PSU voltage figures

[VWWall]

That resistance is measured at negligible current draw. As current goes
up, so will the resistance. Pretty soon (especially on such small wire),
you hit a point where even a small increase in current starts to cause LARGE
resistances - which mean significant voltage drops across the cable.

Copper has a temperature coefficient of 0.00393. This means the
resistance will increase by about 4/10% per degree rise in temperature.
Even if the wire did reach 100 degrees C, this is only a 32% increase,
which will reduce the current for a constant voltage input.

With 21-gauge aluminum wire at 10 amps, you're going to have a voltage
drop of around 0.85 volts just over a 2-foot length of cable. That means
you're dissipating 8.5 watts of power over the length of the cable, and it
means that you're really only getting 11.15 volts at the far end.

Don't know why we're talking about 21 or 22 gauge wire. (Sorry for my
misplaced decimal point). :-( The ATX/ATX12V Design Guide shows AWG 18
wires for the 12V and 5V leads. Looking at several ATX PSUs, I find the
leads are made up of 49 strands whose combined circular mils equal
~1624, which corresponds to 18 gauge wire. This has a resistance of
6.385 ohms/1000 feet. At 10 A, this is a drop of 0.128 V over a 2 foot
length with a loss of 1.28 W, with ~119 W delivered at the cable end.

Most electrical codes allow 18 gauge wire in extension cords rated at 10
A, even with rubber insulation. I've drawn 10 A from the 12 V rail of
an ATX PSU during a test, and the wire didn't get near 100 degrees C.

Virg Wall

 

[w_tom]

In order to get two wires into one Molex connector, those
power splitters use 22 AWG. Aaron S said "had to "daisy
chain" three
(and maybe soon four) of the HDDs off *one* 4 pin Molex power
connector."

 

[kony]

In order to get two wires into one Molex connector, those
power splitters use 22 AWG. Aaron S said "had to "daisy
chain" three
(and maybe soon four) of the HDDs off *one* 4 pin Molex power
connector."

Some do use 22 AWG, but not necessarily, I have some with
18AWG, not to mention that the PSU leads themselves also are
(usually) 18AWG and have two wires per connector on all
connectors until the last.

 

[VWWall]

In order to get two wires into one Molex connector, those
power splitters use 22 AWG. Aaron S said "had to "daisy
chain" three
(and maybe soon four) of the HDDs off *one* 4 pin Molex power
connector."

I have several power splitters with 18 gauge wire. All of the "in line"
Molex connectors on my PSUs have two #18 gauge wires per pin with no
problem. I will admit that if one uses 22 guage wire it ought not be
used to supply high current devices. A standard hard drive requires ~3
A at start-up; this is what power splitters were designed to supply. I
suggest a little re-arrangement of the drives, or finding someone who
knows how to use a soldering iron, would avoid using any wires beyond
their capacity.

I agree with you that "crimped" connector pins can be a problem. Even
modern appliances seem to be using these. I had a toaster which had a
crimped mains line connection which got so hot it melted the plastic
part to which it was attached. Your famous 3 1/2 digit DVM will show up
this situation. ;-)

 

[Franklin]

I think this article answers most questions:
http://www.dansdata.com/spcables.htm

Dan has a nice written style.

I reckon this (slightly off-topic) article on how not to build a computer is
one of his funniest:

http://www.dansdata.com/sbs3.htm

 

[VWWall]

I think this article answers most questions:
http://www.dansdata.com/spcables.htm

For those not "down under", Jameco has power splitters with 18 or 20 AWG
wire. They also have the Molex plugs and sockets if you really want to
roll your own. The 18 AWG wire splitter is $2.95, but you can save
$0.10 by getting the 20 AWG one! www.jameco.com

Cyberguys have them for $0.95, but don't state the wire size. A six inch
extension cable is $1.27. This may be what the OP needs plus the
splitters. www.cyberguys.com

Prices for these sorts of things vary all over the map. I've paid
almost half the price of a floppy drive for the power cable for it.
Making your own doesn't save much if you buy in small quantities.

