8uF to ground.

Now go and read the servo thread, particularly Tom's contribution.

This T-link will create a small instability depening of how close it's pole is to the servo's pole. This small resonance will also create a small hump in the frequency response. It's very obvious if you run a simulation which I recommend.

[Ted205]

i found the post andrew as hinting at -> DC Servo question...

Very helpful indeed. Has this been written up formally anywhere?

as it needs to be

they only thing i noticed is the diodes are not included in the above link. What if anything are these for ?

[peranders]

These diodes are for faster servo recovery when it's overdriven.