PSU

  • Thread starter Thread starter Colin Bearfield
  • Start date Start date
C

Colin Bearfield

My PSU isn't showing any output. A friend suggested it might be an AT
instead of an ATX. How can I tell?

If it is ok and the failure is elsewhere, what voltage should I expect
from the red, black, and yellow conductors?

There isn't a squeak from anywhere. I know that there is power input.

Colin
 
My PSU isn't showing any output.
Click to expand...

Under what conditions are you testing it?
An ATX won't ouput on it's main power rails until logically
turned "on" by the motherboard or shorting the PS_On line to
ground. To short it, anything metal (like a small
paperclip) can be connected to pin 14 (usually green wire)
and ground (any ground point but the black ground wires in
picture below are easiest).

"Some" PSU require additional load on 5V rail to stay on
(turn off so quick it may seem like it didn't turn on at
all). A hard drive is sufficient load, but if the power
supply is questionable, it would be good to use an
old/worthless drive, not a good/valuable one.

Some newer ATX12V PSU can require a load on the 12V rail.
It is even less common but again connecting a hard drive
will usually suffice. To be techicially correct and follow
the typical manufacturer suggestions for minimal load, the
ideal would be a dummy load (like power resistors) that
cause 2A flow from both 5V and 12V rails. This is probably
an unrealistic thing for you to do at this point, but I
mentioned it simply because there are many potential
scenarios where one might have less common equipment
including mobile drives or unique power supplies and need
another method to test them.

See the following picture of an ATX motherboard connector,
it is the far most common arranagement used on any system
from around 1998 to very recently (in the past year, with
some now using 24 pin connectors)

http://69.36.189.159/usr_1034/atx_on.gif

A friend suggested it might be an AT
instead of an ATX. How can I tell?
Click to expand...

AT has a single row of flat-shaped pins, usually as two
connectors (but sometimes both are combined into one but
it's still a long, single row of pins). ATX is always two
rows of pins. Note the information on the power supply
label as well, and the coloring of the wires. Some OEMs
have used deviations from either spec, including popular
OEMs like Dell or Compaq.

If it is ok and the failure is elsewhere, what voltage should I expect
from the red, black, and yellow conductors?
Click to expand...

First we would have to assume the manufacturer used standard
wire color conventions- most do, but some don't. For
example, blue is often substituted for yellow. "Often"
means IF a different color were substituted, usually it does
follow the standard color scheme which is:

Black- Ground (near (if not exactly) 0V relative to another
ground point in the system

Yellow- 12V, +-10% but ideally within +- 5%

Red - 5V, +- 5%

Orange - 3.3V +-5%

There isn't a squeak from anywhere. I know that there is power input.

Colin
Click to expand...

If you describe the system it (was) used in, came out of,
(including the motherboard make & model or the OEM) that
might help identify the power supply, or at least it's
necessary electrical configuration if not the exact make and
model of it.
 
Colin said:
My PSU isn't showing any output. A friend suggested it might be an AT
instead of an ATX. How can I tell?

If it is ok and the failure is elsewhere, what voltage should I expect
from the red, black, and yellow conductors?

There isn't a squeak from anywhere. I know that there is power input.

Colin
Click to expand...

The ATX PSUs don't show output if a specific couple of pins don't show a
load on them and I assume you, like most sensible people just plug the
thing to the mains on its own, switch on and put a multimeter to it ?

I know I have and have thrown a few away before knowing about the following.

<http://modtown.co.uk/mt/article2.php?id=psumod>


Richard.
 
The ATX PSUs don't show output if a specific couple of pins don't show a
load on them and I assume you, like most sensible people just plug the
thing to the mains on its own, switch on and put a multimeter to it ?

I know I have and have thrown a few away before knowing about the following.

<http://modtown.co.uk/mt/article2.php?id=psumod>


Richard.
Click to expand...
These are both excellent answers.

I have used a paperclip to short out the green and one of the blacks.
I now know what voltages to expect and how to tell AT and ATX apart.

Taking the black as common the only pins showing a voltage were:

Pin 9 Purple 5v

Pin 14 Green 5v

To the best of my knowledge there is nothing on the other pins but I
could check for continuity.

