Single or Dual Op-Amps in Active Filter

[jpanhalt]

My project will use a one- or two-stage active band-pass filter (single amplifier biquad) for 500 kHz.
Schematic for a single stage is attached.  The other sage is similar.

Op-amp is a AD8055/8056.  Analog Devices seems to recommend using a dual op-amp (AD8056), but that does not allow decoupling of the supplies for each section.   

Size and cost are not much different regardless of what I use.   I am just asking for other opinions on best practice.

Regards, John

EDIT: Fixed schematic
 

[suicidaleggroll]

My project will use a one- or two-stage active band-pass filter (single amplifier biquad) for 500 kHz.
Schematic for a single stage is attached.  The other sage is similar.

Op-amp is a AD8055/8056.  Analog Devices seems to recommend using a dual op-amp (AD8056), but that does not allow decoupling of the supplies for each section.   

Size and cost are not much different regardless of what I use.   I am just asking for other opinions on best practice.

Regards, John

If you were filtering two different signals, I'd say use two single amps to reduce crosstalk (assuming crosstalk is important in your application).  For two filter stages on a single signal, I see no harm in using a dual amp.

[jpanhalt]

Same signal, just two stages to get the Q I want.

John

[Benta]

John, I'm curious here: how did you select this topology?
It's honestly a new one to me (multiple-amp biquads I know), and I'd be interested to learn more. Can you provide a good reference/introduction/whitepaper?
Sallen-Key and the like land by me directly in the garbage, but I've worked a lot with MFB filters (lowpass mainly) with good results (knowing the limitations)

Thanks.

[edavid]

1. Your op amp is wired backward.

2. Why attenuate by 300X and then amplify by 40X 

3. Why use 4 resistors in parallel instead of a single 430ohm?

[Kleinstein]

This is about the frequency where one normally uses LC filters. Especially if a high Q is needed. The GBW requirements for a high Q active filter is pretty high.

Single OPs might have a slight advantage because of less local heat and less parasitic coupling.

[Benta]

1. Your op amp is wired backward.

2. Why attenuate by 300X and then amplify by 40X 

3. Why use 4 resistors in parallel instead of a single 430ohm?

For 2 and 3, I agree 100%

For point 1, I got curious about this filter topology and did a bit of reseach.
What I found was this:

http://www.globalspec.com/reference/71146/203279/4-3-general-biquad-circuit-using-a-single-op-amp

Which looks a bit convoluted (eg, parallel resistors from the input). I haven't done the calculations, and don't want to, as the advantages over an MFB filter are unclear..

The other one is this:

https://www.ganino.com/files/BSTJ/Vol62/bstj62-6-1399.pdf

This looks a lot like the OP's filter, but has one essential input marked 1'
This is a feedback signal from the following filter stage, and without this, the OP's schematic will not work, as it's then just a positive feedback amp = Schmitt trigger.

I'll stick with MFB for the time being... 

[jpanhalt]

@Benta,
1)  Original design was from the Analog Devices filter design app.   I, too, didn't know what it was called specifically.  AD calls a MFB topology.  Discussion of that topology is in  Schaumann and Van Valkenburg's textbook on analog filters.  There are no reviews with the table of contents and excerpts that I could find.  It has been uploaded in its entirety by someone here: https://www.academia.edu/15449434/Design_of_Analog_Filters_Rolf_Schaumann_Mac_E_Van_Valkenburg

I always hesitate to give that and similar links to copyrighted materials.  I did report it to OUP incidental to a question about the book I bought.  Since the book is out of print, OUP has apparently done nothing about it (yet).   As I pointed out to the person at OUP, I found the book such a pellucid read that I bought it ("new"  for about $40 UDS).   It is honestly my bedtime reading. It compares favorably to Horowitz and Hill, 2nd edition.  There is a later "international" edition at a reasonable price, but again, I could not find substantive reviews of it.  The third author on the newest edition apparently was a trainee of Prof. Schaumann.

2)  I had the op-amp reversed in the original post.  That has been corrected.   Signal input should be to the inverting terminal.  The error in my post resulted from my working immediately before on routing a two-IC version using SOT23-5 chips.  I was too lazy to make a new symbol and just grabbed an ordinary dual op-amp symbol.   As soon as I posted, I notice the inputs were reversed and tried to fix it in time.  You guys are just too quick, which is good.

3) The feedback to the non-inverting input is for "peaking" the Q.   Obviously, oscillation is an "added feature" if you try for too much Q wit one stage.  The genesis of my question is related to that propensity to oscillate.

Regards, John

[edavid]

It's called a Deliyannis-Friend filter.

I don't see any version of it with DC positive feedback though.  I don't see how that is going to work.

Did you simulate it?

[jpanhalt]

Yes, I simulated it (LT Spice) and AD simulates it too.  I built it.  Works great on a solderless breadboard no less.   Single stage (carefully tuned) gives a Q of about 100.   Conservatively, I will probably go to two sections.  I covered the feedback to the non-inverting terminal in #3.  The book I referenced discusses it in more detail.  And yes, it is also called a Deliyannis-Friend filter.  But if you build it, you will see the important effect of the feedback to the non-inverting input, which typically is not in the Friend or Deliyannis-Friend filter.

I am still interested in empirical information about power supply coupling in dual packages with a device such as this.

John   

[Benta]

@John
Thank you for new insights on filters here.
Don't be shy about linking to material (at least not to me), I've skimmed the PDF and decided to buy the book (if I can find it).
I prefer having the real book on hand, and have no problem giving out the 50...100 $ it costs.
The problem is always: which books are really good? There is so much crap out there, that finding the good stuff is hard.

Good luck with your filter, I understand the circuit now.

As to the original question about crosstalk between amps on a single die? Ignore it. I've never heard of anyone having a problem, and I'm 30 years in semiconductors.

[jpanhalt]

I bought mine off Amazon.   If you want details about why I contacted OUP, PM me.  I don't think my issue with the book should affect you in the least.  There is a dealer on Amazon with a "new" book that is only a little more expensive than the used ones.  I suspect it may have the same "issue," but shipping from the US would be expensive.

Thanks so much for the advice on the die.

John

[Lee Leduc]

Here's a couple Mini-Tutorials from Analog Devices for Sallen-Key and Multiple Feedback active filters. Check out websites like TI, Analog Devices, Burr Brown for all kinds of tutorials like this. Hope these help.

[Benta]

Here's a couple Mini-Tutorials from Analog Devices for Sallen-Key and Multiple Feedback active filters. Check out websites like TI, Analog Devices, Burr Brown for all kinds of tutorials like this. Hope these help.

Yes... well... my feeling is that 'jpanhalt' knows exactly what he's doing...

(Hint: read previous posts)

[LvW]

It's called a Deliyannis-Friend filter.

Yes - the shown filter is a modification of the well-known MFB topology - introduced by Deliyannis.
The advantage of the additional pos. feedback is: Q enhancement without excessive gain increase (slew rate!!) with an acceptable componenet spread at the same time.

To jpanhalt:  The shown filter stage will work with good/acceptable properties - as long as a second-order bandpass can fulfill your requirements. However, I doubt if this will be true also for a two-stage 4th order filter.
If you really need a 4th-order high-Q bandpass you should also consider alternative topologies - preferrably, no cascade design but DIRECT realizations (FDNR design, leapfrog, PRB-structures). 

[Benta]

@John:
Found the book and ordered it. 60 Euro (used, good condition) is OK for me.
But it seems to be very desirable, prices at $300+ are around at Amazon.

Weirdness: it's listed as "Casebook in Contemporary Fiction" Go figure...