Why Aren't PC Power Supplies External? 67
An Anonymous Coward asks: "After years in the computer industry it just occurred to me: Why don't computers have external power supplies? The lions share of the heat and noise from a computer are generated from the power supply. Why not take it out of the case and run just the 12 and 5 volt lines into the case, leaving the heat and noise tucked down under the desk somewhere? Perhaps there is some electrical law that makes that a bad idea, but I can't think of it." One stumbling block is that a power supply brings in more than the 12V and 5V lines, it also provides the power connectors for the motherboard, internal peripherals and external fans. However, these could be provided by the case. If a standard form connector from external PS to the PC is designed, might we be able to avoid further incompatibilities like AT/ATX? It's an interesting idea for future designs, but I wonder if there are other stubling blocks out there. Thoughts?
DC outlets (Score:3)
Look around your house/work at how many separate AC -> DC converters there are. I think centralizing would be more efficient. You could have battery backup and very clean power.
DC power is easier to use (most of the time). Telco racks are DC for a reason.
Re:Cost? (Score:1)
When the PSU on my PC went a few months ago I found I could get a new PSU for 40 quid or a new case, with a better PSU, for 30 quid. Nobody buys PC PSU's without case or uses X-Terminals, hence the high price.
If Intel decided that an external PSU was required for ATX2, prices would plumet.
Been done! (Score:4)
I can't wait for my new PC to have the same feature.
Re:IMHO (Score:2)
--
most heat & noise from a P/S? Nope. (Score:5)
These days, Most heat is generated from the CPU, and most noise, believe it or not, is generated from those cruddy little CPU fans. A lot of noise is also generated from extra case fans you may have trying to improve airflow in your case. P/S fans are often much quieter than those other fans in your system.
I've thought of this idea before, and I also think it'd be a fantastic idea. Just keep in mind, the P/S also needs to be cooled, so that external box will have fans in it, and once it's outside the case, it'll be making some noise, too. I remember my Amiga 500 fondly, though, as it's power needs were such the external P/S didn't need any fans...
Re:Clustering/Rackmounting (Score:1)
(Since most of the telephone switch equipment runs on -48V DC...)
They're a bit more expensive than the normal ones...
Then just use a big 110VAC to -48VDC converter, tie in a bank of batteries, and you've got the UPS end of it handled as well.
Re:Clustering/Rackmounting (Score:1)
the phone lines. It helps prevent corrosion. (Alot like anodic protection for ships and such)
Cost? (Score:5)
Anyway, the thing had the pinout of the connector printed on the bottom, so I hacked off the cables, soldered on some standard molex connectors, and wired three of them up to a standard PC power supply running in an empty case.
Back to the question. An Xterminal is a pretty low-power-usage device (no moving parts, and the monitor was on a mains cable) and that seemed to need an expensive PSU. A comparable unit to power a PC as well would be ridiculous...
Baz
Re:Clustering/Rackmounting (Score:2)
Part of why I ask is that the local telco seems to do the which polarity is the red wire and which is the green wire backwards from most of the rest of the USA, and I'm looking for clues.
Re:IMHO (Score:1)
The crappy old laptop I've got uses a power supply like this, as does its external CD drive (creative labs). The only thing on my computer power board with a stupid power adapter is the modem, so it has to sit on a wonky angle in the board.
New Moderator options (Score:2)
Still, I wish there was a moderation option called "The Horse's Mouth" or "The Straight Dope" so that people could filter on primary source comments.
That's one of the e-PC concepts (Score:2)
Tis doesn't really apply to the "dual 1GHz processor 1GB RAM, 4x30GB 10krpm SCSI" crowd, but for a lot of users, these smaller e-PCs are an attractive alternative, being more reliable and less intrusive.
Don't forget, though, that noisy PCs have other uses. In my case, any troublesome infants from visiting couples are ceremoniously placed in the computer room during dinner. Most pretty instantly fall asleep from the hypnotic noise of the fans!
DS
The simplest answer ... (Score:2)
Glad I could help,
Seriously though, I don't think the components care where the power comes from, just as long as it gets it. They are most likely placed there because ideally computers should be self contained boxes ... who wants things dangling from them? Just a thought.
HO Scale (Score:2)
--
Re:IMHO (Score:2)
Re:Actually... (Score:1)
Apple's G4 Cube (Score:2)
-Roy
Expense, and No Need For It (Score:4)
I know PCs don't draw very much power, either, but if the power supply were external, you'd have to include special internal cables, a power socket (or two or three) on the case for these connectors, and some stylish, fancy case for the external power supply. All of which would drive up the cost of the machine $20 (which is a lot these days).
