5
\$\begingroup\$

We have been designing ESD protection circuits for MCU exposed pins. Here comes the reset pin protection:

  1. The pin is connected with a ESD suppressor at the connector.

  2. The pin is connected in serial with a 600Imp ferrite bead and is parallel with a 0.1uF capacitor.

  3. The pin is pulled up with a 100kOhm resistor.

Here comes what happened when the reset pin is striked by ESD:

A. The system resets at 3.5kV contact.

B. The MCU is killed at 7.5kV contact.

\$\endgroup\$
10
  • \$\begingroup\$ Are these direct ESD hits i.e. is the spark directly contacting the RESET line? Also, do you have spark gaps? And what ESD suppressor is it? \$\endgroup\$ – uint128_t Mar 4 '16 at 15:44
  • \$\begingroup\$ yes, it is contact. I did not test air gap mode. A polymer esd suppressor was used. \$\endgroup\$ – richieqianle Mar 4 '16 at 15:46
  • \$\begingroup\$ Err, so you don't have spark gaps? That's probably the easiest/cheapest thing to add, it's probably worth a shot. \$\endgroup\$ – uint128_t Mar 4 '16 at 15:51
  • \$\begingroup\$ Do you mean the structure on PCB having to sharp points pointing to each? \$\endgroup\$ – richieqianle Mar 4 '16 at 16:00
  • \$\begingroup\$ @richieqianle what sort of case is your equipment in? Metal or plastic? Is it connected to earth/ground? \$\endgroup\$ – Steve G Mar 4 '16 at 16:02
9
\$\begingroup\$

What I think you need to do is make sure that the charge from the ESD pulse does not reach the MCU. You have done all the right things to try and absorb this charge.

However, the charge comes from a capacitor with a very low series resistance so you would need devices with a very low series resistance as well to be able to absorb the transient of the ESD charge. Also these devices need to be fast.

What I would try to do is to prevent the charge to flow into the MCU. So I would add a resistor in series with the MCU's reset pin. The MCU will also have ESD protection inside, this combined with this external series resistor and the components you already have should prevent the ESD pulse from resetting the chip. I would start with a 1 kohm resistor but increase it's value if needed.

As the reset input of your MCU will be high-input impedance the extra resistor will not influence normal performance. But it will increase it's ESD handling capability !

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
6
  • \$\begingroup\$ Beat me by a couple minutes. \$\endgroup\$ – Spehro Pefhany Mar 4 '16 at 16:14
  • 1
    \$\begingroup\$ Typed as fast as I could to beat you :-) \$\endgroup\$ – Bimpelrekkie Mar 4 '16 at 16:15
  • \$\begingroup\$ Thanks for your answer, which is very nice! How shall the resistor be connected with the MCU pin in relation with capacitor and ferrite bead? \$\endgroup\$ – richieqianle Mar 5 '16 at 7:00
  • \$\begingroup\$ Very simple: just leave everything as you have it but only add a resistor in series with the MCU reset pin. That's all. \$\endgroup\$ – Bimpelrekkie Mar 5 '16 at 12:55
  • 1
    \$\begingroup\$ I have seen esd suppressors that take up to 500nsec to fully clamp (a very long time as far as silicon is concerned), so the series resistor also helps damp the leading edge of the pulse until the suppressor fully engages. \$\endgroup\$ – Peter Smith Mar 5 '16 at 17:06
4
\$\begingroup\$

You need some series impedance to limit the the current that gets into the chip. ESD immunity is a matter of shunting away energy and limiting the amount that gets into the sensitive bits.

For example, a TVS to ground followed by a series few-K resistor to limit the current. The TVS might be able to clamp 100-150A to a voltage of around 10V, so your current might be limited to <10mA. The uC would reset but would not come close to being damaged.

For higher potential currents (induced lightning or whatever) you can consider opto-isolation and spark gaps, followed by a TVS or similar.

Of course ESD-related resets are not necessarily related to the reset pin itself, which is rarely brought more than a few mm from the uC in good designs.

\$\endgroup\$
3
  • \$\begingroup\$ Crucial bit: You need some series impedance to limit the the current that gets into the chip If you do that, the chip itself will eliminate anything that is left of the pulse. \$\endgroup\$ – Bimpelrekkie Mar 4 '16 at 16:15
  • \$\begingroup\$ Thanks for your reply. Is it possible to prevent the chip from resetting? Thanks! \$\endgroup\$ – richieqianle Mar 5 '16 at 7:24
  • \$\begingroup\$ Yes, with similar ideas. The details depend on the nature of the disturbance, but a capacitor can work well once the current is limited. As I said though, most often resets come from other than the reset pin and the problem is much deeper. \$\endgroup\$ – Spehro Pefhany Mar 5 '16 at 12:55
2
\$\begingroup\$

Why would you connect a capacitor on parallel with the ferrite bead? You are inviting the ESD jolt to come right into the MCU.

Put the capacitor from the inside end of the ferrite bead to GND.

\$\endgroup\$
1
  • \$\begingroup\$ Sorry I described wrongly. What I did is what you described for the capacitor. \$\endgroup\$ – richieqianle Mar 4 '16 at 16:03

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.