Ed Cryer <
ed@somewhere.in.the.uk> wrote:
I'll have a go with cleaning the PSU as you suggest. Enough dust gets
into the CPU heatsink (which I clean regularly), but I've never thought
to clean the PSU.
The box doesn't boot. I switch it on and the PSU whistles; and that's all.
My modifications are;
Large SSD.
DVD-writer.
Memory 8GB.
All been in situ for several years, and working excellently.
Tis possible the dust buildup inside the PSU is the culprit, but also
possible the PSU has been damaged from overheating. Typically the
intake air into the PSU is pre-heated (CPU, memory, drives, mobo, etc).
The PSU isn't 100% efficient, so it generates heat, too. ATX PSUs are typically box shaped, but the ITX PSU you have is elongated. That setup
uses a mini-ITX motherboard and the ITX PSU to reduce size.
Comparison of mobo sizes:
https://upload.wikimedia.org/wikipedia/commons/thumb/1/1e/Comparison_ATX_%C2%B5ATX_DTX_ITX_mini-DTX.svg/330px-Comparison_ATX_%C2%B5ATX_DTX_ITX_mini-DTX.svg.png
https://en.wikipedia.org/wiki/Mini_ATX
After removing the side panel from the case, get a flashlight to watch
the PSU fan. Press the power button, and check if the fan spins or not,
if it jiggles but doesn't spin (siezed fan), or if it never moves. With
the PSU case dismantled, you can flick a fan blade with your finger. If
it halts very quickly (doesn't spin more than once), its bearing is
shot. A good fan won't spin many times when you flick its blade with
your finger, but it should spin. You can also sometimes feel a bad
bearing by slowly rotating the fan using a finger. With a bad bearing,
you might feel a bit of resistance at the bad spot in rotation. If the
fan is bad, likely you can replace the fan, but may need some solder
w/rosin core and heatshrink tubing to snip the wires to the fan, and
splice in the new fan. If you use electrical tape instead of heatshrink
tubing to cover the wire splices, use a cable/zip tie around the tape
since it will unwrap over time.
From some pics of that PSU, there is a voltage select switch (red) at
the back near where the power cord plugs into the PSU. Even if it seems
to be at the correct position to match the input voltage, I would flip
it to the other position, and then back. Make sure it is solidly into
the indent for the correct position, and flipping the switch also cleans
its contacts.
After dusting out the mobo, and CPU and heatsink, and dusting the PSU by blowing through the case grills, if you get to dismantling the PSU to
dust better inside of it, take a look inside to see if you can see a
fuse. Often, instead of a fuse holder, there is just a couple of clips soldered into a PCB into which a fuse is pushed. However, sometimes the
fuse is just soldered onto the PCB (it has leads soldered onto the ends
of the fuse, and those leads are soldered into the PCB).
Does this PSU have a power brick in its power cord? That is, does the
power cord that plugs into the wall outlet go to a brick, and then
another cord runs from the brick to the PSU? If so, and if the output
voltage is labeled on the brick, check the output voltage with a
multimeter. Make sure the cord that plugs into brick is fully seated,
like unplug and replug that cord into the brick. Some folks will do a
lot of work without checking power is reaching the PSU, like they kick
the plug that is under their desk, or the brick hangs from the desk
instead of sitting on the floor. I hate power bricks. I fabricate a
set of wire ties to eliminate the cords walking out of the brick: two
loose on each cord on each end of the brick, loop a cable tie between
the two loose ones on the cords, on each side of the brick, and tighten
down the ties as a bail to pull on the cords into the sockets in the
brick. Typically the plugs are too tight into the brick to use hot-melt
glue. Super glue doesn't work on nylon in a somewhat flex connection.
Bricks are often found with laptops to reduce size of the laptop (but
then the size difference is consumed by the brick which is external),
but yours is a desktop.
