nanosecond measurement cycle

Discussion in 'The Projects Forum' started by amrosik, Jun 11, 2016.

  1. amrosik

    Thread Starter New Member

    Jun 11, 2016
    Hello forums,

    that's my first post, and I hope its not going to expose my complete amateurishness.

    I want to take fluorescent measurements, which can be done with phototransistors for example. And the phototransistor will be partly exposed to the excitation light, which comes from an highpower LED (can easily be fed with 1,5 Amps for a short time). Fortunately the fluorescence has a live time of more than 10 nanoseconds (after 5 livetimes the fluorescence has completely decayed). I want to excite the fluorescence with a nanosecond pulse, and then start the measurement by the phototransistor after the LED has been shut down. Idealy the measurement of the phototransistor goes into an integrator, which accumulates the intensity over 5 livetimes.
    Thats one cycle.

    So my first question is: how can I pulse a highpower LED at pulse durations of less than a nanosecond (i.e at frequencies more than a Gigahertz)? I am fiddling around with raspberrypi and arduinos for the data aquisition. But this measurement cycle has to be done on an external circuitry. Rasperry pi does not tick at gigahertz.

    thank you in advance,

    kind regards from germany

  2. #12


    Nov 30, 2010
    This would require one of our best people and I'm not sure it is even possible.
    If nobody answers, it's not because we hate you. It's because this is a difficult job.
  3. amrosik

    Thread Starter New Member

    Jun 11, 2016
    Picosecond pulsing is actually possible with analog circuits, see this:
    But I doubt that one can run a LED current with it.
    Basically I just need a fast switch, which turns off the LED within a nanosecond. Only the fast turning-off is important, because then the measurement follows. And only one measurement if sufficient, I dont need permanent measurement. Just one shot.
    Last edited: Jun 11, 2016
  4. RichardO

    Late Member

    May 4, 2013
  5. #12


    Nov 30, 2010
    That is certainly easier than trying to teach you gigahertz printed circuits over the Internet!
    I had very little information as to what level of expertise you have.
    Last edited: Jun 11, 2016
  6. ian field

    AAC Fanatic!

    Oct 27, 2012
    Zetex (used to be Ferranti) have an application note for an avalanche transistor circuit that can generate pulses down to pico seconds - AFAICR: their chosen device was the ZTX451. The late Jim Williams who wrote application notes for LT (among others) described having used the 2N2369 in avalanche circuits - but it requires device selection for best results.

    One application for avalanche transistors is in sampling oscilloscopes up to tens of GHz.
  7. AnalogKid

    AAC Fanatic!

    Aug 1, 2013
    Are you sure the LED can respond in one ns? Normal LEDs have rise and fall times of 10's of nanoseconds.

  8. OBW0549

    Distinguished Member

    Mar 2, 2015
    And phototransistors have rise and fall times on the order of microseconds (often, many microseconds), not picoseconds.
  9. amrosik

    Thread Starter New Member

    Jun 11, 2016
    I found a photodiode with a rise time of 1 nanosecond, fortunately with the right wave length sensitivity.

    Concerning the LEDs... I do not see any specifications in the data sheet. Is it possible to forcefully stop the LED from radiating, by not only shorting it, but applying a reverse voltage?

    I am wondering how professional fluorometers are made, which work with pulsed light?
  10. MrAl

    AAC Fanatic!

    Jun 17, 2014

    You can find some nice pin photo diodes that go faster than 1ns, but they could run 100 bucks for one. Maybe you can find one for 40 bucks, but others could be 200 bucks (USD that is) probably used for very high speed fiber optic communication. There are some high speed comparators that might help in detecting the on and off times.

    Generating fast pulses for LED's isnt just about turning on a transistor. The current has to be there immediately, which means the LED capacitance has to be overcome. That could mean a special circuit to give it a behavior that is sometimes referred to as "snap on". As the name implies, it helps the transistor turn on very hard, then backs off. The other problem is getting the transistor to turn off, where you would need a "snap off" circuit, and a very fast transistor, and if a bipolar it must be biased and driven such that it can never go into true saturation or it will take a year and a day to get it to turn off.
    Maybe one transistor to turn on, then another to clamp the output to turn off.

    Another point is that some high power LED's are not made to be turned on and off repeatedly (PWM). Check the data sheets for more info on that.

    You might look into using LED's as camera flashes where the goal is to get a quick high intensity flash. There's been quite a bit done in this area already.