Do you think this PSU is US. It would be coincidental - I've just
changed the motherboard and was preparing to install new drivers. It
didn't do a thing. Not even a beep.

BTW the internal fan didn't run when I powered up the PSU.

many thanks

Colin
 
Taking the black as common the only pins showing a voltage were:

Pin 9 Purple 5v

Pin 14 Green 5v

To the best of my knowledge there is nothing on the other pins but I
could check for continuity.
Click to expand...

There cannot be 5V on the green wire if you have a paperclip
shorting it to ground. You need to measure the other
voltages while that green wire is shorted to ground.
 
Colin said:
These are both excellent answers.

I have used a paperclip to short out the green and one of the blacks.
I now know what voltages to expect and how to tell AT and ATX apart.

Taking the black as common the only pins showing a voltage were:

Pin 9 Purple 5v

Pin 14 Green 5v

To the best of my knowledge there is nothing on the other pins but I
could check for continuity.
Click to expand...

As Kony says in his reply if you short any of the other positive lines
to black (0v usually but internationally agreed standards means that
someone will go their own way), which should be ground then you are
'grounding' that line to 0v also!
Do you think this PSU is US. It would be coincidental - I've just
changed the motherboard and was preparing to install new drivers. It
didn't do a thing. Not even a beep.
Click to expand...

You may have one of those very rare breeds of PSU that has a 'crowbar'
to save itsself. Basically it goes to sleep until it thinks you've gone
away and repaired the fault then switches itself on again. Almost all
PSUs I've had will destroy themselves rather than blow the fuse, out of
spite, well, except for the nightmare one which you will have sometime
in the future that will burn the 5v regulator sending the rail to 12v
and take any plugged in unit (motherboard, drives, cards...) that also
uses 5v also, with it.

Sorry, just preparing you for the future!

Okay, you've just changed motherboards and that's all. If there was a
fault then usually the fan runs up for a fraction of a second then
stops. Whether there's a smell of bakelite or not is whether you are
lucky or not.

Did you look at your pin connections (usually bottom right of the
motherboard) and checked that your connecting pin sets were located as
per manual ?
BTW the internal fan didn't run when I powered up the PSU.
Click to expand...

The fan usually runs from the 12v rail but the whole set of regulating
circuits won't kick in if it doesn't receive its wake up state I assume
by a latch circuit that, as soon as it receives a short pulse from your
on/off switch, electronically closes the same lines as shown in that URL
above, in effect putting a bit of wire across.

Somewhere else in these newsgroups I did read of someone worried about
the same thing and had a reply about blowing on the fan to get it to
rotate. For some reason I didn't read into, it got the thing working.
This isn't a joke but it was an off topic post in a newsgroup that had
nothing to do with PSUs so can I find it ? Can I f...!

I can see it making sense as if the fan was jammed or a bit dusty it
might put an extraneous load on the circuit so the circuit stays asleep.

Before you get back to us, smell the box (with power off of course) and
if you smell bakelite then buy another $5 PSU as it should only smell
dusty at most.
many thanks

Colin
Click to expand...

Good luck and God speed!


Richard.
 
Found another URL just to compare.

<http://www.virtual-hideout.net/guides/atx_psu_mods/index.shtml>


Richard.
Click to expand...

Thanks again. You know 3 decades ago I was an electronics constructor
and in 1981 built a computer. My soldering iron rarely got warm after
that. In all that time I always ignored PSU's and until yesterday
knew next to nothing about a PC PSU. They are fascinating creatures,
aren't they.

Well, I suspect my PSU is ok after all.

I confess that I think that I had the PSU switched off when I noticed
there was no fan working. There is now. I had already checked that
there was power in the power supply cable.

Looking for a short to earth, I have checked the mobo side of the
molex with a digital multimeter. Here are the results:

All the 5v pins and 12v pins are open circuit in respect to the case.
Only the video card and memory are plugged into the board, oh and the
power switch.. The 3.3v pins are around 80ohms and the ground pins
are about 0.7 ohms.

I think that the 0.7ohms means that there isn't a direct short to the
case.

What do you think?

Colin
 
Colin said:
Thanks again. You know 3 decades ago I was an electronics constructor
and in 1981 built a computer. My soldering iron rarely got warm after
that. In all that time I always ignored PSU's and until yesterday
knew next to nothing about a PC PSU. They are fascinating creatures,
aren't they.
Click to expand...

I'm the same! I do still dabble but that's mainly making up specific
leads for the computer. A couple of years ago I did knock up a small
circuit that the joystick port could read from so that a software synth
(SynthEdit - build your own creations) could make crappy noises from but
now I think I'm in line for a pipe and slippers and asking young girls
of 25+ if they want a sweet!
Well, I suspect my PSU is ok after all.

I confess that I think that I had the PSU switched off when I noticed
there was no fan working. There is now. I had already checked that
there was power in the power supply cable.
Click to expand...

You've just reminded me of a tutor many years ago at college who
couldn't strongly state enough about protecting those long thermometers
that come in those plastic tubes with the bungs at each end. After
telling us to put the thing back in its protective case, yep, he took
out the bung forgetting that he'd already done so at the other end and
crash!

We all get those days!

Looking for a short to earth, I have checked the mobo side of the
molex with a digital multimeter. Here are the results:

All the 5v pins and 12v pins are open circuit in respect to the case.
Only the video card and memory are plugged into the board, oh and the
power switch.. The 3.3v pins are around 80ohms and the ground pins
are about 0.7 ohms.

I think that the 0.7ohms means that there isn't a direct short to the
case.
Click to expand...

0.7ohms is quite low, if by that I assume you are touching one pin of
your multimeter to chassis and one to the PSU earth line ? That means
that the amount of metal over that distance from chassis to case is one
large resistor of 0.7ohms. The resistance across my steel computer desk
is 0.2ohms and that's only got two screw joints in it. Damned near a short!

What do you think?

Colin
Click to expand...

You've got a rotating fan right, and volts out ? So you're not getting
POST where all the motherboard checks say "yep, we've got a screen, give
the man a beep, we have some memory, etc" ?

I did come across a board someone wanted looking at that wouldn't wake
up (no fan of course) and it turned out that the CMOS clearing pin combo
was left shorted at 'clear CMOS' so the thing wouldn't start up and not
a recommended setting. Just a thought and one to add to the "doh!" list.

Tell us what state it's at now, okay!


Richard.
 
I'm the same! I do still dabble but that's mainly making up specific
leads for the computer. A couple of years ago I did knock up a small
circuit that the joystick port could read from so that a software synth
(SynthEdit - build your own creations) could make crappy noises from
Click to expand...

Why not make a card that will read a thermistor and relative humidity
sensor and will issue contact closures to run the air conditioner.
That way the user could build custom protocols to save on electricity.

I am using a Carrier Termidistat and have been able to lower the daily
runtime significantly with no real loss of comfort. The reason is
because the thermidistat controls on dew point and not temperature
alone. You set the temp and RH and it computes the DP and keeps the
environment at a constant absolute moisture level. This can add up to
several hundreds of dollars in the cooling season, which lasts 7
months in Houston and most of the Gulf Coast and places several
hundreds of miles inland that succumb to Gulf weather.

The limitations of the thermidistat is that it is a closed system
which prevents the user from tinkering with things like cycle
parameters. With a good minimizing algorithm, the user-assessible
system could learn to seek out the best parameters consistent with
stated comfort in terms of the DP.

PCI board, LAN-connected sensors and access software for $100 retail
would sell. I would be a beta test site for a free unit.
now I think I'm in line for a pipe and slippers and asking young girls
of 25+ if they want a sweet!
Click to expand...

"Candy is dandy, but liquor is quicker."

Does anyone know who is the author of that famous quip?

--

Map of the Vast Right Wing Conspiracy
http://home.houston.rr.com/rkba/vrwc.html

"The possession of arms is the distinction
between a free man and a slave."
-- Andrew Fletcher, Discourse on Government (1695)
 
Bob said:
Why not make a card that will read a thermistor and relative humidity
sensor and will issue contact closures to run the air conditioner.
That way the user could build custom protocols to save on electricity.
Click to expand...

