Why do PSUs have fans????

[BobDog]

Hi, I am sorry to ask what may seem like an obvious (nu-bee) question to some… but WHY TO PSUs HAVE TO HAVE FANS Or, to put it another way, why do they get so hot that they require serious cooling solutions?

External PSUs (e.g. the one used by Shuttle, reviewed here: http://www.silentpcreview.com/article139-page1.html ) need no fans and, indeed, are reported to get “only slightly warm,” is there something magic about putting a PSU into a case that makes it so much hotter?

One might say that Shuttle (and laptop) external PSUs are low-power and thus need not get so hot. Observe, however, the truly massive PSUs on stereo power amplifiers which, I can attest, do not get even the slightest bit warm under heavy load (thought the amps’ output devices can get quite toasty, of course). While most amp. PSUs are toroids or C-core designs (whereas PCs use switch-mode PSUs), I do not think this is the answer either. Audio companies such as Linn and Chord use ludicrously powerful switch-mode PSUs in their amps and I do not believe they get hot—at least not hot enough to need fan-cooling. I owned an EMM Labs DCC-2 digital-to-analogue converter and CDSD transport, both of which used low-power switch-mode power supplies and I can attest that these did not get very hot….

So, if we can make an external PC PSU cool and fanless and if we can make a power amplifier’s internal PSU cool and fanless, why are we having to dick around with these stupid noisy PSUs IN our PCs (or being forced to settle for more dicey fanless solutions from Silverstone, Antech, et al.?). I am very interested to learn . Thanks for your replies.

(I hope this has not been answered elsewhere, but I checked the SPCR PSU article, http://www.silentpcreview.com/article28-page1.html, and about Mike C's attempt to run a PSU fanless, http://www.silentpcreview.com/article15-page1.html , but neither offered any clues to my question... I could not even think of the right search terms to use to look for an answer in the forums.)

[Splinter]

Musical amps are generally many times larger than computer PSUs. If you crack one open, you will generally not find a ton more stuff in there than a computer PSU, but you will find huge heatsinks.

The heatsinks in a PSU are only a few dozen CCs, versus hundreds for music amps.

[BobDog]

Musical amps are generally many times larger than computer PSUs. If you crack one open, you will generally not find a ton more stuff in there than a computer PSU, but you will find huge heatsinks.

The heatsinks in a PSU are only a few dozen CCs, versus hundreds for music amps.

Sometimes... the toroid in my Spectron amp is indeed many, many times the size of that of my PC.

On the other hand, the switch-mode PSUs in Linns, Chords, and EMM Labs gear (mentioned above) are often SMALLER than those found in PCs (and the Shuttle and laptop external PSUs are not much/any bigger either). Also, although I have never unwound a toroid or opened a C-core trannie, I am pretty sure that they have NO heatsinks. Amps do have massive heatsinks, of course, but these are for the output devices, not their PSUs.

I'll buy that size may matter sometimes... but I am not sure that that's the answer this time. Thanks for your thoughts--any others are much appreciated.

[Splinter]

Toroids are extremely efficient, something on the lines of 98%

However, you can't use a Toroid to convert AC to DC

[frostedflakes]

Well you can't use any transformer to convert AC to DC AFAIK; they just step the voltage down.

It's an interesting question, hopefully somebody with more knowledge of electronics could explain it to us.

Power supplies are slowly becoming more efficient, though, with many ATX12v v2.0 PSUs peaking at 80%+ (I think the min. efficiency for this standard is 70-75%, compared to what, 60-65% for ATX12v v1.3?). But I think even when we start to see extremely efficient power supplies (90%+), they will still be actively cooled, because the PSU fans also play a role in CPU/system cooling.

[Splinter]

It's a role that could be more effectively played by a more specified item, such as a case fan.

[BobDog]

I agree with Splinter that case cooling cannot be the sole reason we continue to subject ourselves to whiney fans on our PSUs... plus you don't really NEED to have a PSU fan to cool your case (I will be building a case around a Silverstone ST30NF shortely).

On the other hand, I still am not so sure about Splinter’s argument that toroids are more efficient than switch-mode PSUs… as I recall from speaking to audio manufacturers, it is the other way around.

In any case, setting the issue of toroids aside COMPLETELY for the moment, this still doesn’t get at my original question: why is it that outboard COMPUTER PSUs (and laptop PSUs) have no heat/fan issues, while those inside the case do? (I only brought up the audio gear--which, again, often uses switching PSUs as well--to give another example of powerful, small, and silent PSUs, not to muddy the waters further).

If anyone could give me a simple straight answer on this, it would be much appreciated.

