Buck convertor responds differently on first time power up

[Jezza717]

Hi all,

I am designing a simple buck convertor that takes 24V or 12V and converts it to 5V (technically 5.28V with the the feedback resistors chosen) to power a 5V fan.

I received a few 100 pieces for testing and had some unusual and non-repeatable results.

In ~1/10 convertors on power up, has a voltage spike; see image "Faulty first time power up" where the yellow trace is the 12V input and the blue trace is the 5V output of the board.
However, this only happens once. I have left the boards overnight, discharged the capacitors using some tweezers and can't replicate the issue on the same board. The response after the first time start up is more as expected with a soft-start (see ME3116 datasheet). The image "Normal(or post-first power up) shows the usual response of most boards. This is the same response as the faulty one on the second (or more start up).

Does anyone have any thoughts on what is going on here?

I am currently thinking something physically is changing with the capacitors when charged for the first time...

I have also attached an image of my setup and the circuit of the board below.

[Siwastaja]

Not sure if related to the issue, but massive Cin capable of seriously overshooting the input. If ceramic, add damping, for example a 47uF electrolytic in parallel.

[Jezza717]

Thank you for the suggestion. I have been reading that the ceramic capacitors can cause this, so adding an electrolytic might be useful in parallel.

What I can't figure out now is why it only happens once (on the odd few boards), but isn't repeatable. It only ever happens on first time power up...

[KT88]

Although the ME3116 features soft start, it will draw some inrush current at start-up. It looks like you are using a wall wart as supply - together with the quite small input cap it may be involved in the obseved effect...

Cheers
Andreas

[thinkfat]

Add a small inductor in series to the input (before the input cap, after the diode), 2.2µH or so might be good, also, add an electrolytic cap to ground in parallel to the switchers input capacitor.

[Siwastaja]

Adding series input inductance to already underdamped and clearly overshooting system might not be the best idea... The added electrolytic likely saves the day, though.

[Jezza717]

Thanks for everyone's replies. I agree there is some changes to be made in terms of normal use.

However, I am currently trying to figure out why it only happens on the first time powering up and then never again (as of yet). Because the soft start actually works on the second and future times powering up of the board. So the boards work fine and have no weird spikes (as shown on the second scope picture) majority of the boards first time. Then working fine on all boards after a second power up. Not a single board has caused a fan to blow up after the first time powering up the boards...

Could it having something to do with the IC internally and a default state? I have contacted the manufacturer to see if they have seen this before.

[KT88]

I've never seen something like that with semiconductors - electrlytics however show some initial formation effect of the dielecric after a long period of not being used. Together with the quite small value it might be the root course. Once charged the capacitance would increase and be large enough to operate properly. Try to measure the capacitance before and after the first use. You could also try a smaller cap to see if that exaggerates the effect...

[Jezza717]

Thanks for your insights. I will have to give that a go!

[Siwastaja]

It is possible that a voltage spike exceeding ratings may cause permanent changes on the IC, other than destruction. For example, I have seen a slight overvoltage spike change an UVLO limit of a gate driver from the specified (some 5-6V IIRC) to around 7V but remain otherwise operational.

We don't know the peak voltage because the scope trace is clipped.

[KT88]

yes!

[Jezza717]

hmmm... interesting, perhaps that could be it.

The voltage spike tends to be between 17-26V at the input. I have observed voltage spikes of up to 11V on the output.

[Siwastaja]

Have you verified the EN pin doesn't exceed 5V? The datasheet just "recommends" using 1M series resistor to Vin, but says nothing about conditions when this can be done. If you want to be safe, create a voltage divider there adding another resistor to ground, or clamp it with a TVS.

Just a hunch, could be anything else.