Author Topic: LPF for op amp feedback  (Read 2353 times)

0 Members and 1 Guest are viewing this topic.

Offline electrolustTopic starter

  • Supporter
  • ****
  • Posts: 562
  • Country: us
LPF for op amp feedback
« on: July 18, 2017, 01:30:01 am »
In an audio application, for an inverting op amp  in which you might want to add a LPF to the feedback section (a capacitor in parallel with the feedback resistor), how do you get say 100x gain?

Let's say your input resistor is 2k, you want G=100 so your feedback resistor is 200k.  For 200k cutoff
(is 10x a reasonable point for audio?) that's a filter cap of 4p.  Doesn't that approach PCB parasitics?

Would you instead limit your gain to say 10 (40p cap) and use a second 10x stage to get overall G=100?  Or use a passive RC filter on output?

Maybe you can select an op amp so that the GBW limit comes into play.
 

Offline electrolustTopic starter

  • Supporter
  • ****
  • Posts: 562
  • Country: us
Re: LPF for op amp feedback
« Reply #1 on: July 18, 2017, 10:37:42 am »
Thank you so much for all the links.  One problem I had with this is that there are so many op amp discussions out there that it's hard to find the ones that would cover my question.

Also I will mention there's a whole slew (pardon the pun) of considerations about things like microphonics and nonlinearity and distortion and such wrt. capacitors in audio processing.

You're reading my mind!  So this is for a preamp (hence high gain), and I was indeed worried about the microphonics of the capacitor.  I wasn't sure if that was a real thing I should be worried about in a practical circuit, or was it a minimal and more theoretical problem.  The beginner/general information I could find on it didn't even quantify the extent of the problem so I thought maybe it's just something to know about if you're a cool kid.

Not only is this for a sensitive (precision) application, the environment is harsh and by that I mean there is high vibration and high temperature.  With that in mind, are class 2 caps not appropriate even for bypass applications?  I'm thinking that the microphonics would create more of a problem than the cap is trying to solve.
 

Offline danadak

  • Super Contributor
  • ***
  • Posts: 1875
  • Country: us
  • Reactor Operator SSN-583, Retired EE
Re: LPF for op amp feedback
« Reply #2 on: July 18, 2017, 11:11:11 am »
Since this is a precision application you need to consider circuit sensitivity.
Here is an ap note for analog filters.


http://www.ti.com/lit/ml/sprp524/sprp524.pdf


An alternative is to do a digital filter, where sensitivities largely dependent
on clock accuracy and math precision.

Here is an example with a PSOC, all done onchip (except for coupling caps) -





Regards, Dana.


Love Cypress PSOC, ATTiny, Bit Slice, OpAmps, Oscilloscopes, and Analog Gurus like Pease, Miller, Widlar, Dobkin, obsessed with being an engineer
 

Offline Audioguru

  • Super Contributor
  • ***
  • Posts: 1507
  • Country: ca
Re: LPF for op amp feedback
« Reply #3 on: July 18, 2017, 05:39:13 pm »
Why do you need a lowpass filter to reduce high frequencies? An opamp already has a capacitor inside that cuts high frequencies. Most audio opamps (TL07x) with a gain of 100 cut frequencies above 35kHz.
 

Offline dmills

  • Super Contributor
  • ***
  • Posts: 2093
  • Country: gb
Re: LPF for op amp feedback
« Reply #4 on: July 19, 2017, 04:01:05 pm »
I do quite a lot of audio professionally.

100 times (~40dB all else being equal) is a LOT from a single stage, which is why you are finding such silly values.

That cap across the feedback resistor is mainly useful to cancel the cap from the summing node to the reference plane), as an RFI cap it is way too late as the output impedance of the opamp will not be that low up there where the open loop gain is low, so it is mainly a stability and peaking reduction thing.

For a 40dB stage (Which is almost certainly a mic preamp or something of the sort), I would be reaching for an instrumentation amp style three opamp architecture as it keeps the resistor values low, and resistor thermal noise is usually the dominant noise source in such things, gets you a balanced input for free and asks much less of the opamps.

A 100kHz or so corner is a good thing close to an input stage, but be a little wary of putting it right at the input, as anything done to a differential pair must be very careful about not converting common mode noise to differential noise.

My usual input topology is an RFI filter network (directly to chassis ground, right at the connector), a network to protect the electronics against EFT (Or Phantom power accidents, same sort of thing),
a diode clamp to the rails and then an instrumentation style gain stage (THAT Corp make some excellent single chip mic amp parts for this), I might wrap the first pole of a 1-200kHz LPF around this, then follow with another couple of poles around another opamp.

C0G is essentially blameless, and is good for filters and suchlike (Arguably better then polystyrene, and very much easier to handle!), for inter stage coupling (as opposed to filtering) I actually favour LARGE electrolytics, very much not the fashion I know!
My reasoning is that if you make the cap large (so it's corner frequency is WAY below the audio band, then the signal voltage developed across the cap will be negligible), this is clearly not workable in filters or (Worse) speaker crossovers. 

SMT Film is trouble from a manufacturing perspective (Very picky thermal profile), avoid if you can.

Regards, Dan.

 
The following users thanked this post: electrolust

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21606
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: LPF for op amp feedback
« Reply #5 on: July 19, 2017, 05:19:17 pm »
Note that you need enough GBW to get there, in the first place.

If you want 200kHz BW and 100 gain, you need 20MHz GBW.  You might add a little extra headroom, to shape the band edge properly, say if you want to sharpen it with a little peaking, or use a second or third order filter for a more rapid cutoff.

Or just to get lower distortion, since anywhere you're pushing GBW, that means the feedback is nearly zero, and the amp's quirks show through at full magnitude (distortion, frequency response, PSRR, input noise).

But really, for audio, that kind of filtering would be fine to employ down at 40kHz to 100kHz.  A Bessel type filter, cutoff around there, will give plenty of audio bandwidth, flat phase response and low sensitivity to component variations (in other words: easy to get consistent phase between multiple channels of the same circuit).

And yes, if you use overly large value resistors, you'll get overly large feedback values, and small capacitance values.  In high gain circuits, very small input resistors are typical, which also keeps the noise low, because remember, the noise of that resistor directly adds to your signal.  (The feedback resistor is "refrigerated" by the amp, however, so its noise does not contribute.)  If your source isn't that low impedance, consider using a noninverting, buffered, or instrumentation amp configuration.

20MHz GBW isn't extreme, and there are plenty of low noise, low distortion audio amps suitable for that.  If you needed any more gain, or bandwidth or flatness or lower distortion, it would be wise to use multiple stages.  Then the total GBW is effectively multiplied, and you're golden.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf