Very worth reiterating to the newbies that the PSU is connected to a bench AC supply and NOT directly to mains. You should not connect the ground clip AT ALL if the power supply is not isolated!
Nice video and another way when working on Government power supplies as I did in the past as was in the test procedures for external noise when using something like the paper-clip method, was to connect a (0.1uf cap) between the probe tip and ground lead. Of course having a high voltage ceramic cap way above what was being measured for test purposes, this worked very well for removing stray & unwanted noise and gave true output noise readings.
That's the nicest way to point out the most hilarious thing about this otherwise informative video. To me it's most obvious what noise is.😂 Get an audio engineer in to fix the audio please.
sir, why 10uF electrolyte and 1uF ceramic capacitor was connected across the output before probing? What are the precautions to be taken for measuring ripple in HV output (1000V, 1mA)
Thanks for asking, Ankit! The 10uF and 0.1uF capacitors are typical for low voltage power supplies and represent the typical bypassing that might be applied by the application to eliminate high frequency noise picked up by output cabling. The datasheet value for ripple typically specifies this external capacitance, which is necessary to apply if you want to compare measurements to the datasheet value. In terms of the precautions you asked about, for high voltage rails you should observe proper high-voltage safety guidelines, including making sure your equipment is properly rated and isolated.
A couple of questions. I extended the test point via a coax cable with a bnc connector and a small capacitor at the end. Is that Ok. Also should the input impedance of the scope be at 50 ohms or 1 mega ohm? I made a 5 volt Flyback smps and loading it with 5 amps. I figured with 50 ohm input you get rid of noise which is not part of the power supply. I noticed that some youtubers used this method. What do you think? thanks
Thanks for the question, Danny! Yes, using a coax or twisted pair of wire is acceptable to extend the test point. Both 1 MΩ or 50 Ω input impedances are acceptable, however with 50 Ω a dc blocking capacitor must be placed between the output measurement and the scope. A 1 MΩ termination will tend to be noisier than a 50Ω, but it is generally not too difficult, with good probing, to get a good measurement on a standard ac/dc power supply. For very low ripple measurements, such as multi-phase point-of -load regulators, the 50 Ω termination may be necessary due to the low signal level. If you would like further information upon this topic, be sure to check out this Analog Devices application note: www.analog.com/media/en/technical-documentation/application-notes/AN-1144.pdf
Are those probes available? Or the schematics or mechanical 3d models, say in Rhino, or SolidWorks? What min bandwidth scope is required? Thank you in advance!
Sanjursan, thank you for asking! The probes we used for this video are the Tektronix TPP0500 (500MhZ, 300V, 3.9pF/10MΩ). Scope bandwidth should be at least 20MHz, and set to a 20MHz limit, if higher.
Hi, this video is quite educational. I have a doubt. Here it looks like you are using a single-ended probe/non-isolated probe. What if you have to use a differential probe and still you need to suppress noise pick up? I have used the Yokogawa differential probe and its loop area is significantly high. At least three times more than the single-ended probe.
Vakacharla, thank you for noticing! Yes, we used single ended non-isolated probes in this video. Differential and/or isolated probes could have been used for this test, but attention should still be paid to the loop area.
Good video. But I do question the relevance of the measurements when a PSU is connected to a circuit (in use). Any wires away from the PSU output are likely to introduce the same issues as the ground clip method. So on the one hand you can validate the spec sheet, but it does little to assist with the actual business of getting the "lab" results in practice.
Not sure I am right, but the noise that would be picked up on the wires between the PSU and the load are met with the low impedance of the PSU and the load so they won't matter much. The oscilloscope is high impedance so that path is disturbed much easier.
He didn't mention it, but note that in the video the wires going to the load are twisted and always kept close to each other. This provides some inherent shielding, and also ensures that any external interference they encounter will be picked up and transmitted equally down both wires (common mode interference) and thus not affect the (differential) signal measurements of the scope. It is only when interference is picked up more by one side than the other (i.e. you have a short probe tip but a long ground clip, so the ground picks up the noise more strongly than the tip does) that it will affect the oscilloscope readings. (Of course you do also want to ensure that your load is sufficiently high-quality that it does not introduce differential noise of its own into the power lines...)
IMHO, electronic load also makes the difference. On my setup, the electronic load will make things more complicated. The waveform is heavily disturbed by it. Finally, I used pure resistor load.
Thanks for your comment, Po-Ting! Electronic loads can emit noise that will get picked up by the scope probes, however this effect should be minimal if the probe loop area is small.
Exactly! A great lookin' RMS spec can hide large peak-to-peak spikey crap that causes havoc - board coupling, PSRR issues, etc. If I can't have both numbers, I'll take Vpp any day. It's nice of Cui to be specific about how they measured Vpp, but many manufacturers leave it vague. I never fully trust these numbers till I've looked at them directly.
@@wormdamage Good point. I realize the point of RMS is to measure power delivered. But ripple isn't about power (since it's riding on DC!), it's about what must be filtered out. Thanks for convincing me that Vpp is the better of the two.