While we've been focused on the PSU, it's possible the problem is on the
mobo. There is logic on the mobo that tells the PSU if it powers up or
not. With the old AT-style PSUs, the power switch was hardwired into
the PSU (it was a switch on the front of the case that was wired to the
PSU). With ATX-style PSUs, the power switch goes to the mobo, and logic
on the mobo tells the PSU when to power up. While the power cord is
attached to the PSU, and even with the computer powered down, the PSU
still generates a 5V STBY (5 VDC standby output voltage) in the 24-pin
harness from PSU to mobo. That powers the 3VDC "power-on" logic on the
mobo. As a test, with the 24-pin PSU harness connected to the mobo, you
could short the Pwr-On lead (green wire) to ground. That lead dropping
to zero volts is what tells the PSU to power up; else, it floats high.
https://www.lifewire.com/thmb/se_V_Mh6qrGZHtG5hGu3YQFVnRA=/1500x0/filters:no_upscale():max_bytes(150000):strip_icc()/atx-24-pin-12v-power-supply-pinout-2624578-bd809482596447b7afa7c63232719560.png
Pin 15 is PS-ON, and shorting to ground will turn on the PSU. This line
must be kept low to keep the PSU turned on; else, the only output
voltage from the PSU is the 5VSB line (pin 9) to power the on-board mobo
logic for this line. If the PS-ON line doesn't go low, the PSU won't
come on. When I do this test, I get a sewing needle to jab into the
backside of the 24-pin connector to contact the female pin in there, and
use a jumper cable (alligator clips on each end) to connect the needle
to a ground point. You could use another sewing needle in any of the
GND leads (pins 3, 5, 7, 15, 17, 18, 19, 24) in the 24-pin connector.
If you don't have a jumper wire, you could try jamming sewing needles
into pin 16 (PS-ON) and a GND lead next door (pin 15 or 17), and
crossing the needles to touch each other to make the short across pins
(PS-ON to GND). Once PS-ON is shorted to GND, the PSU fan should spin.
Tis possible the system is overloaded. PSUs wan in capacity over time.
No matter what is their rated power output, that lessens over time.
What the PSU was capable of handling early on may become too much later.
I forgot where I read the tested reduction, but recall it was about 5%
capacity loss per year for typical-grade hardware, less for higher
quality hardware. Expect 20% capacity loss over its lifetime (which
cheap PSUs having a shorter lifetime). As the components inside the PSU
age, they become less efficient which means less delivered power and
more generated heat. A reduction in input voltage to the PSU causes
less efficient conversion, and more heat. Dust buildup reduces thermal transfer away from components in the PSU decreasing efficiency which
generates even more heat. Running a PSU at its real maximum capacity
(which is often less than its rated capacity, especially for cheap PSUs)
will reduce efficiency (which is not linear to the load). Running a
sustained total load of 75% of its rating will reduce capacity. Most degradation is from high temperatures, and why you need to dust out your computer, heat sinks, memory modules, video card, and the PSU. I've
seen folks try using estimators on how much power is needed from the PSU
based on what loads it will supply and efficiency of the PSU, and then
buy a PSU that barely exceeds that power draw. To hell with that. I
figure out what might be the power draw if everything I could attach to
the PSU were populated, like when later upgrading the system, and then
double that for what size of PSU to get. If the computations show a
400W PSU is sufficient, I get an 800W PSU, or bigger depending on the
depth of my pocket at the time. Getting a PSU "to size" is a mistake
that you'll make once. Guess what happens to your power computations
when you later decide to add a 200W video card? You'll spend again to
get a better PSU. Pre-builts are designed to their specifications, not
on how you might upgrade them later. They sell by spec, not quality.
Also, be aware that the ratings on PSU is often for non-sustained
maximums; i.e., they over rate their capacity. Some PSUs are under
rated, but many are over rated. The protections are often over rated,
too. Some PSUs will cut out when overloaded, but some will literally
burn up. When you open the PSU, and after dusting it out, use your nose
to determine if the PSU burned up. In pre-builts, typically the PSU is
the minimum needed for the factory config, and not much leeway for an
enlarged power load (and their power factor curve has a much sharper
drop at the max load end so efficiency drops a lot so more heat), and
longevity isn't that long, like about 5 years, or just enough to get
past the warranty period. There are a lot of crappy PSUs out there, and
many won't meet or exceed their ratings shown on a label on the PSU.
Another test would be to disconnect the 24-pin PSU connector from the
mobo, and short PS-ON (pin 16) in the connector to GND (pin 15 or 17) in
the connector (the pins will be exposed to make contacting them easier). However, some PSUs will still not power on unless there is some power
draw. That is, even with PS-ON shorted to ground, the PSU wants to see something drawing power from it. So, leave a drive connected to the PSU
wiring harness. The drive needs to power up to draw power, so an
optical drive won't do. Leave a spinner or SSD attached. After
shorting PS-ON to ground, the PSU fan should spin.
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)