The real issue here is that Intel and the hardware manufacturers are deliberately picking case designs that require several fans to cool correctly. It's as if these engineers never heard of the word convection.
The answer? Buy a decent, quiet power supply [pcpowercooling.com], and some cooling fans. Note that pcpowercooling.com has a new cooling fan that includes a thermistor. The fan spins only fast enough to keep cool, so it actually spins slower, and quieter, most of the time.
Most people don't care if they're going deaf from that little whirring noise, and OSHA's rules don't apply to the home....
Re:IMHO (Score:2)
Re:IMHO (Score:2)
Re:no more dongles! (Score:2)
I like the solutions for guitar pedals out there. My brother has one pedal that provides power to a bunch more. He only needs three or four wall warts (one for the pedals, one for the Ground Control, one for the DSP, and one for the digital recorder, if he's using it). Now he's obsessed with the Voodoo Pedal Power [voodoolab.com] but he wants to build his own with pots for simulating weak batteries, etc.
I really wish there were some standard for local DC power distribution. I envision a data-capable bus where a device could request the voltage it wants, so that the regulator wouldn't dissipate too much heat if it operated on 3.3V instead of 12V.
With something like that in my power supply, I'd be happy. Only X-10 stuff would waste space then.
Actually, that could be workable with existing equipment using small boxes on "legacy" devices... Sounds like a good application for a PIC12C508 for each device, with a PIC16Cxxx as the main power controller...
If anybody else is interested, feel free to email me. We could draft up something pretty quickly with this.
Re:It's a little more complex... (Score:2)
Your right - for the budget, you can't beat what's out there. Now, if you want to spend BUCKS, I can point you to some folks (Did I mention I used to work for a power supply company)
Re:It's a little more complex... (Score:3)
Not arguing your line loss figures, but you can always use a power supply with remote sense lines to make up for the line loss. I used to work in electronics, and we HAD to do it on some of our higher power boxes. When your pushing 120 amps of 5VDC with a +- 50mV spec, you have NO choice
No thanks. (Score:2)
---
Re:It's a little more complex... (Score:1)
You're right, remote sense can correct for the drops. But then you have to face all of the output impedance / load transient issues that the bandwidth of the remote sense can't correct for. And that takes me back to my point that with enough $$ these problems could be overcome, but who would want to pay for it?
It's incredibly difficult to come up with a power system that can displace the $15-20 commodity power supplies that are in most computers. The commodity power business ranks up there as one of the more stressful ways to earn a living. They have razor thin profit margins with no room for error, with most errors resulting in either shock, fire, or smoke issues.
Cheers,
Jim
It's a little more complex... (Score:5)
There have been some good answers, but the biggest reason PC power supplies are internal rather than external boils down to $$. Well, that and safety, but then safety boils down to $$. And then technical, but that's also solved by $$.
There are at least four voltages produced by PC power supplies, with currents on the 5V line up around 12-20 Amps. The regulation of the 5V line measured on the motherboard is about +-5% or better (+-250mV), which at 20A would work out to about 0.012 Ohms of resistance split between cable and connectors. 14 guage wire (similar to an extension cord) has a resistance of 0.0026 Ohms per foot. So, about 5 feet of wire would chew up all of the allowed resistance with nothing left over for connectors. I've been loose with my numbers, but you get the idea.
Now imagine a cable from the power supply to the computer with 8-10 wires, all the size of those in a standard extension cord. Now, imagine paying for all of that copper.
The old AT-class and now the ATX class supplies are commodity products. This means that the design is stable, has been engineered and cost reduced to death, and are now being manufactured in low labor rate locations with component parts being negotiated to sub-fractional pennies.
Custom power supplies used to be priced in the $1 per watt range (in large volumes). PC power supplies are well down below the $0.10 per watt range and going lower. Anything new will have to survive being compared against a mature $15-20 solution.
Cheers,
Jim
Re:IMHO (Score:3)
--
Actually... (Score:4)
Trying to find a place for it was always a pain, because the cables came out front and rear. It was a nightmare. If it weren't for the location of my desk at the time, it could've been stepped on or tripped over. It was the only thing I hated about my A1200.
As far as noise is concerned, I doubt the fan in the power supply is noisy. I have a test setup right now with an ATX powersupply sitting out openly on my desk, and when it is on, it is all but inaudible. Even in a case, I doubt it it very loud (BTW - why all this talk about loudness? I don't mind the sound of computers - give me the thrum of a computer room A/C and powersupply system for company any day!).
