How can I drive a laser diode at 1MHz for a speed of light experiment?

[akarion]

Hey guys, I was thinking of building a speed-of-light experiment where I pulsed a laser at around 1MHz (I don't think the duty cycle would matter too much, as long as the on/off is very noticeable) so that I can reflect the pulsing light off a mirror some distance away (maybe 30m), and then I can use an oscilloscope connected to photodiodes to visualise the phase shift of the light leaving and coming back.

My idea is to use an ESP32's PWM output to provide the pulses since it can provide a very high quality pulse. If possible I want to avoid using commercial laser drivers as they can have too much capacitance so that my 1MHz pulse turns into a smudge. Is it possible to use a resistor instead? Correct me if I'm wrong, but my idea is that since I'm pulsing the laser really fast with a low duty cycle, I could probably get away with the resistor providing a little more current than the diode is rated for in case that ever happens. I think I'll just need to use a transistor to control the switching because the ESP32 can only handle 40mA of output current, and I'll need a very stable power source, maybe batteries?

Is my idea viable? I've got SFH203 photodiodes and the datasheet says that it is most sensitive to red/infrared, so I will probably use a red laser for this project. Would a diode like the ML101U29 be good?

 

[RedCowboy]

https://www.lostgallifreyan.net/Hardware/Laser%20Diode%20Driver/Laser%20Diode%20Driver.htm

 

[paul1598419]

This experiment was originally done using a fanlike structure to bounce light off a mirror at a distance known and to blolck that returning light. This turned out to be fairly accurate for the time.

 

[atomd]

Well let's do some number crunching. You want to reflect light form mirror 20m away. That gives you 40m optical path length, which translates to roughly 120ns delay. To be able to see any delay you'd want your driving edge to be significantly faster than that, let's say 40ns. Because of both nonlinear dynamic response of LD and LC low pass filter made from parasitics you probably want to pulse your diode with some overshoot to get sharp optical pulse. The simplest way would be to parallel a bunch of fast 5V CMOS inverters and pulse 808nm diode. ML101U29 is much too high power for this project and will be pain to drive without bias-tee. I'd recommend any 5mW 808 diode.

The photodiode you mentioned proudly displays 5ns rise and fall times. That'd be plenty for this application but only if you can connect it properly. You need to load it with 50R (diode capacitance + load resistance creates a low pass filter, slowing pulses down). Preferably using scope's built-in in terminator but for this frequencies passthrough terminator should be fine too. The critical part is that you need to provide -9 to -20V negative bias voltage. It's critical that the bias voltage is low noise in your entire measurement spectrum (so roughly 100kHz to 100MHz), especially at high frequencies because its capacitively coupled to your oscilloscope (through the diode). So 7909 and plenty of RLC filtering before and after.

 

[kecked]

That’s not a bad thought to use a chopper. Reason is the rise and fall of the diode will then be perfect since it is just blocked. The speed of the chop doesn’t need to be that fast. You just are measuring the phase difference between the pulses which is relatively easy. You could eliminate that entirely and use a beam splitter and measure the time difference between a reference and the reflected beam in CW. It will beat as an interferometer. Ie the Michelson Morley experiment.

 

[atomd]

I think you totally underestimate requirements of chopper needed to modulate the entire beam <<1us

 

[kecked]

What I’m saying is you don’t need to chop that fast. You just want a pulse train with shape edges.

 

[atomd]

yes, exactly. And getting sharp edges on chopper isn't trivial.

 

[kecked]

More….why it blocks or it doesn’t. I have not used one for decades. I’d think it be instant on instant off. I however am assuming the speed remains constant.the blades are not bent, ect….

Now thinking of it you’d want to place it at the center of a set of lens too for the smallest spot size. Yea bit more to it had not thought it through. Still doable.

 

[atomd]

More….why it blocks or it doesn’t. I have not used one for decades. I’d think it be instant on instant off. I however am assuming the speed remains constant.the blades are not bent, ect….

Now thinking of it you’d want to place it at the center of a set of lens too for the smallest spot size. Yea bit more to it had not thought it through. Still doable.

Confocal setup is indeed significant better, but even the finite beam waist + finite rotation speed makes modulation take some time. If we assume reasonable 80um waist and take thorlabs chopper as an example (100 rotations per second, 100mm wheel diameter) then outside of the wheel moves at 100*100*2*pi = 60m/s, transitioning through 80um beam in 1.3us. A lot slower then even slowest electronic modulator

 

[kecked]

But the edge is perfect no? See speed doesn’t matter for this experiment as your comparing the phase of two pulses. So slow is fine. Actually CW is fine in an interferometer. Fringes. Getting the path length right is most important. Thanks for the explaination. I never need a chopper for my work. Actually I’m setting up a pulse picker this afternoon on an old Mira Ti:Saph. Doing a continuum through sapphire.

I missed the last line the 1 us yes that would mess it up. Electronic even if not as clean would be better.