Strangely enough I've got a kit-build 24 channel IO ISA unit that I
bought from Maplin and a Compaq 386 in the attic awaiting first orders
to do something.
I am using a Carrier Termidistat and have been able to lower the daily
runtime significantly with no real loss of comfort. The reason is
because the thermidistat controls on dew point and not temperature
alone. You set the temp and RH and it computes the DP and keeps the
environment at a constant absolute moisture level. This can add up to
several hundreds of dollars in the cooling season, which lasts 7
months in Houston and most of the Gulf Coast and places several
hundreds of miles inland that succumb to Gulf weather.
Click to expand...

I wonder what that sounds like when plugged into a soft-synth and BTW,
can it stretch time so that there's more of it in a day ?

There is a point when we have to say "look, I don't have time for
building plastic kits, Meccano, play with my trains, etc so I might as
well stick to one or two things and do that."

Electronics have given away to my some time hobby of using my home built
3D camera for airshows and other exhibitions. Then there's the
gardening, washing, cooking..... A man's work is never done!
The limitations of the thermidistat is that it is a closed system
which prevents the user from tinkering with things like cycle
parameters. With a good minimizing algorithm, the user-assessible
system could learn to seek out the best parameters consistent with
stated comfort in terms of the DP.
Click to expand...

You are Stanley Unwin and my long lost dad! You'll be impressed that
I'd set up over 50ft of wire as an aerial and connected it to my old
radio just the other day.
PCI board, LAN-connected sensors and access software for $100 retail
would sell. I would be a beta test site for a free unit.




"Candy is dandy, but liquor is quicker."

Does anyone know who is the author of that famous quip?
Click to expand...

Would it be the old Hollywood actor (forgotten his name) who used to act
drunk and whether he was actually drunk all the time I don't know but
I'd guess it was him. He played alongside that "come up and see me some
time" woman. Even my memory's going now!

Richard.
 
You are Stanley Unwin and my long lost dad! You'll be impressed that
I'd set up over 50ft of wire as an aerial and connected it to my old
radio just the other day.
Click to expand...

No, I am Donald Trump trying my best to be as frugal as possible.

Electricity in Texas costs a lot during the summer and with $60 oil,
it is prohibitively expensive. The main provider wants 13.5 cents per
kW-hr. That means a standard 4-ton SEER 10 air conditioner will cost
over 60 cents per hour. If the unit runs 10 hours per day, that's
$6.00 per day. Multiply that by at least 200 days per a/c season
(May-Nov.) and you get $1,200.

If I can get that down to 7 hours controlling on DP and I go with a
cheaper provider at 9.6 cents, then I save $570 per season. It takes
$750 in earnings to have that much money in your pocket.

So this is not some Mr. Wizard project - it's a cost reduction project
that results in non-trivial savings.
Would it be the old Hollywood actor (forgotten his name) who used to act
drunk and whether he was actually drunk all the time I don't know but
I'd guess it was him. He played alongside that "come up and see me some
time" woman. Even my memory's going now!
Click to expand...

Nope. Not even close.

Bennett Cerf was the head of Random House, a publisher of Dr. Seuss'
books. He was a frequent guest on game shows many decades ago, like
What's My Line.

--

Map of the Vast Right Wing Conspiracy
http://home.houston.rr.com/rkba/vrwc.html

"The possession of arms is the distinction
between a free man and a slave."
-- Andrew Fletcher, Discourse on Government (1695)
 
Looking for a short to earth, I have checked the mobo side of the
molex with a digital multimeter. Here are the results:
Click to expand...

It is often that a motherboard short may not be a direct
rail-to-ground short but rather one of the data lines or
motherboard-post-regulated power lines. While one can probe
around and learn a bit about the board's layout, easiest is
to simply leave the board out of the case to get it working
initially, and then before installating it, note how many
studs it uses and where they are located, then verify that
the case has all of but only these studs installed.

All the 5v pins and 12v pins are open circuit in respect to the case.
Only the video card and memory are plugged into the board, oh and the
power switch..
Click to expand...


You will need a CPU, and sometimes a heatsink fan with RPM
signal plugged into the appropriate fan header.

The 3.3v pins are around 80ohms and the ground pins
are about 0.7 ohms.

I think that the 0.7ohms means that there isn't a direct short to the
case.
Click to expand...

