Bad PSU ?

[Graeme]

I have a PC which has a Winpower VV400 PSU. Is the following indicative of a
crap PSU?
When I launch any application, or get the CPU to do some work, then there is
an audible change to the case fan speed. I have investigated slightly
further, and have measured the 12V rail to the fans (the fans are supplied
via the connection to the HD). This, when idle, sits at 12.29V. When the CPU
starts to do something, then this drops to 12.05V, and results in a lowering
of fan speed.
The system;
P4 3GHz HT
Gigabyte 8IPE1000 mobo
2x512M RAM
120G 7200 HD
DVD
CD-RW
nVidia FX5700Ultra
NIC
Modem
2x80mm case fans @ 2000rpm
WinXP
Zalman super flower CPU cooler @2500rpm

 

[kony]

I have a PC which has a Winpower VV400 PSU. Is the following indicative of a
crap PSU?

Well, that "winpower" part already indicates crap.

When I launch any application, or get the CPU to do some work, then there is
an audible change to the case fan speed. I have investigated slightly
further, and have measured the 12V rail to the fans (the fans are supplied
via the connection to the HD). This, when idle, sits at 12.29V. When the CPU
starts to do something, then this drops to 12.05V, and results in a lowering
of fan speed.

Generally a power supply with such a high 12V reading, means it's 3V +5V
rail is inadequate for the system... the power supply raises the
inadequate rail, and the 12V rises also, as a result. The fact that the
12V rail later dropped closer to spec'd level could be taken as a sign
that the system is pretty much at the max capacity of the power supply.

The system;
P4 3GHz HT
Gigabyte 8IPE1000 mobo
2x512M RAM
120G 7200 HD
DVD
CD-RW
nVidia FX5700Ultra
NIC
Modem
2x80mm case fans @ 2000rpm
WinXP
Zalman super flower CPU cooler @2500rpm

The curious part is that your system isn't particularly demanding on the
3V, 5V rail, beyond the FX5700 video card. I don't know what safety
features are incorporated into the Winpower unit, as that may be more
significant than the voltage variance on the 12V rail, _IF_ the system
remains stable. On the other hand, it's likely you're seeing more 3V, 5V
ripple to the board & other components, so in the long run it might be
best to replace the power supply regardless, especially if you expect to
get a number of years out of the system rather than trashing it after a
couple dozen months.

 

[Graeme]

Well, that "winpower" part already indicates crap.


Generally a power supply with such a high 12V reading, means it's 3V +5V
rail is inadequate for the system... the power supply raises the
inadequate rail, and the 12V rises also, as a result. The fact that the
12V rail later dropped closer to spec'd level could be taken as a sign
that the system is pretty much at the max capacity of the power supply.



The curious part is that your system isn't particularly demanding on the
3V, 5V rail, beyond the FX5700 video card. I don't know what safety
features are incorporated into the Winpower unit, as that may be more
significant than the voltage variance on the 12V rail, _IF_ the system
remains stable. On the other hand, it's likely you're seeing more 3V, 5V
ripple to the board & other components, so in the long run it might be
best to replace the power supply regardless, especially if you expect to
get a number of years out of the system rather than trashing it after a
couple dozen months.

The PSU came with the case, termed P4 compatible! I hate it when my IT dept
just buys cheap. I like the case, I'm not so sure about the PSU. And the PSU
fan is loud.
I am seriously thinking about changing the PSU just because of its noise
level. If it is also not going to last long, or cause other long term
reliability issues, then it's going to go.
The system would normally be replaced in 2-3 years, but would go to another
home afterwards. So it has to still be working.
I had looked at some low-noise PSUs from QTechnology, but, to be honest, I'm
not sure what Wattage is required. They have a 350W unit (3.3V@28A, 5V@30A,
12V@18A) which uses a 120mm fan and appears to be very quiet.

 

[kony]

The PSU came with the case, termed P4 compatible! I hate it when my IT dept
just buys cheap. I like the case, I'm not so sure about the PSU. And the PSU
fan is loud.
I am seriously thinking about changing the PSU just because of its noise
level. If it is also not going to last long, or cause other long term
reliability issues, then it's going to go.
The system would normally be replaced in 2-3 years, but would go to another
home afterwards. So it has to still be working.
I had looked at some low-noise PSUs from QTechnology, but, to be honest, I'm
not sure what Wattage is required. They have a 350W unit (3.3V@28A, 5V@30A,
12V@18A) which uses a 120mm fan and appears to be very quiet.

