"Fanless" PSU mod.

PSUs: The source of DC power for all components in the PC & often a big noise source.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee, Devonavar

Post Reply
qstoffe
Posts: 84
Joined: Tue Apr 25, 2006 1:32 pm
Location: Sweden

"Fanless" PSU mod.

Post by qstoffe » Sun May 14, 2006 2:02 pm

I'm planing of building my own wooden case. In some recent testing I've found that the noise of fan-vibration is a real big deal for slow spinning fans. However, suspending a slow 120mm fan using rubber strings can make it virtually inaudible.

Anyway, this means I will have to remove the fan from the PSU and use an external "rubber-suspended" fan instead. In order to achieve sufficient airflow for the PSU (that will peak around 300-350W) I think I will remove the entire casing of the PSU. Instead I will build a bigger casing of wood to place the PSU components and a suspended fan in.

Is there anything specific I should think of when doing this that might leave me with a dangerously running PSU ? :!: :(
For example:

- How about grounding ? Do have to connect the PSU to some bigger metalplate in order for it to work (since I will remove the original metal casing completly) ?

- Airflow (high power components will probably make the PSU peak at 300-350W). Will a Nexus 120mm fan @ constant 6V be able to cool this relatively "open" PSU ?

- PSU-fan disconnected. Will a PSU start if I cut the wires to the original PSU fan ? Do most PSUs even have rpm status or are they controlled by voltage only ?

I first considered the fanless Phantom 350W. But after hearing that they are instable in high power systems that may peak at around 300W for longer times, I've pretty much discarded this option.

Finnally. I will probably buy a new PSU for this, so are there any you can recommend that may be a little more suited for this kind of mod ? It will have to be atleast 400W so that I can manage 1 or 2 upgrades without changing PSU.

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Re: "Fanless" PSU mod.

Post by dukla2000 » Sun May 14, 2006 11:39 pm

As long as you are competant with a screwdriver and soldering iron you should be OK. But of course take care and as soon as you remove the first screw the manufacturer's warranty evaporates.
qstoffe wrote:How about grounding ? Do have to connect the PSU to some bigger metalplate in order for it to work?
No - as soon as you open any psu you will see the ground from the mains cable connects only to the psu case. Although I have checked on a couple of PCs that chassis ground and DC ground are the same thing I never worked out precisely how come: it may well be that the psu mobo has earthing rings around its mounting screws that you should replicate.
qstoffe wrote:Airflow (high power components will probably make the PSU peak at 300-350W). Will a Nexus 120mm fan @ constant 6V be able to cool this relatively "open" PSU ?
Potentially yes. If you are drawing say 300W DC and the psu is 75% efficient at that load, the psu will be dissipating 100W of its own heat. A Nexus @ 6V is shifting 18.4cfm (in free air, with any static pressure it is shifting much less) which results in an overall 'system' temp rise of 10 degrees C. (Where the system is the psu, assuming all cfm is effectively cooling the psu.) The huge health warning is static pressure: even relatively low static pressure slugs the performance of low speed DC fans.
qstoffe wrote:Will a PSU start if I cut the wires to the original PSU fan ? Do most PSUs even have rpm status or are they controlled by voltage only ?
No problem - AFAIK any psu that has a 3 wire fan passes the 3rd wire outside the psu to the main mobo for monitoring. Most have 2 wire fans. I have run most (all?) of my psu without a fan connected to the psu fan header. (I have sometimes removed the fan completely, at others wired it to an external controller or alternatively wired it to 5V inside the psu itself.)

Good luck.

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Mon May 15, 2006 1:58 am

If you are drawing say 300W DC and the psu is 75% efficient at that load, the psu will be dissipating 100W of its own heat. A Nexus @ 6V is shifting 18.4cfm (in free air, with any static pressure it is shifting much less) which results in an overall 'system' temp rise of 10 degrees C.
1) What are the calculations behind this result? Show your workings! :wink:

2) I really strongly doubt that a Nexus 120mm @ 6V can transport 100W of heat.

