Power up transient voltage spikes and linear regulators + taming power up spikes

[MarkM]

I'm still testing my custom Atmega328 board.  It's being powered with a 3s lipo and a linear regulator.  There's a power up transient voltage spike of about 25v that was just noticed.  The board was tested with 2x 10uf ceramic caps on the 12v input.  In all of my testing the linear regulator powering the µC hasn't had an issue yet.  It's been tested quite a bit.

I have a couple questions. 

1.  How do you think a linear regulator would hold up to the constant transient spikes from plugging and unplugging battery packs back in over time?  It's an NCP1117 3.3v(SOT223).  I Googled about linear regulators and voltage spikes, but didn't come up with anything.  I know people run arduinos with 3s lipos as well and never heard of any issue.  PS. It doesn't get hot or anything.  Very light load on it.

2. I read this link about power up transients.  Do you think it's worth taming the spike a bit for the regulator?  I might be able to help it with a 100µf cap like they did, but board real estate is lacking.  I might be able to make that work somehow, though. A series resistor might work as well, but I'm not sure what value would be good. I saw someone recommended 1mohm per volt, but other people say "a few ohms."  They also mentioned a MOSFET, which I could probably fit on the board.

What's your preferred way to tame these power on/off spikes?

https://www.pololu.com/docs/0J16/all#3







 

[Simon]

You should have just continued the other thread that way we have the full picture. can you post a schematic of the power section and tell us where the spikes occur.

Is your load switching when you power up ? If there is load interaction causing instability I'd put a diode in series followed by a capacitor to isolate the reg input from the nasties and to break up oscilations on the reg input.

For example i have a mechanical clock with a coil in it to click the clock round, it took up to 2A pulses, so I used a 300mA supply with a big cap to provide the burst current, but I put the control circuit behind a diode and it's own bypass capacitor to keep the voltage up during spike and stop any crap from the clock coil.

[MarkM]

This is a separate issue, but I see what you're saying about continuing the other thread.  I'm hoping someone else could find the thread with a similar question, as I couldn't find much on this topic online at all.  I can delete this and add to the other one if needed.  Just let me know so I can copy the posts.   

There is no load switching when it powers on.  All the regulator is powering is the µC, a little OLED display and a shunt monitor.  The input from the battery was tested without the ceramic caps and there was virtually no spike.  Add the caps back in and there's the huge spike again.  After reading the link and the test, I'm sure the caps are what's causing the spikes.

The linear regulator doesn't seem to mind the spikes.  I'm really wondering how long the linear regulators could handle these spikes for, as they're over the rated max voltage for the regulator.  Are linear regulator max rating more for power dissipation or is there another reason? Maybe it'll be OK for a long time.  I'm not sure.   

[Simon]

Ah, the capacitors, yes, ceramic capacitors have a very low ESR, the cabling and traces to the capacitors have inductance, this forms a "tank" circuit with is an LC circuit, given a pulse an LC circuit will oscillate until the power dissipates, but the capacitors have very low ESR and usually the ESR in the capacitor uses the power in the oscillation up quickly but with no ESR it takes longer for the oscillation to subside.

try a diode into the capacitor close to it or some series resistance which will only help out your voltage regulator in dissipating the power.

[MarkM]

Hi, thanks for the tip.  I'm going to continue with the other thread.  Feel free to delete this one.  You're right, it's the same issue.  Sorry about that.