Power-OFF Transient....... Do you think this will do any long term damage?

I'm working on a low current 5.4V power supply based on the LM5008. It works fine on power-up and when running, however on power down there is a fairly nasty looking burst that lasts about 1.5ms (see image below).

My concern is that one of the IC's in the circuit is rated 6.5V absolute max for Vcc, and it will see perhaps 50 short duration spikes 100ns long in the 6.5V to 10V range during the 1.5ms power-down period. This is a penny pincher design so I would prefer to not add an additional clamping device if reliability is unlikely to be effected. FWIW, the device will normally run 24/7 not constantly be powered up and down.

Thoughts?

If a clamp should be added, what would be best; Zener, TVS, MOV?

A SMPS shouldn't behave in such a way during shutdown. Rather than simply trying to suppress those spikes with clamps/filters, I think you should determine what's causing them and eliminate them at the source. I'm guessing they're not generated by the LM5008 itself, and are coupled from some other part of the system. They may not actually exist at all on the supply line...

mtwieg said:

A SMPS shouldn't behave in such a way during shutdown. Rather than simply trying to suppress those spikes with clamps/filters, I think you should determine what's causing them and eliminate them at the source. I'm guessing they're not generated by the LM5008 itself, and are coupled from some other part of the system. They may not actually exist at all on the supply line...

Click to expand...

Thanks for the quick response. Looks like the ground bounces at the same time. I connected a second probe to ground and then did a A-B capture and it looks much better.

connect scope probe tip and scope probe gnd clip together, then attach both to the offending rail....see if it still occurs.

Anyway, its noise induced into the scope probe...its too spikey for it to be real.
What size ceramic caps have you got at th output?
Mind you, make sure your current loops are well narrow...and on a proper PCB as you know.
The noise happens as the vin falls and the input current gets more erratic and larger due to less vin

Jester said:

Thoughts?

If a clamp should be added, what would be best; Zener, TVS, MOV?

Click to expand...

The noise pattern at the above picture does not look like the characteristic shape of a damping nor oscillation, but rather like a white noise generated by a high impedance bus. Without seeing your circuit it is a bit difficult to understand the cause, but as treez pointed, I would also consider adding a low value capacitor at the point where you are measuring (if not present).

Jester said:

Thanks for the quick response. Looks like the ground bounces at the same time. I connected a second probe to ground and then did a A-B capture and it looks much better.
View attachment 143535

Click to expand...

Not surprising. Definitely no reason to add clamps to the supply rail then. But this may be a symptom of some other problem lurking in the larger system....

A blind hint: The duration of this artifact, so well delimited in time, as well as the white noise pattern suggest that this is probably happening due to one or more switching elements of the circuit at the dead time when none of them are driving current. As pointed out above, the source of the noise must not necessarily be internal to the circuit, but enhanced by its high impendance, but again, without seeing any scheme it is just a guess.

The burst can be probably identified as arcing time of the mains power switch (or whatever you use to disconnect the power supply). It's almost obvious that the actual DC voltage doesn't change during the burst, it will start to drop later.

This kind of measurement artifacts is an everyday problem in power electronics.

FvM said:

The burst can be probably identified as arcing time of the mains power switch (or whatever you use to disconnect the power supply). It's almost obvious that the actual DC voltage doesn't change during the burst, it will start to drop later.

This kind of measurement artifacts is an everyday problem in power electronics.

Click to expand...

That very thought went through my mind as I repeatedly captured the transient while flipping the switch on the power bar, I was even tempted to try pulling the line cord instead to see if the burst would disappear or look different.

Something to be aware of for the future. Perhaps I should make a "Clean" triac switch for testing power down.