Power Supply Voltage Tolerances and Measurements

Electronics Computer Programming Q&A
User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Power Supply Voltage Tolerances and Measurements

Post by MrAl »

Hello,


Recently i had a problem with the computer where it would shut down suddenly without
any warning. At first it happened once or twice, then it started to happen once a
day, then maybe twice a day, etc., until it started happening almost any time.
It even started to happen when the computer was first powered up in the morning,
and it would start to turn on and then immediately turn right off. Sometimes it would
wait until it started to boot first, then turn off, but other times it would make it
through the boot and then later turn off. I figured that since it could turn off
immeditately after turn on sometimes that it was probably the power supply caps
failing and it was detecting a fault condition and powering down, and other times
just barely passing which allowed boot up. Anyway, i decided to try a new power
supply first and work from there.

I took a trip down to the local computer shop and found a power supply that was
appropriate and the price was $49 USD and they had a nice big case there too for
$59 USD that came with the power supply. I decided to go with the new case because
of the similarity in price to the new power supply plus the old case has a power
switch that was sticky and so getting a new case would fix that problem too.

Ok, so now i am home and have everything changed over like the motherboard and
hard drive. I boot up and notice that the power supply voltage levels are a bit
different than with the old power supply. The older power supply had voltage
levels a little higher except for the Vcore which was a tiny tiny bit lower.
Anyway, my main concern was with the +12v supply which was considerably lower than
with the previous supply. It was down to 11.26v while the old one was 11.71v.
The new PS will also load down a little to as low as 11.14v which is quite a bit
under spec (the spec being 11.40v minimum). The +5v supply is just barely over
the min being 4.78v while the minimum is 4.75, but at least it is within spec.
It's the +12v supply that has me wondering if this supply is good enough for the
system.
I checked the rating of the power supply and the +12v line is rated for 2 amps
above the recommended requirement for the motherboard.

What i didnt know was the spec on the measurement system used to measure the +12v line,
which came with the motherboard system of course so i decided to measure the power
supply lines with my best meter. The measurements were quite surprising:
The +12v line measured +11.82 volts, and the +5v line measured 5.02 volts.
These two measurements basically say that the power supply is operating within spec,
and that the measurement system used by the ASUS motherboard is not as good as it
could be, measuring everything a bit low.
The measurements with the meter were all at the end of an unused hard drive cable
and compared with the ASUS software measurements at the same time.

I thought i would let everyone know about this.

Comments welcome.
LEDs vs Bulbs, LEDs are winning.
User avatar
Janitor Tzap
Posts: 1707
Joined: Sat Aug 12, 2006 5:17 pm
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by Janitor Tzap »

Hello MrAl,

What was the wattage of the original power supply?

Most computers now a days require 350 Watt or higher,
to be able to handle all the power hungry devices that are in them.

Thou, with the newer Video Cards that are coming out.
They require a minimum of a 550 Watt Power Supply for them.

When replacing a ATX Power Supply.
I normally get one with more wattage than the previous one.
That way if I need to add more fans, replace the video card, or CPU that needs more power.
I don't need to replace the power supply as well, at the same time.

As for the 12V and 5V Lines being low.
I believe that 10% below, or above the voltage rating is normal for most ATX Power Supplies.

If you still have the old ATX Power Supply.
I'd see if you can't rebuild it, and keep it as a spare.
{Unless its too small a wattage, or you can get a new one fairly cheap.} :wink:


Signed: Janitor Tzap
reloadron
Posts: 519
Joined: Sat Jun 28, 2008 8:57 am
Location: Cleveland, Ohio
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by reloadron »

My experiences with software that measures motherboard parameters like voltages, temperatures, fan speeds and such is that the software gets its information from hardware monitoring chips like the Winbond Series common to Asus motherboards. Most of these chips have a 10 bit A/D with a Vref of 4.096 Volts offering 4 mv steps of quantization. Voltages like Vcore and the 3.3 volt PSU rails are directly fed in while voltages of 5 Volts and above are divided. The Winbond link above shows some examples. It isn't unusual for inaccuracies to happen. Between components used and the software generally looking at and reading from the SMBus (System management Bus) errors arent all that unusual. I find I place much greater faith in a good DMM to measure PSU voltages.

Most ATX and BTX form factor power supplies for home computers are at best fair quality. To simply state that a PSU is a 350, 500, or 750 Watt PSU really doesn't tell the whole story. The issue is how the power is divided out of the PSU. Years ago PSUs placed large emphasis on the 3.3 Volt rail because it looked like that was the main rail that would power the CPU and other devices of hardware. However, that has changed as now most hardware relies on 12 Volt power. A 350 Watt PSU that supports say 30 Amps of 3.3 volt power only leaves about 250 Watts available for the remaining 5 Volt and 12 Volt rails. Additionally most PSUs won't deliver their rated power on all rails when loaded. Power supplies that deliver more current on the 12 Volt rail(s) are more in line with today's home computer needs. With the exception of some extreme gamming systems with multiple high end graphics cards my experience is that a good brand name PSU rated at 550 Watts is more than adequate. Today's CPUs for example deliver more bang for the buck using less power than those of a few years ago. Seems only the higher end graphics cards are using more power.