I can only see one way having an external supply would be a good thing, and that would be if it had plugs for everything - no more wall warts, only one AC plug, everything else plugs into the powersupply. I am imagining something like a beefy looking power strip - it could even have a built in UPS for certain components. I would be even willing to bet the monitor could run off of such a system (maybe - I have a portable tube based TV that runs off a 12 volt/1 amp wall wart - of course, it has a small screen - maybe a larger monitor couldn't do it).
I would think you could actually build something like this yourself - getting large amperage bare power supplies isn't a big problem, so just build a case, add some custom cables... The only issues I can see you running into would be possibly noise (electrical) and voltage/current drops/sagging (from excess length of runs for the cables)...
Worldcom [worldcom.com] - Generation Duh!
Re:It's a little more complex... (Score:2)
------
Re:most heat & noise from a P/S? Nope. (Score:1)
On my systems, I use mini or full tower cases, and I mount an inward-pointing auxilary fan on the lower front of the case (there's almost always a fan holder on the speaker mount or card holder cage). That way, cool air is drawn in from floor level, and once it gets heated up inside the case, it'll be blown out by the PS fan.
Re:IMHO (Score:1)
G4 Cube (Score:4)
Mine is (Score:3)
So when this system was new, I gradually decked it out with one too many gadgets. Multiport serial boards, CD-ROMs, hard drives, tape drive, I don't know, but at some point it started needing more power than the dinky little supply in the mini-tower case could provide. And I couldn't find a supply with enough juice that would fit the enclosure.
So I got a big supply, set it on top of the case with the cover off, and ran the wires down to power the motherboard and all the rest. With a spot-fan on the CPU, no worries there, and nothing else gets hot enough to cause any trouble.
Maybe this approach wouldn't work on these newfangled, high-MHz, noise-sensitive systems they've got these days. Or maybe it wouldn't be a problem, as it isn't for me. Someone would have to try it.
Incidentally the system is still overpowered for what it does... it runs a dialup BBS and has a custom alarm clock from hell to get me up in the AM. (Wasn't that one of the apps CmdrTaco wanted in the BorgBox?)
--
Re:Mine is (Score:2)
I'm reading this with Opera on a 486 DX2/66 with 32Meg of RAM running Windows NT 4 right now, and I have used it to play the occasional game of Doom against the Duron in the other room...
Re:G4 Cube (Score:1)
Re:most heat & noise from a P/S? Nope. (Score:2)
Re:most heat & noise from a P/S? Nope. (Score:1)
There you have it, there is hope.
Re:most heat & noise from a P/S? Nope. (Score:1)
It seems that often the fans are back to front - i believe its more optimal to have the air being sucked out of the case - if you have the fan pulling air in, then surely you are heating up that air in the psu, and passing it on to the other components?
Re:It's a little more complex... (Score:5)
If only everyone here thought like that!
IMHO (Score:1)
--
Re:IMHO (Score:1)
--
Re:Interfernece, cost (Score:1)
Interfernece, cost (Score:4)
When you have an internal power supply, the case shields against outside and emitted interference. You can shield the external power supply, but it is harder to effectively shield the multiple DC conductors that must go from the power supply to the PC. Inside the case, it is a much more controlled environment. This helps with stability a lot, considering components in modern PC's have to have over 1GHz of timing accuracy. Also, you don't have to make a cosmetically pleasing case for an internal power supply (unless your chasis is translucent), but you have to make it less than an eye-sore if it sits on the floor or desk. That extra casing introduces more costs.
Re:G4 Cube (Score:4)
Re:most heat & noise from a P/S? Nope. (Score:1)
Re:Interfernece, cost (Score:1)
Re:Interfernece, cost (Score:1)
Clustering/Rackmounting (Score:4)
A hardware hacker's dream (Score:1)
The idea of one standardised external power supply shouldn't be dismissed so readily. It could end all this hassle with so many different connector types on laptops - if there was a standard for desktops and towers, laptops would surely follow. As anyone who has ever picked up a cheap second-hand laptop - invariably with a dead or dying battery and no ac adaptor - knows, it can be hell first tracking down the right type for your model, then actually finding an example for less than £100. And let's not forget it would also either allow a reduction in PC case sizes or else leave a good chunk of room inside the case for extra add-ons. As anyone that likes to mess about with old x86s and Pentiums, a few centimetres can make all the difference in what you can do with the case on and the case off.