[Splinter]

Power converters on power poles use massive toroids, they're cooled only with heatsinks drowned in a heat conductive oil.

They put through tens of thousands of watts, if their effiency wasnt above 95% they would explode.

[NeilBlanchard]

Hello:

They have used fans (probably) because it far less expensive to make an inefficient PS and just keep it cool with a fan, than it is to make a really efficient one that only needs passive cooling.

Audio is all about high quality power supplies -- and they have to drive some pretty strange impedances and reactive loads with wild phase behavior -- and do so quietly!

[BobDog]

Hello to Splinter and Neil...

(1) I agree Neil, few things are more important to audio than a high quality power supply... but why would this necessarily make it cooler? (BTW, I am attempting to make a high quality music and movie server, hence my new-found interest in silent computing, and one of my main concerns is that the limiting factor will be the sucky quality of my Silverstone PSU... but I digress .)

(2) Further, why do outboard computer PSUs have no need for fans?!? I doubt they are constructed with any greater attention to quality than those designed to go inside our PCs, no?

(3) Splinter, I don't want to argue toroids with you (although switch-mode PSUs are indeed favored by audio designers for their very high efficiency--and every much reviled for their generally poor sound quality), I just wanna know: SETTING THE ISSUE OF TOROID EFFICIENCY ASSIDE FOR NOW, why do computer PSUs need fans when the PSUs for virtually everything else on earth does not

Thanks for all your input thus far….

[sthayashi]

why is it that outboard COMPUTER PSUs (and laptop PSUs) have no heat/fan issues, while those inside the case do? (I only brought up the audio gear--which, again, often uses switching PSUs as well--to give another example of powerful, small, and silent PSUs, not to muddy the waters further).

If anyone could give me a simple straight answer on this, it would be much appreciated.

I'm going a little beyond my own field of knowledge here to answer this one, but power supplies are a function of power capability and size.

The heat that they (can) generate needs to be dissipated somehow and fan cooling aids in this. Laptop power supplies don't have the same power capacity that regular desktops have. Even regular 'desktop replacement' laptops rarely exceed 100W on their power converters. Even the lowliest desktop power supply is supposed to be capable of 300W.

The ATX specification regulates the size and shape of computer power supplies. This is a restriction much tighter than most other devices, including amplifiers. Now in theory, if you could put a power supply outside of the computer case and removed all size restrictions, you could almost easily make it passive.

There's also another specification problem of the past. Previously, the intended ATX airflow was to have almost all the air exit via the PSU. I have computer cases which are a testament to this design philosophy. This is less of a problem these days with 120mm fanned cases, but the design legacy still haunts us today.

Oh yes, and one last thing. Operating environment makes computer PSUs a tougher challenge as well. Computer PSUs are expected to operate continuously at 10-20C above room temperature, giving them less room to work in.

There are multiple answers here. Take any one of them if you want a simple answer. Take all of them if you want the straight answer.

[BobDog]

Hummmm, so if sthayashi is right, all we'd have to do to have hi-quality, hi-power, hi-reliability PASSIVE PSUs is do away with the ATX spec?

Considering the fact that so many builders ahve done this in the past with PSUs (mostly SSF case makers), this seems almost TOO easy. Why has no one done this with a HTPC case yet--some are small but a number of them are large enough to fit PSUs of any reasonable size....

If anyone will build such a thing, I will buy it. Now.

[Green Shoes]

BobDog, if you're running some of the wiring for this behind theater equipment, have you considered a power brick? Most HTPCs are pretty low powered, and (like the mac mini) can get away with a high-quality external AC/DC converter. You can always rig a 120-mm fan in place of the PSU to still comply with the ATX standard. I don't know if there are any bricks specifically tailored for the PC market, but if you're not scared of soldering molex connectors directly into a power supply you might be able to do it yourself .

But I do seem to recall someone finding an external brick for PCs....it's too late for me to try and find it now, though.

Oh, and since no one has said it yet, Welcome to SPCR!!!

[BobDog]

Thanks for the welcome Green Shoes. Hummm, I did not think of trying a power brick (I already got a Silverstone fanless PSU)... though I AM scared of soldering anything onto a device that will plug directly into the wall .

BTW, I am still not sure anyone has really answerd my origonal question, though thanks to sthayashi for coming the closest.

