To overload or not to overload?

[Behemot]

Re: To overload or not to overload?

Didn't I mentioned that like ten times already?

[KeriJane]

Re: To overload or not to overload?

Behemot

If you're probing the ATX connector with your scope, there's something you should know.
It's right here:
http://www.jonnyguru.com/forums/showthread.php?t=5626

I figure you probably already know about this but if you don't your scope readings will be very inaccurate.



Удачи!

[Behemot]

Re: To overload or not to overload?

I am not sure what do you see there, but it does not tell anything at all about place where to measure, and as for capacitors, why the hell would I add caps? They will only filter out the ripple and noise, won't they. Everybody is interested in raw data, what goes from the PSU. It has it's own filtration responsible for clear output.

By this logic I could say there is no need for any filtration at all (some manfucturers do actually ) cause every appliance in PC has it's own caps, right? That's not what we want to know. We want to know whether the output from PSU is clean enough even on dummy load with the PSUs own filtration only.

[KeriJane]

Re: To overload or not to overload?

If you read the whole thread, Oklahoma Wolf explains the reasoning behind the caps.

The short version is something like this.
1- The PSU is not engineered to be run without being attached to a Computer Motherboard.
2- The Motherboard is effectively part of the power supply output circuit.
3- The Motherboard needs to be simulated for accurate PSU testing.
4- The purpose of the caps is to provide this simulation.

If you are using a probe, you can wrap the leads from the caps around the probe tip and shield and get the same results.


I tried it with and without caps and they are necessary.
I did this on my 2-channel tester by leaving the caps off of one of the channels. There was a HUGE difference.
https://www.badcaps.net/forum/showth...5212#post65212

Good Luck!

[Behemot]

Re: To overload or not to overload?

ATX spec defines what the PSU should provide. Not that the PSU may be PoS, relying on motherboards filtration taking over PSU's filtration role.

[budm]

Re: To overload or not to overload?

"to simulate system loading"
Intel knows something that we are still learning.
From Intel power supply requirement spec.
"3.2.6. Output Ripple/Noise
The output ripple/noise requirements listed in Table 11. should be met throughout the load
ranges specified in Section 3.2.3 and under all input voltage conditions as specified in
Section 3.1.
Ripple and noise are defined as periodic or random signals over a frequency band of 10 Hz
to 20 MHz. Measurements shall be made with an oscilloscope with 20 MHz bandwidth.
Outputs should be bypassed at the connector with a 0.1 uF ceramic disk capacitor and a
10 uF electrolytic capacitor to simulate system loading. See Figure. "

[mariushm]

Re: To overload or not to overload?

I would add that if you really want accurate measurement, you should drop the ground lead from the probe and connect the probe as close as possible to the measuring point.

Even that tiny piece of ground wire on the probe is enough to cause some inductance or interference with the ripple measurements. It can pick interference like an antenna.

[KeriJane]

Re: To overload or not to overload?

Thank you for that .pdf, Budm!

And thank you for pointing out the relevant section, 3.2.6 so as to save time.

This is exactly it and better explained than I could.

They call for a differential measurement using a 2 or more channel scope as well as using the two caps in parallel

[budm]

Re: To overload or not to overload?

Like I said, never stop learning. It makes me feel I still know so little instead of just brushing it off when someone bring up the information. Thanks to you, keri, I look into it and learn more.
and as mariushm pointed out, how you attach you measuring equipment is VERY important for correct reading also. Ground points are not ground points. 1/2 inch of 30mil copper trace can make a big difference due to inductance, stray capacitance. There are ton of info on proper grounding for PCB and circuit design as a whole system.

[Behemot]

Re: To overload or not to overload?

I would add that if you really want accurate measurement, you should drop the ground lead from the probe and connect the probe as close as possible to the measuring point.

Even that tiny piece of ground wire on the probe is enough to cause some inductance or interference with the ripple measurements. It can pick interference like an antenna.

Do I decode that right as having as short wire as possible from scope to probe?

ADD// there is some bullshit in that white sheet…United Chemi-Con 293D106X0025D2T? That is Vishay SMD tantal cap.

Anyway, usefull material, I may implement some of it…

[budm]

Re: To overload or not to overload?

From TEKTRONIX:
"As a rule, when making any kind of oscilloscope measurement, you should use the shortest possible grounding path.
The ultimate grounding system, is an in-circuit ECB (etched
circuit board) to probe-tip adapter. This is shown in
Figure 4.11. The ECB adaptor allows you to plug the probe
tip directly into a circuit test point, and the outer barrel of the
adaptor makes a direct and short ground contact to the
ground ring at the probe's tip."

[mariushm]

Re: To overload or not to overload?

