Measuring power supply ripple with Siglent 1202X-E

[BillyO]

.
.
E) You gotta use 20 Mhz bandwidth limit
F) You gotta use 1x probe
.
.

If you want to match the fairy tale the manufacturer is telling, that's the way to do it.  If you want to see what is really going on use a 10X low capacitance probe and as much BW as you have available.

I also really hate that they all require additional filtering to be added for the noise/ripple tests.  The noise/ripple figures they provide should be as the power supply is.  If they require that filtering, then they should include it in the power supply.

Seems like it's all a smoke and mirrors fairytale throughout the PS industry.  Why can't they just be honest?

[MarioBros69]

.
.
E) You gotta use 20 Mhz bandwidth limit
F) You gotta use 1x probe
.
.

If you want to match the fairy tale the manufacturer is telling, that's the way to do it.  If you want to see what is really going on use a 10X low capacitance probe and as much BW as you have available.

I also really hate that they all require additional filtering to be added for the noise/ripple tests.  The noise/ripple figures they provide should be as the power supply is.  If they require that filtering, then they should include it in the power supply.

Seems like it's all a smoke and mirrors fairytale throughout the PS industry.  Why can't they just be honest?

My English is not too good..., but I understand that you advise me to use the probe at x10 and not activate 20Mhz?

[BillyO]

My English is not too good..., but I understand that you advise me to use the probe at x10 and not activate 20Mhz?

It is certainly up to you, but you need to decide what it is your after.

If you want to match the manufacturers specifications you will need to use a 1x probe and use the 20MHz BW limit.  Those along with other precautions like finding the and eliminating the source of ambient noise you are experiencing.  Most PS manufacturers also apply additional filter capacitors too, like both a 10uF electrolytic and a .47uF ceramic across the output then attach the probe directly (no ground wire) as close as possible to the output.

However, if you want to see the real noise and ripple you PS is generating use a 10X probe, full BW and do not put on the additional filter capacitors.  You will still need to eliminate any ambient noise and attach the probe directly to the output.

[BillyO]

For this test I'm using a small 5V 3A Mornsun SMPS.

In both cases the supply was loaded to 2.5A

The first two pictures (next post) show the added capacitors and the 1X probe attached within 1cm of the PS terminals

The green traces are with: Added 10uF and 1uF capacitors, 20MHz BW limit, and 1X probe.  The first of these traces shows the ambient noise with this setup.


The blue traces are with: No added capacitors, full 600MHz BW, and a Tek 10X P6205 750MHz 1.5pF probe.  Again, the first of these traces shows the ambient.

Note also the time base and vertical changes between the green traces and the blue traces.

You can see that using all that filtering you really don't see the actual noise being produced by your PS.  The ripple is essentially the same but there is a significant amount of HF noise that you would not otherwise see.  You are figuratively sweeping the dirt under the carpet when using the ancient "industry standard" method.

[BillyO]

Other pictures...

[nctnico]

But how much of that noise comes into your scope through the ground and how much is actually between the +/- of the PSU?

And measuring with a capacitor in parallel is not a bad idea because that is how most power supplies are used in the end.

[BillyO]

But how much of that noise comes into your scope through the ground and how much is actually between the +/- of the PSU?

I have shown ambient traces (everything connected but PS off).  I have also ensured that the ground from the probe to the PS is less than (much less than) 1cm.  The noise being shown is being created by the PS.

[nctnico]

But how much of that noise comes into your scope through the ground and how much is actually between the +/- of the PSU?

I have shown ambient traces (everything connected but PS off).  I have also ensured that the ground from the probe to the PS is less than (much less than) 1cm.  The noise being shown is being created by the PS.

Yes, but is the noise between +/- or ground induced? It is quite possible the PSU is pushing HF current through the ground connection of the probe which then shows up at the input. A differential measurement is much better.

[BillyO]

Yes, but is the noise between +/- or ground induced? It is quite possible the PSU is pushing HF current through the ground connection of the probe which then shows up at the input. A differential measurement is much better.

I don't have a differential probe and one that could catch those spikes would likely cost more than a new car so won't be happening anytime soon..

What I can do is ground the negative on the DC side and see if that makes a difference.

[nctnico]

Yes, but is the noise between +/- or ground induced? It is quite possible the PSU is pushing HF current through the ground connection of the probe which then shows up at the input. A differential measurement is much better.

I don't have a differential probe and one that could catch those spikes would likely cost more than a new car so won't be happening anytime soon..

What I can do is ground the negative on the DC side and see if that makes a difference.

What you can try is reverse +/- and see which part of the noise flips up-side-down and what remains the same.

[BillyO]

What you can try is reverse +/- and see which part of the noise flips up-side-down and what remains the same.

Well, grounding one side makes things significantly worse (800mVp-p).  I guess since the PS is already grounded that just created a big loop antenna.

