Use DC servo as a Rumble filter ?

[gurevise]

Guys,
What do you think of using a DC servo as a Rumble and/or high pass filter?
DC servo effectively filters low frequency signal, if corner frequency is set to 10...20 Hz.
I guess the question is: are there any downsides to this type of rumble filter implementation? Suggested phono solid state design (minus DC servo) does not have 20Hz roll off and flat down to almost DC, so some kind of high pass is probably a good thing. The question is how much of low pass do I need.
Here is the schematic that shows it implementation



Let me know what you think.

Thanks
Sergy

[gootee]

Have you tried running a representative rumble-like sine signal through it, with a time-domain (transient) simulation?

A DC Servo usually takes only very low frequencies and creates a signal that is added back into the signal path, to try to "cancel out" the DC.

It sure SEEMS like a servo could be a nice efficient way to make a high-pass filter with a very low cutoff frequency.

Look up the specs for typical rumble frequencies and amplitudes, and see if you can design a servo that will be effective in canceling them, while also NOT affecting the desired low audio frequencies. Audio goes pretty low. I don't think that you would want the servo to have ANY effect at 20 Hz, for example (any effect should be WAY below audible, at least), and I would probably say 15 Hz, to be safer. So your cutoff frequency might need to be "much less than" 20 Hz, depending on how you implement the servo and any post-filter.

You will probably want a good low-pass filter after the servo, before the signal is injected back into the signal path, in order to make sure that you don't affect any low audio frequencies.

Remember, too, that you might want a voltage divider just before the resulting filtered servo output signal goes into the opamp input, (back) in the signal path. (And a unity-gain buffer just after the divider will help to see what's actually happening, during simulations.) Also note that a DC Servo usually uses a "differential integrator" topology.

See the major DC Servo thread at:

DC Servo question...

In the later parts of that thread we went kind of crazy with servo and AC-cancellation designs. But you can probably learn a lot about what you need for this design, there.

And there is another DC Servo thread at:

PA100 with DC SERVO?

P.S. What's up with the middle part of your schematic? It looks like you have something resembling an instrumentation amplifier but it's connected with single-ended instead of differential inputs. What is its purpose?

Cheers,

Tom

[gurevise]

Tom,
First 3 op-amps are approximation of INA103 instrumentation amplifier IC.
I do not have model for it so this is close "enough" to simulate EQ.

Good comments. I'll see about analyzing transient behavior.

Thanks
Sergy

[DF96]

You would need to ensure that your servo is well behaved, otherwise you could get wiggles in your frequency and phase response. Adding extra rolloffs to the servo could make things worse. You would need to be up to speed on servo design and stability. Any caps in the servo would need to be chosen carefully, as they will be setting the LF point of the amplifier so need to have linear dielectrics and be matched between channels. Much simpler to just use a coupling cap to do it!

[Ken Newton]

Originally Posted by DF96

You would need to ensure that your servo is well behaved, otherwise you could get wiggles in your frequency and phase response. Adding extra rolloffs to the servo could make things worse. You would need to be up to speed on servo design and stability. Any caps in the servo would need to be chosen carefully, as they will be setting the LF point of the amplifier so need to have linear dielectrics and be matched between channels. Much simpler to just use a coupling cap to do it!

Not to mention the power supply rejection/interaction concerns of the the servo's error amplifier. Better to use a coupling cap., IMHO, with relatively high impedance circuits such as this. Servos make much more sense for low impedance circuits, such power amplifers, where the required coupling capacitor would otherwise be a very large electrolytic.

[Calvin]

Hi,

You can do just what You proposed. Especially easy with the INA103/163.
You can drive the Ref-Pin which is typically tied to gnd with a dc-servo signal.
If the bandwith of the servo is set to 20Hz its a rumble-filter (after RIAA/IEC).
The DC-servo should be related to the first stage of the Phono-Pre (the INA) because of the large amounts of gain needed, which result in large output-offset-values.

jauu
Calvin

[cbdb]

Seems like a bad idea to use a complicated replacement for the simple RC filter. Is the DC servo linear enough or will it introduce extra low freq. distortion?

[Calvin]

Hi,

this implementation of the servo to the INAs reference pin works like a charm without any issues or flaws. Not only in my own circuits, but in a couple of DIY incrnations and well recommended commercial designs too.

jauu
Calvin

[gurevise]

Calvin,
I'm a bit concern about doing servo around the first gain stage. I'd think, it would add some noise to the signal path. I'm hoping that INA103 (first 3 Op-Amps) would work good enough without servo and create some reasonable DC offset. Than the last gain stage would correct DC offset at the output.

Sergy

[Calvin]

Hi,

the INAs work best with gains around +40dB and quite well even up to +60dB.
With such high gains, Offset easily becomes an issue. The input stage of the INAs is bipolar in nature, hence rather large bias currents are needed for low voltage-noise figures. The difference in input bias currents generates an offset which appears amplified at the output. Since the INAs gain is linear over frequency, an Offset of the first stage reduces headroom.
It´s the lower gain stages that are less prone to this problem.
The servo does indeed inject a small amount of noise, but this is negligible and sonically irrelevant. A phono-stage implemented with the INA103 or INA163 and FET-OPs as servos (OPA134, AD8610, etc) is dead quiet and allows to drive low output MC pickups to their best.

jauu
Calvin