Photodiode Question

The very concept of a photoresistor seems strange to me. I've read up on the effect and don't understand it. I think their history effect is worst at low currents.

People used to use photoresistors as choppers. HP made a voltmeter that used CdSe cells as both the chopper and demodulator, illuminated by an incandescent lamp with a motor-based chopper wheel.

John

The old HP Nixie tube frequency counters actually use CdS photoresistors illuminated by neon bulbs to drive the Nixie segments! (I have a few in my basement.)

Photoconductors exhibit gain equal to the carrier lifetime divided by the transit time--essentially you get to re-use the carriers many times. That's also the origin of the speed/gain tradeoff.

Unfortunately both the photogeneration and recombination are stochastic, so they have twice the shot noise of a photodiode. This is sometimes called 'generation-recombination noise', but that causes confusion with normal thermionic G-R noise in IR photoconductors.

I don't know the origin of the photoresponse hysteresis of CdS and CdSSe, but a SWAG would be that there are long-lived trap states that get saturated in very bright light, and that this effect makes the carrier lifetime longer.

Cheers

Phil Hobbs

I couldn't agree any more. Your decision is well understood! ;)

Oh, ya know, for a physics lab we exposed CdS photocells to a strobe light and watched the "exponential" decay. Now, theory claims recombination is proportional to concentration, so the exponential feeds back into itself effectively as a varying tau. We were supposed to measure this 1/R correspondence (where R is inversely proportional to carrier concentration, hence the measurement), but when I ran through all the plots and such, it looked stupid. On a whim, I took sqrt(tau) instead and that was linear. I pretty well stumped the professor with that one. ;-)

Weird? Ya...

Oh, and don't forget you can use CdS photocells as MOSFETs.

Tim

It doesn't have to be, feed a square wave clock through a small capacitor to the gate if there's a particular frequency you want.

--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---

Hello,

Go to

Best regards,

Wim PA3DJS

to

's

The document that I have is no longer present on the Perkinelmer website. Here:

Best regards,

Wim PA3DJS

ve to

Se's

.)

so.pdf

Thanks for the document, page 35 talks about shot noise in LDR's. Are they just throwing equations around or is there shot noise in LDR's?

George H.

Free-free recombination is proportional to carrier density squared, so dG/dt = -kG**2 for some k. There are solutions proportional to 1/t.

On the other hand, if the recombination happens mostly at the ends of the elements, it'll be proportional to the current. Did your experiment run at constant current or constant voltage?

Cheers

Phil Hobbs

Not really much of either, it was a voltage divider.

Tim

Interesting. I've never really understood the finer details of the physics. For instance, the books say that photoconductive gain still goes as the carrier lifetime even when that's longer than the transit time--IOW when the a carrier gets to the bond pads, apparently another carrier is magically injected to compensate.

Some charge neutrality incantation is apparently responsible, but that's pretty hard to believe when the element is a long skinny serpentine thing like a Vactrol, and is made of nasty polycrystalline stuff. Also, any such process would have to be stochastic, which would give rise to additional noise over and above the 2x shot noise from stochastic generation and stochastic recombination. (It would be an interesting measurement, if I had any reason to care about photoconductors.)

Cheers

Phil Hobbs

Here:

Since photons make current, probably so. But LDRs are so messy it's likely buried in other gunk.

One of my guys is building a test setup to measure shot/excess noise in resistors. We need a ~~ 100M resistor to create a small (50 nA maybe) bias current with below shot-level noise. It's not clear if high-value cermet resistors have shot noise or not, so we plan to measure a bunch.

Metal film resistors don't go to very high values; we can probably get

It's an interesting problem, trying to generate a nA-range DC current with low noise. Low voltage across a low-value metal-film resistor doesn't work because of Johnson noise. High value resistors are noisy in their own right.

John

There are also photoresistors made from radiation-damaged GaAs. You can make a ps-speed electrical signal sampler from a GaAs photoresistor banged by a fs-range laser. That would make a nice sampling scope if the laser weren't so big and expensive, and if you could trigger it without huge amounts of jitter.

John

Here:

The conductivity comes from photogenerated carriers, so there's one copy of the shot noise there. Another copy comes from the stochastic recombination, so a photoconductor with unity gain (i.e. where the transit time equals the carrier lifetime) has exactly twice the shot noise.

If there's a lot of gain, i.e. you get to reuse the same carriers lots of times before they recombine, the shot noise gets amplified too--just like an APD or PMT.

Cheers

Phil Hobbs

People have been doing that sort of sampling for 25-30 years--first using coplanar transmission line on an electrooptic crystal like LiNbO3, where you look at the polarization shift in the light, and somewhat later using the plasma-optical effect, which can be either an optical-readout sampler or a switch. Dan Grischkowsky and Jean-Marc Halbout were doing it when I joined IBM, and before that, Dave Bloom was doing it at Stanford (he might have started it at Bell, I forget).

Using a femtosecond laser (a Hall-Haensch comb generator) will get you unmatched stability, but of course the experiment has to be synched to the laser rather than the other way round.

Telecom modulators are getting up above 50 GHz bandwidth, but only in the 1.5 um band, which is too long for GaAs.

Cheers

Phil Hobbs

have to

CdSSe's

ory.)

am.

at

toiso.pdf

"> Since photons make current, probably so. But LDRs are so messy it's

Hmm, Yeah this is a bit confusing. The photons make the charge carriers and reduce the resistance... and as Phil H. says there is generation and recombination noise. But this should only depend on the light level and not on how the device is biased. For a given light level if I double the bias voltage (and thus double the current) does the noise go up?

I'm using 100M and 1 G resistors from Ohmite to make a 10nA current source. I'll look at the noise next chance I get.

George H.

't have to

=A0(CdSSe's

story.)

spam.

that

Optoiso.pdf

re

sorry for the VERY naive question, but would some form of junction leakage have lower noise? Or are junction leakages at and above johnson?

ave to

SSe's

y..)

toiso.pdf

t -

Thanks Phil, I must admit I don't quite 'see' how that works out. If I 'reuse' a photo generated charge carrier say ten times I get ten times the shot noise? (OK twenty times since I get twice the shot noise to begin with.) I'll have to think about that. The bias current divided by the photo current is something like the gain of the LDR.

Say (speaking of noise) I think I remember you mumbling sometime ago that you had a nice derivation of the Johnson noise formula. Something that didn't involve the counting of modes in an infinite transmission line as Nyquist first derived it. Have you written this up anywhere?

George H.