I am building a circuit for sensing photocurrent from a photodiode. The diodes are manufactured with selection transistors to form an active matrix and below is a model of a single pixel consisting of the selector transistor and the diode that is followed by a simple transimpedance amplifier. Later I plan to and multiplexers to choose the selection transistor on one side and diode anode on the other and this way choose the diode for readout. Note that the transistor is an experimental device in organic material technology, which for the time being is modelled with a LEVEL=3 model:

.model OFET_L2 PMOS (Level = 3, L = 5u, W = 1m, ETA = 0.2649, VTO = 3, NFS = 10, UO = 2, TOX = 20u)

enter image description here

The circuit intends to reach readout speeds of a commercial camera for the whole matrix, so hopefully up to a working frequency of 100kHz (10us per pixel readout). The problem is the capacitance of the photodiode which seems to be causing a current surge the moment the transistor opens. That phenomenon has to be verified experimentally (to what extent is is a real problem and to what extent is it an artefact of the way I chose to model the diode), but I am fairly sure this is an accurate representation of it.

I tried increasing the feedback capacitor on the amplifier to absorb the current, but it results in longer times required to reach steady state.

enter image description here Selection frequency 33kHz, Iphoto = 100nA, Rrev = 100Mohm, Cf = 5pF

Cd = {0.005pF, 0.05pF, 0.5pF, 5pF)

Making it too small leads to op-amp instability which was the original reason I put it there. Plus, the spike becomes even bigger.

enter image description here Selection frequency 100kHz, Iphoto = 100nA, Rrev = 100Mohm, Cf = 1pF, Cd = {0.005pF, 0.05pF, 0.5pF, 5pF)

Finally, the problem persists until a fairly large current is provided by the photodiode (much light is incident).

enter image description here Selection frequency 100kHz, Rrev = 100Mohm, Cf = 5pF, Cd = 5pF

Iphoto = {10nA, 100nA, 1uA, 10uA}

My questions are: is there anything I can do about it other than improve the photodiode? I was thinking of adding a more sophisticated filter after the transimpedance amp, but I need some advice how to go about it and whether it is even possible to tackle the problem this way. I don't want to drown hours into developing a solution that is doomed from the start.

Furthermore, if that speed of switching cannot be achieved then how harmful can such spikes be? Are they likely to cause instability in the circuit when more diodes are hooked up?

I read on possible solutions to similar problems, but they are overwhelmingly concerned with power MOSFETs in motor driving circuits where the spike arises from inductance in the motor and can be mitigated with a parallel diode. It doesn't seem to me like that solution could be implemented here.

Thank you very much for any feedback.

  • \$\begingroup\$ Given that photodiode current will happily pass through the leakage conductance of a supposedly "OFF" MOSFET, how will controlling the MOSFET help functionally irrespective of glitches due to C? In other words, it seems your approach is flawed. \$\endgroup\$
    – Andy aka
    Commented Jul 23, 2018 at 14:26
  • \$\begingroup\$ How come? The leakage current of that particular FET is in the order of 10's of pA, while the smallest current we hope to measure from the photodiode is 1-10nA. The leakage shouldn't be a problem in my mind, but perhaps I'm misunderstanding you comment. \$\endgroup\$ Commented Jul 23, 2018 at 15:23
  • \$\begingroup\$ No, you have understand it and have justified your position. \$\endgroup\$
    – Andy aka
    Commented Jul 23, 2018 at 15:33
  • \$\begingroup\$ What might happen if you diode-orred each PD current into a virtual earth and disabled all but one photodiode with a MOSFET across each photodiode i.e. you shunt away the current in all but one PD. \$\endgroup\$
    – Andy aka
    Commented Jul 23, 2018 at 18:19
  • \$\begingroup\$ Yeah, that's the plan, but when switching even one diode I get the current surge. \$\endgroup\$ Commented Jul 24, 2018 at 9:33

1 Answer 1


The "spikes" cause no issues as far as the health of the circuit. I'm not sure of the voltage levels for V3 and V4, but you are nowhere near saturation, and your input voltage spike is less than 1 volt.

The classic approach for video is to use blanking. If you look at your last illustration, you switched at about 5.5 uS you have the appropriate response between 12 and 13 uS, approximately 6.5 uS from when you hit the switch. On the second pulse, you switched at 25.5 uS, and the response is appropriate between 32 and 33 uS. There may be some other technique, but this blanking has been around as long as there have been video outputs. An improved device or circuit might speed up the settling time, but blanking is still in wide use to prevent switching noise from getting into the video output.

So the approach is to gate the amplifier output with a 1-uSec gate that occurs 5.5 us after the switch.

  • \$\begingroup\$ Thank you very much. Could you suggest some ways to implement this? I tried to find resources, but neither "blanking" nor "gating" give much in a search. Perhaps just some keywords that would orient me in the right direction? \$\endgroup\$ Commented Jul 24, 2018 at 12:04
  • \$\begingroup\$ A search on "high speed multiplexers" should get you there. \$\endgroup\$ Commented Jul 24, 2018 at 12:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.