Tests on an ATX power supply

image of an ATX type power supplyI found some interesting things on the net ... to think about ...
These are tests of ATX-type power supplies in a lab , with all the explanations you need to know better ...
C ' is very informative!

As a personal test, I carried out some visual analyzes at first of my " 400 W MAX " power supply but whose model is specified as follows: ISO-230
As described in their document, I actually have a fuse of 5A in primary, and an electromagnetic interference reduction filter on the printed circuit, + a 0.47µF / 275V ~ condo directly into // on the mains input socket.

Here is the wiring of the distribution plug of the different power supplies:
image of ATX connector wiring
The indicated voltages are as follows:
+3.3V/10A +5V/16A +12V/7A
+5V StandBy /1.5A -5V/0.3A -12V/0.3A

So I performed a simple test with a 10 Ω / 10 W vitrified wound resistor on +5 V, and a dichroic halogen lamp of 12 V / 50 W on the +12V.

I placed my probe on one of the secondary windings of the main transformer, and here is the result:

 oscillogram: Main transformer secondary

This power supply operates at a frequency of 25.543 KHz.
I am somewhat surprised by the area in red ( period during which the current is zero ... ) which has a duration of 9.9 µs !!!
But why this mess?
Funny way to order a switching power supply ... I don't even dare to charge it at 150 W ...
Hmm ... no need, the power supply remains safe with 100 W of load on the +12 V ...

picture of the wiring of my small isolation transformerYes, good .... I probably went a little too hard, it is given for 7 A under 12 V which gives us 84 W maximum ....
We can never do wonders with this type of power supply. .. or else the design of the primary control should be completely revised!
... without me, not want to take it all in the face!

Well, ultimately, what interests me is not the maximum current or power that it is supposed to deliver, but the different voltages available at the same time such as +3.3 V / ±5 V / ±12 V

But the test was to be done, just to know what I can get from it, and to feed models under development, it will after all be quite sufficient.

Finally to finish with this, I was able to measure ( thanks to a small isolation transformer for CH1 ... ) the triggering time of the " Power Good " info , after several tests, it is 11.5 ms with my 50 W / 12 V load ... I remind you that this info is used to indicate to the processor of the motherboard of the computer that there was a power failure, and the normal average duration should be around 16ms before the PC does not go to sleep. If the sector returns before this time limit, nothing will happen, the PC will continue to operate as if nothing had happened.

oscillogram: Info Power GoodI explain :

so, with regard to switching power supplies, there is ONE essential thing to keep in mind, HAZARD !!!
Why ?
Here are some guidelines and advice to follow carefully:

  1. The oscilloscope must always be connected to the earth of the mains network ( if the power outlet allows it of course, but in general, they are all equipped ... )
    This precaution can save your life ... I will explain why ...
  2. In this type of power supply ( see alim_primaire.png), there is no transformer as found in traditional power supplies, which implies that the voltage coming from the sector of your lab, often 230 V, is rectified directly through the sector filters LF1 and LF2. ~ ...
    We are in the presence of part of the rectifier bridge of a rectification of the double half-wave type, which induces the following universal formulas:
  3. if we connect the mass of the oscillator probe simply to the primary mass of the power supply which is NOT connected to the earth, the carcass of the oscillator is found at the potential of the phase ( or of the neutral , it's the lottery ... ) of the sector, and under these conditions, the differential circuit breaker in your home breaks!
What is the role of the isolation transformer? Can't we measure live?

and quite precisely, to isolate the assembly on which you are working from the EDF network!
Your oscilloscope being connected to the ground, you will run absolutely no risk, except if your fingers accidentally touch this rectified voltage which is worth approximately + 325 V == ( measurable with the voltmeter )
Why 325 V and not 650 V?
Because the voltmeter measures the rms voltage, which is nothing other than the rectified value (c/c) divided by 2

In half-wave rectification mode, the positive half-waves are rectified, and the negative ones are ignored, which makes that at the oscillo we would only see one alternation out of 2, and the resulting voltage would therefore be: 230 V (rms) x 1.414 = 325 V (c/c)
In full-wave rectification mode, the negative half-waves are also rectified which then return to the positive level, so that we are with double the rectified voltage ( for the peak-to-peak value ), which effectively gives 650 V (c/c) Also note this : in half-wave mode, the output voltage of the rectifier diodes is at a frequency of 50 Hz , while in double-wave mode, given that the negative half-waves are switched back to positive, we has thus doubled the alternations and we therefore obtain 100 Hz !

Switching power supplies, I know by heart, it was my daily "hobby", since I repaired them, both in primary and secondary, with regulations in primary, secondary, self-oscillating, etc ...

So to come back to my small isolation transformer ( 220 V at the primary, and 2 x 6 V at the secondary ), I was used precisely to isolate my oscilloscope from the sector whose waveform I wanted to view in order to check the value of the "Power-Good" voltage ...

Normally, as soon as you want to measure anything on the primary power supply of any device, you should feed it to the mains through a transformer. 'isolation.
This makes it possible to guard against accidents during sometimes hazardous handling and therefore to be secure, but the problem which appears in this case is that the voltage on the primary circuit that one would like to measure would be "floating" in relation to to the potential fixed by that of the earth. Thus, we could very well want to measure a voltage of +325 V, but unlike a conventional power supply with its transformer before the diode bridge and which serves precisely in this case as an isolation transformer, we end up with a so-called floating voltage, that is to say that compared to the human who is more or less connected to the earth by the feet (...), the magnitude of the voltage to be measured becomes unknown, and it may very well be zero as the value of several thousand volts ... ofwhere a second danger!

About this :

I remind ( or inform depending on the case ... ) that the Neutral and directly connected to the Earth at the output of the distribution transformer of the EDF network.
In your home ( house or apartment, it doesn't matter ... ), there is indeed a difference between Neutral and Earth .
Depending on the distance between your home and the EDF distribution transformer, there may be several tens of kilometers, or even more.
This implies that you can measure a significant potential difference between Neutral and Earth.. Moreover, if you tried to connect a device between Neutral and Earth , the GFCI in your home would trip immediately, instantly depriving you of electrical power!
This is perfectly normal, and precisely aims to protect people on the one hand, but also your installation ( and incidentally that of EDF ... )

Switching power is very special, I could talk to you about it for hours.
If by any chance you wanted to try the experiment and take measurements on this type of power supply, please note that I cannot in any way be held responsible for the damage, both physical and material, that your manipulations could cause.
I have written this article for informational purposes, DO NOT TRY ANYTHING if you are not familiar with this type of diet.


Add comment
  • No comments found

Quelques liens vers les sites de mes amis...

Licence Creative Commons

L'ensemble de ce site hormis une notification spécifique est mis
à  disposition selon les termes de la Licence Creative Commons
Attribution - Pas d'Utilisation Commerciale - Partage dans les Mêmes Conditions
3.0 France.
Licence Creative Commons

Les Fabricants de Circuits Imprimés que j'utilise

Mes fournisseurs favoris

Vérifiez la disponibilité de votre nom de domaine :