This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

OPA356: Current or voltage limiter for TIA to In-

Part Number: OPA356
Other Parts Discussed in Thread: OPA355, OPA627

I design a TIA with OPA356 and SFH250V(Photodiode):

Since maximum photocurrent of SFH250V is 100uA, 

and I hope higher gain of photocurrent for my usage.

Thus, 100k Rfb is necessary after I experiment.

But max of voltage output would be 100k*100uA = 10V.

10V seems to over absolute maximum of signal input a lots

(My power is provided by USB-5V)

How do I select/design current/voltage limiter?

  • Your design is not particularly stable right now, I have the source C as 8pF from the diode datasheet, 

    You might want to increase the feedback C, here is 2.2pF, 

    To go further, really need your load defined, 

    Limiting? depends on how much you care about fast recovery = if you don't care, just put some clamp diodes on the inverting node, very low C ones preferred, Transimpedance from SFH250V with OPA356.TSC

  • Here is an example protection diode, might be too much for your needs, but 2- in a package, You might only need one polarity, 

    Did not seem to change the frequency response at all, not sure if the C is in the model, in TINA the model is locked and you cannot see the netlist, 

  • Hi Dewei,

    also, when the output of OPAmp clipps, the -input of OPAmp becomes negative (!) and you may run into a violation of common mode input voltage range of OPA356 which can make the OPAmp latch-up or lock-up:



  • Hi: @Michael Steffes

    Actually, I've read tidu535 before design this circuit.

    Yes, I care about fast transient.

    My purpose is detect a pulse which width about 100ns.

    For the tidu535 with a target -3dB frequency of 1MHz, I tried to modify tidu535 to reach 10MHz bandwidth.

    I think modify capacitor seems to be a good idea.

    My object is detect a pulse which is small width (~100ns), so if system has a little overshoot, it's fine.

    0.1pF of Cfb(where Rfb=100k) for OPA355 seems to has a oscillation after I do transient simulation

    0.5pF is better

    I know 0.5pF of Cfb, I have to take care about layout.

    Because little parasitic capacitance would impact Cfb.

  • Hi @kai klaas69:

    Thank you for your hint.

    Yes, I need to consider how to limit input of V-.

    If add resistor or diode would impact transient response,

    I think maybe I can reduce Rfb to 50K or lesser,

    and make second stage amplifier and make third stage for filtering amplified noise.

  • Hi Dewei,

    the main issue here is the big detecttor capacitance which makes a rather big feedback capacitance necessary to restore the phase margin. And in order to achieve a bandwidth of >10MHz the only way is to decrease the feedback resistance:



    A good remedy was to bias the SFH250V with a higher reverse voltage: A reverse voltage of 10V would decrease the detector capacitance from about 8pF to 2.5pF which would allow you to run the TIA with a higher feedback resistance and a smaller feedback capacitance.


  • Many thanks @kai klaas69's reply!

    I'll also try solution for reverse bias.

    According the SFH250V datasheet, capacitance is 4pF when adding -2.5V reverse bias, I though it's enough.

    Oh! An importing thing I forgot to mention.

    I bought 10 OPA356.

    I assemble the circuit using a breakout board and a breadboard for light  testing.

    7 of 10 OPA356 are fault when I do the light test with above circuit.

    I guess that the reason of fault is latch-up you mentioned.

    (Not sure)

    I tried 100M and 50M feedback resistor at that time,

    And exposed to high intensity light during the test.

  • Hi Dewei,

    then you should urgently add the two BAV99 protection diodes, Michael already mentioned:

    R4 = 100R additionally helps to limit the current into the OPA356.

    To keep the phase margin sufficiently high, the feedback capacitance has to be increased to 1.2pF:

    Take care, further increasing R4 can destabilize the circuit by ruining the phase margin. In this case the feedback capacitor would have to be furtherly increased. It's better to use R4 in combination with Michael's BAV99 protection diodes. Then R4 can be kept small enough.


  • Hi: @kia klaas69:

    I read slva898 mentioned:

    Should I connect one diode to GND and connect another diode to Vcc?

    Your picture connect both diodes to GND.

  • Hi Dewei,

    the scheme Michael has shown is the standard protection scheme for inverting amplifiers and TIAs. Clamping the -input to signal ground with two antiparallel diodes makes sense, because the nominal input voltage of an inverting amplifier or TIA is "fixed" on virtual ground. (Also see figure 33b of datasheet of OPA627.)

    That you provide a bias voltage of 240mV and shift the virtual ground by this amount does not change much, because 240mV is below the turn-on voltage of a pn-diode like the BAV99. You would only have to modify the protection scheme when the bias voltage comes closer to the turn-on voltage of protection diode or even exceeds it.


  • Hi @Kai klass69:

    Regarding BAV99, if I understand correct:

    One side of the diode in the BAV99 is grounded,

    so if |Vi| > 0.7V(forward voltage of diode), then one diode of BAV99 will shunt(sink) current with 250mA(forward current of diode), and input voltage still remain |Vi|.

    If |Vi| < 0.7V, then R4 can be limited input current.