I've built an LED driver and a photodetecting circuit.

Basic forms of the driver and the detector were made.


enter image description here


enter image description here

FYI: Parts used and their datasheets

In this configuration, I want to driver an LED and detect it as fast as hundreds of kHz.

But, output from the photodetector shows a sawtooth form not a square form even with a very low frequency, 10 Hz:

enter image description here

The blue line is the V_OUT in the dirver circuit, the photodiode outputs, V_OUT in the detector circuit drawn with the yellow line.

As you can see a **slew rate of the detector is not fast as the driver. So I made a list of candidates that may cause this problem.

Candidates of the slow slew rate problem

  1. V_OUT in the driver is different from an actual voltage value across the LED. (i.e. I am probing a wrong point. And brightness of the LED may not be a square form as the blue line.)
  2. Reverse voltage applied on the photodiode is not sufficient. In datasheet, BW > 10 MHz when V_R = 12 V. Thus, I need to apply 12 V as the reverse voltage across the photodiode.
  3. R1 (gain) in the driver is too large. So, I need multistage low gain op-amps.
  4. Photoconductive mode doesn't suit. It's better to use photovoltaic, etc. (But, I think photoconductive mode is nothing wrong.)
  5. (Si PIN) photodiode doesn't suit. It's better to use photoresistor, phototransistor, etc.
  6. (Feedback or any) capacitors have too low/high capacitance.

Which one do you think is the keypoint of the problem? If I got everything wrong, please correct me.

  • \$\begingroup\$ How did you size C6? \$\endgroup\$
    – jippie
    Commented Jul 20, 2015 at 8:42
  • \$\begingroup\$ Well, actually there is no reason for me to choose size of C6 as 1 uF. Would changing C6 increase BW, slew rate? \$\endgroup\$
    – Jeon
    Commented Jul 20, 2015 at 9:51

4 Answers 4


You are using exactly the same circuit you used in Ambient light rejection from photodetecting circuit, and I will repeat what I said in that question.

Reduce C6. You have gotten better results by reducing C6 from 1 uF to 0.01 uF. Reduce C6 to 1000 pf (.001 uF) and you will get even better results.

C6 and R1 are combining to form a low-pass filter. With 220k and 1 uF, response will drop to 1/2 of DC when the signal frequency f $$f = \frac{1}{2\pi RC}$$ For your original circuit, this was 1.3 Hz.

When you reduced C6 this became 130 Hz.

Note that these numbers apply to sine waves. It's more complicated with your nominal square waves.

As I said in my earlier answer, you can certainly reduce C6 to 1000 pf (.001 uF). You can very likely reduce it to 100 pF, and get even faster response. Much below 100 pF and, considering your physical layout, you may have stability problems.


The feedback loop RC network is a low pass filter. To a first approximation, the -3dB point (1/(2*piRC) is currently 72Hz. With the 1uF device, that point was less than 1Hz.

Your plots show the hypothesis to be accurate. I was going to run a simulation, but the TI SPICE model breaks in one of simulators.

You can either reduce the resistor size or reduce the capacitor size or both.

Keep in mind that at higher gain, there is a pole formed at the inverting input by layout parasitics and a small capacitor across the feedback resistor is not merely prudent, but necessary for proper operation of the amplifier.

If you want operation to 10kHz (-3dB bandwidth), then start with a RC product of 1.6E-06.

Note that 220k is a bit high for a feedback resistor; normal practice is to keep this at 100k or less.

As an experiment, try 82k for the feedback resistor and 180pF for the capacitor. You will have lower gain, but the output should not be in slew rate limit (which is what you are currently seeing). These values yield a nominal -3dB point of about 10.7kHz

The maximum frequency to not be in slew rate limit is given by fmax= Sr/2πVp where Vp is the peak voltage you wish to get from your amplifier and Sr is the slew rate available from the data sheet.


Answer to myself. This can be regarded as a sort of a progress report.

With a help of jippie, I got it fixed by replacing C6 in the detector with one of 10 nF. But still slow when the frequency becoms faster.


  • LED: Blue
  • Photodiode: Yellow

10 Hz

enter image description here

60 Hz

enter image description here

160 Hz

enter image description here

320 Hz

enter image description here

I will post another if I make further progress.


There are two topologies for a transimpedance amplifier

Grounded ANODE


This topology is great if you require low impedance load for the photo diode.

Rail Tied Cathode


simulate this circuit – Schematic created using CircuitLab

This is great if you require higher bandwidth but can live with lower amplification.


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.