I'm working on the power supply of a temperature controller which sends about 5kW of power maximum (230VAC 50Hz) into 3 sets of 12 heaters (actually, 3kW + 1kW + 1kW). I'm at the stage where I need to detect if any of those heaters burns out or shorts. I am currently considering current monitoring with a software algorithm to detect that.

  • I have 3 RS232 lines available, and 3 analog inputs (0-10V or 0-20mA) total (unless I add something else but I'd rather not to).
  • That leaves me with 1 current monitor for the whole set, or one per heater set. Hence the specs of the current sensor: >21A full scale or >13A full scale, <100mA accuracy.
  • Non-invasive technique is required (intrinsically isolated). I don't trust chips like the Allegro ACS709 to be completely short-free over the entire lifetime (15years) because it may be isolated, but still within a very small package (but maybe I should?).

Clampmeters are very appealing because they're definitely non-invasive and they generally have good accuracy. I have found only one which has a RS232 interface though, and the datasheet is quite poor.

Any suggestions?

  • \$\begingroup\$ Is it AC or DC? (50/60 Hz?) By non-invasive you mean that a small current sense resistor is not allowed? \$\endgroup\$ Mar 31, 2015 at 18:06
  • \$\begingroup\$ @GeorgeHerold: Thanks, I've updated my post. Current sense is indeed to be avoided, the failure detection should be completely independent (here or not it should not matter, whereas if the current sense breaks the whole chain is down). \$\endgroup\$ Mar 31, 2015 at 19:32

1 Answer 1


When it comes to something that appears to be as critical as this I would still go with an invasive hall-sensor but use two of them per outlet. I would also consider ones with a serial bus capability like the Allegro ACS764.

There is also an intrusive hall sensor with two port RS232 connection which allows for daisy chaining several devices here. Might be worth looking at.

  • \$\begingroup\$ Thanks Andy. So you would personally trust those chips to work without putting the power line at risk for > 10years (I'll still ask Allegro for MTBF if that's the case)? Any particular reason not to go for non intrusive devices? \$\endgroup\$ Mar 31, 2015 at 19:36
  • \$\begingroup\$ I really share your concern and would either have some form of trip mechanism or double on things - if things looked wrong on one of them then flag it up to an operator - do a risk analysis too. \$\endgroup\$
    – Andy aka
    Mar 31, 2015 at 19:38
  • \$\begingroup\$ Apparently the Allegro's are Automotive qualified, so that's actually an excellent idea. The ACS764 is not meant for 230VAC though, I'll probably have to go for the ACS712 and use a RS232 ADC on a PCB but that sounds simple enough. Easier to wrap in a DIN rail module as well, but I'll keep a note of your second suggestion for personal projects. \$\endgroup\$ Apr 1, 2015 at 17:56
  • \$\begingroup\$ It looks like it's going to be 3 boards each with a Atmega328P sampling 8 ACS712 +/-5A and communicating through RS232 via a MAX232. \$\endgroup\$ Apr 1, 2015 at 21:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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