Virg Wall

 

[Aaron S]

In order to get two wires into one Molex connector, those
power splitters use 22 AWG. Aaron S said "had to "daisy
chain" three and maybe soon four) of the HDDs off *one*
4 pin Molex power connector."

Hi Tom (and others), I am the original poster. In this post below I will
clarify my wiring and give some voltage readings. Please recall that my
question was to ask if my PSU is getting into trouble. I am amazed it works
at all but strangely it seems to work quite well.

Just to record all the details: Coming off a 250W PSU I have six hard
drives and a Duron 700 on a Via266-based mobo with 768MB 133MHz SDRAM. Plus
the usual extras like floppy, CD-RW, modest graphics card, PCI modem and IDE
adaptor card.

The PSU is a model FSP250-60GTV(PF) and the specs for this PSU are here:
http://snipurl.com/954t The PSU at this link is almost completely identical
except for mine has an active Power Factor control.

People here had suggested I get my digital multimeter out and measure
voltages at the power plugs rather than trust to what the software package
MotherBoard Monitor says. So I have done that.

First of all and for convenience I am copying the original data I posted of
what MBM reported:

---------------- START MBM DATA ------------
When the PC booted up and all the hardware devices were attached
then a screen in the BIOS showed that the nominal 12v voltage was
2.0 % down at 11.77 and the nominal 5v was 2.2% down at 4.89v.

When I booted into XP and using MBM I saw the figure for 12v went
up to 4.93v (1.4% down) and 12v went down to 11.81 (1.6% down).

Then I saw some other figures in MBM:
+3.3v was 3.31v

-12v was between -5.90v and -10.71v with an average of -7.74V.
This seems like a huge difference.

-5.0v was showing in MBM as 0.00v. Maybe is not used or does
not exist.

---------------- END MBM DATA ------------

NOTE: When I measured actual voltages with a meter I also re-checked the
MBM readings and saw that MBM was reporting 11.88v and 4.91v (also 3.31v).


My power leads are as follows: there are three main sets of peripheral power
leads which come out of the PSU. If you want to see all the gory details of
the manufacturer's diagram of the external wiring layout then there is this
one page PDF file: http://snipurl.com/9amg

My own diagram below is in fixed space font and uses the following notation:

4-wire leads coming out of PSU PSU-X, Y, Z
4-pin PSU power connectors MOLEX-A, B, C, D, E, F
extension power connectors E-MOLEX-1, 2, 3, 4
hard drives hard-drive-1, 2, 3, 4, 5, 6
voltage readings 12.00 & 5.00

The arrangement attached to MOLEX-B is a "W"-like extension which I have
added to the original wiring.



240mm 155mm 155mm
PSU-X=========MOLEX-A=============MOLEX-B==============SMALL-CONNECTOR-C
hard-drive-1 E-MOLEX-1 unused
12.01 & 4.95 || 11.98 & 4.91
| |
| |
140mm | | 140mm
| |
| |
E-MOLEX-2 E-MOLEX-3
hard-drive-3 hard-drive-4
11.98 & 4.91 11.98 & 4.91
|
|
| 140mm
|
E-MOLEX-4
hard-drive-5
11.98 & 4.91


240mm 155mm
PSU-Y=========MOLEX-D=============SMALL-CONNECTOR-E
hard-drive-5 floppy
12.02 & 4.97 12.01 & 4.97



240mm 155mm
PSU-Z=========MOLEX-F=============MOLEX-G
hard-drive-6 CD-RW
12.02 & 4.97 12.02 & 4.97



So, is my PSU is getting into trouble?

Why does it work at all because all this should be too much for a 250W PSU?

In fact, why does it work quite well?

 

[w_tom]

Those Molex pins, if I remember, are not speced and
therefore will not properly crimp two 18 AWG wires. Largest
sockets were only for one 18 AWG. Previously problems were
encountered even with two 22 AWG wires.

