Hey anyone accessed the built in temp sensor?

http://www.atmel.com/dyn/resources/prod_documents/doc8271.pdf

ATmega48A/48PA/88A/88PA/168A/168PA/328/328P

Page 262

No, but it's not very accurate: +/- 10C if I recall.

Ah yes... here it is: http://www.arduino.cc/cgi-bin/yabb2/YaBB.pl?num=1274403811/0

Cheers,

No, but it's not very accurate: +/- 10C if I recall.

It is ok if you calibrate it. They state +/- 10C as it changes with input voltage and is slightly different on each chip.

I have had it running accurate to about 1-2C within the range of 18-26C

Just load up the code and compare it to a known correct sensor. I'm not sure what you would want to use it for though...

Mowcius

Thank you pocketscience and mowcius. I’d searched for it but could not find it.

I have access to a thermocouple simulator calibration standard so calibrating it should be reasonably easy.

I think a major problem with this sensor is that the AVR chips actual die temperature is probably going to be more a reflection of how much current is being sunk or sourced from the output pins. If the AVR chip has a static current consumption of say 10ma but you are supplying 100ma to drive led in five output pins, then the load may well have a 10:1 influence of the die temp? Not sure on the numbers, so the basic questions is to what use would this sensor be in a typical project application? I don't think you could assume that it would make a decent ambient temperature sensor.

Lefty

I think a major problem with this sensor is that the AVR chips actual die temperature is probably going to be more a reflection of how much current is being sunk or sourced from the output pins. If the AVR chip has a static current consumption of say 10ma but you are supplying 100ma to drive led in five output pins, then the load may well have a 10:1 influence of the die temp? Not sure on the numbers, so the basic questions is to what use would this sensor be in a typical project application? I don’t think you could assume that it would make a decent ambient temperature sensor.

Agreed.

If the current is constant though then you could use it as such. It would be able to tell you if a unit was overheating but that’s probably about it.

Mowcius

<snip!>

I got a cheap LM35 up and going with -55 → 150 Celcius range - pretty easy and off the bat seems quite accurate. I hooked it up based on figure 7 in the LM35 datasheet fwiw.

Well this was a thread on the internal temperature sensor…
Did you connect it up to a negative voltage supply? Otherwise it is a 2 → 150C sensor.

I recommed the TC74 I2C temperature sensor as it is I2C and will easily measure negative temperatures without some silly input voltage.

Mowcius

It would be able to tell you if a unit was overheating but that's probably about it.

I use it for that in one little project, as a "free" extra safeguard. However, the returned value as described here: http://code.google.com/p/tinkerit/wiki/SecretThermometer seems to be in 0.1 milli-°C steps, rather than 1 milli-°C.

However, the returned value as described here: http://code.google.com/p/tinkerit/wiki/SecretThermometer seems to be in 0.1 milli-°C steps, rather than 1 milli-°C.

Ahh.

I have always calibrated it to a known sensor because they are so far off anway.

I recommed the TC74 I2C temperature sensor

Well this is a thread about the internal sensor… sheesh!! I was simply offering an alternative, much like you did immediately after chiding me.

I guess I’m not worthy, unlike your holiness… :o

<bowing, scraping, backing away…!!>

much like you did immediately after chiding me.

I was not trying to chide you, just making it clear. The comment I made was also not a recommendation about using a temp sensor rather than the internal temp sensor, merely a comment to you about that particular sensor in comparison the the sensor you recommended due to the lack of negative temperatures when connected to an arduino (simply via 5v, gnd and an analog pin) :)

Mowcius