Let's Discuss Servos (Previously OPA188 Thread)

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: OPA188, OPA2188 and OPA4188 Low Drift Amps for Servos

Post by mediatechnology »

We've been discussing the Bandwidth/slew rate limitations of the DC-precise but slow op amps in this thread as well as here.

I ran some more quick tests using an OP07 servo-ing a 5534.

The servo type was simple, inverting.

Two sets of components were used to set a 1.6 Hz corner: 100K/1uF (film) and 1M/100nF. The input to the 5534 stage (inverting gain of -1) was 20V p-p. The OP07's input resistor also "sees" 20V p-p.

I monitored where the servo response stopped falling at a -6dB octave rate ("capacitive") and started rising again ("inductive").

With the 100K/1uF combination the OP07 began to produce an "inductive" response at approximately 20 kHz.

With the 1M/100nF "inductive" behavior began at ~100 kHz.

Lowering the level below 20V p-p made measurement difficult, due to noise, RF ingress and the 5534, but it appeared that the change from capacitive to inductive response shifted upward in frequency for the same RC combination. This would be expected behavior. FWIW the OP07 has an open loop Rout of 60 Ohms.

I can draw two maybe three conclusions already.

The OP07 has sufficient BW, current capability, or whatever, as long as the input resistor remains fairly conservatively large e.g. 1M.

An output RC network really should be used having a fairly low corner because the -6dB octave response of the integrator just isn't enough and there's a noise advantage.

If low-value input resistors are required, <<1M, having a passive LPF at the input isn't a bad idea.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: OPA188, OPA2188 and OPA4188 Low Drift Amps for Servos

Post by JR. »

How is the servo loop closed? I think I follow the forward path 5534 is unity gain inverting, and it's output drives an inverting servo stage with alternate R and C values. The output of this servo section then gets routed to the 5534 + input (100%) ?

It might be interesting to see how it looks with a square wave... Obviously rise time limit the square wave so the 5534 can keep up, and see how well the servo keeps up with the 5534,,, if the 5534 is allowed to slew limit the servo behavior doesn't matter since the waveform could introduce some DC content.

BTW thanks for investing bench time in this.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: OPA188, OPA2188 and OPA4188 Low Drift Amps for Servos

Post by mediatechnology »

The output of this servo section then gets routed to the 5534 + input (100%) ?
The loop is indeed closed from the 5534 output to the OP07 inverting input via Rin.

From the OP07 output, there is a voltage divider of 23 (220K/10K) feeding the non-inverting input. I did this to magnify the OP07's response.

Sorry, forgot to mention that this 'round - that part of the test circuit was in an earlier post.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: OPA188, OPA2188 and OPA4188 Low Drift Amps for Servos

Post by mediatechnology »

Servo Test Circuit

Image

The 5534 serves as a "loop closer" for the device under test.

Although off-topic, this circuit provides a very clear indication of when the 5534 (or the source) reaches the onset of asymmetric slewing and the resulting THD even-order products. The DC term at "TP" rises quite quickly.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

What About A Deboo Servo?

Post by mediatechnology »

This non-inverting integrator dates back to 1967. It meets one of John's requirements: A passive pole on the input.

It would appear that resistor match is important.

Image
Deboo non-Inverting Integrator

See: http://www.maxim-ic.com/app-notes/index.mvp/id/1155

More Howland and Deboo: http://www.national.com/an/AN/AN-1515.pdf

A Deboo alternative: http://webfea.fea.aub.edu.lb/dsaf/Publications/20.pdf
stoeffle
Posts: 6
Joined: Sun Apr 10, 2011 7:32 am

Re: Let's Discuss Servos (Previously OPA188 Thread)

Post by stoeffle »

Hi all,
may I throw in an aspect which seems to go by unnoticed in all the posts posted ahead?
Fact: An audio circuit with a DC-servo to squeeze out that DC offset is, by all means, a HPF.