Have extensive experience with products where the assembly
house took the "screw the specs - it works for me" attitude.
Violating the manufacturer specs for wire size and number of
wires does cause numerous failures. A previous intermittent
was traced to a Molex connector that used two 22 AWG wires in
a Molex socket rated for one. The solution was to convert the
whole defective crimp into a conductive solder lump.

Two 18 AWG wires can be pushed into that square hole? Maybe
if wire insulation is trimmed back to be outside the white
nylon Molex body. But that crimp is designed to crimp a bare
wire AND to mechanically clamp onto wire insulation (strain
relief). The later required so that movement does not cause
connector failure. But two 18 AWG wires cannot be held by
their insulation - meaning intermittent failure. Again, all
this from distant memory. What do I remember the most? All
those failures due to a 'screw the specs' attitude.

 

[VWWall]

Hi Tom (and others), I am the original poster. In this post below I will
clarify my wiring and give some voltage readings. Please recall that my
question was to ask if my PSU is getting into trouble. I am amazed it works
at all but strangely it seems to work quite well.

Just to record all the details: Coming off a 250W PSU I have six hard
drives and a Duron 700 on a Via266-based mobo with 768MB 133MHz SDRAM. Plus
the usual extras like floppy, CD-RW, modest graphics card, PCI modem and IDE
adaptor card.

The PSU is a model FSP250-60GTV(PF) and the specs for this PSU are here:
http://snipurl.com/954t The PSU at this link is almost completely identical
except for mine has an active Power Factor control.

People here had suggested I get my digital multimeter out and measure
voltages at the power plugs rather than trust to what the software package
MotherBoard Monitor says. So I have done that.

First of all and for convenience I am copying the original data I posted of
what MBM reported:

---------------- START MBM DATA ------------
When the PC booted up and all the hardware devices were attached
then a screen in the BIOS showed that the nominal 12v voltage was
2.0 % down at 11.77 and the nominal 5v was 2.2% down at 4.89v.

When I booted into XP and using MBM I saw the figure for 12v went
up to 4.93v (1.4% down) and 12v went down to 11.81 (1.6% down).

Then I saw some other figures in MBM:
+3.3v was 3.31v

-12v was between -5.90v and -10.71v with an average of -7.74V.
This seems like a huge difference.

-5.0v was showing in MBM as 0.00v. Maybe is not used or does
not exist.

---------------- END MBM DATA ------------

NOTE: When I measured actual voltages with a meter I also re-checked the
MBM readings and saw that MBM was reporting 11.88v and 4.91v (also 3.31v).


My power leads are as follows: there are three main sets of peripheral power
leads which come out of the PSU. If you want to see all the gory details of
the manufacturer's diagram of the external wiring layout then there is this
one page PDF file: http://snipurl.com/9amg

My own diagram below is in fixed space font and uses the following notation:

4-wire leads coming out of PSU PSU-X, Y, Z
4-pin PSU power connectors MOLEX-A, B, C, D, E, F
extension power connectors E-MOLEX-1, 2, 3, 4
hard drives hard-drive-1, 2, 3, 4, 5, 6
voltage readings 12.00 & 5.00

The arrangement attached to MOLEX-B is a "W"-like extension which I have
added to the original wiring.



240mm 155mm 155mm
PSU-X=========MOLEX-A=============MOLEX-B==============SMALL-CONNECTOR-C
hard-drive-1 E-MOLEX-1 unused
12.01 & 4.95 || 11.98 & 4.91
| |
| |
140mm | | 140mm
| |
| |
E-MOLEX-2 E-MOLEX-3
hard-drive-3 hard-drive-4
11.98 & 4.91 11.98 & 4.91
|
|
| 140mm
|
E-MOLEX-4
hard-drive-5
11.98 & 4.91


240mm 155mm
PSU-Y=========MOLEX-D=============SMALL-CONNECTOR-E
hard-drive-5 floppy
12.02 & 4.97 12.01 & 4.97



240mm 155mm
PSU-Z=========MOLEX-F=============MOLEX-G
hard-drive-6 CD-RW
12.02 & 4.97 12.02 & 4.97



So, is my PSU is getting into trouble?

Why does it work at all because all this should be too much for a 250W PSU?

