What makes a good Darlington pair (as sensor)?

[mishima]

I wanted to try using a Darlington pair with an ion chamber (small tube) to detect alpha particles, as this site explains: http://www.techlib.com/science/ion.html

Before I order any components, I was wanting to throw something together with some normal transistors I already have. I was curious what qualities make a transistor more or less desirable to use as a Darlington pair.

I have some 6027, 3904, 2222A, 4401, and 2907A transistors. Which would be best? Worst? Why? Thanks for any insight.

 

[jim hardy]

No takers yet ?

You are trying to measure miniscule currents, meaning your transistors are just barely conducting if at all, so probably you'll need good "off" characteristics.
I_CBO is a measure of the transistor's leakage current.Look at the mpsw45's cutoff currents of 100 nanoamps http://www.onsemi.com/pub_link/Collateral/MPSW45-D.PDF
2222's of 10 at room temperature http://www.fairchildsemi.com/ds/PN/PN2222A.pdf
3904's of 50 http://www.fairchildsemi.com/ds/2N/2N3904.pdf

In a Darlington arrangement, the leakage of the first one gets multiplied by the Beta of the second one. It'll be interesting to see if your home-made darlingtons can compare with the MPS45 at ion chamber current levels. (I assume you're not playing with real strong sources - don't carry an Americium smoke detector source pellet in your pocket it can hurt your skin)

Looks like that MPS45 is hard to find in hobbyist quantities.
See what you think of the 2N5306 datasheet. They're 58 cents and in stock at Digikey.
http://www.centralsemi.com/PDFs/products/2N5306.PDF

There's better circuit designers than me on board here. Ask again if what i said isn't adequate.

old jim

 

[mishima]

Thanks, is it right to think of leakage current like noise for my signal in this particular situation? Like with a higher value, I might be tricked into thinking a decay event occurred when it hadn't?

Also, I notice that leakage current is proportional to temperature. I assume this is related to the fact that conductivity increases with temperature for semiconductors. I was wondering more about this relation, as it doesn't seem to be linear (from the datasheet). For example, if I cooled a 2222 to 0 C what decrease in leakage current could I reasonably expect? Could doing that even be worthwhile?

 

[jim hardy]

Noise isn't a bad way to think of it. Your transistor can't tell whether current comes from the ion chamber or from leakage. So it'll respond to either. But the term "noise" usually refers to AC voltage that you could hear, like the "hiss" in cheap audio gear so i'd call it what it is - leakage current.

In that "Experimenter's Ion Chamber" schematic observe they build two identical amplifiers , give one no input current and measure the difference in their outputs. The plan is they'll both have about same leakage current and if mounted very close together so as to have same temperature, the "leakage" part of the signal at each side of the meter is equal so it cancels . Well, just how equal depends on how well matched the transistors happen to be...

Leakage current in silicon transistors changes by a factor of 2 for about every 15 degrees C .
So cooling might help.


good luck !

 

[skeptic2]

Have you considered using an opamp with mosfet inputs such as the CA3140? With it you can trim the offset current and you'll have a better idea of the gain of the circuit.

 

[jim hardy]

Another good amplifier is TI OPA128.
http://www.ti.com/lit/ds/symlink/opa128.pdf

 

[mishima]

I think I will order a few of both of those, thanks for the suggestions. The 128 in particular looks tailor made for this application.

I made the tube yesterday out of a 2 cm diameter copper coupling 4 cm long attached to a copper plate with a hole in it. I hooked it up to a Darlington pair on a breadboard of 3904s. My multimeter was jumping even without any sample, so I think I need better shielding. I am going to try the 2222s and see if there is any increase in stability, but I don't really have a way to quantify it without logging software or something, it jumps faster than my multimeter can account for (or seems to). The range seems to be between 0.1 and 1 volt.

My sources are not particularly strong, I did not have the guts to rip out some americium. I have a few thorium lantern mantles which were part of a cloud chamber kit from an educational supply. I also have a chunk of bismuth from the same place.

Also something strange that I'm sure has an obvious answer...I wrapped the circuit in aluminum foil to see if that had any noticeable shielding effect. After doing so, when I took a few steps toward the table I would get a jump to about 1.4 volts but only when my body was in motion. The foil was not touching anything but the table.

edit: with the 2222s I am seeing less frequent jumps, but the max has tripled to about 3 V...

 

[mishima]

So with my bismuth and thorium I was never really certain it was working at all, there was not a noticeable change in meter behavior between when the sample was close or not. I eventually broke down and got a smoke detector and confirmed without a doubt the functionality of the normal transistors (I used 2 2222s in the end). Even without shielding it was obvious the sample was setting it off.

 

[jim hardy]

Be careful with that smoke detector source. Dont carry it in a pocket next to your skin.
Store it in a jar, perhaps wrapped with aluminum foil.

for some reason EPA allows them to be disposed of in landfills. I'd mail mine back to manufacturer.

When it's installed in the smoke detector the few cm of air is adequate shielding.