temp sensor

[zkt]

What do you think about this Maxium chip. Looks to me like it would take a lot of the work out of it. http://datasheets.maxim-ic.com/en/ds/MAX6675.pdf about 6 USD from Digikey

[LGR]

That only works with type K, but that should be fine. It's internally compensated, which makes it a lot easier. Just be sure that the chip is very close to the TC terminals, and use type K wire up to the terminals.

[FRM]

What do you think about this Maxium chip. Looks to me like it would take a lot of the work out of it

That's why I mentioned it in the first place.

Incidentally I use in-house manufactured micro-thermocouples (made from 46swg wire!) to fit inside a 0.1mm micro-capillary tube for some micro-engineered pressure-vessels (oil analysis), because:
1. you end up with a very small sensor.
2. The reactance is faster than any other sensors I know about.
True, they are nasty non-linear things to deal with, but using good op-amp technology (LTC1050) and companion CJC chip (such as LT1025).
Couple that with a heated voltage reference (and a temperature controlled environment), we manage to get down to +/- 0.1°C combined accuracy over a LIMITED temperature range of 0-60°C with careful calibration.

Temperature sensing is very much 'horses-for-courses', and this is one example where a thermocouple is the only option. No one in their right mind would bother to go to all this trouble if there is an alternative...

[zkt]

I`m sorry fmr I completely glossed over your suggestion.Thanks for the help. And everyone else too.
LOL Thise guys at Omega Engineering are really on the ball. I got to the desk early today to call them up and change the order to K type and its already shiped out! Anybody want a couple J type TC`s? Yours for the shipping cost
Edit
GRRRR I knew this was going too smoothly. Seems that none of the usual suppliers D-k, mouseer, newark have the max6675 in stock. Lead time even at Maxium is 21 WEEKS. Does anyone know of a supplier (in prototype quantities) or a similiar IC that will work with a J or K TC or can lend me a max6675 until my order comes in ?

[FRM]

I normally request (free!) samples...you need to sign up for free samples but if you do then you can go:

https://shop.maxim-ic.com/storefront/se ... ftolbGmkTy

and order a few for free...

They have one variant of this chip available as a sample...but it will take a little while to get it as all samples are flown out of the phillipines/malaysia!!!

[FRM]

For reference...
http://ww1.microchip.com/downloads/en/A ... 00844a.pdf

http://www.picbook.com/downloads/doc/PI ... brator.pdf

and I don't know about this sort of company, and I don't know where you live but this one is in americas..

http://www.1sourcecomponents.com/partinfo/max6675.htm

Hope this helps you.

[LGR]

FRM - You can get get low mass Pt RTDs with that kind of response time, and they're more accurate, and much easier to connect. If you're above something like 500 C, then the thermocouple is the only sensor left on the table, but in the 100-200 range, the RTD is superior. Thermocouples were the standard instrument 30 or 40 years ago, but are only competitive at the very high end of the termperature scale now.

I see no reason to use a thermocouple in the 0-60 C range; they offer no advantages over low-mass RTDs, or even thermistors.

[zkt]

Seems that I left the word DON`T out of the sentence "Seems that none of the usual suppliers D-k, mouseer, newark have the max6675 in stock."
Called onesourse. Best they could do was 10 pieces at $25 each. I cant justify that for a prototype. Guess I`ll request samples from Maxium and put this project on the back burner for a while. On the other hand, I could try to build it without the Max chip. I have a feeling that would be VERY challenging.
edit:

Re the picbook reference. "Due to the unavailability of the
MAX6675, the functionality of the interface with this chip could not be tested." LOL

[FRM]

I see no reason to use a thermocouple in the 0-60 C range; they offer no advantages over low-mass RTDs, or even thermistors.

These low mass Pt sensors - will they fit into a 0.1mm diameter tube? (and only occupy small area)...plus will they handle 700BARG?
If so, then please let me have some more details as I am genuinely interested...

We also build class-A Pt100 based laboratory thermometers that have a calibrated accuracy of +/- 0.01°C from -50 to +150°C...using a relatively complex linearising analog front-end & simple 7seg (no microcontroller!). Hopefully one day I'll get around to coupling the measuring circuitry to an AD7715 I had ear-marked for the job...add remote logging, digital calibration, battery operation, etc.

I feel another project coming on.....

[LGR]

You didn't mention that about being Pt. Those are a very specialized instrument that may very well be the exception to the rule. I'm sure you get a very good price for them, but they're not for everyone....

I think the assumption regarding zkt's project is that you want to avoid calibration, if at all possible, because, as you well know, that's a very time-consuming and expensive process. If you're going to do that, the type K thermocouple would be ok, but the other choices would be good too. If he can tolerate swings of 10C, anything that can handle the temp will work. But as I said before, it has to do with whether this is a one-of-a-kind, or a limited production, or mass production. If it's one-of-a-kind, it's not worth beating it to death. If it's mass produced, it is.

[janni]

Hi zkt,

One thing I don't really get is why you have to measure the air temperature instead of specimen temperature which should be more interesting. Why not attach the sensor to the brass screen under the specimen instead. Small active volume of termocouple junction gives good response in contact measurements, but flowing air temperature measurements may pose a problem - heat flow through the wires significantly affects the accuracy.

I would be also afraid of IR radiation heating the screen (and the specimen as well) making the air temperature measurement pointless. Placing another fine mesh screen between the sample holder and heater, or placing the heater 'out of sight' of the sample should prevent it.

And I'm not sure you won't have temperature regulation problems due to low temperature resolution of the MAXIM chip (c.a. +/- 3K in the range of interest, if I understood the data correctly). You may not need high accuracy, but low resolution is quite another thing.

[LGR]

Janni - all good points, but you have to be careful when connecting a T/C to anything metal that you don't create a ground loop. If you're going to do that, you either need to be certain that the input won't be affected by grounding the T/C, or you have to use an insulator (which can be tricky). Unfortunately, all good heat conductors are also good electrical conductors.

[janni]

That's a good point as well, but not that hard to overcome in this case. Having all circuitry supplied from mains through transformer gives the isolation. There are no connections to a PC or other devices and the heater may be controlled with a relay or optotriac.

Anyway, I still think that the easiest solution would be with KTY84 resistive sensor. It may be linearized with one resistor and gives response in volts, not mV. Calibrating it in two temperatures (room and boiling water) should pose no big trouble and be accurate enough.

[LGR]

Exactly. Thermocouple signals are very weak millivolt signals that are very suceptible to noise pickup. RTD signals are more robust, and can tolorate noise much better. They're essentially a current signal, which is much more noise immune. Even with galvanic isolation, T/C noise pickup can be a real problem, if the sensor is touching a significant mass of metal. It can act like an antenna, and pick up 50/60 hz noise.

BTW - I don't think resolution is a problem in this case, because it's on/off control. There's a deadband around the setpoint, anyway. That extra resolution would only matter if he were trying to do PID.

[zkt]

janni- I think youre right. It goes to 300C, is cheap and available. I didnt see any graph of R vs temp. Is it linear ? I`m guessing that a constant voltage source and a/d pic would do it ?