Power supplies labeled as "Q-Tech" are also junk, but there may be an
entirely different line of power supply called "QTechnology", that I am
not familiar with. _IF_ the spec'd, labeled capacity is accurate, that
should be enough for the system you described. Fortron also makes a 350W
unit with 120mm fan that's pretty quiet, usually priced at around $40.

Due to the way your system's power usage is distributed, you could
probably get by with a good 300W unit, but for greater 12V rail I'd go
with at least a 350W, or 420W if there are plans for further upgrades,
beyond another drive or two.

 

[Graeme]

Power supplies labeled as "Q-Tech" are also junk, but there may be an
entirely different line of power supply called "QTechnology", that I am
not familiar with. _IF_ the spec'd, labeled capacity is accurate, that
should be enough for the system you described. Fortron also makes a 350W
unit with 120mm fan that's pretty quiet, usually priced at around $40.

Due to the way your system's power usage is distributed, you could
probably get by with a good 300W unit, but for greater 12V rail I'd go
with at least a 350W, or 420W if there are plans for further upgrades,
beyond another drive or two.

QTechnology do try to stress that they are in no way connected with Q-Tech.
Which, judging by your comments, is no bad thing. QTechnology are one of the
3 PSU makes that quietpc.com/uk sell. Quietpc seem to carry the Fortron line
in the US and the QTechnology one in the UK.

 

[ric]

I have a PC which has a Winpower VV400 PSU. Is the following indicative of a
crap PSU?
When I launch any application, or get the CPU to do some work, then there is
an audible change to the case fan speed. I have investigated slightly
further, and have measured the 12V rail to the fans (the fans are supplied
via the connection to the HD). This, when idle, sits at 12.29V. When the CPU
starts to do something, then this drops to 12.05V, and results in a lowering
of fan speed.

Well, you don't say where or with what you are measuring the +12v. The
difference could be due to anything from a poorly regulated PSU, to
excessive drop in some MB traces. In any case, the change in the +12v
is only 2%, and should not cause an audible change in fan speed.

I would measure the +12v and its regulation with a DVM at the PSU and at
the fan connection. If the +12v is really that poorly regulated at the
PSU (with only CPU activity change), perhaps you would enjoy a better PSU.
[or change to a quieter fan.]

 

[Graeme]

I have a PC which has a Winpower VV400 PSU. Is the following indicative of a
crap PSU?
When I launch any application, or get the CPU to do some work, then there is
an audible change to the case fan speed. I have investigated slightly
further, and have measured the 12V rail to the fans (the fans are supplied
via the connection to the HD). This, when idle, sits at 12.29V. When the CPU
starts to do something, then this drops to 12.05V, and results in a lowering
of fan speed.

Well, you don't say where or with what you are measuring the +12v. The
difference could be due to anything from a poorly regulated PSU, to
excessive drop in some MB traces. In any case, the change in the +12v
is only 2%, and should not cause an audible change in fan speed.

I would measure the +12v and its regulation with a DVM at the PSU and at
the fan connection. If the +12v is really that poorly regulated at the
PSU (with only CPU activity change), perhaps you would enjoy a better PSU.
[or change to a quieter fan.]

The12V was measured with a DVM at the fan connector (which is taken from the
HD connector - you know the type, it's inline with the power to the HD). And
no, it's not the HD activity that's causing the voltage drop. I've got my
fan speed monitor open and another CPU hungry application already open and
waiting for me to press the 'go' button. When idle, the fan is going at 1896
rpm. When I press the go button the fan speed drops to 1854. This change in
speed is audible which was the indication that something may not be quite
right. The hardware monitor application (the same one that gives me the fan
speeds) also shows a drop in 12V (from 12.16 to 11.84). I will scope out all
of the power rails when I get the chance to borrow a suitable instrument (in
search of any ripple). Other than the 12V dipping, the system is stable.
I think this will end up as a PSU change, and I'll get a couple of quiet
case fans too. Thanks for the reply.