In fact, why does it work quite well?

It looks like your wiring is doing fine. If the supply is truly 250W,
you don't have any real power hungry devices. The high performance CPUs
and video cards are the power users. Hard drives need only ~2-3 A at
12V. I think the spec you cite has the wrong number,(2.25 A), for the
"normal" 12 V supply, with 13 A shown for maximum. I don't think the MB
requires a lot of 12 V power.

You do show a slightly higher voltage drop at the end of the "split"
connections, but well within specs. Of course you can't easily measure
the 12 volt line at the time the HDs all spin up, when they take maximum
current. If they all start OK, I wouldn't worry.


Virg Wall

 

[w_tom]

In a world of clones, we don't have a clue what the actual
wattage is for a supply. Did they provide output wattage or
did they list what the supply might consume on input at
maximum? In short, clone power supplies tend to do what we
should call 'lie'. IOW the amperage ratings for each DC
output is far more informative.

Motherboard voltage monitor is completely useless until you
have calibrated it with the 3.5 digit multimeter.
Furthermore, those voltages as provided are questionable. But
then, until we have multimeter readings where connector
attaches to motherboard, then we really have little idea what
voltages really are.

My guess from these readings is that when you run the system
with every peripheral in operation, then the voltage will drop
significantly. Therefore PSU is undersized. However is
readings stay low but steady, then ripple (internal filter
components are in the process of failing) might be indicated.

With the voltmeter, you may also want to measure voltages
(both red and black leads make connection) at where Molex
connector attaches to peripheral. Then make same measurement
with black lead moved back at where power supply connect to
motherboard (red lead still measures at peripheral). This
would eliminate extender cables as a reasons for problems.

Just a few things to first look at - to quickly eliminate
some of the more obvious reasons for intermittent failures.

BTW, don't believe all this hype about wattage. Most every
computer should work just fine with 250 watt supplies.
However, I made an important point that many (who recommend
supplies) do not understand. More important is the actual
amperage for each voltage. So instead, the naive use the Tim
Allen solution of "More Power" rather than first learn basic
principles. At 250 watts, that supply may be sufficient. But
those multimeter voltage readings both on power up AND when
all peripherals are being accessed would be more informative.

Actual readings are essential to confirm theoretical
(manufacturer's claimed) numbers. This necessary because you
are designing a unique system - using components that we have
no previous knowledge of when used together.

Hi Tom (and others), I am the original poster. In this post below I will
clarify my wiring and give some voltage readings. Please recall that my
question was to ask if my PSU is getting into trouble. I am amazed it works
at all but strangely it seems to work quite well.

Just to record all the details: Coming off a 250W PSU I have six hard
drives and a Duron 700 on a Via266-based mobo with 768MB 133MHz SDRAM. Plus
the usual extras like floppy, CD-RW, modest graphics card, PCI modem and IDE
adaptor card.

The PSU is a model FSP250-60GTV(PF) and the specs for this PSU are here:
http://snipurl.com/954t The PSU at this link is almost completely identical
except for mine has an active Power Factor control.

People here had suggested I get my digital multimeter out and measure
voltages at the power plugs rather than trust to what the software package
MotherBoard Monitor says. So I have done that.

First of all and for convenience I am copying the original data I posted of
what MBM reported:

---------------- START MBM DATA ------------
When the PC booted up and all the hardware devices were attached
then a screen in the BIOS showed that the nominal 12v voltage was
2.0 % down at 11.77 and the nominal 5v was 2.2% down at 4.89v.

When I booted into XP and using MBM I saw the figure for 12v went
up to 4.93v (1.4% down) and 12v went down to 11.81 (1.6% down).

Then I saw some other figures in MBM:
+3.3v was 3.31v

-12v was between -5.90v and -10.71v with an average of -7.74V.
This seems like a huge difference.

-5.0v was showing in MBM as 0.00v. Maybe is not used or does
not exist.

---------------- END MBM DATA ------------

NOTE: When I measured actual voltages with a meter I also re-checked the
MBM readings and saw that MBM was reporting 11.88v and 4.91v (also 3.31v).