IME most PSU's will start even if their fan is disconnected. For example my S12-330 runs fine with a Yate Loon swapped in there running off a Molex connector.

justblair
Posts: 545
Joined: Fri Jul 15, 2005 12:33 pm
Location: GLASGOW, SCOTLAND, UK
Contact:

Post by justblair » Mon May 15, 2006 2:02 am

Just a couple of other points to consider.

PSU's hold onto fairly massive amounts of charge even when switched off. Be very carefull to discharge the PSU before you work on it. Leaving it for a few days will work, alternatively you can disconnect it from the mains and start the computer up a few times to dissipate it. When you hit the on off switch you will see the fans/lights etc fire up for a second as the caps discharge.

I'm not an electrician, my knowledge on discharging psu's has come from what I have read on forums such as this. I am however still alive despite modding the odd PSU.

Please still treat the thing with caution afterwords. Especially the large capacitors.

Use an insulated screwdriver,

Work with one hand only.

If lifting out the gubbins, handle the board by the edges.

If you are building a wooden enclosure for it, bear in mind that wood is flamable. Consider what will happen for instance if the fan you are fitting fails. Are you generating a fire hazard? Home insurance for instance may be invalidated if you are using modified appliances that have been moddified.

If it were me, I would consider making the moddified enclosure from metal.

Le_Gritche
Posts: 140
Joined: Wed Jan 18, 2006 4:57 am
Location: France, Lyon

Re: "Fanless" PSU mod.

Post by Le_Gritche » Mon May 15, 2006 2:43 am

dukla2000 wrote:
qstoffe wrote:How about grounding ? Do have to connect the PSU to some bigger metalplate in order for it to work?
No - as soon as you open any psu you will see the ground from the mains cable connects only to the psu case. Although I have checked on a couple of PCs that chassis ground and DC ground are the same thing I never worked out precisely how come: it may well be that the psu mobo has earthing rings around its mounting screws that you should replicate.
I moved my PSU out of my PC case and let it stay on the painted top panel of the PC case. The first couple of times I touched the PSU casing after that, I got a feeling of a very little electric shock.
I don't know if that was psychological or real, but anyway I quickly put a little wire between the PSU casing and a non-painted conductive part of the PC case. No weird shock feeling since that.
So I agree with dukla2000 that you should ground the PSU mobo. Problem is with a wooden case, there's not much you can ground it onto.

There is another probelm with remplacing the metal casing wth a wooden one : it's EMI.
The PSU casing stop Electromagnetic Interference, if you remove it and your PC case is also made of wood, you have no protection from it. Besides the health hazards which I'm not qualified to write about, there's also the potential effects on the hard drive, your screen, your TV...
You better put some tinfoil in there :wink:
If I were you I would keep most of the PSU casing, just removing front and back panels to use the remaining as a wind tunnel, and duct it to your fan.
Maybe you could explain a bit more the design of your wooden PC case and PSU enclosure.

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Mon May 15, 2006 7:30 am

jaganath wrote:
If you are drawing say 300W DC and the psu is 75% efficient at that load, the psu will be dissipating 100W of its own heat. A Nexus @ 6V is shifting 18.4cfm (in free air, with any static pressure it is shifting much less) which results in an overall 'system' temp rise of 10 degrees C.
What are the calculations behind this result? Show your workings!
Hopefully the 100W heat dissipation is self evident.

The cooling is simple: most literature on forced air-cooling (including several fan manufacturer web sites, couldn't be bothered to google one now) will have the formula
Q = (1.76 * H) / dT
where
Q is airflow required in cfm
H is Heat dissipated in Watts
dT is temperature rise above inlet temperature in Centigrade.

In this case ( :roll: )
dt = 1.76 * 100 / 18.4 = 9.565 degrees Centigrade.

In reality this is very much a best case (not again!) - apart from static pressure on the fan it completely 'averages' any hot spots and also ignores any airflow dead spots. However assuming the temp in the case isn't going much above 30 Centigrade all should be well.

BTW that does raise an old chestnut that has been discussed here before: if the 'ambient' for your psu is anything above 40 Centigrade then all bets are off: certainly Seasonic and (IIRC) FSP have specs that say if the ambient is above 40 (or so) then the efficiency drops and the max loads are reduced.

IsaacKuo
Posts: 1705
Joined: Fri Jan 23, 2004 7:50 am
Location: Baton Rouge, Louisiana

Post by IsaacKuo » Mon May 15, 2006 7:54 am

I've done various PSU mods, but I have so far always done so within the context of keeping the PSU components within the original case. I may have cut out fan grills, removed and/or flipped the fan, but I always keep the PSU more or less intact.

Why? Well, there are a couple safety issues.

First, and most obviously, there may be a heatsink which isn't ground. I'm not comfortable with the possibility of any stray wire or finger touching that.

Second, the only place in the computer where you can get really high voltages is inside the PSU. With ATX safety specifications, this high voltage is almost guaranteed to never "escape" the PSU enclosure. The PSU enclosure is grounded, so if any sort of failure causes any non-ground wire to touch it, there will be a short-circuit and the PSU will shut down (either via safety shutdown or permanent failure).

In contrast, if the PSU components are taken out of the enclosure then there's the possibility of high voltage getting transfered anywhere--including the outer computer case (if metal, of course). In the worst case scenario, the computer is operating just fine and outwardly looks normal--but the outer case is dangerously charged to a high voltage!

I'm perfectly happy living within the constraints of keeping the PSU components inside the original ATX enclosure.

qstoffe
Posts: 84
Joined: Tue Apr 25, 2006 1:32 pm
Location: Sweden

Post by qstoffe » Mon May 15, 2006 12:10 pm

Thanks for the good replies to everyone :D

Will I have to say I'm a bit sceptical about doing this kind of mod because of the potential danger...

But since I'm "poweruser" (play demanding games, run CPU intensive tasks) and I usually have powerhungry components: any PSU, no matter how silent, will eventually ramp up and be the loudest noise-source in the entire case!! Right now I have a Zalman ZM300A-APF (300W) that ramps up like fucking tornado after about 40+ minutes of heavy load on my P4 3.4GHz with a 6800GT. The PSU is by far the loudest component in my current case and since I will be building my next system from scratch (including the case) I will NOT tolerate any fan ramping up like that. In fact I would prefer only to have fans running at constant (low) speeds.

If a look at an alternative to modding the PSU it would be a PSU that is "semi-passve" (fan only starts when it's hot enough). The design of the custom woodcase I'm building will be similiar to the Antec P180 case in reguards of the PSU chamber. Question is: will a good airflow (an external Nexus 120mm @ 6V) be able to cool a semi-passive PSU good enough so that the fan won't start at all !?!? Anyone who has tried this ?

If I on the other hand do this mod I think I will only remove part of the cover so that I get an good airpath straight through the PSU (running like a convertible car at summer :) ). For "some" protection I will have to add everything (remaining PSU + Nexus fan) in a large grating of some sort. But I'm still a bit concerned about grounding. If I run convertible (or "cab" as we say in Sweden :) ) will there be enough metal on the remaining part of the PSU for sufficient grounding ?!? Also the PSU bottom will only have contact with wood or soft foam so the case won't help any with grounding as it does in a "normal" metal case. If you "deem" it necessary I could install a big metal plate at the bottom of the case and connect the grounding from the PSU to this. Will this work ? Or is it even necessary ?

PS. I will be unable to post anything here for the next couple of days - so don't think I've gone missing just because I started some dagerous PSU mod or anything :wink: 8) .

justblair
Posts: 545
Joined: Fri Jul 15, 2005 12:33 pm
Location: GLASGOW, SCOTLAND, UK
Contact:

Post by justblair » Mon May 15, 2006 1:55 pm

You dont need to add a plate to ground the psu. The Psu is already grounded to the mains electricity. In a metal case, the case is then grounded to the PSU.

ie.

Ground - Mains cabling - PSU - Case & components.

If you use a wood case it goes...

Ground - Mains cabling - PSU - components.

The case doesn't help ground the PSU the PSU grounds the case.

If you take the cover off, the remaining parts of the PSU case will remain grounded. You will see a strap linking the case to the mains cable somewhere in the PSU design.

Now having less metal for the electricity to flow through would perhaps become dangerous if the electricity flowed through for a period of time (Think about an overloaded plug catching fire) however the PSU will have fuses inside that will protect against this, blowing as soon as too much current passes through them.

The issue here is different. In a metal enclosure if something overheats it is not really an issue. The metal may bend, but if components catch fire they will short out to the ground long before heat generated becomes dangerous to the environment outside the PSU, or run out of combustable material and the fire will extinguish. In a wooden case the concern is that other material (i.e the wood) will continue to burn after the safety protection of the PSU has activated.

That is the real risk with using wooden (or other combustable materials) for a case. Now if the PSU is in tact then I think that most people (including me) would consider the risk suitably managed. Take the shroud off the PSU and you are removing a barrier between your combustable case and your potential source of ignition. You would have to decide whether that is a risk that you would consider reasonable. However your home insurer may consider that you have introduced an unacceptable risk if you ever had a fire... refusing to pay.

I personally would not be comfortable with an open PSU in a wooden case. I would either keep the PSU in tact and duct it to prevent it ramping up or modify its shroud using mild steel or alu. Your idea about copying the 180 duct into your design I think is reasonable.

If it were in a metal case I would be far less concerned about having an open PSU.

BrianE
Posts: 667
Joined: Tue Mar 29, 2005 7:39 pm
Location: Vancouver, BC, Canada

Post by BrianE » Mon May 15, 2006 3:52 pm

If you're really serious about taking apart a PSU and messing around with how the housing and other parts are grounded, I would get a reasonable quality multimeter (not the cheapest one you can find) if you don't already own one, and learn the basics of using it.

Then you can measure the resistance in ohms between a good known ground and your grounding connections/wiring. I have no idea what the standards are for computer grounds, but the closer you are to zero the better.

Tibors
Patron of SPCR
Posts: 2674
Joined: Sun Jul 04, 2004 6:07 am
Location: Houten, The Netherlands, Europe

Post by Tibors » Tue May 16, 2006 12:36 am

Every time when a person starts talking about a wooden case there is at least one overly paranoid person replying about firehazard and another one about EMI. Just ignore them. There is nothing in your PC that will ever get hot enough to get solid wood burning. (There is a reason people use tinder to start burning wood.) That EMI is not a real problem is shown often enough too. (Windowed cases or cut out fan grilles anyone?)

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Tue May 16, 2006 2:13 am

That EMI is not a real problem is shown often enough too. (Windowed cases or cut out fan grilles anyone?)
EMI from PC's is only a problem if the user intends to run a TV or radio or wireless phone in close proximity to the unshielded PC. Otherwise, as you say, the effect is negligible.

justblair
Posts: 545
Joined: Fri Jul 15, 2005 12:33 pm
Location: GLASGOW, SCOTLAND, UK
Contact:

Post by justblair » Tue May 16, 2006 8:41 am

Tibors wrote:Every time when a person starts talking about a wooden case there is at least one overly paranoid person replying about firehazard and another one about EMI. Just ignore them. There is nothing in your PC that will ever get hot enough to get solid wood burning. (There is a reason people use tinder to start burning wood.) That EMI is not a real problem is shown often enough too. (Windowed cases or cut out fan grilles anyone?)
I would hardly consider myself overly paranoid. I have built and use a wooden case myself. I have also seen a PSU catch fire... and go out again.

If you read what I say, it is not that wooden cases per se are dangerous, but that using an exposed PSU within a wooden case is a risk that I would consider unacceptable.

I thought I had made a clear explanation of the risks without screamin "We are all doomed"

There are so many other ways to do this that dont increase the risk (however small) that I would be looking to them rather than exposing the innards of a PSU.

larrymoencurly
Posts: 170
Joined: Tue Dec 03, 2002 11:29 pm

Post by larrymoencurly » Sat May 20, 2006 12:04 pm

qstoffe wrote:Also the PSU bottom will only have contact with wood or soft foam so the case won't help any with grounding as it does in a "normal" metal case.
Isn't that soft foam really flammable? One person had a HD burn in an external enclosure burn one of its chips, and in turn it burned a hole in the plastic case (claimed to be polycarbonate but seems to be acrylic, which burns much more easily). At the very least, I would leave some air space between the PSU botom and the foam or wood to keep the temperature down.

Post Reply