Electronics > Metrology

Measuring Ripple

(1/1)

moosebiscuit:
We have a ton of power supplies we work on in my lab. Typically we get our PKPK reading from our MDO series oscilloscope. I've not been formally trained on output noise and am often left interpreting procedures. So I have a few questions.

1.) With your typical switching power supply, am I better off using a differential preamplifier connected to my scope? Or just use a passive probe? I'm a little confused on the gain settings and haven't used it near enough to be confident in my measurement with it. It's an ADA4000A. I use the TPA-bnc adapter and it auto configures with the oscilloscope. Tips?

2.) Are banana leads sufficient for connecting to a load from your typical output terminals? I don't always have a good power resistor and sometimes have to rely on an electronic load in CR mode. Is this still an effective measurement?

I have seen some of the videos from this website and have learned a lot. But I still have these questions. Thank you for your time!

EC8010:
Crikey, you're asking a ton of questions there! ;D

Realistically, the time domain is quite poor for describing noise/interference. You're better off looking in the frequency domain via the oscilloscope's FFT. But that assumes you know how to drive the FFT so as to get a proper answer, and that's not trivial. Switching supplies are likely to produce interference spikes at set frequencies. Except that doing so makes them fall foul of EMC conducted emissions tests that sweep and dwell on individual frequencies for 3s. So what the IC manufacturers do is to produce variable frequencies (that don't show up in the 3s dwell time). Although it could be argued that it's against the spirit of the legislation, it's perfectly within the stated rules.

A passive probe has the problem that it can't distinguish between common-mode and differential-mode interference. And probe lead earth inductance is always an issue unless you use one of those springy things that came with the probe directly at the tip (and have a relevant OV point for it to contact). Switchers produce lots of common-mode, so you almost always need a differential probe, with a carefully chosen 0V point.

An electronic load is almost certainly bad news. A switcher will generate interference >20kHz. Is an active load via banana leads a pure resistance at frequencies up to (say) 10MHz? Almost certainly not. The ideal load would be purely resistive connected by extremely short leads.The RF people use lots of small film resistors in parallel in a can full of oil as a dummy load, connected via a transmission line.

In the end, it all depends on which frequencies you are concerned about. If the supply is merely charging a battery, then high frequency interference matters not a jot (unless it upsets other nearby equipment). If the supply powers sensitive analogue electronics, then interference matters. But if the interference is out-of-band then it may not. So a noisy switcher might cheerfully power sensitive analogue electronics only operating up to 5kHz if its interference is above that, or it might supply sensitive analogue electronics operating between 100kHz and 200kHz if its interference is below that. Horses for courses.

Conrad Hoffman:
You should read AN-47 by Jim Williams, especially the probe section written by Tektronix. A ton of good info on high-speed probing there.

Navigation

[0] Message Index