2
\$\begingroup\$

I am designing a heater circuit that will be used in a product to be produced in small to medium quantities (hundreds to thousands of units, not tens of thousands). This heater will be PWM controlled to maintain a set temperature in the heated medium, however as an additional safety measure I would like to include a separate over temperature cut-out to prevent fires or high surface temperatures in case of a fault in the control loop (failed sensor, switching FET shorts out, load is removed from the heater element etc. etc.).

The cut-off should return to an operating state once the over-temperature condition is resolved (ie. normal, trip-once thermal fuses are not appropriate). Ideally the solution would be as simple as possible and contain few parts, the device would be used in remote locations so reliability is a concern.

The heater would be driven from 12V, maximum power would be 40W (ie 3.3A). The cut-off should operate at 50 to 60 degC.

The options I am considering and their relative merits are below:

a. Normally-Closed Bi-metallic Switches, placed in series with the heater element (eg. Klixon 7AM):

  • For:
           + Simple, single component.
           + Seems to be the most common solution for this application.
           + Significantly more expensive than PTC (>$3.00/pc).
  • Against:
           - I have not been able to find any surface mountable options.
           - Has moving parts which raises reliability concerns in my mind although
              they do seem to have quite acceptable lifetime ratings.

b. Positive Temperature Coefficient (PTC) Resistors, placed in series with the heater element:

  • For:
           + Simple, single component.
           + No moving parts.
           + Surface mountable.
           + Cheap (less than $0.40/pc).
  • Against:
           - Generally designed for use in over-current protection and data is given to suit
              eg. available data does not give resistance vs temperature curves.
           - Temperature trip point would be dependant on current through the resistor.
           - Few options available for use over 3A.

c. Temperature Switch IC (eg. Analog ADT650x Series) driving a FET:

  • For:
           + No moving parts.
           + Surface mountable.
  • Against:
           - Multiple parts (reliability).
           - Additional design work (ok, it's negligible but I'm lazy).
           - Requires additional power supply (5v).
           - Total BOM cost is similar to bi-metallic switch.

My questions are:

  1. Any feedback on the merits of these options?
  2. Can anyone suggest alternative options for implementing this cut-off?
  3. Is anyone aware of any PTC resistors designed for this usage or higher current applications?
\$\endgroup\$
  • \$\begingroup\$ Apologies for the slightly funky formatting, I couldn't get the markdown list functionality to properly nest my for/against items. \$\endgroup\$ – Matt B Nov 20 '13 at 9:51
  • \$\begingroup\$ Bi-metallic strips are usually very reliable aren't they? The one in my home thermostat seems to have lasted a long time. What MTBF are you seeking? \$\endgroup\$ – RedGrittyBrick Nov 20 '13 at 9:56
  • \$\begingroup\$ You are probably right (I don't have any experience with them), the bi-metallic strips I was looking at were rated to 10,000 cycles. We're looking for a five year lifetime, so at very worst case with a couple of cycles per day the part should be good enough. The comment was based more on a general suspicion of anything mechanical/moving and may not actually be a valid concern. \$\endgroup\$ – Matt B Nov 20 '13 at 10:21
  • \$\begingroup\$ For a fail-safe my gut feeling is the choose the most simple device, which would be I think a bimetal thermostat. When I worked for a space company these things were used all over the place in satellites, so I think the reliability (at least of the parts they used) can be quite good. But, being space, they were used in groups of 4 (series+parallel) to achieve 1-fail operational. \$\endgroup\$ – Wouter van Ooijen Nov 20 '13 at 10:26
  • 2
    \$\begingroup\$ Bimetal stats crop up in a hell of a lot of mass-produced devices - electric cookers, microwave ovens, kettles, Webasto/Eberspacher heaters, etc. - all safety critical & must pass safety tests / avoid lawsuits, I'd go with the herd there. \$\endgroup\$ – John U Nov 20 '13 at 11:40
1
\$\begingroup\$

bimetal is in my opinion the best for you because when it comes to approvals which can be the most expensive part of the product development cycle you are much better off Now the mechanical life isn't going to be a problem because the bimetals that I use in TO220 packages have lots of hysteresis built into them so they cycle slowly AND the bimetal only operates under abnormal and therefore unlikely conditions THE bimetal is immune to RF fields unlike the chip so it gives less likelihood of having to re lay your PCB I have used and specified thousands of these over about 12 years and have not lost one Also I cant remember replacing one in equipment that I have repaired IF you must use a PTC then beware of them shattering on excessive fault currents I have blown up two series connected 265Vac PTCs on 230Vac but didn't put this in my badbeetles website because at present it only covers ICs

\$\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.