Why does a Mylar cap cause trouble as a coupling cap (high pass filter) but not as a low pass filter in a DC servo? Could it just be that one arrangement has less signal voltage across the cap because the CR time constant is bigger? Clearly any audio distortion created by the cap is fed back into the amplifier input by the servo so in this respect the servo is no different from using a coupling cap.
It IS a single op amp PACKAGE. Just a dual in a mini-dip. Think it through. Moving to a dual in the same space, is cheaper than using 2 Mylar caps. Today, the offsets on fet input op amps are good enough that the added stage will not significantly add to the inherent offset. (sorry about the pin numbers, pushed the wrong key)
Because of the long time constant there could only be harmonics of a sub 1Hz tone from the servo cap. Theoretically, that is.
OK, now it is time to address Wavebourn's statement.
The ADVANTAGE of a servo is that it ONLY operates to any significant extent BELOW the audio bandwidth, and the attenuation of the final resistor (adj) with the feedback to ground resistor (not shown), will reduce the sensitivity another 20-40dB further from ANY aberration due the IC or the cap. Therefore, the cap and IC are buffered from the audio path.
The ADVANTAGE of a servo is that it ONLY operates to any significant extent BELOW the audio bandwidth, and the attenuation of the final resistor (adj) with the feedback to ground resistor (not shown), will reduce the sensitivity another 20-40dB further from ANY aberration due the IC or the cap. Therefore, the cap and IC are buffered from the audio path.
Last edited:
You can't; that was my point. Returning DC to the + input requires a cap between the input jack and the return point of the servo, which defeats the whole purpose of putting the servo in for this sort of design. Otherwise, you inject DC into the source and if the source has a low DC impedance (e.g., a DC output from a preamp or DAC), the servo's voltage will be shorted out, causing the output offset to rise.
those of us who had been tortured with analog video, know that the DC-level after after the coupling cap follows the average level instead of being zero, thus 'clamping' was invented...a DC-servo may have its merits in audio, however the DC-level, whatever it may be, is unlikely be heard anyway. 
It's just the thought of DC through speakers .......those of us who had been tortured with analog video, know that the DC-level after after the coupling cap follows the average level instead of being zero, thus 'clamping' was invented...a DC-servo may have its merits in audio, however the DC-level, whatever it may be, is unlikely be heard anyway.
A DC servo is somewhat an easy trick whose worst purpose
is actually to keep the input differential(s) unbalanced.
I could understand that such a circuit is useful to get rid of the last few mV
DC offset wich can yield high current in case a very high OLG amp output
is shorted , even with no input signal.
As such it is useful in PA gears but surely not in home amplifiers.
is actually to keep the input differential(s) unbalanced.
I could understand that such a circuit is useful to get rid of the last few mV
DC offset wich can yield high current in case a very high OLG amp output
is shorted , even with no input signal.
As such it is useful in PA gears but surely not in home amplifiers.
We use servos for the same reason that many people use coupling caps. We want to keep DC off the outputs and any volume pots. We also want to have lots of AC gain, perhaps up to 100 dB, and still direct couple. Without a servo, we would HAVE to use coupling caps. Good coupling caps are large and expensive, so we use servos, instead.
I use servo designs that allow balanced from input to output, when necessary.
I use servo designs that allow balanced from input to output, when necessary.
John; it sounds silly.
I can say the same about coupling caps, that AC voltage on them operates to any significant extent BELOW the signal bandwidth, and open loop gain of the amp is much less than open loop gain multiplied by gain of servo.
Kinda like sticking a transformer on the outputs ...
Right, and McIntosh did that with great success.
Yes, servos do look sexy.
I think in JFET based fully symmetrical front end amps you cannot get away without a servo. This is the real reason you need them (initial offset even after matching and temp drift). For bipolars you do not need a servo, just initial offset adjustment which is typically in the region of 10 to 15mV but can be as high as 25mV. WRT drift, I measured levels from ambient to circa 60 degrees on two prototype power amps recently that was under 1mV.
If you do use a coupling cap in the feedback network, then it has to be oversized to avoid LF distortion, and I would of course say it needs to be decent quality, branded type.
The AP shows 0 (zero) distortion through an oversized electrolytic, and Edmond and Ovidui proved with the PGP amp.
I think in JFET based fully symmetrical front end amps you cannot get away without a servo. This is the real reason you need them (initial offset even after matching and temp drift). For bipolars you do not need a servo, just initial offset adjustment which is typically in the region of 10 to 15mV but can be as high as 25mV. WRT drift, I measured levels from ambient to circa 60 degrees on two prototype power amps recently that was under 1mV.
If you do use a coupling cap in the feedback network, then it has to be oversized to avoid LF distortion, and I would of course say it needs to be decent quality, branded type.
The AP shows 0 (zero) distortion through an oversized electrolytic, and Edmond and Ovidui proved with the PGP amp.
- Status
- Not open for further replies.