My power leads are as follows: there are three main sets of peripheral power
leads which come out of the PSU. If you want to see all the gory details of
the manufacturer's diagram of the external wiring layout then there is this
one page PDF file: http://snipurl.com/9amg

My own diagram below is in fixed space font and uses the following notation:

4-wire leads coming out of PSU PSU-X, Y, Z
4-pin PSU power connectors MOLEX-A, B, C, D, E, F
extension power connectors E-MOLEX-1, 2, 3, 4
hard drives hard-drive-1, 2, 3, 4, 5, 6
voltage readings 12.00 & 5.00

The arrangement attached to MOLEX-B is a "W"-like extension which I have
added to the original wiring.

240mm 155mm 155mm
PSU-X=========MOLEX-A=============MOLEX-B==============SMALL-CONNECTOR-C
hard-drive-1 E-MOLEX-1 unused
12.01 & 4.95 || 11.98 & 4.91
| |
| |
140mm | | 140mm
| |
| |
E-MOLEX-2 E-MOLEX-3
hard-drive-3 hard-drive-4
11.98 & 4.91 11.98 & 4.91
|
|
| 140mm
|
E-MOLEX-4
hard-drive-5
11.98 & 4.91

240mm 155mm
PSU-Y=========MOLEX-D=============SMALL-CONNECTOR-E
hard-drive-5 floppy
12.02 & 4.97 12.01 & 4.97

240mm 155mm
PSU-Z=========MOLEX-F=============MOLEX-G
hard-drive-6 CD-RW
12.02 & 4.97 12.02 & 4.97

So, is my PSU is getting into trouble?

Why does it work at all because all this should be too much for a 250W PSU?

In fact, why does it work quite well?

news://news.nni.com/ciuvuo%24n6b%241%40atlantis.news.tpi.pl

 

[David Maynard]

Those Molex pins, if I remember, are not speced and
therefore will not properly crimp two 18 AWG wires. Largest
sockets were only for one 18 AWG. Previously problems were
encountered even with two 22 AWG wires.

Oh horse hockey. Those pins are spec'd for up to 14 guage.

http://www.molex.com/cgi-bin/bv/mol...nel=Products&chanName=Series&pageTitle=Series

If anyone want to know what the real rating is, instead of all this bad
memory 'guessing', it's

14 AWG - 10 amps
16 AWG - 10 amps
18 AWG - 8 amps
22 AWG - 6 amps

Prior source page:

http://www.molex.com/cgi-bin/bv/mol...nel=Products&chanName=Series&pageTitle=Series

There are two sizes, one for wire crimp in the range 14 to 20 AWG and the
seond wire crimp in the range 18 to 22 AWG.

The cross sectional area of 2 18 AWG wires is within the cross sectional
area of a single 14 AWG. I.E. They fit.

http://www.molex.com/cgi-bin/bv/mol...nel=Products&chanName=Series&pageTitle=Series

The 'family' for the disk power connector is here

http://www.molex.com/cgi-bin/bv/mol...x&chanName=family&pageTitle=+Selection+Matrix

 

[VWWall]

14 AWG - 10 amps
16 AWG - 10 amps
18 AWG - 8 amps
22 AWG - 6 amps

Prior source page:

http://www.molex.com/cgi-bin/bv/mol...nel=Products&chanName=Series&pageTitle=Series


There are two sizes, one for wire crimp in the range 14 to 20 AWG and
the seond wire crimp in the range 18 to 22 AWG.

The cross sectional area of 2 18 AWG wires is within the cross sectional
area of a single 14 AWG. I.E. They fit.

http://www.molex.com/cgi-bin/bv/mol...nel=Products&chanName=Series&pageTitle=Series


The 'family' for the disk power connector is here

http://www.molex.com/cgi-bin/bv/mol...x&chanName=family&pageTitle=+Selection+Matrix

Your links go only to:

http://www.molex.com/cgi-bin/bv/molex/index_login.jsp

I didn't take the time to check the details, but you just confirm my
opinion. I do note that they are rated for only 25 connect-disconnect
cycles. That's probably more than most of us will give them! Or maybe
that's when you yank them apart by the wires! ;-)
The ones I have with two 18 AWG wires per pin seem to hold the wires
firmly enough by the insulation.
See my post of 5:20 PM today.

