Photomultiplier trans-impedance amplifier - DC elimination?

Hi!

Recently I was involved into design of the amplifier for photomultiplier(PMT). The problem that appears is the fact of having relatively high ambient light on top of usefull signal.

By connecting PMT directly to oscilloscope I measured 40V of DC and 1V peaks of the signal on top of that.

I want to use transimpedance amplifier to improve the bandwidth of the system, to be able to pick up gaussian pulses of 6us duration. Moreover I would like to get rid of the DC without drastic degradation of SNR.

Could you suggest me the best way to achieve my goal?

AC coupling configurations usually seem to result in a noisier system whenever I have looked at it. A tuned circuit load actually works quite well, but is only really applicable to certain types of signal. It might be applicable to your circuit if you have a fixed pulse shape you are looking for. I guess an inductive load would work, but I am not sure that the values required would end up with a practical circuit.

A DC offset feedback loop may work, but you would have to inject current into the input of the transimpedance amplifier and I can't help but think you will also create noise.

In general I have usually overcome such problems by optical filtering and masking as much as possible, then make sure the first stage can handle the maximum DC current you expect. You can then AC couple to the next stage.

Keith.

Hi!
I'm looking for particular shape - gaussian. About the inductive load - would you suhhest gyrator?

What is the DC offset feedback loop?

Would there be any use in trimm/balance pins of an opamp?

Unfortunately in this measurement system I'm looking for the fluorescence, which I need to stimulate by the laser. I cannot get rid of all the stimuli signal even with usage of razor edge filters.

best regards,
Micard

Re: Photomultiplier trans-impedance amplifier - DC eliminati

Micard,

While you get 40V when the PMT is coupled to an oscilloscope - what load do you have? Normally I would expect a transimpedance amplifier to be used. The the problem is how high you can have the transmipednace before the DC saturates the amplifier.

micard said:

Hi!
I'm looking for particular shape - gaussian. About the inductive load - would you suhhest gyrator?

Click to expand...

No, I was thinking of simply an inductor. As I said, I am not sure whether it is ideal for your circumstances, but it has been done in some published papers, I and have done it that way once before. The problem is that you end up with a resonant circuit. In my case I actually used that and tuned it. In your case that may still work. You would need to add a resistor in parallel with the inductor to dampen the Q and stop ringing. The best way to experiment with the idea is with simulation.

micard said:

Hi!
What is the DC offset feedback loop?

Click to expand...

It is not uncommon to design circuits where you take the output voltage (after a low pass filter) and feed the inverse back to the input to cancel out whatever is causing the offset. In your case you would need to sink the excess DC current from the PMT on the input to the transimpedance amplifier. It will work, but I would be concerned about injecting noise in to the input.

micard said:

Hi!
Would there be any use in trimm/balance pins of an opamp?

Click to expand...

No. They will have very little effect as the offset isn't amplified in a transimpedance amplifier.

micard said:

Hi!
Unfortunately in this measurement system I'm looking for the fluorescence, which I need to stimulate by the laser. I cannot get rid of all the stimuli signal even with usage of razor edge filters.

Click to expand...

All the fluoroscopy systems I have designed use photodiodes of various sizes. I don't remember having large DC signals because the optical filtering removes it. I am guessing your excitation and fluorescence wavelengths must be very close if that isn't the case with your system.

What wavelengths are you using?

I presume you have filtering of both the excitation and fluorescence wavelengths?

Is the DC offset actually optical or caused by leakage?

I have usually ended up with at least two stages - a transimpedance stage and a voltage amplifier. Total gains have been up to 2V/nA (using 3 stages in total). Leakage (dark current) becomes a problem at those sort of levels.

Keith.

The problem i have is that lots of stimulus signal is coupled into optical fiber which pick's up the output signal. Moreover the fiber (multimode) is just "gathering" the light without any optical system (and that is not about to change). Therefore I have reasonably big acceptance angle, so even with usage of razor filter (at 500nm to cut out the laser) doesn't filter perfectly (it requires perpendicular light to have nice cut-off).

the wavelengths I use are 480nm laser for stimulus, and about 600nm is the peak of fluorescence signal.

The laser is passing through sample, and the light is picked up by closely placed optical fiber. Subsequently the pick-up fiber is connected to PMT at which window the 500nm LP filter is mounted.

The DC I encounter is so far only caused by light, because when I turn off the stimulus - the output disappears.

I assumed that multiple stages will contribute, when the measured signal will already be low, and higher gain in TI-AMP will suppress the bandwidth. Now I have the problem of too big signal (DC+AC), so I need to handle this one first.

I'm really thankful for your advices!

Best regards,
Micard

Added after 37 minutes:

You have mentioned inductive load - but I feed the current directly into the amplifier.

Could you sketch some scheme to show me what you meant?

best regards,

Mike

Micard,

I think that in your circumstances I would use a TIA with as much gain as I can without the DC saturating the output. I would then AC couple to a voltage amplifier. So, you are likely to want a voltage amplifier with a gain of 50x to 100x. Do you think that will work from a signal to noise point of view? I seem to remember that earlier you had a lot of noise, but that may be avalanche noise (or whatever it is called in a PMT) due to the high static light level.

My idea with an inductor would be to use an inductor to ground then capacitively couple to the TIA. I am not a fan of the idea, although it may be possible to get it to work.

Keith.

Then I believe I need to experiment a little more wirt simulations.

I spent some time finding the previous case with noise. the issue appeared to be in the light source which was "noisy".

Best regards,
Mike

Re: Photomultiplier trans-impedance amplifier - DC eliminati

Mike,

This was your circuit before. With only 52uV of total noise there is a lot of scope for adding some extra gain without the amplifier/resistor noise dominating, I think.

Keith.

Once again thanks!

As soon as I will make some progress with that I will show the results

Mike