0.7 Ohm is close enough, could be trace resistance or a
coating on the case metal. It is not a sign of a problem.
 
Bob said:
.... snip ...

"Candy is dandy, but liquor is quicker."

Does anyone know who is the author of that famous quip?
Click to expand...

I believe it came out of the twenties, and for some reason I
suspect Gypsy Rose Lee.
 
I'm the same! I do still dabble but that's mainly making up specific
leads for the computer. A couple of years ago I did knock up a small
circuit that the joystick port could read from so that a software synth
(SynthEdit - build your own creations) could make crappy noises from but
now I think I'm in line for a pipe and slippers and asking young girls
of 25+ if they want a sweet!


You've just reminded me of a tutor many years ago at college who
couldn't strongly state enough about protecting those long thermometers
that come in those plastic tubes with the bungs at each end. After
telling us to put the thing back in its protective case, yep, he took
out the bung forgetting that he'd already done so at the other end and
crash!

We all get those days!



0.7ohms is quite low, if by that I assume you are touching one pin of
your multimeter to chassis and one to the PSU earth line ? That means
that the amount of metal over that distance from chassis to case is one
large resistor of 0.7ohms. The resistance across my steel computer desk
is 0.2ohms and that's only got two screw joints in it. Damned near a short!



You've got a rotating fan right, and volts out ? So you're not getting
POST where all the motherboard checks say "yep, we've got a screen, give
the man a beep, we have some memory, etc" ?

I did come across a board someone wanted looking at that wouldn't wake
up (no fan of course) and it turned out that the CMOS clearing pin combo
was left shorted at 'clear CMOS' so the thing wouldn't start up and not
a recommended setting. Just a thought and one to add to the "doh!" list.

Tell us what state it's at now, okay!
Click to expand...
I haven't changed anything since the last report with the pin readout.
The 0.7 was variable; it might have settled on 0.6 or 0.7. I just
reported 0.7 because it reminded me of the forward bias of an npn
transistor. Do they still do transistors?

I'm going to try it tomorrow as is because apart from moving the board
out of the case I can't think of anything else.

I might move the mobo out of the case.

Thank everyone.

Colin
 
Bob said:
No, I am Donald Trump trying my best to be as frugal as possible.
Click to expand...

Your house is being watched by American satellites as you are not meant
to care ("give a rat's ass") how much you burn! ;-)

[snippage]
So this is not some Mr. Wizard project - it's a cost reduction project
that results in non-trivial savings.




Nope. Not even close.

Bennett Cerf was the head of Random House, a publisher of Dr. Seuss'
books. He was a frequent guest on game shows many decades ago, like
What's My Line.
Click to expand...

I thought you were asking a question because you had forgotten!

Okay, what are the proper names for elements in the US flag known as
"The Stars and Stripes" ?

Damn you "Ripley's Believe It Or Not" !


Richard.
 
I thought you were asking a question because you had forgotten!
Click to expand...

No, I asked it because the poster said something about treating some
young ladies to candy. I was trying to be helpful by advising him
about the relative merits of liquor vs candy when it comes to women.

Actually, owning a house is the number one way to get a woman's
attention, according to a survey posted in the Telegraph (London).
They want something tangible to take with them after they divorce you.


--

Map of the Vast Right Wing Conspiracy
http://home.houston.rr.com/rkba/vrwc.html

"The possession of arms is the distinction
between a free man and a slave."
-- Andrew Fletcher, Discourse on Government (1695)
 
I'm the same! I do still dabble but that's mainly making up specific
leads for the computer. A couple of years ago I did knock up a small
circuit that the joystick port could read from so that a software synth
(SynthEdit - build your own creations) could make crappy noises from but
now I think I'm in line for a pipe and slippers and asking young girls
of 25+ if they want a sweet!


You've just reminded me of a tutor many years ago at college who
couldn't strongly state enough about protecting those long thermometers
that come in those plastic tubes with the bungs at each end. After
telling us to put the thing back in its protective case, yep, he took
out the bung forgetting that he'd already done so at the other end and
crash!

We all get those days!



0.7ohms is quite low, if by that I assume you are touching one pin of
your multimeter to chassis and one to the PSU earth line ? That means
that the amount of metal over that distance from chassis to case is one
large resistor of 0.7ohms. The resistance across my steel computer desk
is 0.2ohms and that's only got two screw joints in it. Damned near a short!



