Best Filter Type to Remove DC Offset in Biomedical Signals (0-20Hz) After IA

Hi,

For any biomedical applications, the sensor output has to be amplified and digitized.
For amplification, we use Instrumentation amplifier. But the output of IA has some dc voltage. So, how to suppress the dc voltage. At the same time, all biomedical signals are low frequency signals in the order of 0 to 20Hz. Which filter technique is better to use?
Is it good to go for Low pass/High pass/Band pass filter? Can you pls throw some light on this?

What you can do is to pass the signal through a stable level shifting circuit. This is simplyan extra OPAMP with a voltage is summed in the opposite sense (perhaps using a trimpot) such that the output of your circuit is zero when there is no signal input.

can you provide a schematic?

If the signal being measured ranges from 0-20 Hz, it implies that you need DC amplification also (or very slow variation which can be considered to be almost DC)...and therefore using any sort of filter to remove unwanted DC would be counter-productive. You need an active offset nulling scheme - a sort of integrator based feedback that compensates for DC at the output of the IA. HOWEVER it will still not be able to differentiate between DC of the IA versus DC from the sensor. You could use a fixed or manual trimming scheme too...but that will mean that the trimming will be valid for only a small range of operating conditions (for example ambient temperature)...and for biomedical use I would surmise that this will not be sufficient or acceptable from a reliability perspective.

IMO, the best recourse would be to digitise the IA's output as it is and then deal with any DC offset issues in firmware or software. For example a small routine can be run at fixed intervals that disconnects the sensor and reads the IA's DC level and compensates at the data level for that (add/subtract the offset value from the digitised number). Anyways, for the purpose of reading the IA, the system would have to poll it...so every time as part of the polling routine, it can hold the last value, disconnect sensor (by this, I mean not a physical disconnect but via electronics - like an analogue switch), read offset, compensate, read sensor, update value, display it and so on - repeatedly till the unit is turned off. This, I believe would be a lot more robust and reliable.

Also, consider if _this DC component, assuming it exists, is negligible [in all environments this device will be exposed to] or not_. If it's negligible, then no worries.

I have not done research on a type of OPAMP which may also be of help. Look ar chopper stabilised OPAMPs I believe that the DC stability can be measure in microvolts. Worth a look.
The other thought of digitising the signals, and disconnecting the sensor etc sounds good as well but you will need to substitute an impedance possibly a resistor equal or close to that of the sensor. One good thing about all of this is that everyone can learn from what you do and how you do it. Cheers.

something like an OP 07 assuming you dont need low input bias current AD707?