It is the matter of group delay distortion. I have been dealing a lot with this subject, as some musicians gave me a feedback (specially bass players) that they very often find it annoying that (on reproduction) part of the audio frequency range gets differing delay values when passing through a piece of equipment.
These people have kind of sensitive hearing regarding phase relationships. :o
Lately I designed a servo for a DC-coupled audio power amp and it was quite a piece of work in simulation to get, for the overall performance, a flat group delay over frequency response curve from 10 Hz way up and beyond 20kHz.
All those coupling capacitors, used either to pass AC-signals from one end to the next mounted onto differing DC levels, be it to avoid scratchy pots, generate a HPF characteristic with serious implications regarding group delay. Even if designers boast that the -3dB point of it is well below 1 or even 1/10Hz, just take a look at the phase over frequency (=group delay) plot: it is smashing, what happens over there...

Maybe I am completely mistaken and wrong, but I wanted to throw this hat into the discussion arena to find out what is the opinion of the pros in this area.
Thanks.

stoeffle++++.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Let's Discuss Servos (Previously OPA188 Thread)

Post by JR. »

stoeffle wrote:Hi all,
may I throw in an aspect which seems to go by unnoticed in all the posts posted ahead?
Fact: An audio circuit with a DC-servo to squeeze out that DC offset is, by all means, a HPF.
true fact dat...
It is the matter of group delay distortion. I have been dealing a lot with this subject, as some musicians gave me a feedback (specially bass players) that they very often find it annoying that (on reproduction) part of the audio frequency range gets differing delay values when passing through a piece of equipment.
hmm... OK... I don't argue with people about what they claim to hear.
These people have kind of sensitive hearing regarding phase relationships. :o
Lately I designed a servo for a DC-coupled audio power amp and it was quite a piece of work in simulation to get, for the overall performance, a flat group delay over frequency response curve from 10 Hz way up and beyond 20kHz.
A servo tuned for a pole frequency way way below 20Hz should be of little consequence to the audio passband.

For beyond 20kHz I wouldn't expect anything significant from a very very low frequency servo. If there is some statistical quirk about looking at low frequency phase shift in terms of delay in the context of very HF passbands it is probably just that and not significant.

All those coupling capacitors, used either to pass AC-signals from one end to the next mounted onto differing DC levels, be it to avoid scratchy pots, generate a HPF characteristic with serious implications regarding group delay. Even if designers boast that the -3dB point of it is well below 1 or even 1/10Hz, just take a look at the phase over frequency (=group delay) plot: it is smashing, what happens over there...
The use of a DC servo is to eliminate poorly implemented and behaving simple blocking capacitors from the audio path.
Maybe I am completely mistaken and wrong, but I wanted to throw this hat into the discussion arena to find out what is the opinion of the pros in this area.
Thanks.

stoeffle++++.

If a 1 Hz servo bothers you retune it for ,01 Hz. The use of good film caps and proper topology is precisely so it will be well behaved in the audio pass band and above.


JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Let's Discuss Servos (Previously OPA188 Thread)

Post by mediatechnology »

Thanks for joining us stoeffle.
Fact: An audio circuit with a DC-servo to squeeze out that DC offset is, by all means, a HPF.
One analog I draw is that servos are a way to make little capacitors act like big ones.
A servo tuned for a pole frequency way way below 20Hz should be of little consequence to the audio passband.
Except that we found that a simple servo can, depending on component values and op amp chosen, have an output that does not fall indefinitely with increasing frequency.
JR: You made a good case for the passive LPF at the input. (And also at the output.)
My point is that we may think the pole is <<20 Hz but is it really?
What happens when our big synthesized capacitor behaves like an inductor?
Lately I designed a servo for a DC-coupled audio power amp and it was quite a piece of work in simulation to get, for the overall performance, a flat group delay over frequency response curve from 10 Hz way up and beyond 20kHz.
Can you show us?
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Let's Discuss Servos (Previously OPA188 Thread)

Post by JR. »

Indeed lets see some smoking guns.

The best cap is no cap, after that give me a good film cap maybe with a good opamp wrapped around it. IIRC I was arguing that the performance of the servo amp can matter.

I know some folks who use lots of servos with no problems, I haven't used one since my last phono preamp and that design was over designed in many regards.

JR
Cancel the "cancel culture", do not support mob hatred.
stoeffle
Posts: 6
Joined: Sun Apr 10, 2011 7:32 am

Re: Let's Discuss Servos (Previously OPA188 Thread)

Post by stoeffle »

@ Wayne: .. show us?...

I am trying to extract the part of the schematic...have to do this on the other PC, sorry.

stoeffle.++++
Post Reply