Virg Wall

 

[Steve Wolfe]

In order to get two wires into one Molex connector, those

power splitters use 22 AWG. Aaron S said "had to "daisy
chain" three
(and maybe soon four) of the HDDs off *one* 4 pin Molex power
connector."

And that's still probably alright. HDD's don't use much power at all.
Maybe a 15K SCSI drive, but not an IDE drive.

steve

 

[David Maynard]

Your links go only to:

http://www.molex.com/cgi-bin/bv/molex/index_login.jsp

Woops. I guess the links are my 'session' only. Sorry about that.

Easy enough to find though. They're right there on the first page over on
the right under 'industries and applications' - disk drive connectors.
Then, on the page that goes to: 'disk drive power connector' (3'd one
down). Then use the Selection Matrix on the left from the summary page. The
individual crimp terminals and shells are shown there.

 

[Aaron S]

It looks like your wiring is doing fine. If the supply is
truly 250W, you don't have any real power hungry devices. The
high performance CPUs and video cards are the power users.
Hard drives need only ~2-3 A at 12V.

The data I have seen are these:

http://www6.tomshardware.com/howto/20040122/power_supplies-04.html
http://www6.tomshardware.com/howto/20021021/powersupplies-02.html
http://www.pcpowerandcooling.com/maxpc/index_cases.htm
http://takaman.jp/D/?english

tables 3 4 & 5 on pages 6, 7 & 8 of the excellent PDF (6 MB):
http://www.amd.com/us-
en/assets/content_type/white_papers_and_tech_docs/26003.pdf

My graphics card is reckoned not to consume more than 4W max at load
(GeForce2 MX).

I think the spec you cite
has the wrong number,(2.25 A), for the "normal" 12 V supply,
with 13 A shown for maximum. I don't think the MB requires a
lot of 12 V power.

The PSU data at http://snipurl.com/954t says only that 2.25A might be normal
but it says the PSU can deliver up to 13A.

On the 5v rail my PSU sticker show a max of 27A (rather than the lower 21.8A
value next to it).

You do show a slightly higher voltage drop at the end of the
"split" connections, but well within specs. Of course you
can't easily measure the 12 volt line at the time the HDs all
spin up, when they take maximum current. If they all start OK,
I wouldn't worry.

That may be my next measurement.

 

[Aaron S]

In a world of clones, we don't have a clue what the actual
wattage is for a supply. Did they provide output wattage or
did they list what the supply might consume on input at
maximum? In short, clone power supplies tend to do what we
should call 'lie'. IOW the amperage ratings for each DC
output is far more informative.

My PSU is not a clone. It is made by the manufacturer whose link I gave.
The only reason I sad it was almost identical to the model I linked to was
that the manufacturer (for some reason!) does not have my model listed.

Motherboard voltage monitor is completely useless until you
have calibrated it with the 3.5 digit multimeter.
Furthermore, those voltages as provided are questionable. But
then, until we have multimeter readings where connector
attaches to motherboard, then we really have little idea what
voltages really are.

My guess from these readings is that when you run the system
with every peripheral in operation, then the voltage will drop
significantly. Therefore PSU is undersized. However is
readings stay low but steady, then ripple (internal filter
components are in the process of failing) might be indicated.

With the voltmeter, you may also want to measure voltages
(both red and black leads make connection) at where Molex
connector attaches to peripheral. Then make same measurement
with black lead moved back at where power supply connect to
motherboard (red lead still measures at peripheral). This
would eliminate extender cables as a reasons for problems.

I am not clear what you mean when you say "Then make same measurement
with black lead moved back at where power supply connect to motherboard".
Do you mean I shoud break into the black lead near the PSU and use that as
one of the two lines to measure voltage across?

