0
\$\begingroup\$

I'm looking into creating a solution to monitor the temperature of a chip which has no temperature sensing available for me to use. I would like to use some sort of simple through-hole temperature monitoring component, yet I am unsure of what exactly I need to look for. It needs to be a cheap solution (under $5) and able to stand temperatures from at least 50-200 Fahrenheit. I have a few questions about this that I feel would be helpful in speccing out a component:

  • What are the different types of temperature sensors available?
  • What are ones that are low-cost and can measure the temperature of a flat surface (chip, simplistic heatsink, etc)? Would something like the TMP36 work?
  • Is there any go-to solution for monitoring the temperature of a chip without any accessible temperature detection?
\$\endgroup\$
  • \$\begingroup\$ I'm presuming the sensor would be mounted on the circuit board (hence the through-hole statement). If so, then I also presume that you do not wish the "sensor" to influence the temperature of what you wish to measure i.e. it's thermal inertia is small. Based on what I've said I'd go for a thermocouple but any sensor may be ok if accuracy requirements you want are not stringent. \$\endgroup\$ – Andy aka Aug 24 '13 at 11:27
  • \$\begingroup\$ The go-to solution is to use a PN junction on the chip as a temperature sensor. If you can't do that then you aren't measuring the temperature of the silicon integrated circuit but rather the temperature at the outside surface of the package which is generally lower. In that case you would need a good mechanical/thermal connection to the "chip" and that implies some conductive epoxy or mounting hardware. What are you trying to accomplish? \$\endgroup\$ – Joe Hass Aug 24 '13 at 12:02
  • \$\begingroup\$ @JoeHass I'm overclocking a chip on a closed system - that is, I have no access to the functionality of the chip, yet I can change the clock speed by pulling a couple pins high or low. I'm looking to monitor the temperature of the chip as the clock rate changes. \$\endgroup\$ – TreyK Aug 24 '13 at 18:41
  • \$\begingroup\$ You can't monitor the junction temperature of the chip (the temperature that matters) without having a sensor on the chip itself. If you know the thermal resistance of the package and the power consumption of the chip you could estimate the die temperature from package temperature, but without that information you are just guessing. \$\endgroup\$ – Joe Hass Aug 24 '13 at 19:05
  • \$\begingroup\$ @JoeHass got it, that makes sense. I would accept that as an answer if you put it as such. I would also appreciate being pointed toward any attempts at implementing the method you mentioned if you know of any. \$\endgroup\$ – TreyK Aug 24 '13 at 23:35
1
\$\begingroup\$

If you really want to know the temperature of the chip itself, the go-to method is to use a PN diode junction on that chip as a built-in temperature sensor. These temperature sensors have been built in to Intel i86 processors since at least the Pentium III days. The temperature measurement is done by forcing two different currents through the diode and measuring the two forward bias diode voltages. With a little math, a robust temperature measurement can be made. There are many temperature sensor chips that can use such a "remote sensor". Here is a nice explanation of how they work from Analog Devices: Measuring temperatures on computer chips with speed and accuracy-a new approach using silicon sensors and off-chip processing

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.