If I reverse things the ripple may indeed reverse, but the spikes are quite random and roughly symmetrical.  They'll look the same anyway.

It is quite possible the PSU is pushing HF current through the ground connection of the probe

BTW, the PSU is being supplied with a ground.  The only way to deal with this properly would be a differential set-up.  Would two probes and the math "-" function work?

[dophuc]

You can use an isolation transformer to power the PSU, which ensures the PSU is not grounded. And that's how I do it when testing PSU ripple/noise

[tautech]

You can use an isolation transformer to power the PSU, which ensures the PSU is not grounded. And that's how I do it when testing PSU ripple/noise

Any decent PSU output should be isolated from mains Gnd.

[BillyO]

Would two probes and the math "-" function work?

Nope!

[nctnico]

You can use an isolation transformer to power the PSU, which ensures the PSU is not grounded. And that's how I do it when testing PSU ripple/noise

Any decent PSU output should be isolated from mains Gnd.

Define isolated... You'll still have capacitance towards ground and neutral has a reference to ground as well. Together with the X and Y caps, you have all kinds of ground paths together with capacitive coupling to the surroundings.

[tautech]

You can use an isolation transformer to power the PSU, which ensures the PSU is not grounded. And that's how I do it when testing PSU ripple/noise

Any decent PSU output should be isolated from mains Gnd.

Define isolated... You'll still have capacitance towards ground and neutral has a reference to ground as well. Together with the X and Y caps, you have all kinds of ground paths together with capacitive coupling to the surroundings.

Most of that should be on the primary side = isolated from outputs.

[BillyO]

Even if some of this noise is common mode it's still noise being produced by the power supply and in many cases, like where the powered equipment if ground referenced or will be connected to ground referenced equipment, it becomes a factor.  It will even affect circuitry it is attached to through currents created by radiated energy.  I see no reason to ignore it by filtering it out. 

[nctnico]

You can use an isolation transformer to power the PSU, which ensures the PSU is not grounded. And that's how I do it when testing PSU ripple/noise

Any decent PSU output should be isolated from mains Gnd.

Define isolated... You'll still have capacitance towards ground and neutral has a reference to ground as well. Together with the X and Y caps, you have all kinds of ground paths together with capacitive coupling to the surroundings.

Most of that should be on the primary side = isolated from outputs.

It isn't. Especially for switching power supplies / AC-DC converter modules. Due to parasitic capacitances in the transformer these will push HF current into the secondary side. The X and Y caps can short some of this current back to the primary side but without extra filtering (a common mode choke + some capacitors for example), this HF current will flow anywhere where there is a path of low resistance.

[Fungus]

With 20 Mhz it removes some noise but I still can't see the ripple and the strangest thing is that with the source off it shows almost the same

With this oscilloscope and the default probe, should I be able to see it?

You won't see any ripple unless the PSU is under load. More load=more ripple.

[MarioBros69]

Even if some of this noise is common mode it's still noise being produced by the power supply and in many cases, like where the powered equipment if ground referenced or will be connected to ground referenced equipment, it becomes a factor.  It will even affect circuitry it is attached to through currents created by radiated energy.  I see no reason to ignore it by filtering it out. 

Thank you very much for all your posts, they have helped me a lot to understand how difficult it is to measure the ripple of a power supply.

And it leads me to the next reflection, does it make sense to buy an expensive oscilloscope if you are in a noisy environment?

[BillyO]

And it leads me to the next reflection, does it make sense to buy an expensive oscilloscope if you are in a noisy environment?

Sure it does.  Low level measurements are only one of the hundreds or thousands of use cases for a good oscilloscope.

It would be worth while finding out what is casing your noise though.  Then when you do need to make low level measurements you can disable/disconnect it.

My normal environment is quite noisy to,  in order to take the measurements I did today I had to turn quite a few things of around the house.

[tautech]

And it leads me to the next reflection, does it make sense to buy an expensive oscilloscope if you are in a noisy environment?

None at all. What you have now is fully capable for these tasks and better than most and offering 500uV/div sensitivity.

[BillyO]

None at all. What you have now is fully capable for these tasks and better than most and offering 500uV/div sensitivity.

I don't think he's considering upgrading.  I think he means his Sgilent is the "expensive" scope.

[tautech]

None at all. What you have now is fully capable for these tasks and better than most and offering 500uV/div sensitivity.

I don't think he's considering upgrading.  I think he means his Sgilent is the "expensive" scope.

$379 expensive ? 

[BillyO]

None at all. What you have now is fully capable for these tasks and better than most and offering 500uV/div sensitivity.

I don't think he's considering upgrading.  I think he means his Sgilent is the "expensive" scope.

$379 expensive ? 

It's all relative.  Compared to a Hantek or a Fnirsi it is or even to lower end Siglents and Rigols. It's 10 times the price of a DSO154 Pro!!