Cascading passive filter and Sallen-Key?

[tinska.h]

Hello,

I have never seen any designs that would cascade a passive RC-filter before Sallen-Key filter. So can it be done? I don't see any reason why not if the signal is pre-amplified.

In the application, signal is a pulse of lets say 100µs followed by a 5s pause(=OFF). In application environment there is the normal AC noise+harmonics, 400-800Hz and some noise from 15kHz-40kHz. I have limited op-amps (2), which the first one is used to pre-amplify the signal. I need a strong high-pass filter but then again I need a strong low-pass filter. My thought was to have PREAMP -> RC LPF -> Sallen-Key -> LPF Anti-Aliasing -> ADC. Do you guys see any problems with this?

Btw. due to limited op-amps and requirement to achieve good attenuation, would elliptic/Cauer be good choice? (I have never used one) I suppose since it's a non-continuous and pulsed signal, I don't see that much problem with phase shift or group delay, since signal processing in digital domain is only to catch average peak. Any opinions?


Thanks

- - - Updated - - -

Oh and yes, I'm working on single supply and low voltage (~2V). The need to filter before final stage is that I'm afraid of hitting the rails and loosing the actual signal if none of filtering is done before final op-amp.

 

[LvW]

Hello,

some questions/comments to your inquiry:

* what means "I need a STRONG filter" ? Do you have no specific attenuation requirements?
* Of course, you can use a simple RC stage in front of your 2nd order S&K filter - however, it bust be decoupled using a buffer in front of the 2nd stage
* But you should know that there is a special design strategy for designing a third order S&K using only one single opamp
* In general: Lowpass attenuation for frequencies far above the path band depends on the filter order only - not on the approximation (Butterworth, cauer, etc,)
* The selected approximation determines only if the transition band (between path and stop region) is rather small or not. But you must know that the consequence of a small transition band could be a certain ripple in the pathband)
* To find the optimum filter configuration for a certain application requires a clear specification of the requirements (Filter gain, allowed ripple in the path band, end of the path band, attenuation values for specific frequencies or frequency ranges.

 

[FvM]

I have never seen any designs that would cascade a passive RC-filter before Sallen-Key filter.

It's a standard third order low- or high-pass topology. Of course the passive filter has to be reflected in the values of active second order block. The technique is called "absorb first order". I presume that you are trying to implement a defined filter prototype.

Alternatively, the first order filter can be placed behind the second order block, but you lose the low output impedance of the active filter.