I like the thinking of others here that seems to have spread to the idea of having an external power supply that can power many machines at once....that would be sweet! I'd no longer have to have a daisy chain of surge protected extension chords stretching from every power socket in my house. I wouldn't have to keep changing plugs(buy as many six socket extension chords as you want, you'll always run out of space). It would be great! If it weren't for the fact I've spent all my money on extension chords, I'd be first in line to buy a PC with external power supply. (That, and the fact I could have a fairly good go at hacking one together myself....maybe)
8)
Re:Cost? (Score:1)
Extension cords! (Score:5)
It's called a laptop... (Score:2)
External power supplies cause interference (Score:2)
Unfortunately, power is not the only thing that rides on a power cord. Keep in mind that you have components operating anywhere between 100 MHz and 1 GHz inside of a computer. Given FM radio is from about 85-108 MHz, UHF TV around 500 MHz, your 900 MHz cordless phones, etc., one sees that all you need to get a computer to cause interference is an antenna that takes said signal outside of the shielding metal case. What better way is there to get a signal out of a computer than a power cable? After all, 50/60 Hz cables were never designed to keep 100 MHz+ signals contained.
Granted, you can fight this noise by using shielded cables, bypass capacitors, etc., but adding components comes at a cost. Do you really want the extra cost of shielding power cables/connectors & RFI that made it into the power supply?
This reminds me of story (I don't know if it's true or not). First, they build your US $100+ priced power supply, and call it Class A. Then they take out a few components (a few dollars worth), see if it works, and call it Class B (aroud $30-50). Then they take out a few more, and call it Class C ($15) if it still operates. Most computers have Class C supplies. No wonder bad or unstable power supplies are the cause of so many failures.
Re:most heat & noise from a P/S? Nope. (Score:1)
Unless you're one of those nuts (ahem) who wants to put the biggest, baddest hard drive available in his computer. Adding my 75Gb DeskStar made my computer go from "is that thing turned on?" to "Hey, something's making a horrible racket in the next room."
External supply doesn't just have to be for the PC (Score:2)
I'm sure that one (big / 300w+)external DC supply is a lot more efficient (and cheaper) than an internal one plus 6 wall-warts for accessories...
The problem is that you'd either have to choose a standard connector type for each voltage, or have some auto-sensing system...
MadCow.
The minute ext pwr supplys exist, more bitching... (Score:1)
Re:Actually... (Score:2)
Re:It's a little more complex... (Score:2)
Re:It's a little more complex... (Score:5)
The other thing is, a long distance between the regulator and the load slows down the response to load variations and degrades the voltage regulation no matter how thick the wires are. High-speed rack-mounted electronics systems often put a regulator on each circuit card, with power distributed at an intermediate voltage, just to get the regulators a couple of feet closer to the loads. It might make sense to do this with PC's -- convert AC to 48 and 12VDC in a separate module, then put simple switching regulators on the motherboard to transform 48VDC to +5, +3.3, and whatever other voltages are needed. One part of this is already in place; most Pentium motherboards power the CPU with a switching regulator located right next to or even under the CPU socket.
You don't mean dongle (Score:3)
--
spam spam spam spam spam spam
No one expects the Spammish Repetition!
powered USB/FireWire (Score:2)
Re:You don't mean dongle (Score:1)
ofcourse dongle is far more useful of
a word 'cause it can be a derogatory
word for anything that plugs-in, such
as an evil copy-protection dongle;
those pesky devices which stick off
of parallel ports and let expensive
applications run.
I will however admit that in this case
wall-wart is a wonderful word and I
should've remembered it.
And they should make those things illegal.
However, people love peripherals and
peripheral makers hate UL/CSA/FM listing
120VAC devices.
Re:no more dongles! (Score:1)
would like to draft up something. I'm not
a member of the IEEE or anything, but perhaps
we could even send in something to them
no more dongles! (Score:5)
Re:most heat & noise from a P/S? Nope. (Score:1)
Standardisation of External Power Sources (Score:1)
Seems to me (and also to Douglas Adams - Author of Hitchhikers Guide etc [see here [douglasadams.com]]) that it would be much better to wire houses with extra sockets on the powerpoints - thus the point will have (for example) a 250vac, a 5vdc and a 12vdc socket. Then every manufacturer creates their products to work using these sockets and thus you dont have to own 1000 'dongly things' .
There will be some equipment (your PC may be one of them) that would need to do its own conversions - maybe for getting a 0vdc or some such - I'm no expert. But for the most part there would be no need for damned power supplies for everything!
Re:Clustering/Rackmounting (Score:1)
I think I used too much water.
Next time I'll just pour a little bit at a time.