Weird photodiode step response.

May 29, 2019 4 Replies
G

So testing a bunch of photo diode TIA's today. (opa124, not recommended for new designs.) The low gain (332 ohm) step response had a ~0.5 us delay, and then a sharper edge than expected. (higher gains had similar weird stuff.) (I'm trying to upload dropbox files from home... but it's not working.) The circuit is TIA, no bias, N-I input grounded, opa124, Z_f is



332 ohms and parallel C ~470 pF, (maybe less.)

Here's the normal response (50 ohm termination.)

formatting link



ab-normal, 50 ohm term.

formatting link



and ab-normal with no 50 ohm load on the long (~6') thin (RG-172) coax output... which is how I test it. Typically the 50 ohms makes little difference.

formatting link



I'll replace the opamp, and hopefully that will fix it... But it was interesting to me.



George H.


Vote
Twitter Facebook LinkedIn
J

Underdamped step response, not unusual.

If you back-bias the photodiode some, its capacitance would go way down. Or do one of Phil's fancy cascode+bootstrap things.

Just a cascode can be a remarkable improvement.

What photodiode?

John Larkin Highland Technology, Inc lunatic fringe electronics
Vote
Twitter Facebook LinkedIn
G

Yeah but I've got a bunch of R//C gain stages on a rotary switch. All the Rs, Cs, solder joints look fine. I figure the opamp is fried in some way, it mostly works fine.

Right this is something I did years ago, before I knew as much.

PIN-44D? Big square thing. (osi opto-electronics?)

George H.

Vote
Twitter Facebook LinkedIn
J

PIN-44D has 700 pF at zero bias. Enough to make any self-respecting opamp wig out.

You can Spice it as a 700 pF cap in parallel with a current source.

I'd back-bias it 15 or 20 volts.

OPA124 isn't very fast. You might consider something else.

If noise isn't a big deal, you can back-bias the PD and dump the current into a series R+L to ground, and follow that with a fast voltage amplifier.

The opposite extreme is a bootstrap cascode into a fast TIA.

John Larkin Highland Technology, Inc lunatic fringe electronics
Vote
Twitter Facebook LinkedIn
G

Right.. but hundreds (of opamps) have answered the call before and worked fine. (I soldered in a new opamp and all is good.)

Right old design.

Yeah it's also got this nasty noise peak around 20 kHz (I'd have to check my notes. I found that years ago.)

George H.

Vote
Twitter Facebook LinkedIn

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required