1

We have had some issues with resonance overshoot from our 10W wall supply to a products LDO. The theory is that this overshoot will destroy the LDO once in a while rendering the product useless.

While investigating this I examined the difference between having the wall supply plugged in the wall first and then into the product and the other way around.

So when having the wall supply plugged into the product first, and then into the wall I will get this static on the supply input of the product (see first picture). Down to -5 V for a short time . Second picture shows how long before this small static event is before the voltage ramps up to normal on the product.

Is this problematic for an LDO with Absolute maximum input -0.3 - 6 V? Is there any way to mitigate this in the future?

Static event on wall supply plug-in200ms after the static, the voltage ramps to normal on product

enter image description here

EmilBonnevie
  • 440
  • 2
  • 10
  • Gobbledygook alert: *having the wall supply plugged into the product first, and then into the product* – Andy aka Jun 12 '20 at 11:59
  • Thank you Andy aka ;) I even double checked, i guess I ended up creating another error! – EmilBonnevie Jun 12 '20 at 12:03
  • So, does your product connect to anything else, or is it just a plastic box with power supply input? Is the wall adapter grounded? Does it have isolated or earth referenced output? At least if you connect the scope, it is grounded via scope. – Justme Jun 12 '20 at 12:07
  • It is just a plastic box, with some microcontrollers on. Wall adapter is not grounded, oscilloscope is grounded. It is an isolated SMPS wall adapter. – EmilBonnevie Jun 12 '20 at 12:16
  • What was your scope precisely connected to when making the measurements? What nodes relative to what node? I'm also not really sure what the 2nd picture is showing. Is your product earthed or isolated from earth. If earthed, how is it earthed? – Andy aka Jun 12 '20 at 12:35
  • Please show meausrement setup. – winny Jun 12 '20 at 12:40
  • Did you try differential measurement with scope, so that ground leads of the scope are not connected (except together)? – Justme Jun 12 '20 at 12:41
  • I have added a picture of the probing. I did not try diff measurement, do you think it is because of the ground connection on the scope? Second picture shows faintly the static about the 10ms mark from far left, 180 ms later the power supply ramps up the voltage on the product input. – EmilBonnevie Jun 12 '20 at 13:11
  • It might be the scope grounding the DC output, and the stray and EMI capacitances charge via ground lead when plugging it into mains. That's a fast surge. And the ground lead with the alligator clip has stray inductance through which fast surges cause ground bounce. – Justme Jun 12 '20 at 14:50
  • I will try and do the same measurement with the illegal no-grounded scope then. I do not have differential probes available at the moment. – EmilBonnevie Jun 15 '20 at 08:25

1 Answers1

2

The first scope trace could be a measurement artifact:

enter image description here

When you plug in the wall supply, a large inrush current will charge the input caps. There will also be a current spike through the primary-to-secondary capacitance. This could cause some common mode noise which could be displayed by the scope as an actual signal. Also your probe's ground clip may pick up some of th magnetic field due to the capacitor charging current.

If the input supply voltage was probed on a capacitor, especially a ceramic capacitor which has low ESR and inductance, these high frequency +/- 5V oscillations would look suspicious.

Think about the current that would have to flow in a 100nF ceramic cap to get voltage to spike up by 5V in one scope pixel, say 200ns at 2µs/div. That's more than 2 Amps! Pretty unlikely.

I'd recommend not using the scope probe, and soldering a coax to the input ceramic cap instead, with a 50R resistor in series. Also place a ferrite sleeve on the coax, set the scope to 50R input impedance (make sure it can take the DC voltage), and measure again.

bobflux
  • 70,433
  • 3
  • 83
  • 203
  • What initiated my investigation was exactly the large inrush current charging the input capacitors. This is shown in the second picture, with the ramp up to 5 V. But this event happens about 180 ms before, probably exactly as the wall adapter pins hit the wall plug. Would you conclude that this artefact only appears when my scope is measuring? – EmilBonnevie Jun 15 '20 at 08:22
  • 2
    You can stick the scope probe's tip and ground on the same pad, like the output ground. Since you're measuring ground, nothing should show up on the scope. If something shows up then you know it's a measurement artifact. – bobflux Jun 15 '20 at 09:09
  • 2
    Yes thank you! That was exactly what I needed. I put both probe and ground clip on the ground pin of the product and the artefact still appears! Great verifiable solution. – EmilBonnevie Jun 15 '20 at 09:36
  • 1
    Good! So it's either the probe alligator wire acting as a loop antenna and picking up a magnetic field, or common mode current flowing in the probe cable's shield and generating a voltage. You can determine which is which by using the short ground spring provided with the probe instead of the alligator wire, as that does not act like a loop antenna. – bobflux Jun 15 '20 at 11:30
  • 2
    I did a pigtail measurement and now I cannot get a trigger on the event. So it seems that it was due to the alligatorclip. Thanks for all the input :) – EmilBonnevie Jun 15 '20 at 12:40