How to define opamp specs within an active-RC filter

Hi, all,

If we have to design an active-RC filter with some specs given, such as cut-off frequency, SFDR, and load, then how to define the opamp specs (GBW, slew rate, rin, rout)? It would be appreciated if any of you could provide some literatures on this topic.

Thanks in advance,
abcyin

Hi,

I'd say you can't specify an OPAMP just by the RC.

Usually, if you can use a passive RC, then use it.
But when you think you need an active filter, then specify "why".
Usually this "why" are the most important parameters for the OPAMP.

And there are some "independent" parameters, like supply current, supply voltage, offset error, distortion...
But nobody here can tell you which parameters are important for your application.
I assume you are the only one to know..

There are applications, where distortion need to be low, but offset voltage is no problem, and there are applications the other way round.
Just open any Opamp datasheet and go through the table with the electrical specifications. Decide column by column if this parameter is important for you and where to set the limits.

Klaus

In addition, the answer to your question depends on the topology of the active flter you will select.
For example, some filter structures are based on idealized "infinite" gain stages and some others on finite and fixed gain stages.
Alternatively, you could use impedance converter concepts.
In general, the most important opamp specifications are "transit frequency ft" (unity gain bandwidth) and "slew rate"SR.

SFDR isn't a linear circuit specification (because they don't generate spurious signals).

Here's an article that discusses op amp selection for active filters, and here's an excellent reference on active filter design.

Yes - I only can strongly recommend the above given ref for active filter design (one small drawback: GIC-based designs are not covered).

Thanks for all of you, the suggestions are constructive. I'll come back to the discussion after reading the documents. Thanks.

L.P Huelsman was one of the original opamp gurus at Burr Brown, back in the early 1970s. And the app note that crutschow linked to, is some of its best work.

When Burr Brown was acquired by Texas Instruments, some "golden nugget" app notes were indeed lost.
I'm glad that some have been unearthed.

The short answer is everything. Start somewhere and then follow an iterative process. For example what voltage rails do you want to use? Do you think you need RRout to swing near to those rails. Do you need RRin because your input signal is large (for non RRin input common mode restrictions can be substantial at low voltages)? This narrows down available choices substantially.

Now you can start actually thinking about performance such as GBW product and slew rate. Does DC matter? Then offset, offset drift, bias and offset current matter (and make sure to balance the impact of those, no point having high offset currents dominate a low offset voltage amplifier).

Now did you end up in a place that's acceptable or not? Maybe you could change to an inverting filter topology (Multiple feedback vs Sallen Key) which means input common mode of 0 and ditch the RRin requirement....maybe going up in supply would eliminate both RRI and RRO requirements? Maybe going down in supply would open up higher performance and/or cheaper amplifiers.


Unfortunately everything is interconnected and it turns out there is a reason there are 1000's of opamps on the market.

The last thing I'll mention since its been biting us lately is that its wise to ensure passive components knock out frequencies above the range the amplifier can handle (in terms of GBW and slew). If the amplifier 'sees' frequencies above those ranges it likely won't behave in a 'linear' fashion. For example high frequency AC becomes DC because of mismatched output slew rates etc.


Finally I'll plug LTSpice as a very good platform for looking at many of these things (though there are many others, some specialized for filters). I recommend the "UniversalOpamp2" model which has parameters for GBW and slew rate. Of course you can use the included models (And now LTSpice includes tons of Analog parts), but the universal model simulates those important parameters very efficiently.

Also this is my preferred calculator for component values:
https://www.beis.de/Elektronik/Filter/ActiveLPFilter.html

1) Filter type: HP,LP,BP,notch, All Pass, custom
2) characteristic: Burterworth, Bessel, Linear Phase, Chebychev, Raised Cosine etc
3) Pass Band gain (dB)
4) BW -3dB point
5) Bandstop attenuation (dB) @ f OR filter order
6) Output current, Slew rate, output swing
7) Output Offset error, Gain error , tolerance stackup (Input offset= Iin*R mismatch, Vio, Iio)
8) Output Noise level from gain and input R and input V,I per root (Hz) BW
9) Cost, Package ( Quad, single ) SMT or THT

etc