Fast-acting temperature sensor?

Phil Pilgrim (PhiPi)

I have a friend who's into paragliding. He's expressed an interest in recording temperature vs. altitude during his flights. (He says it will give him information about the stability of the airmass.) The catch is that his altitude can change as fast as 750fpm in a good thermal. With the average adiabatic cooling rate being around 5.5 deg. F/1000 ft., I doubt that the usual suspects among temperature sensor chips (e.g. DS1620) can keep up. Can anyone recommend a temperature sensor with low thermal mass that can track the anticipated rate of change?

BTW, I had occasion a couple days ago to examine a NASA dropsonde used by NWS for probing hurricanes. It had what looked like a teeny glass-enclosed diode, which I assume to be a temperature sensor. But I have no idea what the part number was...

Thanks,
Phil

Post Edited (Phil Pilgrim (PhiPi)) : 7/20/2007 3:58:49 AM GMT

Beau Schwabe

Phil Pilgrim,

"...It had what looked like a teeny glass-enclosed diode..."

That's probably exactly what it was. The PN junction has a very reliable/predictable temperature response.

Read toward the bottom of this link:
http://www.capgo.com/Resources/Temperature/Semiconductor/Semi.html

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Post Edited (Beau Schwabe (Parallax)) : 7/20/2007 4:42:15 AM GMT

Paul Baker

Yup, by placing a very small current source through a diode and monitoring the voltage drop, a very accurate measurement of the temperature is possible.

I will be doing this shortly to get accurate die temperatures of the Propeller, now I just have to figure out how to create a 100nA current source.

There is a Maxim App note on how to use an ESD diode for this purpose, while this isn't your application or help you find a part number, it should provide enough information for your application.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Post Edited (Paul Baker (Parallax)) : 7/20/2007 4:52:32 AM GMT

Beau Schwabe

Paul and Phil,

This might do everything you need...

http://cache.national.com/ds/LM/LM84.pdf

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Paul Baker

Sweet, thanks so much Beau, that will make short work of the problem. Guess it helps to have a former employee of NS around [noparse]:)[/noparse].

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Beau Schwabe

This might be a better fit with slightly higher resolution...

http://cache.national.com/ds/LM/LM95231.pdf

http://cache.national.com/ds/LM/LM95235.pdf

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Post Edited (Beau Schwabe (Parallax)) : 7/20/2007 5:17:49 AM GMT

Phil Pilgrim (PhiPi)

Paul & Beau,

Wow, I've been tag-teamed! Thanks so much for the rapid, informative responses! Yep, this has got to be the way to go. I can envision a remote SOT (or smaller) transistor-as-diode, hanging in the air, soldered to two thin strips of copper foil for maximum thermal response to the ambient. And the National parts make it drop-dead easy to interface. This is gonna work!

Thanks again!
Phil

Tracy Allen

The champ for tiny thermal mass and speed of response is a thermocouple made of fine wire. Also, due to the small size, it has low radiation burden. I regularly make thermocouples from stranded wire by extending out one 32 gage strand from each side to make the sensing junction. There are interesting tricks for coating the tiny junction with silica. A chip like the MAX6675 can easily return the temperature value to the processor.

Beau, that LM84 is interesting -- It looks like a good bet for monitoring the die temperature of the Prop that Paul putting through the paces in the new environmental chamber!

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

Phil Pilgrim (PhiPi)

Tracy,

I assume by "stranded wire" you mean thermocouple wire like this? How do you weld the two strands?

Thanks,
Phil

Beau Schwabe

Tracy Allen and Paul,

"Beau, that LM84 is interesting -- It looks like a good bet for monitoring the die temperature of the Prop that Paul putting through the paces in the new environmental chamber!"

Yeah, I "think" that if you use isolated power supplies, you can use the reverse biased diode(s) available on any of the IO pads. This way you could monitor virtually any chip.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Tracy Allen

Phil, I usually use type T (copper-constantan) TT-T-24S, which is 24 gage stranded 7x32, and solder the ends with a silver solder. The added metal does not affect the output so long as the junction is in the isothermal region and only the two main metals leave the isothermal region. I also keep type K for higher temperatures. I've never tried to weld a junction and have purchased some prewelded beads from Omega.

I've never tried DIY welding on the small wire. I've heard it is kind of tricky, but not hard after all the parameters are adjusted.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

Paul Baker

Beau, Im starting to wonder if the National chips you've pointed to will work for my application. They seem to indicate the need for an dedicated diode connected transitor as done on Pentium chips, but I necessarily must use the ESD diode of an I/O pad as talked about in this application note: http://www.maxim-ic.com/appnotes.cfm?an_pk=3500

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Beau Schwabe

Paul,

"...They seem to indicate the need for a dedicated diode connected transistor as done on Pentium chips..."

I believe this will be more of a calibration issue than anything else... once the parameters are defined, then the number returned would represent the correct temperature.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Paul Baker

The issue I am concerned with is that by using an ESD diode to measure the temperature, one of the terminals is clamped to either Vss or Vdd. Since the docs state there is a diode fault state when D+ is sensed to be equal to Vdd, Vss, D- or floating then D+ must be tied to the I/O pin. So to make this work D- should be tied to Vdd and·D+ would drive higher than Vdd, but according to the ESD diode structure of the chip, it's own ESD diode would start to conduct, so Im not sure if the chip(s) will work on ESD diodes.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Tracy Allen

Paul,

I think that is why Beau specified that the LM84 (and its support processor) needs to be run off of an isolated power supply. There is no connection between the grounds nor between the Vdd's of its power supply and the one running the Propeller under test. From the perspective of the the LM84, it is like it is hooked up to an isolated 1N4148, but it is one of the substrate diodes on the Propeller. The propeller is running on its own power supply, and it does not know that one of its substrate diodes has been taken over. That test pin is not used for anything else, and it seems it could be configured as input or output, it doesn't matter, but Beau would have a better advice about that.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

Paul Baker

Ah, thanks Tracy.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

pjv

Hi Tracy and All;

In joining thermocouple wire into thermocouples, I've seen a procedure where the wires are tightly twisted over a few millimeters, and then the very tip momentarily brought vertically into a very hot oxy/acetylene flame, fusing the metals together.

It creates a tiny fused ball at the end and appears to be quite robust,

Cheers,

Peter (pjv)

Tracy Allen

I found a reference I was looking for, where a capacitor discharge is used to self-weld the wires. This was in an engineering journal and has a table with the different thermocouple combinations with wire gage down to 0.001" diameter, versus the capacitance and voltage to use to achieve a successful weld. The wires are held in clamps in a micromanipulator under a microscope, end to end. The wires are bent in a form so that the butt ends are in compression against one another before pulsing the current. There is a type A and a type B bend, depending on the tensile properties of the two metals. The electrical apparatus consists of a bank of capacitors that can be switched in in a binary weighting scheme, and a power supply of 200 to 500 volts. A footswitch delivers the energy pulse.

There is some discussion on the use groups about this kind of technique, but most of it pretty discouraging:
www.craftkb.com/Uwe/Forum.aspx/welding/4280/How-To-Tac-Weld-Thermocouple-Wire
Not at all systematic.

The oxy/acetylene technique sounds a lot easier, if you have the equipment.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com