Best ripple measuring point and DSO probe setup for ATX PSU ?

[ZoLKoRn]

Hello all, I'm a newbie here and hope is the right section for the thread after several searches but can't find the answer.

I have a project to build a simple electric load for testing PC power supply and will be following intel ATX standard for measuring the ripple but I have a bit concern about point to be should measuring because of intel didn't suggest where is better for a measure just gave the diagram



So I start this thread to need a suggestion where the point is should be correct or better and about the simple circuit to build a junction for connecting the oscilloscope following intel diagram.

From the picture below, intel suggest connecting two capacitors between power and ground line which 10uF/0.1uF and probe from DSO will be connected before 10uF and one after 0.1uF but didn't have any information when we should be connected capacitor to wire cable,
should be close to PSU or close to Load but from I have several searching for, all suggestions close to PSU would be better.



1. Well if put probe close to PSU as possible so where I should be connected capacitor?
2. If I have built a small PCB and using the BNC connector (fig2) for measuring point and put it near load is okay?
3. following fig2 my diagram for connected capacitors is correct?
4. Do I need for 50 Ohms terminator before connecting a probe to DSO because I just using dual prob, not a differential probe?

About DSO, I have watched Dave video how to measure ripple and have some doubts about the method for probe connecting, I saw Dave has put 50 Ohms pass-through adapter before connecting the probe to DSO and I have understood for the reason but...
if following intel standard or ATX, they didn't suggest to using 50 Ohm adapter for cable but HP standard or methodology suggestion in a different way.



HP includes Keysight suggest for put 50 Ohm resistor and 0.01uF capacitor in series before connecting the probe to PSU wire and put 50 Ohms terminator at end of probe cable before connecting to DSO.

1. Which HP method or diagram has shown it means they use the differential probe right?
2. if I have to focus on PC PSU testing should be following intel suggestion?

From all question above, I need to know the way or how to make sure it not wrong and minimum minor error as much as possible if anyone has some more advice which better way or simpler would be great 

Thank you for all answer 

[David Hess]

Forget your little adapter; it will only get you into trouble.  The measurement is differential with a separate chassis ground connection.

The HP example is closer to what I would do however I would probably make the differential probe connections closer to or at the load for convenience.  (1) The HP way with the probe connections as close to the power supply output is the best way to get consistent results for power supply performance.

(1) Thinking about it, what I would actually do for simplicity is make a Molex or SATA to differential probe adapter to connect to one of the unused ATX power supply outputs on a power supply cable which is not shared with any other loads.  This presents complications though if the ATX power supply has multiple separate unshared +12 volt outputs and I do not remember if the ATX specifications discuss this.

[magic]

I don't know, I would think if they ask for those capacitors it means that they want you to measure at the load. What would be the point otherwise (unless the caps are part of the load itself)?
This is also the more realistic and practically relevant measurement if you think about it. Everything connected to those PSUs has 10uF on board or more, usually a lot more. Which is why I think this is what they want you to do.

[BravoV]

As David pointed out you can't use ordinary passive scope probe there, that Intel's document used a differential probe, totally different thing, although the illustration sort of misleading as it looks like using just plain coax cable. 

The devil is in the details, you missed the important tiny footnote down there ... it says ..

Oscilloscope Note:
Use Tektronix TDS460 or equivalent ... and a P6046 probe or equivalent.

The Tektronix P6046 is already very old and discontinued, hence it stated "or equivalent" means newer differential probe also valid as long equal or better than P6046 as its sort of de-facto standard for diff. probes.

A P6046 complete set with it's case looks like this ...

[ZoLKoRn]

@BravoV - Yes, actually I have already checked for the monster on intel footnote but just forgot to mention in my question.

but like I mentioned early I have watched Dave video https://youtu.be/Edel3eduRj4 (during 21.52 min) Dave said if we don't have differential probe we can use a dual normal passive probe instead with no ground connection
and using Math mode on DSO, so if I use this method it will okay or not for following intel setup?

If we can't use a dual passive probe so will better move to HP method and making a custom connector which 50 Ohms/0.01uF series and 50 Ohms terminator before connecting to DSO?

@David - About little adapter, What kind of trouble will have? from I thought it will problem on the ground/earth loop or not? but I think we skip it by didn't connect earth pin on PSU or DSO should be okay.

@magic - I thought the reason they ask for a capacitor because to eliminate some high-frequency signal on the wired or systems maybe.

Thank you for all reply guys 

[David Hess]

As David pointed out you can't use ordinary passive scope probe there, that Intel's document used a differential probe, totally different thing, although the illustration sort of misleading as it looks like using just plain coax cable. 

You could actually make this measurement using plain coaxial cable as shown in the HP example and get pretty good results if the oscilloscope had good variable attenuation adjustments for the vertical channels.  Some DSOs however do variable attenuation on the digital side which is problematic and all but the oldest DSOs will suffer from increased quantization noise after subtraction.  (1) Personally though I prefer using x1 probes because then the oscilloscope's AC coupling function can be used instead of external capacitors and all of the oscilloscope's input protection will be available.

Better these days would be to get a high sensitivity differential probe which supports AC coupling although Pintek has something suitable if a sensitivity of 10mV/div is acceptable.

(1) The oldest DSOs like the Tektronix 22xx series do the vertical channel variable attenuation and the subtraction completely in the analog domain so they do not suffer from increased quantization noise when making two channel differential measurements.  Indeed, for them the differential measurement also removes some of the oscilloscope's front end noise and this can be used to match the sensitivity of the two channels for maximum common mode rejection.  It is a little weird to see it in action because there is an eerie "null" in the trace noise.