RTD linearization

ftp://jjlarkin.lmi.net/RTD_in_air.JPG

ftp://jjlarkin.lmi.net/RTD_on_board.JPG

ftp://jjlarkin.lmi.net/RTD_lotsa_copper.JPG

This confirms my observation that tiny surface-mount resistors can dissipate a lot of power if they are soldered to big chunks of copper.

John

On a sunny day (Mon, 14 Dec 2009 12:38:00 -0800) it happened John Larkin wrote in :

All good and well, but when measuring temperature I would think you wanted a low thermal mass to get fast response?

I have LM135 in an old SMD plastic tube filled with silicone kit, outside,

1 meter from the ground, about 10 cm from the wall, hanging from the wire. Been working now OK for 10 years or so. Nice defined output voltage versus temperature. You can calibrate those too. But you really do not need better accuracy then 1C... ever.

I wrote: All good and well, but when measuring temperature I would think you wanted a low thermal mass >to get fast response?

I have LM135 in an old SMD plastic tube filled with silicone kit, outside, 1 meter from the ground, about 10 cm from the wall, hanging from the wire. Been working now OK for 10 years or so. Nice defined output voltage versus temperature. You can calibrate those too. But you really do not need better accuracy then 1C... ever.

And I want to add something to that. All that talk about measuring to .x digits temperature (substitute for 'x' what you had in mind), will really look funny when: The sun shines on your outside sensor, and the temp is all of the sudden 20 degrees to high. Hail falls on it, and starts melting, and it insists it is zero C. Snow, rain, so: You need a very good place, shielded from wind, sun, away from hot walls, other radiating objects, at the correct height.

I have been measuring outside temp now for many years, and seen some very funny readings.

low thermal mass

But I am an artiste!

degrees to high.

readings.

I plan to put the outdoor sensor (1K RTD in a plastic tube, epoxy filled) on the north side of the cabin, in the niche under the stairs where the gas meter (and all the old skis) is stashed, exposed to the world but protected from rain and snow. The adjacent wall is concrete blocks, the wall of the unheated garage.

Response speed isn't much of an issue here.

John

On a sunny day (Mon, 14 Dec 2009 14:40:42 -0800) it happened John Larkin wrote in :

Are not we all? Designing is an artistic action.

degrees to high.

funny readings.

Well, at least you can exchange it anytime for a LM135 in that same circuit, without needing linearisation, only a software update.

Sounds good, hope the cat does not sleep on it there :-)

Yes that is true, weather does not change that fast.

would be better.

signals

Here we normally use pt1000 so copper resistance is not such a big issue, and driving it with a current source has caused lots of issues, because the common mode noise rejection sucks when one wire sees ground the other a high impedance current source it is possible to get around it with a much more complex circuit but why bother if you can correct in software

-Lasse

On a sunny day (Mon, 14 Dec 2009 15:07:57 -0800 (PST)) it happened " snipped-for-privacy@fonz.dk" wrote in :

source would be better.

of signals

Why not use LM135 or similar?

source would be better.

of signals

might be simpler electrically, but generally the inductrial type temperature sensors in stainless steel meant to screw into e.g. pressuretanks pipes etc. are pt1000/pt100 types

-Lasse

A pt1000 element is available in industrial bolt-on versions and no polarity issues. IOW any fool can install a pt1000 element.

--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
                     "If it doesn't fit, use a bigger hammer!"
--------------------------------------------------------------

degrees to high.

funny readings.

I'm not a big fan of the National LM35 family. They aren't very accurate, have EMI sensitivity, and all the ones I've used have nasty latchup problems. Not a good choice for outdoors at the end of a long cable.

Platinum RTDs don't latch up and don't rectify RF and don't oscillate into capacitive loads and are accurate to a fraction of a degree C.

John

That's what the guys at one shop said. Until I torque'd off a 16mm crane hook ... screeee ... *POP*

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

On a sunny day (Tue, 15 Dec 2009 00:18:56 GMT) it happened snipped-for-privacy@puntnl.niks (Nico Coesel) wrote in :

Yes I know, I have designed input circuit for PT1000. But I actually used constant current. But in many cases you do not need something like a PT1000, I have done industrial temp sensing with a Si diode too... hehe And, usually things like PT1000 are not installed by fools, but qualified technicians. Who know about 4 wire systems for example.

On a sunny day (Mon, 14 Dec 2009 16:01:08 -0800 (PST)) it happened " snipped-for-privacy@fonz.dk" wrote in :

linearizes

source would be better.

sort of signals

True, but this started about J.L's outside temperature sensor for his home automation IIRC.

snipped-for-privacy@puntnl.niks

Give it another few years of immigration and trade unionism. You will=20 be lucky if they ever learn the local language usefully, or much else.

linearizes

widget

I think you slipped some decimal points there. That would require a=20 theta(device-air) on the order of 1000 K/W. I don't think even a bare =

01005=20 could be that bad.

linearizes

out

widget

I would like to see them just as a matter of curiosity.

wanted a low thermal mass

outside,

wire.

=46or household use you are probably right. In some other cases that is = no where=20 near good enough. For example some sensitive chemical processes.

ftp://jjlarkin.lmi.net/RTD_in_air.JPG

ftp://jjlarkin.lmi.net/RTD_on_board.JPG

ftp://jjlarkin.lmi.net/RTD_lotsa_copper.JPG

John

His numbers are about right. My 1206 RTD, in free air with skinny leadwires, was roughly 300 K/W, so 6 mW make about a 2 degC error. On an actual pcb with deliberately fat traces, theta is maybe 5x better. I intend to use a bigger ceramic-slab RTD and epoxy it into a plastic tube... and check it against a good thermocouple anyhow. I expect a self-heating error well below 1C. I think I can use the coax as sort of a thermal antenna, too.

You might check your numbers before you suggest that Sphero has "slipped some decimal points."

John