So ... I really object to the 20MHz bandwidth limit. It may have been the bees knees back 55 years ago when 20MHz scopes were the best there was and when 99.997% of all products were immune to noise much over a few MHz. That is not longer the case. Even me as a hobbyist can put together a small computer that is badly effected to power supply noise well in excess of 20MHz. Indeed even 30 year old microprocessor based systems utilized and were sensitive to frequencies in the 100's of MHz. The industry needs to pull up it's sock on this one and get real. The noise on that small PS before you switched in the 20MHz bandwidth limit was unacceptable. Just to be clear, I have no issue with your probing advice and technics, it's only the outdated bandwidth limitations.
@@arkatub It depends on what you are goin g to use the power supply for and what kind pf power supply it is. If it's a linear supply you will be using with audio and simple op-amp circuits, then it's fine. If it's a switching power supply you will be using with a PC, some other fast digital system or HF/UHF radio, then most probably not.
Thank you so much appreciate it if I may critique because I'm a student of videos I watch. I feel the speed that you were set up Your instrument was for people who already Know how to do what you do! Which kind of I think defeats the purpose of your video I couldn't tell which graph is for what? when you were sent up the scope slow down and explain what you are pressing And elaborate if you can. I wish I knew every button that you press and what they are for.
Best video on youtube about these methods. Thanks
Thank you so much!
Very worth reiterating to the newbies that the PSU is connected to a bench AC supply and NOT directly to mains. You should not connect the ground clip AT ALL if the power supply is not isolated!
Nice video and another way when working on Government power supplies as I did in the past as was in the test procedures for external noise when using something like the paper-clip method, was to connect a (0.1uf cap) between the probe tip and ground lead. Of course having a high voltage ceramic cap way above what was being measured for test purposes, this worked very well for removing stray & unwanted noise and gave true output noise readings.
Very good and concise explanation. Thanks
Thank you, Pedro! We are glad to hear that our video was helpful.
Wow, I just learned a lot really quickly! 👍✌️
Love to hear it, Elsif! Glad we could help.
Appreciate the information.... Great job nailing that 80s handycam audio 🤠 really takes us back a ways.
Thank you for the feedback, Robert! We are working on improving our audio for future videos.
That's the nicest way to point out the most hilarious thing about this otherwise informative video. To me it's most obvious what noise is.😂 Get an audio engineer in to fix the audio please.
Video is great Thank you very much indeed.
My newbee experience(s) exactly. Thank you.
Excellent explanation
Thanks, very good explanation 🙂👍
Nice work
Say I had a 20amp regulated power supply, how do you calculate the values of the Electrolytic and ceramic capacitors?
Much appreciated
Excellence suggestions.
that is very ironic having a video about noise and the audio has almost more noise than information.
😂😂😂
It's unfortunate the audio is poor on this video.
Thank you for the feedback Bob! We are working on improving our audio for future videos.
@CUIproducts maybe Bob should put his hearing aids in.
@@andrewverran3498 Maybe you should if you didn't notice the poor (but entirely intelligible) audio.
Nothing that cannot be fixed with AI today
sir, why 10uF electrolyte and 1uF ceramic capacitor was connected across the output before probing? What are the precautions to be taken for measuring ripple in HV output (1000V, 1mA)
Thanks for asking, Ankit! The 10uF and 0.1uF capacitors are typical for low voltage power supplies and represent the typical bypassing that might be applied by the application to eliminate high frequency noise picked up by output cabling. The datasheet value for ripple typically specifies this external capacitance, which is necessary to apply if you want to compare measurements to the datasheet value. In terms of the precautions you asked about, for high voltage rails you should observe proper high-voltage safety guidelines, including making sure your equipment is properly rated and isolated.
@@CUIproducts Thanks for your response.
@@AnkitGargIndia glad we could help!
A couple of questions. I extended the test point via a coax cable with a bnc connector and a small capacitor at the end. Is that Ok. Also should the input impedance of the scope be at 50 ohms or 1 mega ohm? I made a 5 volt Flyback smps and loading it with 5 amps. I figured with 50 ohm input you get rid of noise which is not part of the power supply. I noticed that some youtubers used this method. What do you think? thanks
Thanks for the question, Danny!
Yes, using a coax or twisted pair of wire is acceptable to extend the test point. Both 1 MΩ or 50 Ω input impedances are acceptable, however with 50 Ω a dc blocking capacitor must be placed between the output measurement and the scope. A 1 MΩ termination will tend to be noisier than a 50Ω, but it is generally not too difficult, with good probing, to get a good measurement on a standard ac/dc power supply. For very low ripple measurements, such as multi-phase point-of -load regulators, the 50 Ω termination may be necessary due to the low signal level.
If you would like further information upon this topic, be sure to check out this Analog Devices application note: www.analog.com/media/en/technical-documentation/application-notes/AN-1144.pdf
Some good ripple and noise in that audio lol.
You stole my joke. I just assumed the video was made 10+ years ago.
