So I have a supply labeled "ATX Switching power supply MODEL:450W",
which is quite suspicious sounding already, but worse still is that
there doesn't seem to be anyone claiming to be the manufacturer.
There is a sticker labeled "200808023681 WP08023 Made In China",
which seems to be the most informative indication on the outside.
Input/output specs are listed however, and this is how they read:
----------------------------------------------------------
| VOLTAGE | CURRENT | FREQUENCY
V~ | 115V~ | 7.5A | 60Hz
INPUT | 230V~ | 4.5A | 50Hz
----------------------------------------------------------
___ |+5V|+3.3V|+12V|-5V |-12V|+5VSB|PS-ON |POL | COM
V--- |30A| 25A | 25A|0.6A|0.6A| 2A |REMOTE|P.G.|RETURN
OUTPUT |
----------------------------------------------------------
| 180W |300W|3.0W|7.2W| 10W | -- | -- | --
MAX | 430W | 20W |
| 450W |
----------------------------------------------------------
(that looked so much better in vi...)
Inside, what immediately catches my eye is the lack of input
filtering, and the jumper wires where the coils should be, and the
use of Canicon and Micon capacitors throughout. The PCB is labeled
YUELIN YLP-013 PCB REV:2.2 070829.
I got some Nichicon PL(M) and LQ(M) capacitors out of a blown Lucent
623W supply (from an SGI Octane -- replaced with Cherokee 747W), so
some of them are, how should we say, a bit overspec'ed for this
supply, but, everybody seems to claim "you can go up in volts", and
"you can go up in capacitance", so, that's what I'll be doing, and
if I have to do point-to-point wiring off the PCB, then so be it.
I salvaged a two 3300pF caps, and installed them in the spaces
labeled "CX1" and "CX2" on the input stage, as well as replacing the
jumper wires for "LF1", "J1", and "J2", with toroidal inductors
which I'd obtained from a blown "Wan Nien/Enhance Electronics"
supply. Also used was the input resistor, coded Green, Brown,
Yellow, Gold, which looks up to mean 560k Ohm, 5% tolerance, which I
fitted to the blank spot labeled "R1".
Because the AC line input from the C14 inlet was using such flimsy
wiring, I went ahead and upgraded the wiring, choosing to install
two short leads to the L(ine) and N(eutral) inputs on the PCB and
wire nut them to the wires from the inlet for easy removal (I would
have used a connector, but the original of either ATX unit didn't
have any, and sifting through my parts I only found a plug, and not
the onboard header, so wire nuts it is -- no way am I gonna solder
direct-to-the-plug like it was originally).
The original bridge rectifier, labeled KBL406, is a 4Amp unit, which
is right on the edge of the specifications: if you take 110V * 4Amp,
you get 440Watts, so obviously they intend you to use 115V * 4Amp,
in order to reach a "450Watts safe" 460Watt rated limit for that
part. A scrap supply I had used a KBU605, slightly better at 6Amps
rating, which, using their 115V * 6Amp formula yields 690Watts (660
if using 110V), and backing off from it by 10Watts like they did,
this particular component ought to be good for 680Watts of output
(supposing I'm properly guessing the way to rate/use these things).
Thankfully, it is a model with a hole in it, which allows for
heatsinking, whereas the original 4Amp part wouldn't have been as
easy to attach a heatsink to. I haven't tried this yet, but this
part is next on my list to do. Question... the board I sourced this
part in question from had burned/melted input switchers, as well as
a charred area around the board nearby (which includes the bridge
rectifier). Supposing this 6Amp bridge rectifier is still functioning,
is it safe for extended use? It doesn't seem melted like the
switchers, so I assume this is just like a "load test", and if it's
still working, then no issue.
I also have two international rectifier 30CPQ150's that I might use
as an ouput-stage upgrade.
At the moment, I've made a small perfboard-with-output-caps, and
wired the Nichicons to the output holes using stranded speaker wire,
which was thicker than any of the other wire I had available. I was
thinking about putting some high-frequency ceramic or such capacitors
on the output lines on the supply's PCB to "make up for" hosing the
electrolytics off-board, but I haven't done this yet.
