Measure Ripple of a DC voltage with Scope

[raresvintea]

Hello, i have a debate related to Ripple of a DC voltage. My method is to set the probe to AC and select from the scope measurements Vp-p and zoom in as much as i can an then capture several frames and that's it. I observed today that if i go to us and a 10mV/div i see a value, but if i take a larger time base i get a diffrent value.
What's a rule of thumb or how do you measure?

[pqass]

Use AC coupling.
Use 1x probe setting.
Limit bandwidth to 20MHz
Keep wires short; use the probe ground spring instead of the long ground lead, nor the spring-loaded hook probe cap.

[srb1954]

Use AC coupling.
Use 1x probe setting.
Limit bandwidth to 20MHz
Keep wires short; use the probe ground spring instead of the long ground lead, nor the spring-loaded hook probe cap.

Don't use a 1x probe. This will quite often have a bandwidth of less than 10MHz so you won't see so much of the higher frequency components present in switching supply ripple waveforms. It is best to use a 10x probe to preserve the full bandwidth of the scope and get a more sensitive scope to compensate.

Limiting the scope bandwidth to 20MHz will also hide some of the higher speed transient signals that might be on the output of a switching power supply. The 20MHz limit comes from decades-old standards mainly intended for the evaluation of linear supplies and shouldn't be used with modern switching supplies. However, many manufacturers still use this old standard to make the noise levels of their supplies look more favourable so the only time it is valid to use the 20MHz limit is when comparing compliance of a power supply with the manufacturer's published spec.

The advice on scope probe grounding is good for general circuits but with switching power supplies you may get common-mode currents from the switcher operation traversing through the scope ground lead producing aberrations on the displayed waveform. Better results can be obtained with a wide bandwidth differential probe which will minimise interference from common-mode currents.

[2N3055]

Hello, i have a debate related to Ripple of a DC voltage. My method is to set the probe to AC and select from the scope measurements Vp-p and zoom in as much as i can an then capture several frames and that's it. I observed today that if i go to us and a 10mV/div i see a value, but if i take a larger time base i get a diffrent value.
What's a rule of thumb or how do you measure?

You need to go as low as to capture at least several periods of mains frequency (50/60Hz). Like others said you need to engage 20Mhz frequency limit. That should get you covered.

[ogden]

Don't use a 1x probe. This will quite often have a bandwidth of less than 10MHz so you won't see so much of the higher frequency components present in switching supply ripple waveforms.

Right. Use 50 Ohm input instead. Further reading, few of many:

http://www.how2power.com/pdf_view.php?url=/newsletters/1712/articles/H2PToday1712_design_TexasInstruments_Part%201.pdf

https://www.ti.com/lit/ml/slyy136/slyy136.pdf?ts=1679908498178

[pqass]

Dave gets into it in video #594.
The 20MHz bandwidth limit is so the value is comparable.
Also can use a "poor man's differential probe"; subtracting 1 'scope channel from another channel and not using the ground leads.

[jonpaul]

in decades of SMPS work,

1. LINE trig to see 50/60 and mains harmonics eg 120 Hz ripple

2. INT trig on SMPS switching rate eg 25 kHz....500 kHz to see HF ripple.

3. Take care of return CM noise, and probe DIRECTLY across output bypass ELKO or HF bypass cap.

4. We used 10x probes no issues as the noise level is generally 5..500 mv p-p

5. Instead of 1x, we just take a RG174/U coax and BNC at one end, termination of coax direct to the PSU output via a 100 nF series cap.

6. rare to have a lot of noise > 1 MHz thus probe and scope BW is seldom at issue.

Jon

[metebalci]

I checked the InfiniiVision (6000 series) Output Ripple Analysis, it sets the probe (I have 10x) coupling to AC, bandwidth to 200MHz, trigger to the same channel rising and trigger level to a higher than mean level. I dont mean it is the way it should be done, I wondered how the application sets it up.

[David Hess]

The duration of the record determines the low frequency cutoff, so for instance if you wanted to measure noise down to 0.1 Hz, then a 10 second record is required.

[Vovk_Z]

Don'tforget that typically digital scopes (at least cheap ones) aren't fine for low-ripple or low-noise measurements because of a very low (8-10 bit) ADC resolution. 

[David Hess]

Don'tforget that typically digital scopes (at least cheap ones) aren't fine for low-ripple or low-noise measurements because of a very low (8-10 bit) ADC resolution.

And modern scopes have high flicker noise making low frequency noise measurements questionable.  Old oscilloscopes with JFET inputs are not good for this either, but modern MOSFET inputs are worse.

[jonpaul]

in DC PS PARD measurement, flicker and low F noise is seldom in question

besides the mains frequency and harmonic of the PFC/ rectifiér on the raw bus, it's Line / load variations, référence noise, loop feedback noise gain, etc.

Suggest you  référence, the excellent Hewlett Packard 1980s application note,on power supply measurements and the Tektronix Circuits Concept books, on power supply.

Bon Soirée

Jon

[Kim Christensen]

Hello, i have a debate related to Ripple of a DC voltage. My method is to set the probe to AC

Right. Use 50 Ohm input instead.

Not a good idea to switch your scope to 50ohm input when measuring a DC supply with a 1:1 probe unless the DC voltage is low enough. 

[jonpaul]

a few more notes on probes and scope setup

1. 10x probe scope 1 M Zin SET GRD >>connect probe to DC bus of DUT

Switch to AC coupling (precharge scope AC coupling capacitor)

1X probe :scope input 1 Meg
 Set scope at max V/dig, same sequence: COUP set to  GRD, attach to DC bus
Set coupling to AC

50 Ohm scope : Do not use on PSU. For Zo probe and wide band use only

hope this clarification assists,


Jon 

[David Hess]

50 Ohm scope : Do not use on PSU. For Zo probe and wide band use only

They are rare however some old oscilloscopes had 50 ohm inputs which supported AC coupling and could be used to make this sort of measurement, but a high impedance input is still better.

[ogden]

50 Ohm scope : Do not use on PSU. For Zo probe and wide band use only

They are rare however some old oscilloscopes had 50 ohm inputs which supported AC coupling and could be used to make this sort of measurement, but a high impedance input is still better.

Analog Devices experts strongly disagree. Keyword: "blocking capacitor":

https://www.analog.com/media/en/technical-documentation/application-notes/an101f.pdf
https://www.analog.com/media/en/technical-documentation/application-notes/an-1144.pdf

EMC [conducted emissions] compliance of power supplies usually is measured using instrument having 50 Ohm input as well.

one of many:
https://www.minicircuits.com/WebStore/dashboard.html?model=BLK-89-S%2B

[Andree Henkel]

1:1 Probe with 38MHz bandwidth; from Rohhde & Schwarz, however not cheap:

https://www.batronix.com/versand/messtechnik/tastkoepfe/Rohde-Schwarz-RT-ZP1X.html