4
\$\begingroup\$

I am building a circuit to heat a very small area to a relatively high (~100C) temperature. To achieve this, I'm using an SMD resistor controlled by a MOSFET, with an adjacent thermistor for temperature monitoring and control. To achieve faster heating I'm considering running a higher power through the resistors than what they are rated for, at least until they reach the needed temperature. As long as I control the temperature (via PWM control) of the resistors and maintain their temperature within what they are rated for (typically around 125C) are there any issues that would result from doing this?

\$\endgroup\$
1
  • 1
    \$\begingroup\$ Are you sure that the temperature inside them is what they are rated for too? \$\endgroup\$
    – PlasmaHH
    Nov 27 '16 at 22:00
5
\$\begingroup\$

I'd suggest that small SMD resistors would be questionable, but you certainly could use very small power resistors to achieve your goal.

This type of power resistor might suit your need: http://www.vishay.com/docs/51055/d2to20.pdf

You have not been specific about your needs in terms of volume, but resistors like this have an SOA out to 140 degC. It would seem possible to drive this sort of device with a PWM to provide the temperature increase needed.

I've also used two TO-220 power transistors clamped around a crystal to make a crystal oven at 80 degC...worked well.

Depending on your application (and since you are already using a FET to drive the resistors) you could simply use the FET as the heat source. By holding VDD constant for an N-Channel FET at some voltage level you can dissipate all the heat required directly in the FET. All the same rules apply in terms of temperature, but most power FETs have an SOA out to at least 125 degC so 100 degC sounds pratical. You could sense the current through the FET using a small value resistor.

You could even use a linear power regulator in anything from TO-92 to TO-220 or even SMD D2PAK to provide the heat. Just configure in a constant current circuit and modulate it from the feedback pin divider.

\$\endgroup\$
3
\$\begingroup\$

Probably not. However keep in mind that the temperature on the surface of the resistor (or at the sensing resistor) may not be an accurate reflection of the core temperature of the resistor. For large, fast increases in voltage or current the resistor temperature gradient (difference between core inner and outer temperature) may be substantial. I suspect that this gradient wouldn't be relevant for an smd part.

\$\endgroup\$
4
  • 2
    \$\begingroup\$ Actually this gradient can be very significant for typical thin-film SMT resistors. Most are laser-trimmed to achieve the specified tolerance. This results in part of the current path being extra skinny, effectively forming a fuse, and they can certainly fail at moderate overloads even of short duration. OP should use a wire wound resistor (which are available in smallish SMT packages if that's desired) \$\endgroup\$ Nov 28 '16 at 5:42
  • \$\begingroup\$ Good point! I hadn't considered the trimming 'channel'. Wirewound sounds like a winner... \$\endgroup\$
    – BobT
    Nov 28 '16 at 15:28
  • \$\begingroup\$ Does this apply to thick film resistors as well? \$\endgroup\$ Nov 28 '16 at 22:16
  • \$\begingroup\$ Would not be possible to "increase the power of the resistors" for example filling the PCB with thermal pads and vias? \$\endgroup\$
    – ErnstOlch
    Jan 30 '21 at 21:12
2
\$\begingroup\$

Lets talk about time constants of components. Ceramic substrates for resistors include silicon atoms. Silicon cubes, 1cm cube, has thermal tau of 1.14 seconds.

Silicon cubes, 1mm cube, has thermal tau 100X faster, or 11.4 milliSeconds.

Silicon cubes, 100micron cube, has thermal tau again 100X faster, or 114 microseconds.

Point being: your thermistor will not report the correct temperature, because there is substantial "thermal distance" between the heat source and monitor.

You can model the heat flow between the Thermistor and the Heat Source, using Rth of copper foil {1 ounce/foot^2} as 70 degree C per watt per square --- any size square: 1cm, 1 foot, 1 mm.

[ before edit, the numbers were for copper: 0.9sec, 9.0mS, 90uS]

\$\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.