CdS Photocells

I wouldn't expect them to have any hard limits. Testing would be really expensive.

The induced carriers have to dissipate, so I'm surmising a Spice model might handle that with the variable-R/C scenario. ...Jim Thompson

A plot of R versus time after drive current is removed would tell me a lot. ...Jim Thompson

I'd guess an exponential, (blinking an led is not that hard. I'm assuming you have a 'scope.)

George H.

One would need a chart recorder for such a slow event. ...Jim Thompson

You could spend a modest fortune acquiring all the IEC specs.

and that's just EMC. How does one know which specs are applicable to which instruments?

Maybe these will help

...maybe he thinks the CdS cells will be used for AVC or summtin lyke thet.

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

Ok, here's the lab:

Fit's not as good as I remember it, but it is suggestive.

Setup:

- CdS photocell to +10V

- Resistor to GND (voltage divider)

- Rigol DSO reading divider

- Xenon photoflash exposes photocell periodically

Obviously, the DSO reads quantization noise, which I'd partially smoothed out in one of the columns, but better smoothing functions might be applicable too.

Tim

Is this an old lab, or something you just did? What's the bump at ~10k ohm in the tau vs R graph. That looks weird.

George H.

To resolve events that need 5mS resolution would need a rather fancy chart recorder, particularly if a direct resistance measurement is needed.

Just reading a single channel, something like the HP34970A gets very strange when asked to record simple 2wire resistance at 50mSec intervals or less, with a fixed (no autorange) minimum-digit range.

I don't know if this is firmware, but the most rapid data report seems to be 200mS intervals. This may be sufficient for turn-off plotting only.

....so a storage scope is probably the best bet, unless a voltage derived data recorder reading develops fewer recording issues.

RL

Thanks, Joe! ...Jim Thompson

Thanks, Tim! ...Jim Thompson

Just drive the element with a current source and plot voltage. ...Jim Thompson

Could you just drive 10uA thru the resistance and measure voltage versus time? ...Jim Thompson

The carrier recombination rate goes approximately as n_E * n_H.

(In real life there are densities of initial and final states to worry about, among other things, but this is a reasonable approximation most of the time.)

In a classical photoconductor, i.e. one with a lightly-doped channel, the minority carrier lifetime is nearly constant until you really blast it. That's true of things like germanium photoconductors used in the far IR. You freeze out most of the dopant states, so the relevant band gap is only a couple of meV.

CdS and its relatives are generally very nearly stoichiometric, iirc, so that the background doping is very small. That gives them a very large on/off ratio, but their recovery is strange. Even in the absence of traps (of which there are many, it being an amorphous or polycrystalline material), the recombination rate is quadratic in the carrier density, leading to long tails in time. It's like a fast pulse in a long transmission line (series resistance going as sqrt(f) due to skin effect)--it sort of drools its way back toward the dark state. The long-lived traps do other things that I don't have a good handle on ATM.

It's horrible stuff. Back when the competition was RCA vacuum photodiodes, there was a decent niche for CdS in night lights, audio compressors, and electric eye doorbells, but I can't think of anything I'd want to use it for today. Even _phototransistors_ are way better as well as much cheaper.

Cheers

Phil Hobbs

Oh, and I used to have an HP counter (5245L or something) that used NE-2s illuminating CdSe cells to control the segments of Nixie tubes. That was pretty cool.

Cheers

Phil Hobbs

Street light photoswitches used to, might still, use a CdS cell and a resistive-heater-bimetal switch. Nice slow response.

There were some heat-sinkable TO3 power CdS cells.

I once had an HP voltmeter that had a 1 mV full-scale range, which was impressive back then. It used CdSe cells as chopper and demodulator stages. A synchronous motor, a light bulb, and a chopper wheel sequenced everything.

HP and other people used CdS cells in self-tuning audio distortion analyzers.

Just.

There doesn't seem to be much concern for illumination source amplitude or control here. Admittedly an on/off type of trial...what would you say if I told you that, at turn-on, the record shows an apparent sudden increase in the 'resistance' measurement, when only a reduction is anticipated/demanded by expectations?

Obviously an artifact generated by the test method. But?

Will the test method ever be good enough, over the orders of magnitude that a simulation will eventually demand (always more than real life it seems)?

So first, a faster-than-measurement constant(quasi-calibrated) light source.

Then possibly a constant voltage, to measure current flow, with log response.

Beginning to sound more like one of the more elaborate photo-amplifiers that some in this thread are dabbling with.....not just an LED and resistor.

Maybe it should be modeled first, before the practical results we do get are trashed.......

RL