 

[w_tom]

Those numbers suggest a bad (or undersized) supply. However
I would be further curious: to also compare those voltage
reading to readings where wires connect to PC board in power
supply - to learn if they used insufficient gauge wire or bad
soldering. Just curiosity.

 

[larrymoencurly]

I have a PC which has a Winpower VV400 PSU. Is the following
indicative of a crap PSU?

When I launch any application, or get the CPU to do some work,
then there is an audible change to the case fan speed. I have
investigated slightly further, and have measured the 12V rail
to the fans (the fans are supplied via the connection to the HD).
This, when idle, sits at 12.29V. When the CPU starts to do
something, then this drops to 12.05V, and results in a lowering
of fan speed.

I couldn't find anything about Winpower, but any safety certification
number printed on it may help you find the actual manufacturer at
www.ul.com, www.csa.com, or one of the other testing organizations.

I found that even tiny changes in RPM resulted in audible changes from
the fan, and the +12V fluctuated slightly with changes in load even in
the best PSUs.

 

[ric]

I found that even tiny changes in RPM resulted in audible changes from
the fan, and the +12V fluctuated slightly with changes in load even in
the best PSUs.

I've never built a system or had a system where any fan's noise fluctuated
during use. Then again, I've used only PC Power & Cooling PSUs, so perhaps
that *is* the problem.

 

[kony]

I've never built a system or had a system where any fan's noise fluctuated
during use. Then again, I've used only PC Power & Cooling PSUs, so perhaps
that *is* the problem.

Fan fluctuation "can" have nothing to do with the power supply, if the
motherboard is poorly designed it will drop the 12V.

PCPower & Cooling do make great power supplies, but even they will just as
easily drop the 12V rail in the same situation seen with generics, when
the actual capacity of the unit is lower than needed... big difference is
that you can trust the label on a PC P&C unit, know what you're buying.

 

[~misfit~]

Fan fluctuation "can" have nothing to do with the power supply, if the
motherboard is poorly designed it will drop the 12V.

PCPower & Cooling do make great power supplies, but even they will
just as easily drop the 12V rail in the same situation seen with
generics, when the actual capacity of the unit is lower than
needed... big difference is that you can trust the label on a PC P&C
unit, know what you're buying.

I'm using a Soltek SL-75FRN2-L mobo, it has the four connector 12v input
from the PSU. I believe it's a good board, I've used quite a few of them
with no problems. Running an Athlon 1800+ Tbred B at 2.1GHz @ 1.825vcore (in
BIOS), 200MHx FSB, 2 x 256MB PC3200 RAM in dual channel mode with a ti4200,
one HDD and two optical drives.

My PSU however is crap, real crap. It came with the case, is supposed to be
400W. I don't have a DVM but Mobo Monitor 5 reports my 3.3v as varying
between 2.96 and 3.10v for this session. Those readings are typical for this
PSU/system. My 12v rail has varied between 11.86 and 12.46v, again using
MBM5's high/low page. When I start SETI and I hear my fans slow down. When I
bought the case/PSU the retailer told me the PSU is crap. It only cost me
$NZ45 (around $US30).

If I had the money I'd certainly get a new PSU. However I'm still struggling
to pay off the mobo. Surprisingly this system is Prime95 stable and is solid
as a rock. If I had built it for someone else I wouldn't have let it out the
door with those readings, however as it's mine and I'm broke I'm stuck with
it until I can get a better PSU. vcore for this session according to MBM5
has varied between 1.81 and 1.86v

Just thought I'd mention it. My fans pitch varies audiably as the CPU load
does, it's actually quite annoying. The three 12v case fans are connected to
the molex connectors. IMO it's definitely due to a bad PSU.

 

[kony]

I'm using a Soltek SL-75FRN2-L mobo, it has the four connector 12v input
from the PSU. I believe it's a good board, I've used quite a few of them
with no problems. Running an Athlon 1800+ Tbred B at 2.1GHz @ 1.825vcore (in
BIOS), 200MHx FSB, 2 x 256MB PC3200 RAM in dual channel mode with a ti4200,
one HDD and two optical drives.

It may be a good board... i have no experience with those, but some boards
have higher margins for CPU power delivery than others... Yours may not be
as suited to >2GHz, > 1.8V, as others are.

