Active Filter Question

Thread Starter

mojo_risin

Joined Jul 3, 2013
35
Hello

I've been doing a project on blood pressure measurement, and I am following some manufacturer's notes, namely Freescale's and Microchip's app notes. I have some doubts in analog signal conditioning, and I've been working all day on this, with no result at all.

I am using MPX2050 pressure sensor, and I am amplifying its output with an in amp, bringing it to acceptable level for the MCU. If I've been able to extract the pulse oscillations using two chained band-pass filters, but both app notes use one two pole high pass filter. I want to try to implement this with OPA277. If you open the Freescale's app note, the filter configuration is shown on the first page. I want to replace the LM324N with OPA277, and I have a dual supply, so that's not a problem at all. The real problem is that I do not understand how this filter works. If someone thinks this is funny, I agree.

I can see that C2 and R3 (I will be grounding R3) are making a basic high pass filter with a cutoff frequency of 0.48 Hz. The circuit is set as non-inverting configuration, so the filter would allow only frequencies above 0.48 Hz, and block dc. Can someone please explain to me what is C1 used for? Is the combination of R1 and C1 going to ground signals above 4.8 Hz?, and thus eliminate them from the feedback path? I really want to understand the concept of "two pole high pass filter", and I want to make it work.

As I said, I've been able to get a good shaped pulse from the pressure sensor, but I think I can do better. If anyone has any other suggestions as to how to filter the signal, I am happy to try them.

Thanks a lot, everybody!

:)
 

crutschow

Joined Mar 14, 2008
34,285
The filter is somewhat unorthodox but should work. It is actually two single-pole HP filters in cascade with different corner frequencies. It doesn't use feedback to give a sharper rolloff at a single corner frequency as standard 2-pole active filters (such as Sallen-Key types) do.

As the second page of the Freescale app note shows, the 1st pole is due to the R1C1 time constant and the 2nd pole is due to the R3C2 time constant. As the frequency increases, the impedance through R1C1 becomes smaller which causes the op amp circuit gain [1 + R2 / (R1 + Xc1)] to increase, from a minimum of 1 at DC (assuming R3C2 is a short), to a maximum of [1 + R2/(R1+0)] = 151 at frequencies well above the corner frequency.
 

t06afre

Joined May 11, 2009
5,934
The LM324 is an opamp made for single supply use, and this is the use of it in that circuit. The OPA277 is a typical dual supply opamp. So I am not sure the OPA277 is the best option to replace the LM324
 

Thread Starter

mojo_risin

Joined Jul 3, 2013
35
The circuit shows DC biasing the LM324. I think the only issue will be that the signal out of the OPA277 will be AC, so I would need to inject DC. I will try it, anyway, hope it will be ok :) Otherwise I would have to buy another op amp.

So, as far as I understand, the circuit first cuts off any signal beyond 0.48 Hz (the C2-R3 pair), and then amplifies the rest of the signal by a certain number. I was thinking that I might need to put a passive RC low pass filter after the in amp stage, with a cutoff of about 10 Hz, just to get rid of any noise. Is it going to be a good idea? Single RC stage will have a gain of (almost) unity for the frequency of interest (around 1 Hz). This low pass filter will feed the MCU's ADC, and also the aforementioned active filter. Please let me know what you think about this. I will try to put the circuit on a breadboard and test it tomorrow or on Tuesday.

Thanks for the input!
 

LvW

Joined Jun 13, 2013
1,752
mojo_risin,

to answer one of your questions in post'1 (role of C1):

For high frequencies the gain of the amplifier is (1+150)=151 and for low frequencies (below the corresponding cut-off frequency) the gain is only 6 db (factor of two) due to 100% feedback (C1 blocks). Thus. it is not really a highpass (which normally attenuates lower frequencies down to zero at dc).
In general, a series connection of both parts can work as a highpass - however, it is a rather poor highpass.
I would recommend a classic active highpass filter (for example in Sallen-Key topology, as contained in the link given in post#2). Another solution would be - as indicated by you - a second-order bandpass (also Sallen-Key). However, in this case you must know the allowable/required bandwidth.
 

AnalogKid

Joined Aug 1, 2013
10,990
The other function of C1 involves the amplifier input offset voltage. Without C1, a forward gain of 151 amplifies everything, including the input offset voltage. This can add a significant DC error to the amplifier output. C1 is open at DC (perfect capacitor that it is), reducing the circuit to a voltage follower. So for example a 5 mV input offset will show up on the output as 5 mV instead of 750 mV.

ak
 
Top