EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: LoveLaika on September 06, 2023, 06:22:41 pm
-
I'm working on a project and I'm trying to get some ideas about how to proceed given one issue. Previously, I built simple SiPM (photomultiplier) boards using the recommended attached circuit schematic. Basically, the SiPM is reverse-biased with a negative bias voltage which passes through a 2nd order filter. When the PM activates, the resulting current is converted to a voltage thanks to the 10-ohm resistor. It works pretty well, but for this circuit, I'm trying to get four of these SiPMs to get their individual outputs (as represented by node A#) as well as sum the outputs together to get one signal. I have an idea of how to do that (through the use of a summing amplifier to sum the voltages together), but I'm concerned about crosstalk. What are some ways I can reduce the crosstalk so that the SiPM signals don't affect one another or the summed-up voltage signal?
Previously, I had one board that used one trans-impedance amplifier to connect four SiPMs at their cathodes and turn their respective currents into one voltage signal. However, doing so introduced some crosstalk where one activated SiPM affected the others which ultimately affected the sum as well. We moved on from TIAs due to issues with speed and space, so we then used the simple readout method as mentioned. However, we still encountered some crosstalk among the SiPMs when they were individually read out (no summing at all). We believe that this is due to the shared connection with the bias voltage (the SiPMs all shared a connection to the bias voltage), and we were able to reduce this particular crosstalk effect by increasing the size of the filter capacitors from 10 nF to 100 nF (even 1 uF). That worked well with no loss in speed. With this new venture, I was wondering what else could be done? Could the SiPM signals be isolated through the use of a small balun/tiny 1:1 transformer? Or, since the SiPM current signal is converted to a voltage already (and we would just be summing the resulting voltages), is crosstalk prevention even necessary? The voltage signals would first go through a resistor before the op-amp's inverting node, and I would assume that resistor to be larger than the 10-ohm resistor. Would the addition of such resistors also help mitigate the effect of the SiPMs on one another?
I'm just kind of at a loss on how to best proceed, and I wanted to brainstorm some ideas here with you guys.
-
Your bias supply polarity appears to be backwards. Why are you loading the output of the SiPM with 10 Ohms? That is gobbling up all the signal. You want the TIA to get as much of the signal as possbible. And, it doesn't need a lot of gain, as the SiPM has the gain in it. You may need more than 30 V to get the SiPMs into their avalanche region. individual SiPMs need their own individually setable bias voltage, as their ideal avalanche voltages are all different. There's a very small range between good gain and excessive dark current pulses. Also R1 and R2 can be made much larger, as the reverse currents in the SiPMs are very small. I have used at least 10K Ohms. Do not connect the bias sides of the SiPMs together, as that will couple crosstalk between them.
Jon
-
Can I suggest that you invest in a copy of
BUILDING ELECTRO-OPTICAL SYSTEMS: MAKING IT ALL WORK
SECOND EDITION
Philip C. D. Hobbs
Electrooptical Innovations
Briarcliff Manor, New York
A JOHN WILEY & SONS, INC., PUBLICATION
-
Thanks for the reply. Apologies for not responding sooner. This is not a TIA circuit. This is a readout circuit that does not use a transimpedance amplifier to convert the current to voltage. The user guide of the SiPM manufacturer suggested this kind of simple readout circuit along with the values shown, and it's worked fairly well so far. Having that 10-ohm resistor would turn the current into a readable voltage faster than it would with a TIA (based on what we were using before), and we found that with this kind of simple readout, we were able to make tiny boards to fit our needs. It also recommended a reverse bias voltage of -25 volts, but we stuck with -30.
Ideally, we would want each SiPM to have its own bias, but we never pursued it. Assuming that we did, are there linear regulators that output the same voltage as the input? There doesn't seem to be any linear regulators that can go as low of a voltage as we need. Perhaps a simple unity-gain buffer would be good to provide an impedance to reduce the crosstalk? Each SiPM doesn't output much current (~20 mA max)
-
Can I suggest that you invest in a copy of
BUILDING ELECTRO-OPTICAL SYSTEMS: MAKING IT ALL WORK
SECOND EDITION
Philip C. D. Hobbs
Electrooptical Innovations
Briarcliff Manor, New York
A JOHN WILEY & SONS, INC., PUBLICATION
This book is excellent!