Ripple

[bigfishe]

what is the best way to measure ripple on a power suppy?

[luka]

put probe of scope on output, set to AC coupling and some small volt/div setting, probably below 1v/div

[janneman]

If you don't have a scope you can use an AC multimeter, but the indication will only be approximate. Most AC multimeters are calibrated with a sine signal and the ripple will be more saw-tooth in shape so the indication will be off, maybe up to 20% or more.
But you will have a ballpark number.

jan didden

[jitter]

Indeed, most multimeters will only give a correct value for a sinewave. Only True-RMS multimeters will give correct values for other signal shapes as well...

[eem2am]

you may measure output ripple of smps like this (page 32)........

http://www.powerint.com/sites/defaul...les/der260.pdf

[AndrewT]

I find with my cheap (not true RMS) DMM Vac that they can measure AC ripple on a DC supply.
When I compare the rounded sawtooth Vpp from the oscilloscope to the Vac reading from the DMM I see a 3times ratio.

i.e. Vpp oscilloscope ~= 3*DMM Vac

For quick comparisons of ripple I just use the DMM and measure the Vac of the ripple and triple it, to estimate the ripple Vpp.
This works for 50Hz and it's harmonics.
I doubt any sensible Vac reading could be had at >=50kHz.

[tvrgeek]

I use TrueRTA spectrum analyzer so I can see not just the ripple, but the harmonics. It is eye-opening to see the second and third harmonics far worse than the 120Hz fundamental, 4ths and 5ths still in the same range! This way you know what you are working on and know what changes have what effect. You also then can see if you have any 60Hz, which is probably inductive and not actual supply ripple. ( substitute values for 50Hz mains for those of you across the way). I made capacitive coupled probes just for this to MEASURE results of things like fast diodes, different caps etc. Better than believing these threads. Yes, I do see a difference in diodes, placement of suppression caps, series of the caps and physical construction, but I recommend you see for yourself.

Remember, in an unregulated supply, ripple is dependent on the load, so how much ripple you can expect falls under "it depends". This is why in the "old" days we used to put in large shunts in our tube amps to lower the ripple on B+. No EU green standards back then.

[gorgon53]

tvrgeek, are you sure your measured hird and fifth harmonic is coming from the analog powersupply? I would like very much to know about the amplitude-relations and where you thinck they could originate since as fare as I know fullwave rectified ac should only giv dc,second, fourth, a.s.o.
Can you please shed more light on this?

Also, I think what you ment is that additional heavy shunt resistor loads improve regulation (but this will increase ripple in unregulated C-input situation wich I presume you are describing).
All true spectrum analyzer data would be wellcome since I dont have such luxory.

[Elvee]

Here is a simulated example based on a real circuit:

In this case, the average ripple voltage is ~640mV.
An AC-coupled rms-calibrated average-measuring multimeter would read 710.6mV.
In the same conditions, a true rms meter would read 721.2mV.
The corresponding p-to-p voltage is 2.128V.
Last pic also shows the spectrum.

[AndrewT]

Elvee,
can your simulation be modified to show the effect of waveform flattening that occurrs on real mains electricity waveforms?

The flattening is in effect adding on odd harmonics to the fundamental.

I suspect the ratio of even harmonics to odd harmonics in your simulated spectrum will change a lot when the flattened waveform is used as an input.