2
\$\begingroup\$

I have a stepper control board with a 12v TVS diode right at the inputs (D11). I added it to the design to protect against static, and I think it's doing that just fine.

enter image description here

The problem is that the diodes often fail shorted and ruin the boards. This happens more often then static damage ever did, so they're actually making the reliability WORSE.

The TVS diode (D11) is a Littelfuse SMAJ12A. D10 is a S5J-E3/57T, which I added for reverse polarity protection. The input is a 12v 5A power brick with overcurrent protection.

I'm not sure exactly what's causing them to fail, but I think it's due to non-ESD events, for example I know that one of the power supplies I have in the shop will produce a 70v + spike for a microsecond or so if the connection to the load is momentarily disconnected, then reconnected (as with a loose connector).

Another common way for the diodes to fail is for a 24v supply to be connected instead of a 12v. Without the TVS diode, this just causes the regulator to run hot, but WITH the TVS diode, the board is fried instantly due to it going into conduction and subsequently failing shorted.

Are there any solid ways to protect the TVS diode? Or at least prevent it from always failing shorted and ruining the board. I've considered added a chip fuse in series with the TVS diode, but I know that will reduce the protection.

Update: I put a scope on the power line with the motors running, it's very clean with no spikes above 13v, so it's not regeneration from the steppers.

Update 2: I tried scoping the power line while toggling the power per Jon's suggestion. There's definitely some noise over 13v, but I don't know if it's enough to be a problem.

enter image description here

Update 3: Solved (probably). It sounds I had the wrong idea when I picked a zener voltage as close as possible to the 12v working voltage. I'll switch to a 30v TVS diode and see if that reduces the failures.

\$\endgroup\$
11
  • \$\begingroup\$ A schematic of the stepper control board would probably be useful. I would guess that when the stepper coils are turned off, the current is recirculated back into the positive supply causing an increase in voltage causing TVS conduction. \$\endgroup\$
    – Tut
    Commented Jan 4, 2017 at 15:44
  • \$\begingroup\$ So you use a 12V working voltage TVS and wonder why it dies on 24V let alone 12V. Note a SMAJ12A has a minimum breakdown of 13.3V. you sure your 12V supply is 12.0V and not something higher \$\endgroup\$
    – user16222
    Commented Jan 4, 2017 at 16:08
  • \$\begingroup\$ Good thought Tut, I'll check that and post back. \$\endgroup\$
    – Drew
    Commented Jan 4, 2017 at 16:26
  • \$\begingroup\$ Update: I checked the power line with a scope, and it's very steady, and not approaching 13v, so I don't think that's it. \$\endgroup\$
    – Drew
    Commented Jan 4, 2017 at 16:38
  • 3
    \$\begingroup\$ Could you post an answer with the solution you found so that the question can have an accepted answer? \$\endgroup\$
    – Kevin Reid
    Commented May 21, 2017 at 0:47

2 Answers 2

Reset to default
4
\$\begingroup\$

I know this question is several months old but I couldn't help but respond.

I had a nearly identical problem with a project at work recently. The TVS diodes originally used on a board were too small (too low of a breakdown voltage) and would fail short-circuit, taking out the entire board. The solution for me involved sourcing a new diode, and there were two parts to this:

  1. Spec a diode with a breakdown voltage as high as possible while maintaining a reasonable clamping voltage (the clamping voltage must be below the absolute maximum ratings of the devices it is protecting). My project ran on a maximum of 15 volts and the chip being protected had an absolute maximum input voltage of 35 volts. I sized the diode to have a breakdown of 18 volts and a clamping voltage of 30 volts. Keep in mind that the breakdown voltage is nominal and the actual voltage may vary, so having it so close to 12 volts could be a problem. I would recommend a diode with a 15V breakdown voltage, but again, make sure the clamping voltage is lower than the absolute maximum voltage of the next device down the line.
  2. Make sure the diode is suitably sized to handle the power properly during a transient event (especially important during nearby lightning strikes but is still applicable for ESD). Instead of using an SMA, try using an SMB or SMC. It looks in your image like you have enough room, so shoot for an SMC.

Something else I notice, you have your reverse polarity protection (D10) going straight between the power and ground. This is very bad form as it will be shorted and damaged in the event that the polarity is reversed. You should consider using a series diode instead, or even better a MOSFET-based reverse polarity protection circuit (these are very easy to implement).

\$\endgroup\$
0
\$\begingroup\$

SMAJ12A has a minimum breakdown of 13.3V. you sure your 12V supply is 12.0V and not something higher, especially during supply transients?

Lightning and ESD events are high pulse power but short duration.

You normally set TVS's as high as you can with regards to the threat that the interface could be exposed to with enough margin around the working voltages, including transients

enter image description here

\$\endgroup\$
1
  • \$\begingroup\$ It's very close to 12v. +-100mv perhaps. I checked with a scope, and during normal operation there are no transients. When I power on the circuit, there ARE transients above 13.3v but they are very short (~1us). I'm not sure if that's causing problems. I tried toggling the power for 10 minutes straight and couldn't cause a failure. \$\endgroup\$
    – Drew
    Commented Jan 4, 2017 at 20:58

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.