Zenering a big transistor

te:

com

iode connected)

miters. (LND150)

so far).

age the junction?

gain after

too?

s would do it.

istors

ause

TDs

For kicks, I took 6 irf520 (N-chan fet) stuck 'em to a hunk of copper, near each other. Shorted G to S (after one measured zero volts) and used my DMM to measure the body diode.

That's not too bad. There must be loads on the web about using the fet body diode as a temp sensor.

George H.

IUI.

Cryo stuff?

.com

te:

hoo.com

r (diode connected)

t limiters. (LND150)

sted so far).

damage the junction?

ge again after

lse too?

iodes would do it.

ransistors

can

because

t. RTDs

s:

ng

ected

ll

ces

he

bias

Right, 77- 400 K. The diode connected transistor is going to 'crap out' somewhere below 77 K. Well it will still be a diode but no 'gain' from the collector.

Yes, thanks. I originally saw that in EDN, I think, where it had a better write-up. I used the technique to stabilize a smoke detector design once.

I think it might have been ?Transistor Sensor Needs No Compensatio n,? Jim Williams, EDN, April 25, 1991, but I can't find it just now.

(Update: Okay, I found it. That was the article, but I don't notice any textual differences between that and the file you posted.)

Woodward has a three-current version to reduce ohmic errors.

Here are some more of the app notes I scanned through yesterday...

AN137 - Accurate Temperature Sensing with an External P-N Junction

AND8494 - Thermal Sensing Methods used in ON Semiconductor Devices

AN10349 Digital temperature sensor accuracy explained

AN12.14 Remote Thermal Sensing Diode Selection Guide

Cheers, James Arthur

Thanks for all those nice links! I collect such things... you never know when such a need will arise. Many of these schemes, difficult in a discrete implementation, are a piece-a-cake in my custom chip world ;-) ...Jim Thompson

Lakeshore has some magical recipe that makes repeatable temperature measurement diodes at liquid helium temps, but it's not easy. Other people have messed up at this.

om

te:

hoo.com

@yahoo.com

nsor (diode connected)

rent limiters. (LND150)

tested so far).

to damage the junction?

ltage again after

g else too?

w diodes would do it.

a transistors

g I can

ool because

g it. RTDs

going

onnected

well

evices

ily

ts,

k the

of bias

yone

Hmm OK, I have no liquid helium here. So I can't do tests below 77K. It'd be fun to look at commercial diodes at low temps... maybe find one that works. (Is there really a special sauce at Lakeshore?)

George H.

Below 20K, diode voltage drop skyrockets (to, like, volts at a couple K) and things get ohmic. Apparently it's erratic on doping or something, because other people have tried to compete with Lakeshore and every batch was different.

We did the cryo instrumentation for the Jlabs/CEBAF machine, and for the helium liquefaction plant for the Supercollider (which may still be there.)

Hmm OK, I have no liquid helium here. So I can't do tests below 77K. It'd be fun to look at commercial diodes at low temps... maybe find one that works. (Is there really a special sauce at Lakeshore?)

George H.

Pull a vacuum on the LN2 and it will go colder than 77K.

Yeah I'd figure there should have been plenty of people trying to make one work, given the price Lakeshore charges. So a project most likely to fail.

Below 20K where are the charge carriers coming from? Very shallow impurities? You could shine a little light on the junction to make carriers.. that's kinda cheating.

George H.

I wanted to use a tuning-fork quartz crystal as a cryo temp sensor, but I couldn't get my customer (who had the facilities to help develop it) interested. They stuck with the Lakeshore diodes, lots of them.

Huh, the 32kHz ones? Aren't they made to have a low Temp-co? I guess away from room temp they would change more. They'd have to live in vacuum or the liquid helium might change the frequency. Most people are afraid to try something new... better the devil you know.

George H.

Right ~65K or something like that. I had a friend who did that for an experiment in grad school. He made a big block of solid nitrogen in the dewar.. it took a while to thaw.

George H.

Their TC is parabolic, intended to have the flat spot at human-wrist temperature. At low temps, it's very steep.

I got some samples of a vacuum-sealed 32KHz tuning-fork crystal, but I never had the facilities to test them.

Most people are afraid to try something

They could have at least got some papers to publish!

This is almost impossible to read, but I think they take one down to

George h.

Do you use 2 with different tempcos & compare f?

NT

COTS crystals don't have the tempco you want. This is the one I recall:

The high Q of a quartz oscillator implies small self-heating of the sensor. The aging of quartz means it needs recalibration to stay accurate. So, how do you calibrate a 0.0001 K precision thermometer?