I am having strange failures in a high-side current measurement circuit that I have not yet understood. The U3 measurement chip is getting damaged and shorts out. Perhaps someone can help identify potential failure points in this type of measurement circuit? Measurement circuit

+VSA is a "safe zone" discrete supply (below) designed to hopefully minimize the chances of this measurement circuit creating a failure point (sigh). It's typically between 3-4V. M3 lets a very small (microseconds) pulse of voltage (up to ~7-12V maybe) through before the gate turns it off... then M3 is turned on in a slow-start (but still fairly quick) manner. Because of this, +VSA is potentially lower than S+ for that very short period (That circuit's input is fed from the same point as the top of M3). I don't see any reason in the datasheet for that to be a problem, though ("Independent supply and input common-mode voltage").

I added D10 and D11 thinking the differential voltage spec of 0.8V was being violated with inrush current... but I've had another failure with even this new design.

It only seems to fail during power cycle events.

The failures seem to be related to parts of the downstream circuitry getting shorted out during/before power on... but in testing only this circuit I can't seem to kill it no matter what I do downstream. It only seems to fail in combination with the rest of the unit. So, I'm wondering if anyone can identify additional failure modes I can look into.

Here's the datasheet for U3: ZXCT1086 datasheet

+VSA circuit: enter image description here

EDIT: Ok, so the +VSA circuit above is perhaps confusing. The +24V node refers to the system voltage input at the top of M3 (well, system input after the fuse, switch, GDTs, and reverse voltage protection). The node after R26 is +24_AUX, which is not the same.

However, I did notice the diodes D10 and D11 are perhaps not clamping to ~400mV like I expected. They're rated for 410mV forward current, but at 1mA. At even 15mA the forward voltage goes up to 1V. I did not expect that, and I can't believe that this far into my circuit designing "career" I haven't heard of how much they vary by current before. I knew they varied a little, but that's way more than a little. I'm adding the datasheet for the diodes as well. It appears I need to re-think the protection the diodes are trying to accomplish.

Another note: In previous testing, I was unable to force the voltage across R26 to go any higher than 100mV.

Diode datasheet: BAS70H,115 datasheet

  • \$\begingroup\$ It only seems to fail during power cycle events. When I measure voltage regulation during power on, I typically get my diff probe between S+ and S- to see if there's any potential noise that occurs. This datasheet indicates the differential voltage should be between 0 to 0.5V. If you find one your oscope that it's very sinusoidal, it could indicate a large amount of noise. I don't think I've ever seen any designs at work that have diodes between your sense lines. I have seen capacitors but not diodes. \$\endgroup\$
    – user103380
    Dec 14, 2019 at 17:44
  • \$\begingroup\$ I didn't realize the diodes probably don't help at all for reasonably high inrush currents. I thought ~400mV Vf diodes would be okay for the 800mV maximum differential... wow, nope. I haven't measured over 100mV between them yet, though. \$\endgroup\$
    – darron
    Dec 14, 2019 at 19:35

1 Answer 1


ABS MAX SPECS: Input current into S+ or S- (†) ±12 mA (-0.3V max)
The differential input voltage limit, V S+ - V S- may be exceeded provided that the input current limit into S+ or S- is not exceeded

Power On surge Current for R14 , I=24V/100 = 240mA into S- , S+ charging up C5.

This appears to be a violation.

  • \$\begingroup\$ Comments are not for extended discussion; this conversation has been moved to chat. Any conclusions reached should be edited back into the question and/or any answer(s). \$\endgroup\$
    – Dave Tweed
    Dec 15, 2019 at 18:08

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.