[yeha]

well, no ac -> dc conversion can be perfectly efficient, some heat is always created. it's possible to get it up to 95% efficiency or so, but that requires extremely expensive parts. if your psu is 80% efficient and you ask 150 watts from it, you'll have 37 watts of heat appearing that you have to dissipate somehow - if you just let it build up, your psu melts and you can't surf the net anymore. all psus are the same, most of the time they're even less efficient, but because they aren't being asked to supply as much juice they don't need to actively dissipate the heat - laptops draw ~30 watts, other dc devices like routers / cell phones / pc speakers / alarm clocks / cable tv boxes / etc. draw far less, thus the power supplies they use (brick style) only have to deal with 5 to 10 watts of heat, which is simple for a fat heavy brick to deal with.

and for that matter, i have seen external brick psus that had 40mm fans in them for heat dissipation, and the external bricks that hp e-vectra pcs i used to administer used got *very* warm if the system was loaded for a while.

summary: all psus create heat, and computer psus supply more power than most others, thus put out more heat. psus die if they overheat. passive cooling of psus has a certain threshold (in the confines of cost/usefulness), and computer psus often fall outside that threshold. laptops, routers, phones, digital clocks, etc. are inside the threshold, thus passive cooling of their power units is the norm.

[BobDog]

Great answer yeha,

I think I can certainly buy that one. Thanks. I am still less clear on why audio equiptment's PSUs do not get so hot, though form Splinter it would seem that their PSUs are sufficently different that this forum is not the place to get into it (and also, come to think of it, the hi-power Linn amps with switching PSUs do use extreme cooling--directly through their exotic and EXPENSIVE casework: http://www.linn.co.uk/spec_sound/produc ... N=19759076).

At a minumum we now have a good answer why "laptops, routers, phones, digital clocks, etc." are cool, so to speak , where PCs are not. Thanks.

[sthayashi]

Hummmm, so if sthayashi is right, all we'd have to do to have hi-quality, hi-power, hi-reliability PASSIVE PSUs is do away with the ATX spec?

Considering the fact that so many builders ahve done this in the past with PSUs (mostly SSF case makers), this seems almost TOO easy. Why has no one done this with a HTPC case yet--some are small but a number of them are large enough to fit PSUs of any reasonable size....

If anyone will build such a thing, I will buy it. Now.

Shuttle did that with their "Zen" computer. It was passive brick. Sadly, it's been a disappointment to me to NOT see that brick come back on any of their other machines.

However, here's the other thing. Part of the reason that the ATX PSU specification is limiting is due to the size and weight required. Would you be willing to purchase a 10kg PSU that was the same size as a SFF computer. Actually, don't answer that question. Do you think you could convince a PSU manufacturer that there is a market for such a PSU?

BTW, I have a couple guesses as to why and how audio PSUs seem relatively cool. First, they ARE on the massive side (they generally seem a lot heavier to me than your average computer PSU). Second and this is pure speculation on my part: I don't think most of the audio PSU's energy is being turned into heat inside the amp.

If you follow the power path of an audio amplifier, you'll see that one of the main goals is to generate a lot of audio power (imagine that...). Well, where are those 300W going to. IIRC, a lot of it is going to the speakers (unless you have a Class-A amp that's not driving speakers). Well then most of those 300W aren't being turned into heat inside the amplifier, but into heat and audio energy inside the speaker.

Like I said, I know very little about this field. I'm not as familiar with analog electronics and signal driving as I am with other fields of EE.

[lm]

If you follow the power path of an audio amplifier, you'll see that one of the main goals is to generate a lot of audio power (imagine that...). Well, where are those 300W going to. IIRC, a lot of it is going to the speakers (unless you have a Class-A amp that's not driving speakers). Well then most of those 300W aren't being turned into heat inside the amplifier, but into heat and audio energy inside the speaker.

Obviously any psu's job is to give power to the components it runs, be it speakers or the cpu... You didn't say anything there imho.

But really, how often are you using 300W on an amp? If your not trying to blow your ears off, the power input of a stereo amp might not be very much.

My amp, that can do 2*75W (more than I ever need) tends to eat 18W from the wall (I measured it with a powermeter), which doesn't really seem to depend on if I play music or not, as long as I am playing it with the volume I like to use. Didn't bother to test how much it eats when playing too loud...

[sthayashi]

Obviously any psu's job is to give power to the components it runs, be it speakers or the cpu... You didn't say anything there imho.

Well, if a PSU could be a lot cooler if it doesn't have to sit in the heat that it produces.

[Green Shoes]

I could see some of the bigger multi-channel amps pulling over 300W, like those 7-channel by 250-watt monsters that some people have in their home theaters. However, the PSU in a box like that is the only component producing a major amount of heat, unlike in a computer; it therefore has the luxury of using the entire case as a heatsink, or any unused space therein.

