Transient Suppression

[Rabid Badger]

I'm building a spot welder for 18650 cells.  I'm shooting for a (slightly improved) knockoff of this guy's v2.2 design:

http://www.instructables.com/id/DIY-Arduino-Battery-Spot-Welder/

He uses a 25V 470uF capacitor to power the microcontroller during welding pulses, but this limits the pulse to 50ms.  I substituted a couple 16V 1700uF caps to extend the maximum pulse length.

The problem is there is a voltage spike of up to 22V at the beginning and end of each welding pulse.  The original design does nothing to mitigate this, but I would like to try to eliminate it to protect the capacitors and spare the microcontroller the extra stress.

I'm planning on adding a 12V Zener/TVS diode ahead of the capacitors to shunt the transients to ground.  Here are my questions:

1)  Will this work?

2)  Is the pulse short enough that it doesn't matter?

3)  Should I use a normal Zener diode or a TVS diode?  I've read that TVS diodes might have too much impedance for low power applications.

4) Should I take a different approach?

[David Hess]

The 22 volt spike is either a measurement artifact or outside the influence of your added shunt suppressor.  If those big capacitors are not stopping it, then the added zener diode is not going to either.

It would be helpful to see exactly what the 22 volt spike looks like.

How exactly is it being measured?

It might be from magnetic coupling between the high current loop and your measurement configuration.  Minimizing loop area and using maximum separate would help.

[Rabid Badger]

The 22 volt spike is either a measurement artifact or outside the influence of your added shunt suppressor.  If those big capacitors are not stopping it, then the added zener diode is not going to either.

I haven't measured anything, I'm taking the word (and scope images) of the guy that made the original design.

It would be helpful to see exactly what the 22 volt spike looks like.

A screenshot showing the spike along with a circuit diagram of what I have planned is attached to my original post.

How exactly is it being measured?

It looks like he used a DS1054Z.  I'm not sure about the setup.

If I can't get rid of the spike (if it even exists) do you think it will actually damage anything?  Should I just find some 25V caps and let them filter it out for the micro?

[David Hess]

It would be helpful to see exactly what the 22 volt spike looks like.

A screenshot showing the spike along with a circuit diagram of what I have planned is attached to my original post.

I mean at a faster timebase so the rise time and shape can be discerned.

If I can't get rid of the spike (if it even exists) do you think it will actually damage anything?  Should I just find some 25V caps and let them filter it out for the micro?

There is no harm in including a big TVS diode.  TVS diodes are zener diodes characterized for high surge current.  I would place it close to the Arduino.

But note that the microcontroller shares its power ground with its signal ground so unless some type of isolation is used, if that spike actually represents the ground being shifted by 22 volts, there could be other problems.

I am inclined to think that the spike is a measurement artifact and perhaps in the Rigol oscilloscope itself.

[Rabid Badger]

I mean at a faster timebase so the rise time and shape can be discerned.

Ah, gotcha.

There is no harm in including a big TVS diode.  TVS diodes are zener diodes characterized for high surge current.  I would place it close to the Arduino.

But note that the microcontroller shares its power ground with its signal ground so unless some type of isolation is used, if that spike actually represents the ground being shifted by 22 volts, there could be other problems.

I am inclined to think that the spike is a measurement artifact and perhaps in the Rigol oscilloscope itself.

I'll go ahead and throw a TVS diode on the Arduino power pins just to be safe.  I doubt the spike is due to a ground shift, this design has been used by a lot of people and if it had a fault like that someone would have noticed by now.

Thanks for the advice!