Photodiode series resistance -> dominant noise source!

I received SFH206K and BPW34s photo diodes today, and will try to spend some time looking at them tomorrow, but I did a quick measurement on the

The Rs is larger than I expected, so take this plot with a pinch of salt until I verify the measurement.

I did a quick TDR on an SFH206K. It's not a great measurement, because the ohmic dip is squeezed between the inductive kick (it's a TO92!) and the capacitive slope.

Ignoring that uncertainty, I see 54 ohms at zero bias and 55 ohms at

-20 volts.

Maybe I'll try a different package. We have some SOD123's in stock too.

A frequency-domain measurement could resonate out the L-C and better resolve the ohmic part.

Now that is a very pretty picture of an ugly device! The capacitance numbers look right, and the behaviour of the resistance is very interesting. If zero resistance is at the bottom of the screen, it looks about right too.

Thanks

Phil Hobbs

Thanks. I'll try that out. 54 ohms is in the ballpark.

Cheers

Phil Hobbs

I'd be impressed if I got within an octave of the actual resistance.

It would be better to connect it to a 50 ohm sig gen, find the series resonance dip, and measure the voltage. Or use one of those VNA things.

No - look at the market readout - Cs 10pF Rs 480 @ 25V reverse bias (with a 1MHz 25mV excitation - similar to your Boonton).

So about a factor of 10 higher than you are extrapolating via noise readings.

I assumed that there was some shunt resistance messing up the reading, but measuring at a lower frequency didn't show this. Since I just have the diode wrapped in a bit of electrical tape I checked the dark(ish) current on the 610C

which shows about 0.55nA at 25V. This is quite a bit less than the datasheet typical value (5nA), but I'd tend to believe the Keithley.

Measurements on the BPW34s give about half the Rs value of the 206K, and about 1pF less capacitance. These figures bracket the value Luna Semi gave you for their device.

If I replace the diode with a 10pF NPO and do the same measurement I get a flat 10pF reading with a few ohm Rs as expected.

Here's a measurement on an HP 4191A at 500MHz

which is close to your 50 ohm. At 1GHz it's reading 33 ohm (the spring test fixture is only rated to 500MHz).

And the BPW34s at 500MHz.

ve

i

Thanks!

This is all a bit mysterious, because the noise voltage is too low for the measured resistances. Of course a 300K diode looks like a 150K resistor of the same value, because only half the zero-bias diffusion current contribut es, so mild weirdness is not unexpected.

I've tried ring-contact diodes (EG&G FND100 and Hamamatsu S1722) with no gr eat improvement. I may have to go to 1-mm class photodiodes with lenses glu ed to them, but I'd really rather not.

Your tester looks very nice--what is it?

Cheers

Phil Hobbs

A lens - get more photons! - is a great way to improve s/n.

Yup. Problem is, the Fresnel reflections from the curved surface can mess u p some measurements.

I've now measured about 20 varieties of JFET, and the best BF862s are still the champs. Some others are in the same range--better than the worst, not as good as the best.

Cheers

Phil Hobbs

An HP 4194A. They are indeed very useful (impedance to 40MHz and gain/phase to 100MHz). But they have held their price very well in comparison to instruments of a similar vintage, so they're in the need to have rather than the nice to have category.

The somewhat related HP 4192 and HP 4195 can sometimes be found for a bargain price (especially the latter). The 4192's are good to 13MHz and perform their measurement in a similar manner to the 4194.

The 4195's go to 500MHz but base their impedance measurement on load reflection (like the 4191) so are only of use for impedance measurements at higher frequencies. This

is a Cs+Rs measurement on a 4195 showing a similar reading to that of the 4191 posted previously.