[IsaacKuo]

Part of the reason that the ATX PSU specification is limiting is due to the size and weight required. Would you be willing to purchase a 10kg PSU that was the same size as a SFF computer. Actually, don't answer that question. Do you think you could convince a PSU manufacturer that there is a market for such a PSU?

The thing is that it wouldn't need to be so large or heavy. There are already some passive PSU's on the market that fit within the ATX specification. At worst, all you'd need to do is take that block and remove it from the case, along with extended power lines from the external block to the case. The part installed in the computer case could just be a backplate, along with perhaps a hole for an optional fan (a de facto case fan).

To me, the bizarre concept is the one that current passive ATX PSUs use--to actually leave the passive PSU in the worst conceivable location for it. It's confined in the hottest part of a computer case, blocked on three sides and heated up by another side. Why? Because they assume the people buying their products are morons?

[mathias]

I'm wondering if audio PSU's use the same type of power, I doubt that they use three to five different voltages, so I'm sure that helps, and I suspect that they use a more efficient voltage, maybe something like 40V DC?(just a guess, I don't know if converting to higher voltages makes efficiency easier)

Does anyone know whether audio equipment needs DC current or if all or part of it can use AC?

---

... through their exotic and EXPENSIVE casework: [url]http://www.lin%20...%20*<very%20long%20link>*%20...%20&CFTOKEN=19759076[/url]).

At a minumum we now have ...

Could you edit that post and not use such long links in the text? It causes side scrolling.(use the (url=____) tag instead )

[sthayashi]

Does anyone know whether audio equipment needs DC current or if all or part of it can use AC?

This one I DO know the answer to. DC all the way. Some people go nuts trying to get the cleanest DC power possible. For these people, even computer's PSU is too noisy.

[BobDog]

(1) sthayashi is correct, all audio gear uses DC only (the Sutherland phono pre-amp, for example, uses something like 14 D-cell batteries for power ).

(2) Clean DC is absolutely, 100%, with-out-doubt, THE most important thing to good audio (and video too, for that matter). I have never tried to use a computer PSU in a audio component (given my electrical skills, I think I'd end up burning down the house), but I am reasonably certain that the results would not be so good. Indeed, considering the amount of garbage that switching PSUs generate, even having a computer on the same circuit as your audio gear (without massive filtration between), is asking for trouble.

(3) Note that even passive computer PSUs still have to use heatsinks, heatpipes, etc., because THEY GET VERY HOT. Audio PSUs (and laptops, etc.) need no such things--perhaps yeha and mathias are right: these other devices need not draw so much power nor need they drive a number of different voltages?

(4) Even if this is true however, one still wonders about the really, really massive power supplies in power-amps the Boulder 2060, for example, can draw 5000 watts AC fer’ God’s sake (and it’s peak power is only 1200 watts… so that is only like 24% efficient—though that is to the speaker taps, which is not the same as the PSU itself). (Please, let's not get caught up in the toroid vs. switching debate again, yes, the Boulder uses a toroid, but other hi-power amps use switching supplies--see, again, the Linns and Chords or the old Carvers too.)

(5) Of course, the 2060 can heat a large room in winter (just look at THOSE heatsinks!), which leads me to my final thought: it is the output devices, not the PSU that get hot. Is it possible that audio PSUs just dump all their energy into downstream components, thus allowing the PSUs to operate at peak efficiency while forcing the other stuff to deal with the excess heat? It’d be like designing your computer PSU to always run full out and then asking your CPU and GPU to deal with all the extra energy. It’s just a thought; if I knew anything about electrical engineering I wouldn’t have had to start this thread in the first place !

[Katharsis]

here's 2 cents from my knowledge. im an electrician. i deal mainly with mains power, building hotels and such. i'm not so hot diagnosing a circuit board.
the transformer on the pole is hot. it's got oil flowing through it to huge fins all exposed to the air. a 480 to 120v 200amp transformer just has a large case around it with holesin it to keep people from touching it and frying themselves. we put them in rooms with large air conditioners to keep them cool. the more amperage you have the hotter it all gets.

computer psu's are getting better than their original bigass transformers, but they still not perfect. if the input is 120v and the comp is drawing 300w the wall plug is only pulling 2.5 amps but the other side is running off of the 12 and 5v rails and could be drawing 20-30 amps from the psu at times. this isn't counting the extra heat from not being 100% efficient.
the psu is rather cramped and there isn't much room for large heatsinks. add to that the heat and low airflow inside the top back corner of a computer case. thats why they either use fans or convert the entire case of the psu into one large heatsink. this is also another reason why intel wants to put the psu at the bottom of the case

sigh, i left and came back and have no idea where i was in my thoughts. i tried to clean it up as best as i could. hopefully i said something usefull and didn't get too far off the topic

[wsc]

I think you guys are trying to make this too hard. Firstly I think the ATX power supply specs were originally developed with the knowledge that PSU's at the time were really inefficient and would therefore benefit (dollar-wise) from active cooling. The spec was probably developed with a "kill two birds" mentality by giving the PSU double duty to act as a case cooler since it was going to need active cooling anyway.

