question about PC power supplies

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

Threaded View
I have a question about a typical PC power supply.  Let's say that they  
are marked 12V 14A, 5V 20A, and 3V 27A.  I know these voltages and  
currents may not be right (just using as an example), but can the amount  
of current indicated per voltage be used simultaneously (in other words,  
can I use say 12V @ 13A & 5V @ 19A at the same time) or are the maximum  
(or near max) currents available for one voltage only at a time?

Re: question about PC power supplies
On 11/8/2020 4:23 PM, Al5 wrote:


Quoted text here. Click to load it

They are not rated to draw maximum rated current from all the
voltage outputs at the same time.  And ... who knows how
accurate the ratings are in the first place?

Ed

Re: question about PC power supplies

Quoted text here. Click to load it

If the power supply "brick" is fixed voltage (non-adjustable), then
the maximum current specification is at that voltage.  A 12V 10A power
supply will deliver 12V at 10A.  However, you'll need to read the fine
print for how long the power supply can deliver that 120 watts.  It
may have a 20% duty cycle, or perhaps 10 minutes on, 50 minutes off,
or som similar spec.  Unless the power supply has plenty of heat
sinking and a big fan, it's not likely to run continuously at the
maximum rated output.

If there is more than one fixed output (not variable), the power
supply should deliver power on both outputs simultaneously at the
rated load (and duty cycle).  Often, the maximum power output and duty
cycle are rated by the total output, where you can trade power between
outputs, as long as the total output is within the maximum rated
output power.  The limitation is usually not the circuitry, but rather
the cooling system.

Power supplies with adjustable outputs usually include a graph showing
how much output current is considered safe at various voltage output
setting.  Of course, nobody looks at the graph until after they blow
up the power supply.

Good luck and try not to get too close to the maximum ratings.
--  
Jeff Liebermann     snipped-for-privacy@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
We've slightly trimmed the long signature. Click to see the full one.
Re: question about PC power supplies

Quoted text here. Click to load it

For UL or CSA safety rated power supplies, the combined output  
powers shouldn't exceed the continuous output power rating of  
the product, as written on the label carrying the certification  
marking.

The unit is safety tested with any/all combination of continuous  
output channel powers that can occur below the total power rating  
for the unit.

Individual outputs are good for their labelled C x V rating.
If there is a special pulsed limit for an output, it should  
be documented on it's spec sheet, but this need not show up on  
labelling. A minimum load condition may be specified to  
maintain the expected regulated output voltage.

You'll get longer life out of the unit if you keep a derating  
on average loads, keep design air flow levels unrestricted  
and minimize the number of times the input power is cycled.

RL

Re: question about PC power supplies

Quoted text here. Click to load it

Yep.


Yes, but not any/all combination, but rather whatever the manufacturer
declares to be safe.  For example, if the power supply shares the +5V
and +3.3V circuitry, the maximum total power will be whatever the
circuit can handle.  Example of such a nightmare specification:
<
https://hardwareinsights.com/wp/wp-content/uploads/2013/02/PSI-Label.jpg

The +5V and +3.3V share quite a bit of circuitry.  The labels says
that the total simultaneous dissipation for the two sections is 130
watts.  Yet, if I calculate the maximum for each section, I get:
   3.3V * 22A =  72.6W
     5V * 16A =  80.0W
        ==============
        Total = 153W
which is more than the rated 130 watts.  Same with the other sections
on the chart.  If I add together all the outputs:
   3.3V * 22A  =  72.6W
     5V * 16A  =  80.0W
   12V1 * 15A  = 180W
   12V2 * 16A  = 192W
   -12V * 0.8A =   9.6W
   5VSB * 2.5A =  12.5W
        ===============
        Total  = 547W
which is more than the rated total output of 500 watts.

Quoted text here. Click to load it

--  
Jeff Liebermann     snipped-for-privacy@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
We've slightly trimmed the long signature. Click to see the full one.
Re: question about PC power supplies
wrote:

Quoted text here. Click to load it

Where the label specifically gives grouped power ratings,  
they should be followed.

The same goes with any other grouped rating, including  
the rating for the full power supply, 500W.

Users are seldom provided with such detailed ratings  
info on the actual label, suggesting that this unit is  
marketed as a stand-alone product for evaluation/use  
by lower-volume end-user/developers.

Safety testing will have evaluated those ratings as  
listed on the label, under any worse-case shift of load.
It can be time consuming to do so, but may reduce the  
actual part cost to the mfr, in the long run.

Too much information often indicates a not-very-rugged  
design, but in cost-critical markets, this is seldom  
a deciding factor.

RL

Re: question about PC power supplies

Quoted text here. Click to load it

There doesn't seem to be much of a defacto standard in computah power
supply labeling or specs.  I picked a suitable label from what Google
Image Search found:
<https://www.google.com/search?q=atx+power+supply+label&tbm=isch
Looking at the various labels, some are overly complexicated, like the
one I offered as an example, while others are sparse.  The sparsely
labeled power supplies are a problem because they will give the
voltage and currents, the total power rating, and nothing else.  As I
demonstrated, just adding up the power for each section results in a
total power that is far greater than the total power rating.  This
leaves the decision for how much power to draw from each section up to
the user, with not guidance from the ratings.  That's probably not a
problem for PC applications.  However, users that like to convert ATX
power supplies into bench power supplies may have a problem.  To make
things worse, few of the specs and none of the labels bother to
mention the rated duty cycle and operating temperature limits.

--  
Jeff Liebermann     snipped-for-privacy@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
We've slightly trimmed the long signature. Click to see the full one.
Re: question about PC power supplies
wrote:

Quoted text here. Click to load it

Duty cycle or temperature limits are usually present on the device  
spec sheet - for commercial product the defacto temp before  
deratimgs is 40C. Duty continuous unless stated otherwise.

As previously mentioned, safety certified units are stressed  
with any possible combination of outputs that do not exceed the  
combined power rating of the unit. For a any test not specifically  
aimed at the effect of output ratings, the outputs will be  
loaded to a derating of Prated / Psum, on each output, as a  
defacto standard 'rated' load.

ATX power supplies are designed to meet the ATX application, which  
requires only the minimum labeling requirements for certification  
and is often a captive or single supply chain spec as far as  
outputs, total ratings and environment/mtbf are concerned.

Refer to the Intel ATX developers guidelines for the copyright  
owner's standard spec. This is largely a mechanical and  
interconnection requirement, not an electrical one..

These products are normally limited by the temperature index of  
their isolation components, fan life and electrolytic cap life.
None of those items are on the label, but will be on the mfr's  
spec sheet. Only the isolation components are safety evaluated.
Anything else is subject to single-fault abnormal testing alone.

The only outputs normally expected to handle pulsed loads are  
those powering motors - fan and disc drive loads typically on  
the +12.

RL

Site Timeline