You've got a rotating fan right, and volts out ? So you're not getting
POST where all the motherboard checks say "yep, we've got a screen, give
the man a beep, we have some memory, etc" ?

I did come across a board someone wanted looking at that wouldn't wake
up (no fan of course) and it turned out that the CMOS clearing pin combo
was left shorted at 'clear CMOS' so the thing wouldn't start up and not
a recommended setting. Just a thought and one to add to the "doh!" list.

Tell us what state it's at now, okay!


Richard.
Click to expand...
Last night I took a chance and powered up and got a sensible POST.
There was no response from the floppy drive but I thought it was a
poor power connection. The BIOS found everything correctly including
the floppy drive.

Of course, the O/S found loads of new hardware and wanted drivers. It
wanted the W98SE install CD but couldn't see it. I put that down to
not having the right drivers installed. I did have the drivers on the
original floppy disk that came with the mobo, but guess what - they're
on a floppy disk.

I made sure that the floppy drive was properly cabled up, even
changing the data cable for good measure. On power up the LED on the
floppy dr9ive did nothing but there was a flash of light when it had a
disk in it.

So everything should be alright, but it still can't lift the drivers
from the CD, not even the mouse driver. It just behaves as if the CD
is absent.

I've got to go and walk dogs now; it's what I do in retirement. I'll
have another go tonight.

Colin
 
Last night I took a chance and powered up and got a sensible POST.
There was no response from the floppy drive but I thought it was a
poor power connection. The BIOS found everything correctly including
the floppy drive.

Of course, the O/S found loads of new hardware and wanted drivers. It
wanted the W98SE install CD but couldn't see it. I put that down to
not having the right drivers installed. I did have the drivers on the
original floppy disk that came with the mobo, but guess what - they're
on a floppy disk.

I made sure that the floppy drive was properly cabled up, even
changing the data cable for good measure. On power up the LED on the
floppy dr9ive did nothing but there was a flash of light when it had a
disk in it.

So everything should be alright, but it still can't lift the drivers
from the CD, not even the mouse driver. It just behaves as if the CD
is absent.

I've got to go and walk dogs now; it's what I do in retirement. I'll
have another go tonight.
Click to expand...

Easiest thing to do would be boot to the Win98SE floppy,
choose "with CDROM support" (or similar meaning if wording
differs), then copy the Win98 CD's "win98" folder to the
hard drive. You need not have all the subfolders so the
basic commands at the prompt would be,

mkdir c:\win98
copy <insert cdrom drive letter here>:\win98\*.* c:\win98


It'll take a few minutes to finish copying.
 
Colin said:
Last night I took a chance and powered up and got a sensible POST.
There was no response from the floppy drive but I thought it was a
poor power connection. The BIOS found everything correctly including
the floppy drive.

Of course, the O/S found loads of new hardware and wanted drivers. It
wanted the W98SE install CD but couldn't see it. I put that down to
not having the right drivers installed. I did have the drivers on the
original floppy disk that came with the mobo, but guess what - they're
on a floppy disk.

I made sure that the floppy drive was properly cabled up, even
changing the data cable for good measure. On power up the LED on the
floppy dr9ive did nothing but there was a flash of light when it had a
disk in it.

So everything should be alright, but it still can't lift the drivers
from the CD, not even the mouse driver. It just behaves as if the CD
is absent.

I've got to go and walk dogs now; it's what I do in retirement. I'll
have another go tonight.

Colin
Click to expand...

Look at your boot order in your BIOS. Go for floppy, CD then other.
From experience after much hair tugging with my sister at the other end
of the phone line, had I realised that the company that sold her the
(locked in) system, that the floppy didn't figure in the boot up
sequence but us old 'DOS die hards' like the old way of floppy first,
what with the Oakdale or whatever they're called, drivers being on floppy.

Simply put, if the floppy has the CD drivers and the Win98 doesn't then
get those drivers to wake up the CD - then - no CD seen by the config
setup on floppy.

Right, I'm off to get drunk and I hope you all do the same!



Richard.
 
Back
Top