0
\$\begingroup\$

I want to make two "matched" 4-order active filters, the -3db frequency will be below 100Hz. I have several choices to choose:

  1. Choose four single version opamps. Each filter uses two of them.
  2. Choose two dual version opamps. Each filter uses one of them.
  3. Choose two dual version opamps. Each filter take two half-of-one-opamp.
  4. Go directly to the quad version.
  5. If there are more better one, please tell me :).

Is it worth to consider this when design an active filter? If so, then which choice are better? The quad version may derate in their performance, but may be routed easier. The single version may can't matched well, etc..

Any suggestions?

\$\endgroup\$
6
  • \$\begingroup\$ If there isn't something special going on, the matching between your filters should depend much more on the matching of the passive components in the feedback network, than on matching the characteristics of the op-amps. \$\endgroup\$ – The Photon Jan 28 '16 at 2:30
  • \$\begingroup\$ Thanks. But I want to know, if use method 3, are there some cons and pros. Even if it won't work more better, it shouldn't work more worse, right? \$\endgroup\$ – diverger Jan 28 '16 at 3:00
  • \$\begingroup\$ Without the frequency range of use, this question is difficult to answer... \$\endgroup\$ – MadHatter Jan 28 '16 at 4:31
  • \$\begingroup\$ @MadHatter: The -3dB frequency is below 100Hz. Because it's an active filter, so the input signal may contain higher frequency components. \$\endgroup\$ – diverger Jan 28 '16 at 7:19
  • \$\begingroup\$ Low pass, band pass or high pass? \$\endgroup\$ – Andy aka Jan 28 '16 at 8:50
1
\$\begingroup\$

Low pass active filters are great but up to a certain point. For instance you may have the 3dB point set at 100 Hz and, for a 10 kHz input you'd expect the output to be 80 dB down. well, don't be surprised if this isn't as clear-cut as you anticipated. Maybe 10 kHz will be OK but at 100 kHz the op-amp may not be such a good active component and the signal at the output starts rising up as frequency increases.

I have no idea what the highest input frequency could be for this application but this could play a vital role in deciding what op-amp to select and, you will find that the better op-amps (i.e. ones that are more suitable to extreme applications) are not available as a quad package.

For "my" type of work the best quad op-amp is the OP4177 - it generally has pretty good specifications all round but it wouldn't suit being the active op-amp in a 100 Hz Sallen Key LPF where the input frequency might be as high as 1 MHz and input signals at 1 MHz are expected to be attenuated by 160 dB (4 decades at 40 dB/decade) - that just won't happen.

So maybe you can find a quad op-amp that compromises DC performance (i.e. produces several millivolts of DC error) but has a much wider gain-bandwidth-product (GBP) so that it adequately deals with high out-of-band frequencies.

My bottom line is this - look at what the circuit is meant to do, simulate it at worst-case high input frequencies and see if it performs. Chances are that it won't do what you want without changing to a device that compromises DC performance. So then you end up using a dual op-amp that has both decent DC performance and high GBP.

\$\endgroup\$
4
  • \$\begingroup\$ Can you give some suggestions about method 3, I'm inclined to use that schema. In that, the two active filters may "matched" better, right? And for the higher frequency well beyond the capacity of opamp, I'll place another R-C pole before the active filter stage. \$\endgroup\$ – diverger Jan 28 '16 at 9:44
  • \$\begingroup\$ I've used method 2 and method 3 with no noticeable difference. I'd also place another RC between stages "just in case". \$\endgroup\$ – Andy aka Jan 28 '16 at 9:51
  • \$\begingroup\$ I've used method 2 before. But haven't tried method 3. Because I need better matching of the two filter, I wonder what can benefit from M3. BTW, the candidate is OPA2277. \$\endgroup\$ – diverger Jan 28 '16 at 9:56
  • \$\begingroup\$ I don't see any benefit of 3 over 2. Op-amp choice is good DC wise but watch the 1nA bias currents thru high value resistors. 1MHZ GBP might be OK but, all round I don't think this device would be as good as an OP4177 (quad) unless you were really into DC accuracy but if you were I feel sure you'd need a device with lower input bias current. It's noise numbers are good too but, with cut-off of only 100 Hz this will never be an issue. \$\endgroup\$ – Andy aka Jan 28 '16 at 10:09
0
\$\begingroup\$

At low frequencies it all doesnt matter .By low frequencies i mean say 70Hz where I did a Biquad in a previous life .At high frequencies coupling between opamp sections may muck things up .Capacitive coupling gets noticably worse at 20KHz so you know when to use seperate chips.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.