The 8x output electrolytics I placed in have the values:
3.3V: 1000uF, 2200uF
12V: 1000uF, 2200uF
5V: 1000uF, 6800uF
-5V: 1000uF
-12V: 1000uF
Except for the 6800uF, which is rated 6.3V, the rest are rated for
35V, and all are Nichicon PL(M) series with 105 degree C temp ratings.
So right now, just the recap, input filter, fan connector, and AC-in
wire thickness upgrade has been done. The fan that is in it does work,
but I'll replace it before I'm finished. I tested with a single hard
disk for load, and measured the following with a digital multimeter:
3.3V output: 3.42-3.43
5V output: 5.20-5.21
12V output: 12.39-12.40
-12V output: -11.69 - -11.70
-5V output: -4.92 - -4.93
I did take some pictures along the way to where I'm at now, so I'll
be posting those soon, just have to sift out what's visible, and what's
fuzzy, etc.
Feel free to critisize this.. it's my first attempt at a power supply rebuild,
and I'm especially interested in comments about off-boarding the output
filter caps like I've done. I know it does increase their impedence, but..
how much? Would some smaller caps right on the output lines overcome
this? Also, I know the "X caps" that I've placed are rated as "Y caps",
so not exactly correct.. but they are behind a fuse, so even if they fail
closed (correct model fails open), that will just pop the fuse. Isn't this
better than not having anything fitted there? I have a proper X2 cap,
which I was thinking to mount on the power inlet, because it is too big
to fit the PCB. If I do that, will it be a problem to leave the PCB as I've
got it now, or should I remove the "Y caps" from the "X cap places" if
I put a proper X cap on the power inlet.
Note... it seems like a user here has done something similar before,
so I'll leave this link here, so you know I'm aware of it (that's the thread
that I "needed to see pics of" and why I created this account:
https://www.badcaps.net/forum/showthread.php?t=7691
which is quite suspicious sounding already, but worse still is that
there doesn't seem to be anyone claiming to be the manufacturer.
There is a sticker labeled "200808023681 WP08023 Made In China",
which seems to be the most informative indication on the outside.
Input/output specs are listed however, and this is how they read:
----------------------------------------------------------
| VOLTAGE | CURRENT | FREQUENCY
V~ | 115V~ | 7.5A | 60Hz
INPUT | 230V~ | 4.5A | 50Hz
----------------------------------------------------------
___ |+5V|+3.3V|+12V|-5V |-12V|+5VSB|PS-ON |POL | COM
V--- |30A| 25A | 25A|0.6A|0.6A| 2A |REMOTE|P.G.|RETURN
OUTPUT |
----------------------------------------------------------
| 180W |300W|3.0W|7.2W| 10W | -- | -- | --
MAX | 430W | 20W |
| 450W |
----------------------------------------------------------
(that looked so much better in vi...)
Inside, what immediately catches my eye is the lack of input
filtering, and the jumper wires where the coils should be, and the
use of Canicon and Micon capacitors throughout. The PCB is labeled
YUELIN YLP-013 PCB REV:2.2 070829.
I got some Nichicon PL(M) and LQ(M) capacitors out of a blown Lucent
623W supply (from an SGI Octane -- replaced with Cherokee 747W), so
some of them are, how should we say, a bit overspec'ed for this
supply, but, everybody seems to claim "you can go up in volts", and
"you can go up in capacitance", so, that's what I'll be doing, and
if I have to do point-to-point wiring off the PCB, then so be it.
I salvaged a two 3300pF caps, and installed them in the spaces
labeled "CX1" and "CX2" on the input stage, as well as replacing the
jumper wires for "LF1", "J1", and "J2", with toroidal inductors
which I'd obtained from a blown "Wan Nien/Enhance Electronics"
supply. Also used was the input resistor, coded Green, Brown,
Yellow, Gold, which looks up to mean 560k Ohm, 5% tolerance, which I
fitted to the blank spot labeled "R1".
