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:

Re: Power Supply Voltage Tolerances and Measuremen

Post by MrAl » Fri Oct 02, 2009 4:05 pm

Hi again,


Ron:
Thanks for the link. I'll take a look.
Oh yeah, this power supply has both 12v V1 and V2 (or whatever they are) running to the same pad on the
pc board too. I think they have to derate the two lines because of the connector type and wire size.
A power supply that puts out 40 amps with a connector that was only rated for 20 amps would have
to use as least four pins: two for ground and two for the supply voltage, and rate those two for
20 amps each. The connectors have an ampere rating per pin so that has to be respected. I think
that's what it is anyway.

Janitor:
Ok that's good to know. Yes, i would like to take a look at that pdf as long as i am doing this.


I pulled the three caps for the plus supplies 12v, 5v, and 3.3v, and they all look physically ok except
the 5 and 3.3v caps look like they are bulging a little at the top, but only at the top, and that could
be the way they were made, it's hard to tell. Nothing leaking from them though. I might test them
tomorrow.

I sourced some caps at Digikey and it looks like it could cost some bucks for these caps. If i get the
better ones that are about $3.80 each i would have to buy 5 cause that's the min quantity, so it
could end up running me over 20 dollars just to fix this thing. Is it worth it?
LEDs vs Bulbs, LEDs are winning.

User avatar
Janitor Tzap
Posts: 1436
Joined: Sat Aug 12, 2006 5:17 pm
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by Janitor Tzap » Fri Oct 02, 2009 8:13 pm

MrAl wrote:Janitor:
Ok that's good to know. Yes, i would like to take a look at that pdf as long as i am doing this.
Ok,
I'll try to leave it in your PM box.
I pulled the three caps for the plus supplies 12v, 5v, and 3.3v, and they all look physically ok except
the 5 and 3.3v caps look like they are bulging a little at the top, but only at the top, and that could
be the way they were made, it's hard to tell. Nothing leaking from them though. I might test them
tomorrow.

I sourced some caps at Digikey and it looks like it could cost some bucks for these caps. If i get the
better ones that are about $3.80 each i would have to buy 5 cause that's the min quantity, so it
could end up running me over 20 dollars just to fix this thing. Is it worth it?
If the cap is bulging at the top.
This is a sign of heating inside the cap.
You put it on a hand held cap tester, it may check ok.
But if you have a better cap tester, like a Z-Meter or such that puts a load current test on it.
You'll see it is falling while under load.
{A good ESR meter will show this too.}
Also...
Check the bottom of the caps.
These are rubber type plugs with the leads sticking out.
They may be cracked, or due to the internal heating.
The holes that the leads stick out have melted, and oil has started leaking out.
{You can sometimes see this film that is left on the PC board from the cap.}

Is it worth repairing?
That's kind of tough to answer.
Normally this comes down to what you originally paid for the Power Supply.
If you paid $40 or more for it.
Then I'd say yes, it would be worth repairing it.
{If this was a Dell, Mac, or an HP type of power supply.}
{It would be cheaper to repair than to replace it.}
{Because they're computers are proprietary in there parts.}
{It's not uncommon to pay $80 or more for one of their replacement power supplies.}


Signed: Janitor Tzap

User avatar
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 » Sat Oct 03, 2009 2:01 am

I have several PSU boards around here including the one from the Dead Cooler Master and if you want one for the caps you are welcome to it. The Cooler Master supply died on the main transformer primary side so the rectifiers and caps are likely all good. Never tested them. However, if you want the board it's yours and I can just mail it to you in a padded envelope.

As to the rails and distribution of the 12 Volt power this is a good read to a point. There is actually an ATX 12 Volt PSU design guide published by Intel which can be found here that covers the multiple rail thing a little. Pretty much runs with what you mention MrAl.

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 » Sat Oct 03, 2009 6:10 am

Hi again,


Ron, thanks for the links. Some very interesting reads there.

Janitor, you hit the nail on the head with the cap thing...
The two caps that were bulging measured pretty bad, here are the findings:

The cap for the +12v supply was pretty much ok, and was not bulging at the top.
The capacitance seemed close to the package marking of 2200uf, and the ESR looked to be in the low
tens of milliohms (0.010 to maybe 0.030 ohms).