My PSU however is crap, real crap. It came with the case, is supposed to be
400W. I don't have a DVM but Mobo Monitor 5 reports my 3.3v as varying
between 2.96 and 3.10v for this session. Those readings are typical for this
PSU/system. My 12v rail has varied between 11.86 and 12.46v, again using
MBM5's high/low page. When I start SETI and I hear my fans slow down. When I
bought the case/PSU the retailer told me the PSU is crap. It only cost me
$NZ45 (around $US30).

Rightabout now I'm wondering if your CPU would be stable at lower Vcore if
you had a better power supply. "Most" Tbreds don't o'c any better past
1.8V, usually not even needing that much unless ~2.4GHz, due to core heat
density, in other words that's rightabout where the heat buildup causes
diminishing returns unless it's exotically cooled, ie - water, freon,
peltier, etc.

If I had the money I'd certainly get a new PSU. However I'm still struggling
to pay off the mobo. Surprisingly this system is Prime95 stable and is solid
as a rock. If I had built it for someone else I wouldn't have let it out the
door with those readings, however as it's mine and I'm broke I'm stuck with
it until I can get a better PSU. vcore for this session according to MBM5
has varied between 1.81 and 1.86v

Just keep in mind that Vcore variations roughly translate into more wear
on the capacitors... it would surprise me if yours aren't a little on the
warm side, though being in a row as they are may help. I play around with
parts all the time, seeing how far I can push them, but for the few
systems I rely on day-in, day-out, I take whatever measures necessary to
be sure the caps aren't more than very slightly warm, at most.

Just thought I'd mention it. My fans pitch varies audiably as the CPU load
does, it's actually quite annoying. The three 12v case fans are connected to
the molex connectors. IMO it's definitely due to a bad PSU.

Do you need three fans that're loud enough to be heard when voltage is at
high-state? Oh, is it summer there now? LOL, still cold here.

 

[~misfit~]

It may be a good board... i have no experience with those, but some
boards have higher margins for CPU power delivery than others...
Yours may not be as suited to >2GHz, > 1.8V, as others are.

It's spec'ed to XP3200 and I've run several 2500+'s in these boards at 3200+
spec (albeit with better PSUs and not quite in the 1.8vcore range) and they
seem fine.

Rightabout now I'm wondering if your CPU would be stable at lower
Vcore if you had a better power supply. "Most" Tbreds don't o'c any
better past
1.8V, usually not even needing that much unless ~2.4GHz, due to core
heat density, in other words that's rightabout where the heat buildup
causes diminishing returns unless it's exotically cooled, ie - water,
freon, peltier, etc.

I believe you are right on that one. I think that if the vcore didn't
fluctuate so much I could get by with 1.80v. It took 1.825v with this PSU to
get it Prime95 stable overnight.

Just keep in mind that Vcore variations roughly translate into more
wear on the capacitors... it would surprise me if yours aren't a
little on the warm side, though being in a row as they are may help.
I play around with parts all the time, seeing how far I can push
them, but for the few systems I rely on day-in, day-out, I take
whatever measures necessary to be sure the caps aren't more than very
slightly warm, at most.

I haven't felt the caps, I will now. Thanks for the tip.

Do you need three fans that're loud enough to be heard when voltage
is at high-state? Oh, is it summer there now? LOL, still cold here.

LOL, yeah. Ambient gets up to 30°C plus. One of those case fans is being
used to augment the HSF, I'm using the standard AMD HSF with an 80mm fan in
the side ducted to supply cool, outside-case air to it. That's the kicker
about being on a limited income, you have to make do with what you can
afford. It dropped the core temp 10°C though, better than a mate got by
spending big bucks on an all-copper 500gm HS.

One thing, the PSU fan seems quite gutless, it's not pushing much air out
the back and what it is pushing is pretty hot. What do you think about me
putting a more powerful fan in the PSU? (I've been inside PSUs before and
know the risks). Do you think it's likely to help? It'll only cost me time,
I have a variety of fans and a slodering iron here.

Thanks for your comments Dave.

 

[kony]

I believe you are right on that one. I think that if the vcore didn't
fluctuate so much I could get by with 1.80v. It took 1.825v with this PSU to
get it Prime95 stable overnight.