In a way I think I kind of did that except that I check some of my readings
by taking the earth off the system case. This showed no difference at all
to the readings taken at the Molex connector. I also checke to see if the
chassis and the black lead at the MOLEX had any potential across them and I
found there was absoultely none.

BTW at the time I also measured the voltage at back of the the 20-pin
motherboard MOLEX and got 12.00v and 4.97v. Also I read -11.60v, -4.91v and
3.37v.

Just a few things to first look at - to quickly eliminate
some of the more obvious reasons for intermittent failures.

BTW, don't believe all this hype about wattage. Most every
computer should work just fine with 250 watt supplies.

Yes, I have heard that stated somewhere. Wasn't there even an article about
it which cause a bit of a debate? I never heard the debate nor knew what it
concluded, if anything.

However, I made an important point that many (who recommend
supplies) do not understand. More important is the actual
amperage for each voltage. So instead, the naive use the Tim
Allen solution of "More Power" rather than first learn basic
principles. At 250 watts, that supply may be sufficient. But
those multimeter voltage readings both on power up AND when
all peripherals are being accessed would be more informative.

Yes, good idea. That's much harder for me to arrange though.

 

[w_tom]

http://www.molex.com/cgi-bin/bv/molex/index_login.jsp

I didn't take the time to check the details, but you just confirm my
opinion. I do note that they are rated for only 25 connect-disconnect
cycles. That's probably more than most of us will give them! Or maybe
that's when you yank them apart by the wires! ;-)

Now there was a significant limitation either I overlooked
or completely forgot.

 

[w_tom]

The 20 pin Molex connector voltage readings suggest Power
Supply is not in a slowly failing mode. If those are readings
when computer is accessing multiple peripherals
simultaneously, then power supply is quite sufficient and
adequate.

We want measurements on power to ground at each disk drive
when disk drive is being accessed (ie running a disk
diagnostic). This to confirm cables are indeed acceptable so
that good voltages even appear at every peripheral. IOW first
measure with both meter probes where Molex connector mates to
disk drive. Then take a second measurement from + power lead
on disk drive Molex to black ground wire at 20 pin 'PSU to
motherboard' Molex connector. The first measurement should be
well within spec - in agreement with voltages previously taken
at 20 pin Molex connector. The second measurement must have
almost no change - should show no difference from first
measurement. A higher voltage when taking second measurement
would suggest one of the two ground wires (black) on those
extender cords is defective.

Most critical for any device that connects to multiple
voltages - the ground (black) wires must always be the best
connection. Any problem with a ground wire can cause damage
to 'multiple voltage' electronics. IOW the second
measurement (from disk + power wire to PSU - wire) confirms
wire connection that can most likely cause intermittent
failure or, even worse, permanent transistor damage.

Technically, the chassis ground is not same as motherboard
digital ground. If no system problems, then measuring from
chassis ground should result in no difference. However,
confirm voltage does not change when using 20 pin Molex ground
and using chassis ground before using chassis ground for any
future measurements. If any measurement does not make sense,
then go back to using 20 pin Molex ground.

Provided first were numbers that say power supply is OK.
Second measurements say that correct voltage is even being
provided to peripherals - that power wires (especially ground)
are constructed sufficient.

 

[kony]

Now there was a significant limitation either I overlooked
or completely forgot.

IMO, the largest failing is when disconnecting and the plug
is not pulled *perfectly* straight out, rather at an angle
or wiggled, causing the female plug insert to enlarge,
visually notable by a wider seam... it can be bent inward
again.

 

[Franc Zabkar]

BTW at the time I also measured the voltage at back of the the 20-pin
motherboard MOLEX and got 12.00v and 4.97v. Also I read -11.60v, -4.91v and
3.37v.

The -12V reading makes much more sense now. Previously you stated that
it "was between -5.90V and -10.71V with an average of -7.74V." You
still haven't given us the BIOS readings for this voltage, though.

Have you checked whether your motherboard is on MBM's supported list?
See http://mbm.livewiredev.com/mobolist.html


- Franc Zabkar