The TWO caps, one for +5v and one for +3.3v BOTH looked bad. They had bulging tops but more
importantly the ESR was around 0.5 ohms for both of them and even more important than that was
that their capacitance measured between 10uf and 200uf, depending on current level (as Janitor
remarked) and this is very bad for both caps with package markings of 2200uf each.
When testing these two it is interesting that the two cap meters i have both read over range because
they each only go up to 200uf, and the two caps are 2200uf, but that's probably because the meters
use a very small current to measure capacitance. Once i beef up the test setup by using a wave generator
and resistor and scope and pump up the current to as little as 12ma peak to peak (yes that is 0.012 amps,
not 12 amps nor 120ma) the capacitance goes way down, like to 10uf or maybe even a little lower. With
a lower current the capacitance seems to rise to about 200uf, but that's still way way off.
In a simple filtering ability test, these two caps were about 20 times *worse* at filtering the same
signal under the same exact test conditions as the +12v cap. This tells me they are bad too.
BTW, 20 times worse means they pass a signal with 20 times the amplitude as another cap would allow
in a low pass configuration. Obviously 20 times is really really a lot different when the SAME cap value
(2200uf) is being used for both tests!

Because of the test results, it's a wonder this thing ever worked. I can see now why it did not work
OUT OF PC CASE when it did appear to work for a while INSIDE THE PC CASE. It's because the caps
on the motherboard were filling in for the bad power supply caps all along. Once out of case, the power
supply had to work with it's very own caps and therefore failed when the +12v line was loaded.

Also because of the tests it appears that this supply will again work properly when the two caps are replaced.

There are also some 470uf 200v caps i wonder if i should replace them too. Maybe i'll test them also.
The remaining caps appear to be for low current apps.

It seems that i had remembered about the ESR change with capacitor age, but i had forgotten that the
electrolyte dries up too and causes the capacitance itself to dip way down as well. Two bad things.

I might add as a final note that although these two caps are definitely faulty, there was absolutely no
sign whatsoever of leaking electrolyte. Apparently it had simply dried up rather than leaked out.
LEDs vs Bulbs, LEDs are winning.

User avatar
Janitor Tzap
Posts: 1436
Joined: Sat Aug 12, 2006 5:17 pm
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by Janitor Tzap » Sat Oct 03, 2009 7:43 am

Coolness. :cool:

I don't think those 470uf 200v caps are going to be bad.
They are on the HOT side{AC} of the supply.
They rarely fail.

Most newer caps have a pattern stamped on to the top of the can.
This is either a V or X.
This acts as a pressure relief valve, so the cap will vent, instead of exploding. :)

Check your E-mail, I sent that PDF.


Signed: Janitor Tzap

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 » Sat Oct 03, 2009 8:35 am

Hi again,


Oh ok thanks much.
I had a feeling that the X on top was to allow the top to expand a little easier but didnt have any written info on this
until you commented. Thanks for that too.

I found a couple other bad caps too, two 1000uf,10v units.

There are a bunch of other caps too, low voltage 470uf, 10uf, and 2.2uf, i wouldnt expect those to be bad though.
The 470's are used for the negative supplies, and the 2.2uf's look like they are for signals. Dont know what the 10uf's
are for yet though. The cases on all these look ok too.

This turned out to be a little more interesting than i thought it would at first. I worked on designs and troubleshooting of
power supplies from about 50 milliwatts (LED drivers) to 30 killowatts (10,000 watts per phase, three phase AC) but this
is the first time i really dug into a computer type power supply like this. If this is the typical failure mode i can see
fixing these things would be a breeze. I didnt check the controller chips yet though, what type they are. Maybe ill
dig into that next.

I found some cheaper caps that i am going to try. The whole parts list should run under 10 dollars plus shipping so
i am going to give this a try. The caps are pretty easy to get out too once the pc board is removed.
I know other people have done this with success too, what i dont know is how long the supply will live after it is
fixed because of possible other things that can go wrong with it later.


A BIT LATER:

That pdf file looks interesting. Im going to read that next.
In the mean time, would you happen to know what kind of connector i would purchase to make a good connection
to the output of the supply when i go to test it? I would need the pin size i guess and spacing, or better yet a
part number. Might as well get a connector too.
LEDs vs Bulbs, LEDs are winning.

User avatar
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 » Sat Oct 03, 2009 9:56 am

MrAl
That pdf file looks interesting. Im going to read that next.
In the mean time, would you happen to know what kind of connector i would purchase to make a good connection
to the output of the supply when i go to test it? I would need the pin size i guess and spacing, or better yet a
part number. Might as well get a connector too.
Generally referred to as a Molex connector the actual connectors are manufactured by several companies. This Link provides the actual Molex part numbers for a 24 pin both male and female. They are shown on the right and left of the page.

