Good temperature sensing chip?

[con-f-use]

Hi, I'm looking for a cheap easy to use IC to sense the temperature of e.g. a MOSFET. Will be used with an ATTiny85 and doesn't need better than 1°C resolution. Should be accurate between 0-150°C. Any recommendations? I can crawl the usual suspects and buy the cheapest, but I don't want to by the cat in the bag. So what have you used in the past?

Thanks in advance.

[Richard Crowley]

What does "Cood" mean?  I'm confused.

[nctnico]

Microchip TC77 is nice. 150 deg. C seems to be a bit high though.

[con-f-use]

"Cood" means I'm dyslexic and English it not my first language. Don't mind me being stupid.
@150 Celsius: not a hard requirement but I might need it for another project, later on.

[ovnr]

The 150°C requirement is perhaps a bit high? I'd probably use a diode (measuing change of Vf) or NTC resistor if I had to go that high.

I really do like the ADT7410 chips though, and they're actually rated up to 150°C (I didn't expect that!). SO8 package, 0.5°C accuracy, and pretty high resolution (I2C interface too) - and it also has configurable alarm outputs. They're kind of overkill and also somewhat expensive, but they Just Work (tm).


(edit: Corrected my wrong assumptions about temp range)

[miceuz]

I've used TC1047 - SOT-23 package, typical 0.5 °C accuracy, goes to only 125°C though.

[Smokey]

What are you sensing?  Does it need to be mounted to something like a heatsink to sense heatsink temp?  Are you sensing air temp?
Do you need analog output or do you want to read the temp digitally?

[con-f-use]

Digitally preferred but the Attiny has an ADC so I don't really care. Mounted directly to heat sink or other IC package via thermal epoxy glue for now. However, I will order a few more for future projects.

[Fank1]

If it's to check a Mosfet you will have to mount it on the case.
By the time the case reaches 150 C the mosfet has already failed.
The 150 C rating for the mosfet is the junction temperature, not the case temp.
The 125 C version would be fine.

[con-f-use]

Yes I know that. For the MOSFET thermal shutdown projects it's not necessary. As I said in my OP, I have another project where 150°C would be great so I don't have to order twice. Guess I'll have to use a thermo couple and amplifier for that.

[dannyf]

A diode (2n3904/6 for example) would fit the bill. 150c is tough, however. It is as accurate as you can calibrate.

[con-f-use]

That was absolutely new to me, but now that I read a little bit it makes sense. Doesn't seem terriby reliable though since it needs at least two point calibration and a stable current source. But interesting, who new!

[dannyf]

The tempco of those devices is surprisingly consistent, assuming that they are from the same batch / manufacturer and biased to the same dc operating point.

The starting point (Vfwd at any given temperature) can be quite different, sometimes 20mv (~=10 degrees). So once the tempco is measured, you just need to calibrate individual devices' Vfwd (aka one point calibration).

[LukeW]

Here, let me DigiKey parametric search that for you

The LMT87 http://www.ti.com/lit/ds/symlink/lmt87.pdf is the cheapest one I could find on DigiKey that is actually rated up to 150C, excluding NTC thermistors or just using a diode or transistor.

With a typical thermistor you'll need to do the maths to convert its resistance function, which is exponential and not linear, into temperature. Having a semiconductor device with a linear output is tempting.

I often like using the Microchip MCP9701 (or others in the MCP970xx family) for non-critical temperature sensing applications accurate to within a couple of degrees, for environments below 125C. They are very cheap, and linear, and without any external components or annoying maths. But only rated up to 125C.

If your 150C Tmax specification can be compromised a little bit, say to 125C, this is likely to significantly open up a wider range of candidate options and cheaper choices.

[tree]

AD590

OR design your own:
See page 7, http://cds.linear.com/docs/en/application-note/an45f.pdf

[con-f-use]

Okay thank you very much. I think, I have what I need now.

[denelec]

There's also the LM35.
http://www.ti.com/lit/ds/symlink/lm35.pdf
It outputs 10mV per  degree Celsius.

And the LM34 outputs 10mV per degree Fahrenheit.
http://www.ti.com/lit/ds/symlink/lm34.pdf

I once used a LM34 tu build a thermostat. Works great.

[Rick Law]

Hi, I'm looking for a cheap easy to use IC to sense the temperature of e.g. a MOSFET. Will be used with an ATTiny85 and doesn't need better than 1°C resolution. Should be accurate between 0-150°C. Any recommendations? I can crawl the usual suspects and buy the cheapest, but I don't want to by the cat in the bag. So what have you used in the past?

Thanks in advance.

How about the LM35.  Per their datasheet, -55 to 150C, accurate to +-0.5C.

[cellularmitosis]

I like the MCP9701A and MCP9700A.  Less than $0.40 each.  I've super-glued them to the backs of heatsinks before.

[con-f-use]

Thanks. There's very little difference between the MCP9700A and the TC1047A except that the latter is a little more accurate, its temp range and current consumption is a bit worse and it costs twice as much as the MCP. Why would Microchip manufacture both, and why would I buy the TC1047A when I can get the MCP9700A at halve the price?