Do I decode that right as having as short wire as possible from scope to probe?

No, the point is to have as little as possible unprotected wire to the scope.

From the probe to the scope you have basically coax cable, it's shielded from the interferences.

But at the probe, you have that thin wire with a crocodile clip at the end that you're supposed to attach to a ground point. That can often be an additional noise source that can mess up your measurements, especially when the psu has a lot of EMI (bad input filtering etc)

So on most good probes, that wire with crocodile connector CAN be removed from the probe and there's a metallic ring or something on the probe body to which you can attach a thicker wire or hold the probe directly against a metal post or something.

Here, this video explains it shows it in practice better than I could say it:

http://youtu.be/6LEN_HNFK40?t=7m24s

[budm]

Re: To overload or not to overload?

"ADD// there is some bullshit in that white sheet…United Chemi-Con 293D106X0025D2T? That is Vishay SMD tantal cap"
That is not BS.
United Chemi-Con is subsidiary of the Japanese company Nippon Chemi-Con, and moved to the Lansing plant when it bought Sprague Electric Company in 1992.
Sprague Vishay Inc join together later on.
And now you know.
Never stop learning.

[Behemot]

Re: To overload or not to overload?

But at the probe, you have that thin wire with a crocodile clip at the end that you're supposed to attach to a ground point. That can often be an additional noise source that can mess up your measurements, especially when the psu has a lot of EMI (bad input filtering etc)

So on most good probes, that wire with crocodile connector CAN be removed from the probe and there's a metallic ring or something on the probe body to which you can attach a thicker wire or hold the probe directly against a metal post or something.

Here, this video explains it shows it in practice better than I could say it:

http://youtu.be/6LEN_HNFK40?t=7m24s

Yeah, that thing is unattaching all the time here. But I have played with that when doing that crapy PSU's I've tested last time. Being nearby the ATX connector or somewhere else, the noise on scope was pretty much the same all the time.

I think you are exaggerating things here. As I said, last time I played with that for a little while, it was like 3-4 mV noise the whoel time. That is not affecting anything at all. I may start thinking about dealing with that when I will reach like 20 mV on full load with some power supply. But I think that won't happen anytime soon so it would be just waste work.

You have to realize the cable is GROUNDED, that is because the very reason it is MEANT to provide ground. It can work as biggest antena in universe, maybe, but as long as it is grounded, almost all signals will disappear on the ground. Nothing is ideal for sure, it will make some affect, but too small error to affect any measurement.

Sprague Vishay Inc join together later on.

I hope you are not telling me Vishay is under Chemi-Con, because Vishay is crappy brand. True is, I haven't seen them for some time, but if I'll see some bad with recent manufacturing date, I'd have to think something very bad about Chemi-Con!

[budm]

Re: To overload or not to overload?

It bought Sprague from United Chemicon.
"Vishay has grown through acquisitions to include such top names as Dale, Sfernice, Draloric, Sprague, Vitramon, Siliconix, General Semiconductor, BCcomponents, and Beyschlag. Vishay's portfolio of brands represents an unmatched collection of discrete semiconductors and passive components. All of these brands and products are part of one global manufacturer: Vishay."

[Behemot]

Re: To overload or not to overload?

So it's the other way around, I see.

[budm]

Re: To overload or not to overload?

"You have to realize the cable is GROUNDED, that is because the very reason it is MEANT to provide ground. It can work as biggest antena in universe, maybe, but as long as it is grounded, almost all signals will disappear on the ground. Nothing is ideal for sure, it will make some affect, but too small error to affect any measurement."
But which end of the ground on that circuit are you using as your ground ref. One end of copper ground trace on the board or on the you wiring will make difference than another ground especially when you have lots of current flowing through that ground, and we not even talking about ground bounce on other layer of the board or any wire next to it, or ground loop. Grounding is very complicate nature to fully understand, do not under estimate. Your products can be a piece of junk if you do not carefully taking care your grounding when they are used in the real world.

[Behemot]

Re: To overload or not to overload?

Do you think I am totally stupid? Also where do you think I'd connect it to when the ground is from the other side of the loader board? Do you think I am deliberately making it difficult to also make it wrong? Come on man, slight assumption about peoples intelligence sometimes…

I use empty peripheral molex's ground and usually take readings on SATA as there are all three positive voltages in one place. +5 V SB and -12 V is a must to take from Main ATX off course, it is not present anywhere else.

[budm]

Re: To overload or not to overload?

"Do you think I am totally stupid?"
Where in hell did you see that in my statement?

[Behemot]

Re: To overload or not to overload?

When you reffered to using common ground, through which all the power flows, as refference. Only idiot would do that, don't you think?