The easy route, however, is to buy a simple 24 pin extension cable. Then cut off the female end that mates with the motherboard. Makes for simple and easy. Then tie all the Red (5 Volts), Yellow (12 Volts) and Orange (3.3 Volts) lines together. Additionally you can place a switch to common on the Green line. Additionally you can place the Green on a scope to see PWR_ON go low when the switch is closed as well as the PWR_OK line go high within a time period (generally under 500 mSec.

This is an example of a 24 pin PSU to motherboard extender cable and at 12 bucks makes for easier than building up a connector. Much like old PSU boards I have several lying around and you are more than welcome to any of this stuff.

Ron

User avatar
Janitor Tzap
Posts: 1436
Joined: Sat Aug 12, 2006 5:17 pm
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by Janitor Tzap » Sat Oct 03, 2009 11:13 am

reloadron wrote:MrAl
That pdf file looks interesting. Im going to read that next.
In the mean time, would you happen to know what kind of connector i would purchase to make a good connection
to the output of the supply when i go to test it? I would need the pin size i guess and spacing, or better yet a
part number. Might as well get a connector too.
Generally referred to as a Molex connector the actual connectors are manufactured by several companies. This Link provides the actual Molex part numbers for a 24 pin both male and female. They are shown on the right and left of the page.

The easy route, however, is to buy a simple 24 pin extension cable. Then cut off the female end that mates with the motherboard. Makes for simple and easy. Then tie all the Red (5 Volts), Yellow (12 Volts) and Orange (3.3 Volts) lines together. Additionally you can place a switch to common on the Green line. Additionally you can place the Green on a scope to see PWR_ON go low when the switch is closed as well as the PWR_OK line go high within a time period (generally under 500 mSec.

This is an example of a 24 pin PSU to motherboard extender cable and at 12 bucks makes for easier than building up a connector. Much like old PSU boards I have several lying around and you are more than welcome to any of this stuff.

Ron
Ron,

Yeah,
Getting an PSU motherboard extender cable, and modifying it would be cheaper. :wink:

MrAl,

The chip is probably a KA7500B Voltage Mode PWM Controller, or maybe a DBL494.
Every manufacturer will use a different chip, normally whatever is cheapest.

As for how long a supply will last....
I purchased {El-Cheapo} 400 Watt ATX Supplies for $15.
These lasted me 4 years till the caps gave out.
Manufacturers expect people/businesses to replace their computers every two, or three years.
Thus, if these {El-Cheapo} supplies are any measure of how long your rebuilt supply will last.
Then maybe 4 years more.
{Not a accounting for acts of God.} :lol:


Signed: Janitor Tzap

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 » Sat Oct 03, 2009 12:12 pm

Hi again,


Ron:
Do you know the pitch of that connector and pin diameter/size by any chance? I couldnt find
that connector listed as that part number. If i knew the pitch and pin size i could look
that up instead. I might however take you up on the offer for a connector with some wires
attached, that sounds even easier to work with. I'll look for one first before i put you
through that trouble though.


Janitor:
Silly me, i looked at the chip part number and saw first "CHMC S6A00" but didnt look at
the "D339" under that, and similarly the other chip marked "CHMC S6B03" and somehow
overlooked the "B494" just below that. I think i was in too much of a hurry :smile:
These are probably type LM339 comparator and type TL494 (or similar) controller ic's.
I'll check some of the connections to verify this unless you recognize those other
numbers. I am thinking of studying this power supply out fully since i have it apart
and everything, except there are quite a few parts in there like lots of small resistors
and several small caps and a bunch of signal diodes...i'd say easy 100 parts in there.
It might not be too easy to trace everything out since i dont have a schematic of this
exact unit.

I was checking the life of the soon-to-be new caps, and i see 3000 hours at 105 deg C,
and 5000 hours at 105 deg C, and the like. I am trying for the 5000 hour type but some
of the caps are only available in the 3000 hour model. 5000 hours at 12 hours per day
is just over a year, but that's at 105 deg C and hopefully this power supply doesnt get
that hot inside because it has TWO fans, one in the back to exhaust air and one on the
bottom to pull in air. It's not a bad supply really, but they must have used really
really cheap caps. I've only used this power supply for a bit over 2 years now.
I figure i should get at least four years from it then. Funny though, i have had other
power supplies last for five years. This is the first time i had one go bad like this
and this is supposed to be a higher quality unit! Strange.