Good video
Are those probes available? Or the schematics or mechanical 3d models, say in Rhino, or SolidWorks? What min bandwidth scope is required? Thank you in advance!
Sanjursan, thank you for asking! The probes we used for this video are the Tektronix TPP0500 (500MhZ, 300V, 3.9pF/10MΩ). Scope bandwidth should be at least 20MHz, and set to a 20MHz limit, if higher.
Excelente
Hi, this video is quite educational. I have a doubt. Here it looks like you are using a single-ended probe/non-isolated probe. What if you have to use a differential probe and still you need to suppress noise pick up? I have used the Yokogawa differential probe and its loop area is significantly high. At least three times more than the single-ended probe.
Vakacharla, thank you for noticing! Yes, we used single ended non-isolated probes in this video. Differential and/or isolated probes could have been used for this test, but attention should still be paid to the loop area.
Would not using coax suffice. Reworked correctly this is the best way to fully minimize the loop area.
GREAT!!!
👍
Good video. But I do question the relevance of the measurements when a PSU is connected to a circuit (in use). Any wires away from the PSU output are likely to introduce the same issues as the ground clip method. So on the one hand you can validate the spec sheet, but it does little to assist with the actual business of getting the "lab" results in practice.
Not sure I am right, but the noise that would be picked up on the wires between the PSU and the load are met with the low impedance of the PSU and the load so they won't matter much. The oscilloscope is high impedance so that path is disturbed much easier.
He didn't mention it, but note that in the video the wires going to the load are twisted and always kept close to each other. This provides some inherent shielding, and also ensures that any external interference they encounter will be picked up and transmitted equally down both wires (common mode interference) and thus not affect the (differential) signal measurements of the scope.
It is only when interference is picked up more by one side than the other (i.e. you have a short probe tip but a long ground clip, so the ground picks up the noise more strongly than the tip does) that it will affect the oscilloscope readings.
(Of course you do also want to ensure that your load is sufficiently high-quality that it does not introduce differential noise of its own into the power lines...)
IMHO, electronic load also makes the difference. On my setup, the electronic load will make things more complicated. The waveform is heavily disturbed by it. Finally, I used pure resistor load.
Thanks for your comment, Po-Ting! Electronic loads can emit noise that will get picked up by the scope probes, however this effect should be minimal if the probe loop area is small.
A good idea. Also it should be better to measure the noise at the load side, not the PS side (with respect to closed loops).
@@la7yka It's adviced to do measurements for minimized ground loop. Best at the bypass capacitors without ground clips.
There was a lot of "ripple" picked up on the audio recording of this video
After watching the video my question is, What must you do to measure the peak-to-peak Voltage of ripple?
Math !
I don't get it; your measuring voltage with one lead?
Why wouldn't ripple be AC RMS? Peak to peak can be exaggerated if it has small spikes.
Exactly! A great lookin' RMS spec can hide large peak-to-peak spikey crap that causes havoc - board coupling, PSRR issues, etc. If I can't have both numbers, I'll take Vpp any day. It's nice of Cui to be specific about how they measured Vpp, but many manufacturers leave it vague. I never fully trust these numbers till I've looked at them directly.
@@wormdamage Good point. I realize the point of RMS is to measure power delivered. But ripple isn't about power (since it's riding on DC!), it's about what must be filtered out. Thanks for convincing me that Vpp is the better of the two.
Surface power supply can filter out all that switching noise . . . .
U should work on your audio. Sounds like u are in a tornado ^^ Maybe u should fix that first ^^
So ... I really object to the 20MHz bandwidth limit. It may have been the bees knees back 55 years ago when 20MHz scopes were the best there was and when 99.997% of all products were immune to noise much over a few MHz. That is not longer the case. Even me as a hobbyist can put together a small computer that is badly effected to power supply noise well in excess of 20MHz. Indeed even 30 year old microprocessor based systems utilized and were sensitive to frequencies in the 100's of MHz.
The industry needs to pull up it's sock on this one and get real. The noise on that small PS before you switched in the 20MHz bandwidth limit was unacceptable.
Just to be clear, I have no issue with your probing advice and technics, it's only the outdated bandwidth limitations.
I have a cheap 10MHz digital scope, I was thinking of testing a PSU with it, would it not be good enough?
@@arkatub It depends on what you are goin g to use the power supply for and what kind pf power supply it is. If it's a linear supply you will be using with audio and simple op-amp circuits, then it's fine. If it's a switching power supply you will be using with a PC, some other fast digital system or HF/UHF radio, then most probably not.
Youre audio is full of interferece as a bacground noice 🙄 something you maby could do something about. Good luck from Sweden. SB.
Thank you so much appreciate it if I may critique because I'm a student of videos I watch.
I feel the speed that you were set up Your instrument was for people who already Know how to do what you do! Which kind of I think defeats the purpose of your video
I couldn't tell which graph is for what? when you were sent up the scope slow down and explain what you are pressing And elaborate if you can.
I wish I knew every button that you press and what they are for.