0
\$\begingroup\$

I have an input circuit like this:

Input circuit with HBM on left and op-amp on right

The op-amp on the right is a buffer to a PIC24F. The IEC HBM is on the left. Elsewhere the circuit uses SMAJ43CAs, so initially the plan was to protect these inputs (there are four) with them too.

The inputs have to survive a constant 24 VAC as an unintended, but possible, scenario, so the TVS is above that peak voltage. Intended operation is a 0-10 VDC signal.

If I calculate what happens at 12 kV on the HBM capacitor, the current stays below 200 mA (the limit of the BAT54S) at its peak (the initial connection), so it looks like the op-amp is being protected by the large size of the resistor, and the TVS would achieve nothing.

I've found a few answers on SE close to this question, but not one on using resistors alone for ESD. So the question is, what do you think of this circuit? Here are some suggestions:

a. Yes it is OK and commonly done.
b. It is not commonly done but should work.
c. It might work but I would never do it like that (I would use a TVS with the circuit).
d. No one would use a circuit like this if they had any idea what they are doing.

Any advice of what you would do would be most welcome.

\$\endgroup\$
4
  • \$\begingroup\$ @mkeith Correct, but then the resistor needs to be rated for the full ESD voltage, which would be fairly expensive. \$\endgroup\$ Jan 9, 2023 at 21:20
  • \$\begingroup\$ Essentially a duplicate of electronics.stackexchange.com/questions/398533/… , though the present answer appears to be low quality. Better drawings and comments here: electronics.stackexchange.com/questions/500954/… \$\endgroup\$ Jan 9, 2023 at 21:27
  • 1
    \$\begingroup\$ Component identifiers like R1 etc. would help you describe what is on the schematic. \$\endgroup\$
    – Andy aka
    Jan 9, 2023 at 21:33
  • 1
    \$\begingroup\$ @mkeith No, I mean just in regards to the "the resistor definitely provides protection" part -- it does, if it doesn't arc over in the process! Just wanted to clarify that. \$\endgroup\$ Jan 9, 2023 at 21:52

1 Answer 1

0
\$\begingroup\$

You're on the right track, but consider that the 75 kΩ likely sparks over when exposed to such voltages; a small chip resistor certainly* will, a common axial resistor might, and a suitable high-voltage resistor won't.

Putting the TVS out in front guarantees safe voltages for everything after. You don't need to use a massive TVS, if all it needs to handle is ESD; note that SMAJ series parts are rated for surge waveforms too (much longer pulses, microseconds rather than nanoseconds; far more energy!). A smaller TVS, or maybe even a Zener, will suffice (TVSs of this type are naught but zener diodes anyway, just optimized for high surge handling).

Otherwise, assume the 75 kΩ arcs over and therefore is effectively short circuited for high voltages. You still have the BAT54S on the front line. While Schottky diodes aren't great in pulsed operation (relatively high internal resistance), they generally handle ESD alright after all, so this will be fine. A PN junction diode would also be fine (BAV99, pair of 1N4148W, etc.), with just a little higher voltage drop for small currents (but probably less peak voltage drop under ESD conditions, because of the lower internal resistance).

Now, the high clamping voltage under ESD (typically ~30 V, due to a combination of internal resistance and stray inductance) may still be troublesome to the op-amp. You can place a series resistor between clamp diode and op-amp, and the amp's own internal clamp diodes (note: assumption) will handle the remainder. To make 30 V look like 2 kV HBM or less, at least 5 Ω is needed here. The input capacitance is probably low enough that over 10 kΩ could be used without impacting circuit bandwidth, and doesn't increase the noise floor much from as-is. So, any value in this range will do about as well.

This then solves the ESD situation, though you might still want to avoid sparking of the resistor (it might fail open some day), so I'd still prefer the front-line TVS option, but only mildly, for this reason.

Also, you may want to avoid Schottky diodes because of their relatively high leakage current, which compared to the ~20 kΩ divider [Thévenin] resistance, will cause at least a few LSBs of error when at high ambient temperature. It's up to you whether this is a problem or not (acceptable signal error / required accuracy, whether smaller resistors can be selected, etc.).

* I've actually seen chips as small as 0603 rated for 10 kV peak, but only when embedded in potting compound(!).

\$\endgroup\$
3
  • 1
    \$\begingroup\$ Thank you for such a complete answer. I had completely missed that 12kV is 4mm in air, so it can easily jump gaps, and then the outcomes become much more uncertain. It now makes more sense to me why TVSs of this type are relatively large (5mm long or so) and the contacts are some distance apart. I think I will go for the TVS approach after all as it removes this uncertainty. Thank you again for putting so much in your answer. \$\endgroup\$
    – REPuzzle
    Jan 9, 2023 at 22:50
  • 1
    \$\begingroup\$ Well; pin spacing doesn't much matter once it's clamped to <rated voltage>! Just that, to handle the surges they do, rather large chips (some mm across) are needed. 1.5KE and SMCJ families for example can handle a lot of power. For just ESD, a much smaller chip will do, easily fitting in e.g. DO-34, SOD-323, and smaller packages. Something like SMF36CA, PJSD36CW, CDSOT23-T36C, etc. \$\endgroup\$ Jan 9, 2023 at 23:15
  • \$\begingroup\$ That is a valid point,- the high voltage can never build up across the TVS. Thanks for adding that and the part numbers too. \$\endgroup\$
    – REPuzzle
    Jan 10, 2023 at 7:47

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.