Something else to note is a good commercial grade 12 Volt 30 Amp well regulated switching type power supply can run around $300. A typical home computer PSU cost much less. The following is a typical example of what was supposed to be a 600 Watt PSU:

THE DEAD COOLERMASTER is a good example of a cheap relabeled junk PSU.

Most PSUs like that in the link would never deliver their rated power.

Time to get back to work & thanks MrAl for some interesting information.

Ron
Robert Reed
Posts: 2277
Joined: Wed Nov 24, 2004 1:01 am
Location: ASHTABULA,OHIO
Contact:

Re: Power Supply Voltage Tolerances and Measuremen

Post by Robert Reed »

But MrAl, did the new supply cure the computer problem? If so, do you have voltage measurments of the old supply to compare?
User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: Power Supply Voltage Tolerances and Measuremen

Post by MrAl »

Hi again,


Janitor:
The original was 450 watts, the new is 430 watts.
The 'video' is the onboard type, no special card.
Keep in mind that the 12v and 5v lines were 'low' but only as measured
by the ASUS software and the chip they use.
I was thinking of looking at the old power supply, but i took a look
and couldnt find anything by eye. I suppose i could check for excess
ripple because it seems like a capacitor problem. I didnt get around
to doing that and didnt think of it until i got the new case anyway.
Another thing that would be nice is if i could remember how to fire
these ATX supplies (24 pin connector) up without a motherboard, like
with load resistors and the like, so that i could test the ripple and
stuff like that. Do you have this information handy?

Ron:
I checked the chips you mention and apparently they are supposed to have
a 1 percent tolerance on the voltage measurements, but that's only 120mv
out of 12.000 volts, so that would make a 12v supply look like 11.88 and
this software is reading a lot lower than that (11.26v). I dont know why
there would be that much difference, and as i said i used a good meter
and that read 11.81 or 11.82 volts.
Yes i know about the 3.3, 5, and movement toward more heavily loaded 12v
supply so i did check that.
Thanks for your useful info too as i was looking at that chip you mentioned
and got some ideas from that.

Robert:
Apparently it seems to have cured the problem. Before it would run for a few
hours at best before shutting down, and sometimes even sooner, but now it runs
all day long and doesnt shut down, even when i pull the line plug out! (ok
kidding there he he). Seriously, it did seem to cure the problem. I guess more
time will tell for sure.
The old measurements i only have using the software, which read higher than
with the new supply. That means that the old supply must have been running high
beacause all of the readings were within spec. If i remember right the old 12v
line read 11.71 volts, and since then i have found the 12v line to be reading
about 0.5v low (approximately) so add 0.5 to 11.71 and we get 12.2 volts. That's
strange too i guess (12.2 would be read with a good meter, 11.71 with the software).


Thanks for the replys and ideas, and if anyone has a good explanation of how to
run the ATX 24 pin power supplies using load resistors and whatnot that would be
good for me to read too, as i can start testing the old supply without having to
connect it to the motherboard again which i dont want to do anymore.



Thanks again...
Al
LEDs vs Bulbs, LEDs are winning.
reloadron
Posts: 519
Joined: Sat Jun 28, 2008 8:57 am
Location: Cleveland, Ohio
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by reloadron »

Hi Ya

OK, you can run the PSU easy. First I suggest maybe a 10 Ohm 10 Watt load on the 3.3 Volt bus which should be any orange wire in the 24 pin connector to ground. Next to start the PSU just place a jumper between the green wire in the 24 pin connector and common. Regardless if you have a 20 or 24 pin main motherboard connector there should be only one green wire. That is the PWR_ON line.

THIS LINK shows an illustration.

and

THIS LINK illustrates the actual startup of a PSU.

Ron
User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: Power Supply Voltage Tolerances and Measuremen

Post by MrAl »

Thanks much Ron, I'll try this out in a few minutes and get back here.
LEDs vs Bulbs, LEDs are winning.
User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: Power Supply Voltage Tolerances and Measuremen

Post by MrAl »

Hi again,


Ok, tried it.

The first problem i ran into was that i didnt have a matching connector. This doesnt
seem like much, but i also dont have a tool to pull out the pin connectors.
Luckily, i found some small brads (nails) that slide into the connector pins just right.
I was then able to connect some alligator clip leads to the nails and make contact
to the load and jumper.