In the mean time, thanks again to both of you for the wealth of information you have
provided here. That's making this fix not only possible in the first place but also
interesting too.
At first i wasnt too keen on fixing this thing, seeing some supplies on the web for like
20 dollars, and 30 dollars, for supplies that dont look too bad either, but that would
be too boring now :smile:
LEDs vs Bulbs, LEDs are winning.

Robert Reed
Posts: 2276
Joined: Wed Nov 24, 2004 1:01 am
Location: ASHTABULA,OHIO
Contact:

Re: Power Supply Voltage Tolerances and Measuremen

Post by Robert Reed » Sat Oct 03, 2009 4:29 pm

MrAl
Been following your post more as a matter of general troublesooting ATXs. Got a couple of questions for you.
How many hours run time do you think the bad supply had on it? Were the fans ever down for a period of tme while it was running? The reason I am asking is I am copncerned about the Caps lifespan. Even the 5000 hr rating you mentioned on new parts sounds absolutely horrible.You mentioned specs at 105 degrees - were these actually 105 degree or 85 degree class caps. Its been my experience that in the last 30 years, passives quality and reliability have improved greatly from the past - all passives ( partly due to lower voltages and operating temperature).As a matter of fact I am amazed at how close the values are on them compared to the stated value (1-5% most of the time even for 5 & 10 & 20% stated values)And they are much cheaper than they used to be. Just look at the average 10 year old TV that may up to 50,000 hours service time on their caps ( my wifes has close to 80,000 hrs.) and are still humming along. Whoops, don't take that as 120 Hz hum :grin: . And lastly, what brand of caps were these so I don't accidently purchase that brand!

Thanx
Robert

User avatar
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 » Sun Oct 04, 2009 3:06 am

Ron:
Do you know the pitch of that connector and pin diameter/size by any chance? I couldn't find
that connector listed as that part number. If i knew the pitch and pin size i could look
that up instead. I might however take you up on the offer for a connector with some wires
attached, that sounds even easier to work with. I'll look for one first before i put you
through that trouble though.
This should be the connector PDF including all the data from Molex including pin diameters and pitch.

Available from Mouser just need to remember the connector doesn't include the pins. I don't think so anyway. Also I think Mouser sells the insertion and extraction tools needed for those connectors and others like them.

this website is a very good source of information for all the connectors used in ATX and BTX 12 Volt form factor power supplies.

Hope that helps the cause along.

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 » Sun Oct 04, 2009 4:22 am

Hi again,


Robert:
The hours on the supply was maybe 6000 hours before it stated to fail. The fans never stopped
as far as i know, and still work today.
The caps i found were from Digikey, and the max hours i found was 10,000 hours although most
are 5,000 hours and less, like 3000 hours. But that's at the max temperature too.
I didnt see any 50,000 hour ones at all. Take a look on Digikey.

Ron:
I checked out the connector but that was the gender that plugs into the mother board.
What i would need for testing the supply itself would be the gender that plugs into
the power supply, or that the supply plugs into (i think it would be called 'male').
Thanks for the data sheet pdf's. I see now that the pitch is 4.2mm, but i was unable
to find any on Mouser or Digikey.
LEDs vs Bulbs, LEDs are winning.

User avatar
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 » Sun Oct 04, 2009 5:27 am

You would think I could get the gender thing correct. :smile:

This link is an example of the pin header that is on the motherboard and could be modified. However, my experience has been that using an extender cable and cutting one end off is a good route to take.

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 » Sun Oct 04, 2009 8:18 am

Hi Ron,


Thanks for the link. I was able to find one now that i think i can use, price about $5.00 USD.

Now that i think back a little about the two caps, i dont think i have ever seen caps get that
bad except for when they exploded or leaked badly.
LEDs vs Bulbs, LEDs are winning.

User avatar
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 » Sun Oct 04, 2009 11:14 am

Hi Ya

Just remember I am not sure that when you buy a connector of that type you get the pins/sockets. Nothing worse than a chunk of nylon and nothing to put in it. :sad:

I haven't mentioned it till now since I doubt your PSU would have been involved but during the early 2000s there was a rash of bad capacitors. I saw my share of them. The Wikipedia sums it up in a nutshell with some pictures. Known to many as the bad capacitor plague it left a mark in history. If you look at the pictures in the link you will see images of exactly what Janitor Tzap was referring to. You got yourself one heck of a project and I can't wait to see how this goes. Should be pretty interesting.

Ron

Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest