psu issue - Page 2

DigitalChaos · 05-17-2002, 12:01 PM

PCP&C turbocool 475
and here is the corner oposite the fan.
im not sure what exactly is being cooled here, but it is using pretty small heatsinks... on copper one on back.. and then an aluminum plate

DigitalChaos · 05-17-2002, 12:22 PM

Quote:

Originally posted by Cyco-Dude
remove the heatsinks compleatly. attach the MOSFETS directly to the waterblock. thats how bladerunner did it.

bladerunners site, ZERO FAN ZONE.

not only do the heatsinks provide structural support for other pieces of the psu, but its too damn cluttered to be soldering

gmat · 05-17-2002, 01:54 PM

Quote:

Originally posted by bigben2k
5% is very standard in most circuits.

Exact.

Quote:

MOSFET do get an input, but it's not exactly a positive half-sine. I seem to remember (from theory) that some of those big caps are there to partly fill those voltage drops (to 0V) at 120 Hz. (60 Hx * 2)

The MOSFET is actually there to remove the 120 Hz "ripple" in an otherwise DC signal, and to level the output to selected voltages.

What's going out of the primary loop (switching) is nearly a half-sine. You're right, a first filter prevents it to drop to zero - for power purposes only, the regulators wouldnt hold otherwise.
If you look at what's given to mosfets in a PC psu you'll going to get a thrill

They give a really sh*tty input to regulators. Giving a better signal at input costs a lot.
The mosfets still output good DC but the result is they'll heat up a lot.
Virtually you can give any voltage to a regulator it will always output it's rated voltage, with heat building up with differences * power drain.

The RC i mentioned was after the mosfets, to smooth out even more the output (mainly removing 100 or 120hz remains) and to prevent power spikes due to wires inductance.

Quote:

-snip-
The R+C is actually a filter. (Correct me if I'm wrong here gmat, 'cause it's still early!). The combination can be calculated to filter out a specific frequency range. It is very commonly used in speakers, in the form of a cross-over, to separate the low frequency signals from the high-frequency signals, in order for the woffer/tweeter to get its rightful part of a sound.

Extacly. The RLC combo (Resistance Inductance Capacitance) is a filter. With it you can build a complex impedance to do anything you want (filtering, oscillating, etc...). Inductance is produced by coils.

Quote:

I'll have to read up on switched supplies, because I haven't dealt with them in 10 years... [/b]

You have one in your own PC

just crack it open and plug in your oscilloscope you're out for a thrill.
I tried that. Now i even wonder how PCs can work with such crappy PSUs.

gmat · 05-17-2002, 02:01 PM

Quote:

Originally posted by DigitalChaos
PCP&C tubrocool 475
and here is the corner oposite the fan.
im not sure what exactly is being cooled here, but it is using pretty small heatsinks... on copper one on back.. and then an aluminum plate

Looks more like EMI shielding for a big coil. I wouldnt worry about that.
The part you need to cool down is on the other side, where the big heatsinks reside.
One more time, extending mosfets wont cost you those 1% ratings at all. First you've got wires sticking out of your PSU to plug your hardware... are those throwing out the reliability ?

If you're not sure about your soldering skills try to find someone who's got a *electronics* (that means, small) iron and a steady hand.

DigitalChaos · 05-17-2002, 04:25 PM

Quote:

Originally posted by gmat

Looks more like EMI shielding for a big coil. I wouldnt worry about that.
The part you need to cool down is on the other side, where the big heatsinks reside.
One more time, extending mosfets wont cost you those 1% ratings at all. First you've got wires sticking out of your PSU to plug your hardware... are those throwing out the reliability ?

If you're not sure about your soldering skills try to find someone who's got a *electronics* (that means, small) iron and a steady hand.

actually.. it may be a EMI shield AND heatsink.. cause on one side you can see the black thing sandwiched between the aluminum and copper... on the other side of the aluminum there is a mosfet attached. (probably for the standby rails or something)

ive done several Vmods before.. which are way more intricate... its just that there are so many little riser boards n stuff that were put on AFTER the mosfets were installed. it would probably take me several hours just to dismantle the damn thing =)

your explanation about how the cables comming out of the psu are about the same as extending the mosfets makes good sense to me. id just rather use the existing heatsink network goin on in there to attach a block to.

gmat · 05-18-2002, 09:02 AM

Quote:

Originally posted by DigitalChaos

id just rather use the existing heatsink network goin on in there to attach a block to.

Now that makes sense

Just post pics for us when it's done...

BladeRunner · 05-19-2002, 12:26 PM

DigitalChaos has a good point as I also had an Antec 400 watt PSU but like the PC power & cooling he has it was far too cluttered internally to use for my project design / solution. The only issue with using the heatsinks that are already in the PSU is often they are live mains voltage (or at least one of them is). This would have the undesired effect of making the coolant live if a metal block was attached to it, (causing an earth trip or fuse to blow in UK mains systems.

I was thinking about a PC power and cooling 600watt or 650watt for future water cooling. Can't remember the model now but it's a double sized unit. not sure if it's server / duel CPU only. anyone have one of these?

DigitalChaos · 05-19-2002, 01:51 PM

hrm... good point about the heatsinks being live... i would imagine i could just connect one of those to ground and see what happens...

most of the mosfets apear to have a thermal pad between them and the heatsink which would stop the heatsink from being live...

also, i could just attach my water blocks with a big thermal pad to prevent electricity from flowing.

BladeRunner · 05-19-2002, 04:35 PM

I'd favour getting a Mulitimeter, (and someone that has the know-how to use it if you don't) to check the sinks for "liveness" rather than deliberately shorting it out.

The fact the mosfets have mica shims doesn't mean the sink is not live. The mosfets in the enermax were all shimmed but the sinks were live. I could find no definite reason why they needed to be live. Now the sinks are not there, and the block in the PSU is earthed with the mosfets attached to it also fully isolated with the same mica shims without any obvious downside. more about this in the PSU cooling build on my site.

gmat · 05-20-2002, 04:39 AM

I gave a thought to this heatsink problem, and the only reason i see for them to be live is EMI noise reduction (unless they're actually connected to the rest of the circuit).
Put 2 opposing sinewave signals in 2 opposing blocks of metal - in the middle you've got a flat line.
In this case maybe the sinks are live to compensate the unavoidable mains noise you've got on mosfets output.
The only way to work this out is with a good oscilloscope.

DigitalChaos · 05-20-2002, 07:57 AM

no, the reason they are live, is the design of most mosfets... the metal plate that attaches to the heatsink is live.

gmat · 05-20-2002, 08:38 AM

Wow never saw this kind of mosfet regulator.... usually the backplate is connected to GND meaning it's at 0V. Like BR said pick up a multimeter and tell us what's going on

BladeRunner · 05-20-2002, 01:35 PM

gmat

I think it must be something like that as there has to be a reason. Like you say the metal back part of the Mosfets on my PSU is now & was before the water-cooling at 0v, it's the heatsink itself that was live mains voltage. The sink also wasn't being used as a bridge joining two parts of the PCB together either, as I checked for this.

The mica shims prevented the back part of the mosfet touching the sink directly, (as far as electrical conductivity is concerned),

I don't pretend to understand fully how a PSU works, but the water block is now earthed and the mosfets mounted in the same way with the same mica shims. It appears to work ok without the original heatsinks "liveness". Like you say a scope and someone that knows how to use it is what's really required

ct100 · 05-21-2002, 04:47 PM

In some PSUs the Sink is used as a bridge between 2 Points of the PCB... I modified already 2 Enermax 550W where I had to soldier a bypass under the PCB...

But other than that reason there is no sense in live sinks...

I modified many Enermax PSU...

Here are some pics:

Heavy modified watercooled Enermax 550W:

Another One:

bigben2k · 05-21-2002, 04:52 PM

Nice!!!!!

BladeRunner · 05-21-2002, 06:24 PM

ct100

Interesting, as with my 550watt the power rail origin was the same point, so the bridging it would not have made any difference.

I notice that you have a copper "Cover" near that coil in mine that gets hot. Is it in contact with the coil?, I was intending to do something similar but wasn't sure if it might short something.

It can be seen in this image, it's the coil shown by the shorter Pink arrow, (click for large image).

If this is possible I could easily run a copper plate to it from the side of the waterblock shown here.

ct100 · 05-21-2002, 07:37 PM

Enermax PSU's are very different, even if they have the same model no... so its possible you have another version of this PSU...

2 of my Enermax needed this bypass... without it, they could not provide 5V Standby Voltage...

This copper "Cover" is made from a 1,2mm thick copper plate, and does his job very good in cooling the Coil... no more 70°C+ now

I used AS2 under this "Cover" to make better contact to the Coil

BladeRunner · 05-22-2002, 02:59 PM

Quote:

Originally posted by ct100

This copper "Cover" is made from a 1,2mm thick copper plate, and does his job very good in cooling the Coil... no more 70°C+ now

I used AS2 under this "Cover" to make better contact to the Coil

Does this not affect the coil. I would have though that it is a coil for a reason, so by connecting it with a copper plate and artic silver will make it less of a coil if you get my meaning, sort of shorting out one part to another.

ct100 · 05-22-2002, 03:13 PM

It does not short the coil!

The copper wires of the coil arent pure copper... they have a very thin layer of shellac arount the copper... so they arent conductive at the surface...

AS2 isn't conductive, too...

so you dont have to worry about shorting something by covering the coil with a copper plate

gmat · 05-23-2002, 02:27 AM

The layer on the coil is a thin varnish. Any scratch on it and the copper will be exposed. For safety reasons i'd put an extra mica shim between the coil and the copper plate.....

DigitalChaos · 05-28-2002, 10:42 PM

well i took a volt meter to this thing.... and this is what i got.

the heatsink on the left has no voltage with the casing (ground)

for the heatsink on the right i got 120AC or 55DC with the negative lead on the heatsink and the positive lead on casing (ground)

these readings are for both the aluminum and copper sides of each set of heatsinks

DigitalChaos · 05-28-2002, 10:43 PM

now this little heatsink set scares me... i got this reading on both the copper and aluminum

750AC or 350DC with the positive lead on the heatsink and the negative lead on the casing (ground)

DigitalChaos · 05-28-2002, 10:45 PM

soooo.... looking at the way things are in there... im thinking about putting a quiet 80mm in there as a replacement.. and then put a hole in the top for a 120 or 92 mm to pull more air out.

Brad · 05-29-2002, 02:42 AM

I dare you to touch it

gmat · 05-29-2002, 02:51 AM

Thats 2 fans, and 1 more than stock

You just lose the purpose of modding that PSU. (reading your original post again...) If you want to get rid of noise, the solution is clearly not adding more fans...
BTW the readings you want are the AC ones. If the heatsink is acting like a bridge just disconnect it and replace it with a wire on the opposite side.
Again i doubt that copper part IS a heatsink. Looks REALLY like a EMI shield. It's hot because of magneto-resistive effects produced by the nearby coil.
And yeah there are some very high voltage parts in a PSU. Thats usually written on a big sticker

(i know i know, us geeks just pop it open disregarding red & yellow hazard signs...)

05-19-2002, 12:26 PM	#32
BladeRunner Cooling Savant Join Date: Jul 2001 Location: Chesterfield Uk Posts: 459	DigitalChaos has a good point as I also had an Antec 400 watt PSU but like the PC power & cooling he has it was far too cluttered internally to use for my project design / solution. The only issue with using the heatsinks that are already in the PSU is often they are live mains voltage (or at least one of them is). This would have the undesired effect of making the coolant live if a metal block was attached to it, (causing an earth trip or fuse to blow in UK mains systems. I was thinking about a PC power and cooling 600watt or 650watt for future water cooling. Can't remember the model now but it's a double sized unit. not sure if it's server / duel CPU only. anyone have one of these? __________________ Zero Fan Zone

05-19-2002, 04:35 PM	#34
BladeRunner Cooling Savant Join Date: Jul 2001 Location: Chesterfield Uk Posts: 459	I'd favour getting a Mulitimeter, (and someone that has the know-how to use it if you don't) to check the sinks for "liveness" rather than deliberately shorting it out. The fact the mosfets have mica shims doesn't mean the sink is not live. The mosfets in the enermax were all shimmed but the sinks were live. I could find no definite reason why they needed to be live. Now the sinks are not there, and the block in the PSU is earthed with the mosfets attached to it also fully isolated with the same mica shims without any obvious downside. more about this in the PSU cooling build on my site. __________________ Zero Fan Zone

05-20-2002, 01:35 PM	#38
BladeRunner Cooling Savant Join Date: Jul 2001 Location: Chesterfield Uk Posts: 459	gmat I think it must be something like that as there has to be a reason. Like you say the metal back part of the Mosfets on my PSU is now & was before the water-cooling at 0v, it's the heatsink itself that was live mains voltage. The sink also wasn't being used as a bridge joining two parts of the PCB together either, as I checked for this. The mica shims prevented the back part of the mosfet touching the sink directly, (as far as electrical conductivity is concerned), I don't pretend to understand fully how a PSU works, but the water block is now earthed and the mosfets mounted in the same way with the same mica shims. It appears to work ok without the original heatsinks "liveness". Like you say a scope and someone that knows how to use it is what's really required __________________ Zero Fan Zone Last edited by BladeRunner; 05-20-2002 at 05:45 PM.

05-21-2002, 04:52 PM	#40
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	Nice!!!!! __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

05-21-2002, 06:24 PM	#41
BladeRunner Cooling Savant Join Date: Jul 2001 Location: Chesterfield Uk Posts: 459	ct100 Interesting, as with my 550watt the power rail origin was the same point, so the bridging it would not have made any difference. I notice that you have a copper "Cover" near that coil in mine that gets hot. Is it in contact with the coil?, I was intending to do something similar but wasn't sure if it might short something. It can be seen in this image, it's the coil shown by the shorter Pink arrow, (click for large image). If this is possible I could easily run a copper plate to it from the side of the waterblock shown here. __________________ Zero Fan Zone

05-19-2002, 01:51 PM	#33
DigitalChaos Cooling Savant Join Date: Jan 2002 Location: Waukesha, Wi Posts: 698	hrm... good point about the heatsinks being live... i would imagine i could just connect one of those to ground and see what happens... most of the mosfets apear to have a thermal pad between them and the heatsink which would stop the heatsink from being live... also, i could just attach my water blocks with a big thermal pad to prevent electricity from flowing.

