Another DC servo question.

[Dxvideo]

Hi all,

This is the servo aided preamp that I am planning to build for my gainclone...
My question is;
In the simulation, there always be a -85mV offset on outputs.. Am I making a mistake?
And, should I put a DC blockage capacitor after the pot? While my gainclone has no Ci cap (I am happy with its DC offset value = 40-50mV)
Thx in advance...

[Mooly]

Hi,
DC servo on a pre-amp. I wouldn't even go there
Why not just use the offset null facility of the OPA627 if it bothers you that much.
As to your servo, you have time constants all over the place, which will cause all sorts of problems. The servo doesn't directly cause the output of the OPA627 to settle at zero, it causes the output to settle at the required voltage to counter the "offset error" of the TL081, which is not the same thing. To get an absolute zero at the OPA627 you would still have to null the TL081
The long settling time of any servo means you would also need a 100% foolproof switch on delay to the speakers to prevent damage.
Null the OPA627 -- much better, and make sure the rails to it are well stabilised and clean.

[AndrewT]

I can't see anything wrong.
But, the two 100k in the servo output filter reduce the correctable offset to <60mV for +-12Vdc supply.
Check what the simulator has for the voltage on the servo output. It may be beyond the range of correction.
It may be that Zin=100k is causing an offset that is too large depending on the simulator model.

The C6 is doing nothing. Either omit it and use the DC blocking in the source equipment to ensure there is no input offset or increase it's value >=1u5F.
Don't fit a DC blocker on the output. There should be no DC to block, that's the whole purpose of the servo if it is operating correctly.

The power amp will still see the variable source impedance of the volume pot.
What about adding a similar DC servo to the power amp? But, reduce the 100k filter resistors to <=51k
If you do go for full DC coupling then do seriously look at DC detection and relay muting of power amp input and output.

edit,
regarding Mooly's point about the servo's offset, both input pins have 1M0 and that should minimise the opamp offset.

[Dxvideo]

Originally posted by Mooly
Hi,
DC servo on a pre-amp. I wouldn't even go there
Why not just use the offset null facility of the OPA627 if it bothers you that much.
As to your servo, you have time constants all over the place, which will cause all sorts of problems. The servo doesn't directly cause the output of the OPA627 to settle at zero, it causes the output to settle at the required voltage to counter the "offset error" of the TL081, which is not the same thing. To get an absolute zero at the OPA627 you would still have to null the TL081
The long settling time of any servo means you would also need a 100% foolproof switch on delay to the speakers to prevent damage.
Null the OPA627 -- much better, and make sure the rails to it are well stabilised and clean.

Hmm...
In fact I am not sure about the long term DC Offset stability of OPA627.. I am planning to use my gainclone as fully DC coupled. So if anyvalue of DC occours on the preamp output then this would be a risk for my speakers. May be I am a bit sceptic.. Are you sure if I trim the OPAs offset with 1-8 pins it keeps itselfs stability at long term (for example in a warm box) ?

Originally posted by AndrewT
I can't see anything wrong.
But, the two 100k in the servo output filter reduce the correctable offset to <60mV for +-12Vdc supply.
Check what the simulator has for the voltage on the servo output. It may be beyond the range of correction.
It may be that Zin=100k is causing an offset that is too large depending on the simulator model.

The C6 is doing nothing. Either omit it and use the DC blocking in the source equipment to ensure there is no input offset or increase it's value >=1u5F.
Don't fit a DC blocker on the output. There should be no DC to block, that's the whole purpose of the servo if it is operating correctly.

The power amp will still see the variable source impedance of the volume pot.
What about adding a similar DC servo to the power amp? But, reduce the 100k filter resistors to <=51k
If you do go for full DC coupling then do seriously look at DC detection and relay muting of power amp input and output.

edit,
regarding Mooly's point about the servo's offset, both input pins have 1M0 and that should minimise the opamp offset.

Sure.. I will put a DC sensing relay on gainclones output.
The C6 is there because if any DC offset from source comes over servos control limit (may be 2v ) then what happens? I am not sure about that..

[Mooly]

Hi Andrew,
You probably recall I used a servo in my MOSFET amp. The problem with introducing more than one time constant is that although the output may settle correctly at zero volts quiescent, under dynamic conditions it causes problems. Put simply it's like a see sawing effect. Say the output of the TL081 goes positive ( and at switch on this effect is a real problem ), C3/R3 adds a further "delay" slowing the correction voltage. So the output of the OPA627 over compensates, the TL081 now reduces it's output but again this time it under compensates. The effect is a nice damped oscillation at the output.
As you say the power amp is the place to do it.

[AndrewT]

if the input offset is 60mV then the servo opamp will send it's output to the supply rail.
This is due to the 200k:1k0 resistor ladder in the inverting input.
60mV * 200 =12V = supply rail voltage.

But, none of your sources should send a DC offset into your system.
Every source should have a guaranteed near zero offset (by design).
Sort the problem at source.

[Dxvideo]

I supposed to have more powerfull DC servo by using it in just preamp stage.. Because its hard to control huge offsets like in the power amps (as discussed before)..
Am I wrong?

[Mooly]

Hi Ozgur,
It really is a non problem this. The offset untrimmed should be negligable anyway. Why are you so concerned over it.
Trim it with a good quality multi turn pot and that's it -- zero come come freezing cold or baking hot

[AndrewT]

Originally posted by Mooly
Hi Andrew,
You probably recall I used a servo in my MOSFET amp. The problem with introducing more than one time constant is that although the output may settle correctly at zero volts quiescent, under dynamic conditions it causes problems. Put simply it's like a see sawing effect. Say the output of the TL081 goes positive ( and at switch on this effect is a real problem ), C3/R3 adds a further "delay" slowing the correction voltage. So the output of the OPA627 over compensates, the TL081 now reduces it's output but again this time it under compensates. The effect is a nice damped oscillation at the output.
As you say the power amp is the place to do it.

I don't agree that a damped oscillation is the consequence of using a DC servo on either the the power amp or the pre-amp.
One set up criteria is that the amp should have zero offset from cold. The servo then attenuates any offset that develops as the temperatures stabilise towards operating condition.
A problem does occur if a fast pulse is sent into the input, but that is not a music signal. I suspect that this problem can be made worse if two servos are cascaded with inappropriate RC time constants. However no one came to our aid in the servo thread at the time to assist with servo stability/design.
Many commercial pre-amps have a servo nowadays, since the chips are so cheap to solder in place.

[AndrewT]

Hi,
a fully DC coupled amplifier and/or pre-amplifier requires the input connection with Rs to be permanent to maintain the correct DC conditions at the respective inputs.
If an interconnect is disturbed/removed that would generate an offset that will cascade through the system to the speakers. That is why DC detect and isolation is so important