Also I think some of you guys are mixing up what heatsinks are used for in audio equipment. Unless you have a switch mode power supply in your audio equipment, the heatsinks are for cooling the power stage, ie, what makes music, not what makes power. Car amps use SMPS (switch mode power supply) and they probably stick the active devices on the same heatsink the MOSFETs for the power stage use. I do not believe toroids are 98% efficient either, but I'd have to look it up.

As far as laptop bricks go, hardly any of them are rated for much over 110 watts. If your efficiency is 70% thats only 33W you have to get rid of at MAXIMUM power draw, which may happen rarely if ever. Its another deal entirely when you have a desktop which sees 110W (and not infrequently alot more) AT IDLE. This is where the bean counters get in a fist fight with the engineers. To decrease the heat output you have to use nicer MOSFETs with lower Rdson (on resistance) among other things. Then you have thermal runaway conditions because power devices handle less power the hotter they get. That can lead to a thing thermal runaway, which leads to a thing called your house just burned down. How can you meet a $25 price point for a device made with maybe $2 worth of materials if your budget is blown on a huge expensive heatsink so you can appease the .01% of the market who cares about getting every last bit of silence?

edit: sorry to ramble

[BobDog]

I don't know.... Audio heatsinks cool amp (and other) out-put stages, not PSUs (as wrc correctly notes and as I've been saying all along. This is becacause their PSUs do not get hot. If their PSUs don't get hot (at all, really), then why do computers'?

Maybe wsc is right and switching PSUs in audio gear do get hot (like PC's PSUs), my only experience with switching PSUs in audio gear is with EMM Labs DACs and Transports which are obviously very low power (but they did not get hot/were not activly or really passivly cooled in any way--they were quite small as well). As I said in an earlier post, Linn, whcih uses switching PSUs in their power amps uses very elaborate case work to cool the machine. I thought it was jsut for the output devices, but perhaps it is needed to cool their PSUs as well.

[yeha]

a lot of audio gear hardly draws any power at all. my father had some monstrous speakers back in australia and he told me they only actually used 10 watts or so. all the speakers i have in my house right now draw 4 watts max according to my kill-a-watt, regardless of how loud they're turned up.

if a part's not getting hot, it's not drawing power, that's all there is to it. that energy they're supposedly drawing has to go somewhere, if it turned even 3 watts into pure sound energy you'd be instantly deaf, so everything else it draws must be turned into heat. since it doesn't heat up, it's not using much power. that's all there is to it.

[BobDog]

lot of audio gear hardly draws any power at all. my father had some monstrous speakers back in australia and he told me they only actually used 10 watts or so. all the speakers i have in my house right now draw 4 watts max according to my kill-a-watt, regardless of how loud they're turned up.

Absolutely correct, speakers rarely use more than a few watts--though yours must be very efficent to NEVER go over four (horns perhaps?).

But as an earlier poster noted, one of the things that speakers do is act as a heat/energy sink for amps; you'd think that the GREATER the current draw by your speakers, the less heat your amp would produce. This is absolutely the case for class A amps, though, again, it is the output devices and not the PSU that gets hot from sinking all the power not consumed by the speakers. Class D amps are so cool (most need few if any heatsinks) because their output devices are used so efficently, their power supplies seem to be unaffected by current draw... but I am not an expert in this so I am less sure.

[BobDog]

I just read this in the Zalman TNN 500AF thread:

I am very impressed with the PSU. It is always cool. Not even warm! And totally noiseless.

...

Zalman's sofisticated PSU which uses separate transformers for all required voltage outputs, i.e. 3.3, 5 and 12V.

That sounds like mathias might have been right when he wrote:

I'm wondering if audio PSU's use the same type of power, I doubt that they use three to five different voltages, so I'm sure that helps, and I suspect that they use a more efficient voltage, maybe something like 40V DC?

If so, that just pisses me off that PSU makers remain so half-ass that they cannot offer us a soultion like this without having to buy the full TNN 500 system. Sure, it would caost more, but I be some of us (like me ) would be willing to drop a few extra $$$ for a PSU that is silent AND cool, others could have PSUs with fans still on them and spend less and there is nothing wrong with that either.