How to Design a Microvolt Pulse Detector Circuit with High Gain and Low Noise

[donpacino]

I am currently working on a small personal project to design a few circuits. I am trying to built a circuit that can detect 100 uV pulses with a pulse width of about 10 uS. I'd like to boost the voltage to a value readable by an FPGA (let say ~2 volts). I'm currently thinking about the architecture I want to use to solve this problem. I'd like your input on my approach and maybe some suggestions for other circuit layouts to use.

Initially I thought about using a standard instrumentation amplifier layout similar to the one seen in the link below. However I am concerned with the non ideal effects of the op amps.

hhttps://upload.wikimedia.org/wikipedia/commons/thumb/e/ed/Op-Amp_Instrumentation_Amplifier.svg/400px-Op-Amp_Instrumentation_Amplifier.svg.png

The required gain to get to 2 volts is 20,000. with a CMRR of 100 that would be multiply common mode noise by 200, which could be fairly significant. With the dc offset introduced by the op amps an subsequent amplifier stages would greatly increase any noise, even with a good CMRR.

That being said, the best I could come up with is a 2 stage instrumentation amp, followed by another diff amp stage. I have large dc offset and would appreciate any architecture ideas.

thanks!

 

[NFuller]

That being said, the best I could come up with is a 2 stage instrumentation amp, followed by another diff amp stage. I have large dc offset and would appreciate any architecture ideas.

Multiple stages is a good idea. A gain of over 1000 for an opamp is likely to produce a fair amount of non-linear distortion. To get rid of the DC offset, use a large value capacitor (100uF should be ample even for fairly low frequencies) at the output of each stage.

 

[berkeman]

100 uV pulses with a pulse width of about 10 uS

How much tolerance is in the "about 10us" number? I'd recommend using AC amplification stages with bandpass filtering around the fundamental frequiency of these pulses. That will help you to reject the noise as well. I wouldn't make the Q too high for the filters, since you want single pulses to get through.

 

[donpacino]

Hey good ideas from both of you.

The 10 us is a hard minimum length (highest frequency) that would be seen. the longest would be 50 us.

Thanks for the advice.

 

[tech99]

The waveform has components down to zero frequency, so a BPF will produce distortion and may result in an incorrect voltage reading. The gain being asked for is very high, 86 dB, and is not easy. I would tend to use a smaller voltage gain of, say, 20 and then use a peak reading square law diode detector. This can then be followed by some DC gain. The diode can be followed by a current-to-voltage converter op-amp circuit, which will work very well.

 

[donpacino]

Thanks for the help everyone, I got it working fairly well using ltspice!

 

[Tom.G]

Thanks for the help everyone, I got it working fairly well using ltspice!

Well, are you going to tell us how you did it? We too like to learn!