.... could just be the particular CPU though, which seems likely as you're
keeping it cool enough and enough voltage, yet 2.1GHz isn't all that high
for a B-Step... then again, around that range it's pretty hard to notice a
couple hundred MHz one way or the other.

I haven't felt the caps, I will now. Thanks for the tip.

I know that some cap manufacturers (for example Rubycon) spec that their
caps' lifespan is cut in half per 10C temp rise. Fortunately they
determine end-of-life pretty strickly, with a lot of function left in the
parts, else boards would be dying left-and-right.

One thing, the PSU fan seems quite gutless, it's not pushing much air out
the back and what it is pushing is pretty hot. What do you think about me
putting a more powerful fan in the PSU? (I've been inside PSUs before and
know the risks). Do you think it's likely to help? It'll only cost me time,
I have a variety of fans and a slodering iron here.

Thanks for your comments Dave.

Sure, sounds like a good idea, especially if that's the original fan, a
cheap sleeve-bearing model that might sieze. Sometimes in order to keep
case noise-emissions low, I'll leave a slow fan on the back and suppliment
it with a pusher on the inside, either hanging down overtop the board
(when I'm comfortable that I'll leave a power supply in a given case and
that case has an extra inch between the power supply and board), or behind
the power supply, when there isn't room below it or with power supplies
like Sparkle or Thermaltake, that have heatsinks with large right-angle
protrusions that would block any airflow blowing straight up from the
bottom. That requires a bit of cutting though, maybe more work that it's
worth for a generic power supply... the simple fan swap may be the best
alternative in your situation, though I'll throw a few more ideas out
there just for the heck of it.

Another option might be if there's a thermal sensor inside, controlling
the fan speed. If so you might be able to slightly bend it toward the
heatsink, throw a blob of epoxy or (other heat-conduction material) on it
to make a bridge to a 'sink), or even solder it's leads together,
effectively shorting it to give the max fan RPM, or a bit fancier is a
rheostat or POT substituted for that sensor, usually around 10K Ohm range,
will allow manual control... just be sure you're looking at the fan
control sensor, not a thermal cutoff sensor for the whole PSU.

Sometimes the thermal sensor is on it's own tiny PCB, obvious by following
the fan header which plugs into it. The same modifications can apply, or
when it's that separate PCB, it's typically daisey-chained to a 2nd fan
header on the main PCB, so the fan can just be unpluggged from the
fan-control PCB and plugged into the unregulated fan header. Then again
I'm thinking of moderately decent power supplies, it's hard to speculate
about what may or may not be found in a generic-comes-with-case PSU.

 

[Graeme]

[ snipped crap PSU rant!]

The12V was measured with a DVM at the fan connector (which is taken from the
HD connector - you know the type, it's inline with the power to the HD). And
no, it's not the HD activity that's causing the voltage drop. I've got my
fan speed monitor open and another CPU hungry application already open and
waiting for me to press the 'go' button. When idle, the fan is going at 1896
rpm. When I press the go button the fan speed drops to 1854. This change in
speed is audible which was the indication that something may not be quite
right. The hardware monitor application (the same one that gives me the fan
speeds) also shows a drop in 12V (from 12.16 to 11.84). I will scope out all
of the power rails when I get the chance to borrow a suitable instrument (in
search of any ripple). Other than the 12V dipping, the system is stable.
I think this will end up as a PSU change, and I'll get a couple of quiet
case fans too. Thanks for the reply.

I've scoped out the 12, 5 and 3.3 supplies. Both of the 3.3 and 5V supplies
look ok. There is no strange behaviour here. The 12V is another matter
entirely. Even under no load (CPU idle) there are dips in this supply. They
only last 15ms (max), occur at irregular intervals, and go down from 12.2 to
12.0V. This is measured using a Tek TDS2024 scope.
Obviously these little dips don't cause the fan noise to change, but I'm
pretty sure that they shouldn't happen. I guess the 12V rail gets the
biggest 'beating' as it's used to supply the CPU with its power (via that
4-pin connector).

 

[~misfit~]

... could just be the particular CPU though, which seems likely as
you're keeping it cool enough and enough voltage, yet 2.1GHz isn't
all that high for a B-Step... then again, around that range it's
pretty hard to notice a couple hundred MHz one way or the other.