I loaded the 3.3v supply with 10 ohms as you suggested, and it worked great.
The power supply fired up. The supplies read pretty good, except the +12v read
as high as 12.5v, but i expected it to be a little high without a load.
Next i loaded the +12v supply with 5 ohms, and the power supply would not
start up (nothing else changed). I then took off that load and connected it to
the +5v line instead, and the power supply again started fine.
I loaded the +12 (in addition to the loads on the 3.3 and 5v lines above) with a
10 ohm resistor, and again the power supply would not start up. I figured
maybe the load was too small, so i went to 2 ohms, but the power supply still
did not start up. I went to 100 ohms, and it again started up and the +12v
line read about 12.3v or so.

I really wanted to test the 12v supply with load but it doesnt seem to want to start
with any significant load on the +12v line. In the computer when connected to the
Mother board, it did start up for a while but then later would shut down, unlike what
is happening with the fake loads and the green wire to ground jumper.

Any ideas how to get this fired up, or do you think the power supply is very
defective now, more so than before when it was inside the computer connected
to the MB ?

Thanks.
LEDs vs Bulbs, LEDs are winning.
reloadron
Posts: 519
Joined: Sat Jun 28, 2008 8:57 am
Location: Cleveland, Ohio
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by reloadron »

My guess is that the PSU is pretty much dead for all intensive purposes. Shorting the green (PWR_ON) to common is actually taking a TTL logic high and placing it to a low. That green line normally has 5 Volts SB (Stand By) on it which is always present. The 5 Volt SB actually provided by a small individual always on (when the PSU rear switch is on) 5 Volt power supply. That 5 Volt SB is always present on the motherboard for features like wake on LAN or wake on whatever.

The fact that the PSU won't come up under even light loads on the 12 Volt rail really points to the fact that the PSU is folding over or failing under even a light load. A few years back I made a proto type ATX PSU load bank as a demo for loading PSU rails. The load bank was to see if the units would deliver as advertised. You would be surprised how many wouldn't.

Actually the basic drawing for the less than perfect demo model can be found here. I only loaded a few rails as I was using a data acquisition system that was limited to 4 channels. I later did a much larger version.

I just replaced the PSU in my wife's computer and the symptoms were a little like yours. The system would "try" to start but wouldn't come up. Removed and replaced her PSU and it has run fine the past month. I have a graveyard of dead PSUs growing. I eventually remove the cables and connectors from them. Anything worth keeping. :smile:

Ron
User avatar
Janitor Tzap
Posts: 1707
Joined: Sat Aug 12, 2006 5:17 pm
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by Janitor Tzap »

Well, from your testing.

I would say the caps in the +12V line have failed in the Power Supply.

If the supply is out of warranty, crack it open and start poking around the 12v section.
More than likely it will be a 2200uF or 100uF cap that has popped.

But since you have it opened up, check the other caps as well.
In many cases the other caps will be weak, but are just functioning.
Replace them any way.
I've gotten caught too many times now.
By trying to replace the minimum amount parts on a power supply.
Then having it come back a week later with the other caps blown. :roll:

Ron,
Nice work you did there. :)

I wasn't aware the voltage tolerances had gotten tighter on the ATX & BTX Supplies.
But it does make sense, now that lower voltage chipsets and CPU's are being used
that require tighter voltage regulation.


Signed: Janitor Tzap
User avatar
CeaSaR
Posts: 1949
Joined: Sat Nov 08, 2003 1:01 am
Location: Phoenixville, PA USA
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by CeaSaR »

Hey Ron,

You also have some info about this on your site, Images 6 page, about 3/4 of the way down,
next to the seasonal themes picture and further below the snow covered minivan.

Also, don't forget you PSU pages.

CeaSaR
Hey, what do I know?
reloadron
Posts: 519
Joined: Sat Jun 28, 2008 8:57 am
Location: Cleveland, Ohio
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by reloadron »

CeaSaR wrote:Hey Ron,

You also have some info about this on your site, Images 6 page, about 3/4 of the way down,
next to the seasonal themes picture and further below the snow covered minivan.

Also, don't forget you PSU pages.

CeaSaR
One of these days I really need to organize and clean up that website. :smile: I think it has been about 5 years since the last time. That web is just a collection of literally stuff much with no rhyme or reason. The images pages are pictures I occasionally link to for assorted pictures. I forgot all about the PSU voltage tolerances table in there. I also forgot that when that page loads the video clip of the power sub station going poof automatically runs. That is annoying!

A few years back I was involved with a PSU testing thing so I was doing all sorts of stuff with PSUs. The idea was that if someone is going to test PSUs they should be tested to a specification using a well defined and published procedure and method. I was butt deep in PSUs for awhile. :smile:

@ Janitor Tzap, thanks for suggesting the caps as I have yet to mess with the unit I removed from the wife's machine. I'll focus on them before I think about guttint it for parts.