Because the AC line input from the C14 inlet was using such flimsy
wiring, I went ahead and upgraded the wiring, choosing to install
two short leads to the L(ine) and N(eutral) inputs on the PCB and
wire nut them to the wires from the inlet for easy removal (I would
have used a connector, but the original of either ATX unit didn't
have any, and sifting through my parts I only found a plug, and not
the onboard header, so wire nuts it is -- no way am I gonna solder
direct-to-the-plug like it was originally).
The original bridge rectifier, labeled KBL406, is a 4Amp unit, which
is right on the edge of the specifications: if you take 110V * 4Amp,
you get 440Watts, so obviously they intend you to use 115V * 4Amp,
in order to reach a "450Watts safe" 460Watt rated limit for that
part. A scrap supply I had used a KBU605, slightly better at 6Amps
rating, which, using their 115V * 6Amp formula yields 690Watts (660
if using 110V), and backing off from it by 10Watts like they did,
this particular component ought to be good for 680Watts of output
(supposing I'm properly guessing the way to rate/use these things).
Thankfully, it is a model with a hole in it, which allows for
heatsinking, whereas the original 4Amp part wouldn't have been as
easy to attach a heatsink to. I haven't tried this yet, but this
part is next on my list to do. Question... the board I sourced this
part in question from had burned/melted input switchers, as well as
a charred area around the board nearby (which includes the bridge
rectifier). Supposing this 6Amp bridge rectifier is still functioning,
is it safe for extended use? It doesn't seem melted like the
switchers, so I assume this is just like a "load test", and if it's
still working, then no issue.
I also have two international rectifier 30CPQ150's that I might use
as an ouput-stage upgrade.
At the moment, I've made a small perfboard-with-output-caps, and
wired the Nichicons to the output holes using stranded speaker wire,
which was thicker than any of the other wire I had available. I was
thinking about putting some high-frequency ceramic or such capacitors
on the output lines on the supply's PCB to "make up for" hosing the
electrolytics off-board, but I haven't done this yet.
The 8x output electrolytics I placed in have the values:
3.3V: 1000uF, 2200uF
12V: 1000uF, 2200uF
5V: 1000uF, 6800uF
-5V: 1000uF
-12V: 1000uF
Except for the 6800uF, which is rated 6.3V, the rest are rated for
35V, and all are Nichicon PL(M) series with 105 degree C temp ratings.
So right now, just the recap, input filter, fan connector, and AC-in
wire thickness upgrade has been done. The fan that is in it does work,
but I'll replace it before I'm finished. I tested with a single hard
disk for load, and measured the following with a digital multimeter:
3.3V output: 3.42-3.43
5V output: 5.20-5.21
12V output: 12.39-12.40
-12V output: -11.69 - -11.70
-5V output: -4.92 - -4.93
I did take some pictures along the way to where I'm at now, so I'll
be posting those soon, just have to sift out what's visible, and what's
fuzzy, etc.
Feel free to critisize this.. it's my first attempt at a power supply rebuild,
and I'm especially interested in comments about off-boarding the output
filter caps like I've done. I know it does increase their impedence, but..
how much? Would some smaller caps right on the output lines overcome
this? Also, I know the "X caps" that I've placed are rated as "Y caps",
so not exactly correct.. but they are behind a fuse, so even if they fail
closed (correct model fails open), that will just pop the fuse. Isn't this
better than not having anything fitted there? I have a proper X2 cap,
which I was thinking to mount on the power inlet, because it is too big
to fit the PCB. If I do that, will it be a problem to leave the PCB as I've
got it now, or should I remove the "Y caps" from the "X cap places" if
I put a proper X cap on the power inlet.
Note... it seems like a user here has done something similar before,
so I'll leave this link here, so you know I'm aware of it (that's the thread
that I "needed to see pics of" and why I created this account:
https://www.badcaps.net/forum/showthread.php?t=7691
Comment