How to Measure Ripple Voltage on a Switch-Mode Power Supply - Workbench Wednesdays 51

I have indeed already done that, although this function on my oscilloscope is very limited. can't even zoom in. If I change the scale, it only changes the current wave and does not change the reference. I don't think that's useful.

But with a longer ground wire you can indeed see more over and undershoot and much more ringing. That makes the measurements a bit unreliable in my opinion. The shorter the ground wire, the less over and undershoot and ringing. If I use an even shorter ground wire, there might be no ringing at all.

But what continues to surprise me the most is that with you, on your video, I see no over and or undershoot and no ringing at all. You even indicated that you used the longer ground wire. Such results are almost impossible, right?

Did you make measurements with the ground clip? It'd be interesting to compare them. (And if you want to get fancy, it's a good chance to learn how to use the Reference functions on your scope...)

Yes all my measurements were with the short ground spring and was therefore surprised at the over and under shoot I saw with a commercial power bank and on your video I didn't see that overshoot at all.

Thanks again for your detailed response and help. Learned something again. It's really great that you respond so quickly and want to be helpful.

The over/under-shoot is caused for the same reason as the ringing.

Unfortunately, you cannot entirely isolate the effect between the probe and the circuit. That's a fundamental truth of all measurements. Your probe becomes part of the circuit. The inductive loading of the ground connection and the capacitive loading of the probe tip (well, its total impedance) are now part of the node you are measuring.

In general, your only practical options are to reduce the probe's loading or use a lower-loading probe. On a passive probe, the only practical way is to use a short probe tip and the ground spring collar. Then minimize the signal and ground measurement points on your device-under-test. (If you make that loop much tighter and the frequency of the ringing doesn't change, that is an indicator it is real--to me, anyway.)

Using a lower-impedance probe is expensive. You'd really have to go to an active probe (which your scope may not even support or offer.)

And, as I said, if you measure around the inductor with a near-field probe and see that same ringing, it is usually a good indicator that it is the circuit. It means there is enough energy in that frequency band to radiate.

Not sure if I understand everything you're saying. I must also admit that I have only been in the electronics hobby for 2 years.

But the rising edge is about 2.2ns. According to your formula you just gave, that would work out at 159.09MHZ.

My oscilloscope is a siglent SDS1202X-E 200 MHZ, so that shouldn't be a problem right?

By the way, my self-made power bank has a rising edge of 1.3ns. If I understand you correctly, A 200mhz oscilloscope would not be enough to measure this correctly? I do indeed see less ringing there so it must be the bandwidth limitation.

But apart from the ringing. What about that overshoot and undershoot that the oscilloscope shows? Is that overshoot and undershoot really there or is this another limitation of the oscilloscope and/or probe?

I'll check out your video, If I haven't already :-)

By the way, I now see the following in the datasheet of the siglent SDS1202X-E: Overshoot (500 ps Pulse) <10% I don't know exactly what this means yet but this might explain the over and under shoot with my oscilloscope.

I should have added. A quick test is to zoom your time base on the edge of the switching signal. The idea is that you want to make sure the oscilloscope is sampling as many points as it can on the edge.

Then turn on the rise time measurement.

If that rise time is nearly the same as the specified rise time for your oscilloscope's bandwidth (plus probe), then you are seeing band limiting effects. And on digital oscilloscopes when you slam the front end with an edge that is too fast, it causes ringing to show up in the digital filters.

So, measure the rise time of that edge and see if it matches the scope's datasheet.

A shortcut is to estimate the bandwidth of the signal as 0.35 / rise_time. (This is a first order approximation).

(You can watch an longer explanation and example of the calculation here.)

This test does not help tell if your probe is introducing the ringing, but it helps point to whether the scope can accurately reproduce what it is seeing.

I still have a lot to learn I notice. And I need a sponsor apparently :-)

Thanks again for your help and quick responses!

Keep in mind that this is not just a "frequency" effect. It's the energy in the rise-time, not the repetition rate of the switching signal.  Nice sharp edges contain many frequency components. (Same reason why the clock or data rate isn't enough to determine the bandwidth needed for an oscilloscope. The rise/fall time of the edges has much high-frequency content compared to either.)

You need a generator with an adjustable rise time (or slew rate) for a comparable test. This ability is usually on high-end AWGs. And even then, once you enable the adjustable rise time, the fastest becomes relatively slow.

Again, a simple near-field probe will prove whether the noise is the circuit or measurement system. If the near-field probe sees it, then something is causing an emission.

But I still don't see an undershoot on your last image.

I will do some more testing and hopefully develop myself.

my next purchase should be a function generator to see and test whether my oscilloscope still has the overshoot and undershoot on the same frequency.

In any case, many thanks for your quick and extensive responses!!

The lack of ringing on the falling edge suggests to me it is the circuit and not your measurement setup.

(I don't remember if I show this in the video.)

Here's the same circuit made with 6mil traces and no ground plane. The probing setup is the same. Here you can clearly see the switching node ringing which causes the output to ring as well.

A near-field probe would tell you for sure since that ringing would show up as emitted EMI (like the purple trace in my captures.)