Here's the numbers off it:

AXDA1800 DUTC3 9417156270093 JIUHB 0311TPMW



Unfortunately it isn't one of the 1.5v CPUs. I can get it to run at nearly
(but not quite) 2.2GHz prime95 stable but that's a bit inelegant, I like to
keep my FSB at 200MHz. I would like to have got more out of it but, like you
say, a couple of hundred MHz isn't that noticable, except for SETI.


The caps sem fairly cool, just warm, and they're positioned in the exhaust
air-flow from the CPU.

I know that some cap manufacturers (for example Rubycon) spec that
their caps' lifespan is cut in half per 10C temp rise. Fortunately
they determine end-of-life pretty strickly, with a lot of function
left in the parts, else boards would be dying left-and-right.



Sure, sounds like a good idea, especially if that's the original fan,
a cheap sleeve-bearing model that might sieze. Sometimes in order to
keep case noise-emissions low, I'll leave a slow fan on the back and
suppliment it with a pusher on the inside, either hanging down
overtop the board (when I'm comfortable that I'll leave a power
supply in a given case and that case has an extra inch between the
power supply and board), or behind the power supply, when there isn't
room below it or with power supplies like Sparkle or Thermaltake,
that have heatsinks with large right-angle protrusions that would
block any airflow blowing straight up from the bottom. That requires
a bit of cutting though, maybe more work that it's worth for a
generic power supply... the simple fan swap may be the best
alternative in your situation, though I'll throw a few more ideas
out there just for the heck of it.

Another option might be if there's a thermal sensor inside,
controlling the fan speed. If so you might be able to slightly bend
it toward the heatsink, throw a blob of epoxy or (other
heat-conduction material) on it to make a bridge to a 'sink), or even
solder it's leads together, effectively shorting it to give the max
fan RPM, or a bit fancier is a rheostat or POT substituted for that
sensor, usually around 10K Ohm range, will allow manual control...
just be sure you're looking at the fan control sensor, not a thermal
cutoff sensor for the whole PSU.

Sometimes the thermal sensor is on it's own tiny PCB, obvious by
following the fan header which plugs into it. The same modifications
can apply, or when it's that separate PCB, it's typically
daisey-chained to a 2nd fan header on the main PCB, so the fan can
just be unpluggged from the fan-control PCB and plugged into the
unregulated fan header. Then again I'm thinking of moderately decent
power supplies, it's hard to speculate about what may or may not be
found in a generic-comes-with-case PSU.

Thanks for the suggestions Dave, I'll have alook inside the PSU a bit later
and see if I can locate any type of control circuitry for the fan before I
change it. As you pointed out though, in a PSU this cheap it probably
doesn't have anything fancy. I got the case and PSU for less than the price
of a budget 350W PSU alone, I guess you get what you pay for. My other
option when I bought it was to only go with one 256MB RAM stick and get a
better Case/PSU (tight budget) but I decided the RAM was the way to go.

I very much appreciate the benefit of your experience.

 

[w_tom]

Sounds like an electrolytic capacitor is failing. IOW you
have discovered a more serious problem before it happened.
The low voltage on digital meter was also a precursor to this
type of failure; enough that using the Oscope was an excellent
next diagnostic step.

Those dips can distort regulator feedback to assume voltages
that did not exist. As load increases, note how the Oscope
waveforms change. Compare that to what digital meter
reports. IOW learn how meter can be used to quickly identify
this type of failure; to avoid crashes before they happen.

 

[Graeme]

Would that be a cap in the PSU or on the mobo? Obviously it'd be easier to
change the PSU. How would I go about getting a replacement (only 2 months
old)? How on earth can you describe this 'fault' to a large distribution
company who only point you to manufacturers' web sites?! (And Winpower has
drawn a blank through my research. And there is no UL etc certification
written on the PSU. Only CE but no certificate number)

 

[Graeme]

I've just tested a nearly identical machine (same PSU and mobo), and its 12V
rail also exhibits these drops.
The only other performance machine I have access to is our dual Xeon unit
which is hardly ever used. I've just probed its 12V supply and it is rock
solid stable. I had to remove the PSU to find out who the manufacturer is -
and - it's a 460W Fortron unit. Me thinks (from what I've heard on the
group) I'll get one of these.