I do have another site I actually use and have organized to some extent.

Thanks
Ron
User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: Power Supply Voltage Tolerances and Measuremen

Post by MrAl »

Hi again,


As i said though, when the power supply was inside the computer case connected to the mother board
it was working at least to some degree so i assume that the plus 12v line came up at least a little and
was able to take a load, and the software voltage measurements said this too.
So it is strange that now it wont turn on at all. Apparently though there is something wrong so
i will take another look inside and see what i can find.

Nice site there Ron, i didnt realize that was your site at first.
I took a look at some of the pages, and i see that you had a power supply that had flux on the bottom
too. Mine does also i noticed, in one place as did yours. This is indicative of an assembly line repair,
where the initial supply did not test properly so some tech in the plant repaired it before it was shipped
and they dont always remove the flux. Removing the flux is certainly a good idea, but i do have various
devices that work for years and years without removing the flux so i dont know if i would worry too much
there. In all of the plants i have worked in in the past (all in USA) the flux was always removed however.
I dont know what their projected lifetime is for these kind of units and it is probably not too long so they
dont worry as much i bet.
I also see some good test pictures and stuff like that so i guess i asked my questions in the right forum huh?
:smile:

One more thing, if i try to replace the electrolytics, what esr should i use? The reason i ask is because
output cap esr in switching power supplies is often a sort of critical thing, where some power supplies
require very low esr but others require at least some min esr to prevent oscillations.
More on this subject, recently i was working with someone from a university in France and we rediscovered
this when working on a boost converter. The articles on the web often skip over this but when i did the
full time domain analysis it became clear that this is not something to be overlooked. With a very good
cap (little or even no esr) the unit would oscillate, while with some reasonable esr the unit was stable.
This is partly because the esr's eat up some power and that reduces the circuit "Q" which cause oscillations
sometimes. I realize however that the computer power supply is not a boost circuit and probably doesnt
contain that kind of circuit anyway, but i have also read about this for the usual buck converter too.
Maybe use some trial and error....ie solder in a new cap and try it again until it works?
I guess another question would be: has anyone here replaced their power supply caps and fixed the
power supply problem yet? I know it works, but some real life experience would be nice to hear too.
LEDs vs Bulbs, LEDs are winning.
reloadron
Posts: 519
Joined: Sat Jun 28, 2008 8:57 am
Location: Cleveland, Ohio
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by reloadron »

This particular link is one I found a few years ago with an actual schematic of an ATX PSU. Granted it is only a 200 Watt older unit but the overall design hasn't changed much and the theory of operation remains the same. That may help you out a little.

Yeah, the flux issue I got bent over. :smile:

Many of my years have included some special government programs and conforming to certain standards. Even today I work in support of the Navy Nuclear Propulsion Program. There is no room for the bigger the blob the better the job when it comes to soldering. I guess it all sort of spoiled me. Anyway, I hope the link helps a little.

Ron
User avatar
Janitor Tzap
Posts: 1707
Joined: Sat Aug 12, 2006 5:17 pm
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by Janitor Tzap »

MrAl wrote:One more thing, if i try to replace the electrolytics, what esr should i use? The reason i ask is because
output cap esr in switching power supplies is often a sort of critical thing, where some power supplies
require very low esr but others require at least some min esr to prevent oscillations.
More on this subject, recently i was working with someone from a university in France and we rediscovered
this when working on a boost converter. The articles on the web often skip over this but when i did the
full time domain analysis it became clear that this is not something to be overlooked. With a very good
cap (little or even no esr) the unit would oscillate, while with some reasonable esr the unit was stable.
This is partly because the esr's eat up some power and that reduces the circuit "Q" which cause oscillations
sometimes. I realize however that the computer power supply is not a boost circuit and probably doesnt
contain that kind of circuit anyway, but i have also read about this for the usual buck converter too.
Maybe use some trial and error....ie solder in a new cap and try it again until it works?
I guess another question would be: has anyone here replaced their power supply caps and fixed the
power supply problem yet? I know it works, but some real life experience would be nice to hear too.
I have found that you can get away with using electrolytics with little to no esr rating.
But, don't expect the voltage to be all that stable, or oscillation {Ripple} free while it is under load.

Many of the {El Cheapo} ATX & BTX Supplies use cheap electrolytics in them.
They state on them 450 Watt, but in reality are only good for about half that much. :roll:

There was an excellent article that TJ Byers did.
PC Power Supply Repair
Nuts&Volts
September 1996

I don't know if you can still find it in the Nuts&Volts Archives.
But I still have a PDF of the article, if your interested.


Signed: Janitor Tzap
Post Reply

Who is online

Users browsing this forum: No registered users and 8 guests