understanding Op amp Filtration Circuit

[rrn]

Can anyone help me out in understanding circuit shown in the image. Little bit about circuit, photodiode output is fed to op amp with negative feedback from R5 and R6.  R3 and R4 form voltage divider to give reference to NON inverting terminal of opamp.  is the opamp 1st order or second order high pass filter? what does C2,C3,C4 does (AC Coupling?).

[Benta]

It's a Sallen-Key high pass filter.

[IDEngineer]

I wanted to say Sallen-Key too, but there are some strange things going on in that schematic. Definitely not traditional SK configuration. I wonder if the schematic is accurate.

[rrn]

What strange things in schematic makes you think that schematic is wrong.just wanted to know.

[IDEngineer]

The feedback loop, for starters. Traditional unity gain SK would have the schemetic's R6 going straight back to the junction between C3 and C4. Instead, there is a "tap" from a very high ratio divider that then also connects to the inverting input. Let's presume that's because they want an SK filter with gain (possible but tricky)... in that case you generally use a more traditional dedicated divider from the output to the inverting input, leaving the positive feedback loop separate to handle its filtering functions.

If R6 and R5 are intended to be a negative feedback divider, that's a HECK of a lot of gain for an SK configuration (A = 330) and R5 would be tied to ground, not in the filter network. On the other hand, if R5 and R6 are part of the filter network, then why is there a connection to the inverting input? That's not part of a normal SK filter design.

Maybe this is some derivation of an SK. Hopefully someone will chime in if so.

[Buriedcode]

Its the schematic for the "heart sensor" (pulse oximeter) that did the arduino rounds a few years ago. Its not a standard SK filter, as it was meant for low PCB area with maximum attenuation.  As the change in signal from the photodiode measuring the pulse is in mV they decided to use a passive filter followed by huge gain.  I vaguely remember reading that this circuit was the result of trial and error, albeit from the tarting point of a standard filter.  Try LTspice and see the response.

[rrn]

Yes, it is from used for pulse sensor.

[IDEngineer]

I vaguely remember reading that this circuit was the result of trial and error, albeit from the tarting point of a standard filter.

I can believe that! "Trial and error" is the standard rule of thumb for SK's with gain beyond about 2. As I recall, there's a TI app note about SK filter design that actually says that in black and white, and then specifically recommends against doing it.

[Benta]

I vaguely remember reading that this circuit was the result of trial and error, albeit from the tarting point of a standard filter.

I can believe that! "Trial and error" is the standard rule of thumb for SK's with gain beyond about 2. As I recall, there's a TI app note about SK filter design that actually says that in black and white, and then specifically recommends against doing it.

Yeah, it's a crap filter. But due to its prevalence in all kinds of data sheets and application notes, people insist on using it. SMH.

[IDEngineer]

I didn't mean to suggest all SK's are worthless. Just that they're best suited for unity gain, or at most 2X. The nice thing about SK's is that they used matched component values, handy in high volume manufacturing where you then have a higher quantity of fewer line items and can benefit from economies of scale. But you're correct, if you need more than casual filtering there are a lot of better options.

Funny comment there about "passive filter followed by huge gain". Textbook example of how to trash your signal to noise ratio... attenuate the signal FIRST, then amplify it.

[Buriedcode]

In this case it works rather well for the application. Detecting the very small change in absorption of green light (used in this case) which requires huge gain, whilst cutting everything above about 200Hz and below 30Hz, using a single opamp and a few common value passives isn't actually that easy.  I've used that circuit to make my own wrist-based monitor and it performs similarly to commercial fitness watches - that's not suggesting this is necessarily great, it's more that the wrist is a terrible place for this measurement and there isn't a "better" or "best" way.

I have seen some shocking analogue designs for Arduino projects, but this one does its job well - probably not ideal for educational purposes, but shows what can be done with a minimalist approach.

[rrn]

Is this Band Pass or High Pass Filter?