I'm looking at building a detection system for fast laser pulses (532nm wavelength for 250 ps duration). It seems to me that this will require a very fast photodiode with sub nano-secind rise time. Are such components manufactured and sold commercially, and if so, what is an estimate of the cost I can expect.

  • 1
    \$\begingroup\$ Why does a web search for ultra-fast photodetector not satisfy your questions? \$\endgroup\$
    – glen_geek
    Sep 25, 2022 at 14:03
  • 8
    \$\begingroup\$ Are you trying to measure the width of the pulses or just detect them? Width will require a fast photodiode but detection (what you asked for) does not. Even femtosecond pulses can be detected by nanosecond or slower photodiode. You just get a longer electric pulse. \$\endgroup\$ Sep 25, 2022 at 14:40
  • 3
    \$\begingroup\$ If you want pulse duration, and the width of every pulse is the same, consider using an autocorrelator rather than a photodiode. These are relatively inexpensive to build and much more accurate than a photodiode. \$\endgroup\$ Sep 25, 2022 at 14:45
  • \$\begingroup\$ @user1850479 I am only interested in detecting them, not measuring their width. Specifically I want to determine where in a 20*20mm area the pulses are hitting. The idea I had was to use photodiodes to construct a target to place in front of my laser and use this to locate the pulse. \$\endgroup\$ Sep 29, 2022 at 7:46
  • \$\begingroup\$ @ThomasHolle Then you don't need a fast photodiode and the answers to your question are not relevant to your problem. Normally beam tracking uses an ordinary CMOS or CCD sensor, or if higher speed is needed, a position sensitive photodiode. \$\endgroup\$ Sep 29, 2022 at 17:12

1 Answer 1


Photodiodes are not specified by rise time. They are specified by capacitance, and this capacitance combined with the gain and stability requirements of the transimpedance amplifier defines their rise time. Then there are tricks like bootstrapping and cascoding to reduce the effective photodiode capacitance so increase the speed for a given feedback resistor but the relationship is not straightforward.

What this means is no one can give you a price range because no oen can even narrow down what photodiode you might even need since you have not provided your level of illumination which means no one can narrow down what photodiodes you may be looking at. A huge part of photodiode cost is active area. Larger active area means more capacitance which means a slower photodiode, but also collects more light so better for low illumination. Since you are shining a laser at it you probably don't need a very sensitive photodiode.

However, you also did not specify detection or measurement. This matters because measurement requires a larger active area to capture the entire laser spot to deal with spatial variations in the laser spot and/or to not saturate the laser spot. The required active area could literally make the difference between needing a $1 photodiode and a $100 or even $500 photodiode.

A 10-90% rise time works out to approximately have the highest Fourier frequency component of the signal be:

$$BW = \frac{0.35}{t_{rise10-90}}$$

So a 1ns rise time would require a bandwidth of 350MHz.

You will need to read how a transimpedance amplifier works conceptually with a photodiode. This is a basic photodiode transimpedance amplifier calculations which you will also definitely need to read: https://www.ti.com/lit/ug/tidu535/tidu535.pdf?ts=1664111058976

It's a bit confusing so I will leave these bullet points:

  1. Reverse biasing a photodiode reduces the capacitance which lets you operate it faster at the expense of some noise
  2. A real transimpedance amplifier required a feedback capacitance to be stable
  3. The feedback resistor in the amplifier determines the gain which means you need to know the photodiode response and illumination to know the gain you require
  4. The feedback resistor and feedback capacitance together determine the bandwidth.
  5. ...but wait...where does the photodiode capacitance come in to determine the bandwidth then? It doesn't really...at least not directly. Two things are at work: required gain and stability. Things that increase stability will put an upper limit on bandwidth but they don't determine the bandwidth.

Here is the convoluted relationship feedback resistance, feedback capacitance, photodiode capacitance, bandwidth, stability, gain, and opamp gain bandwidth product (GBWP):

  • higher value feedback resistors means more gain but lower bandwidth
  • higher value feedback resistors will increase the GBWP required for stability
  • higher feedback capacitances mean lower bandwidth
  • lower feedback capacitances will require higher GBWP to remain stable
  • higher photodiode capacitance and higher op-amp input capacitances will require higher GBWP for stability

If your calculations tell you that to get your combination of gain and bandwidth require photodiode capacitances or feedback capacitances that approach <1pF, or even a few pF then your specifications are not feasible. In that case you need to start augmenting the basic transimpedance amp such as reverse biasing the photodiode (by far the simplest and almost free), bootstrapping it , and/or cascoding or other techniques.

If you don't understand any terms I used that doesn't matter. I didn't explain everything required anyways and what I did explain was in almost zero detail. You are going to need to read about these topics starting with how a transimpedance amp works and how to size components for them when using a photodiode (which is the link I posted). Augmenting the basic circuit is a whole other story: https://zeus.phys.uconn.edu/halld/hpd-3-2005/frontends.pdf

There are many levels of understanding between the two links.

For 350MHz you are going to need augmentation such as bootstrapping and/or cascoding.

  • 2
    \$\begingroup\$ I regret that I have but one upvote to give for this answer. \$\endgroup\$
    – The Photon
    Sep 25, 2022 at 19:41
  • \$\begingroup\$ Yep. Silicon detectors are easy in principle, but a good measurement chain to get a challenging result from them can take years of work. \$\endgroup\$
    – John Doty
    Sep 26, 2022 at 10:59
  • \$\begingroup\$ If you buy a ready-assembled photodiode module (example that might actually work here with enough attenuation - it's very sensitive even though un-amplified) the rise time is specified. For pusle detection as opposed to pulse width measurement I've been known to use deliberately slow photodiodes. Honestly if this is a one-off it's better to just buy something off the shelf - by the time you've designed and built something it will cost more because of the labour. \$\endgroup\$
    – Chris H
    Sep 26, 2022 at 12:26
  • \$\begingroup\$ @ChrisH Yeah. If I knew my laser was strong enough to run unamplified and was able to get my hands on a photodiode that could produce produce enough output and was fast enough without a TIA I would run it bare with just a resistor. At least for detection. Not for measurement where linearity matters. \$\endgroup\$
    – DKNguyen
    Sep 26, 2022 at 17:28
  • 1
    \$\begingroup\$ Detecting pulse position requires actual measurement in most implementations that do not involve a massive grid of photosensors. \$\endgroup\$
    – DKNguyen
    Sep 30, 2022 at 4:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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