05-20-2002, 04:39 AM	#35
gmat Thermophile Join Date: Mar 2001 Location: France Posts: 1,221	I gave a thought to this heatsink problem, and the only reason i see for them to be live is EMI noise reduction (unless they're actually connected to the rest of the circuit). Put 2 opposing sinewave signals in 2 opposing blocks of metal - in the middle you've got a flat line. In this case maybe the sinks are live to compensate the unavoidable mains noise you've got on mosfets output. The only way to work this out is with a good oscilloscope.

05-20-2002, 07:57 AM	#36
DigitalChaos Cooling Savant Join Date: Jan 2002 Location: Waukesha, Wi Posts: 698	no, the reason they are live, is the design of most mosfets... the metal plate that attaches to the heatsink is live.

05-20-2002, 08:38 AM	#37
gmat Thermophile Join Date: Mar 2001 Location: France Posts: 1,221	Wow never saw this kind of mosfet regulator.... usually the backplate is connected to GND meaning it's at 0V. Like BR said pick up a multimeter and tell us what's going on

05-21-2002, 04:47 PM	#39
ct100 Cooling Neophyte Join Date: May 2002 Location: Germany Posts: 5	In some PSUs the Sink is used as a bridge between 2 Points of the PCB... I modified already 2 Enermax 550W where I had to soldier a bypass under the PCB... But other than that reason there is no sense in live sinks... I modified many Enermax PSU... Here are some pics: Heavy modified watercooled Enermax 550W: Another One:

05-21-2002, 07:37 PM	#42
ct100 Cooling Neophyte Join Date: May 2002 Location: Germany Posts: 5	Enermax PSU's are very different, even if they have the same model no... so its possible you have another version of this PSU... 2 of my Enermax needed this bypass... without it, they could not provide 5V Standby Voltage... This copper "Cover" is made from a 1,2mm thick copper plate, and does his job very good in cooling the Coil... no more 70°C+ now I used AS2 under this "Cover" to make better contact to the Coil

05-22-2002, 03:13 PM	#44
ct100 Cooling Neophyte Join Date: May 2002 Location: Germany Posts: 5	It does not short the coil! The copper wires of the coil arent pure copper... they have a very thin layer of shellac arount the copper... so they arent conductive at the surface... AS2 isn't conductive, too... so you dont have to worry about shorting something by covering the coil with a copper plate

05-23-2002, 02:27 AM	#45
gmat Thermophile Join Date: Mar 2001 Location: France Posts: 1,221	The layer on the coil is a thin varnish. Any scratch on it and the copper will be exposed. For safety reasons i'd put an extra mica shim between the coil and the copper plate.....

05-28-2002, 10:45 PM	#48
DigitalChaos Cooling Savant Join Date: Jan 2002 Location: Waukesha, Wi Posts: 698	soooo.... looking at the way things are in there... im thinking about putting a quiet 80mm in there as a replacement.. and then put a hole in the top for a 120 or 92 mm to pull more air out.

05-29-2002, 02:42 AM	#49
Brad Thermophile Join Date: Oct 2001 Location: Nuu Zeeelin Posts: 3,175	I dare you to touch it __________________ 2x P3 1100's at 1400, Abit VP6, 2x Corsair 256mb PC150 sticks, 20gb 'cuda ATA-III, 2x 40gb 'cuda ATA-IV in raid 0. 20" Trinitron. No fans 2x 2400+ at 2288mhz (16.0 x 143), Iwill MPX2, 2x Kingmax PC-3200 256mb sticks, 4x 20gb 60gxp in Raid 5 on a Promise SX6000. Asus Ti4200 320/630. Cooled by Water

05-29-2002, 02:51 AM	#50
gmat Thermophile Join Date: Mar 2001 Location: France Posts: 1,221	Thats 2 fans, and 1 more than stock You just lose the purpose of modding that PSU. (reading your original post again...) If you want to get rid of noise, the solution is clearly not adding more fans... BTW the readings you want are the AC ones. If the heatsink is acting like a bridge just disconnect it and replace it with a wire on the opposite side. Again i doubt that copper part IS a heatsink. Looks REALLY like a EMI shield. It's hot because of magneto-resistive effects produced by the nearby coil. And yeah there are some very high voltage parts in a PSU. Thats usually written on a big sticker (i know i know, us geeks just pop it open disregarding red & yellow hazard signs...)

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)