Hi,

I was thinking of implementing a slim-terminal with a network boot and stuff, for which it would be nice to have an external power supply, due to slim-nature. The requirements are relatively small, because the PC will not even have a hard disk (just motherboard, CPU, Memory and integrated video/wlan card).
So, the idea is to build an adapter from say 12V (24 or any other) DC PSU to ATX motherboard. (req. +12V, -12V, +5V, -5V, +3.3V).

As for the more general picture, the idea is to have a laptopPSU-like thing externally and then some sort of DC/DC converter board inside the case.

Could anyone come up with any ideas?

Sincerely,
Lexter

 

They have 12V Mini ITX power supplies
mini-itx.com - news

__________________
Bill
Smart Kits build Smart People
http://www.blueroomelectronics.com

 

Well, yeah ... Actually, it is possible to buy it for $20+, but the idea is to do DIY DC/DC converter.

 

It's one of those more time than it's worth projects. Sure you could do it, but it's going to be bigger, and likley less efficient, you'll probably have a few initial failures, probably to the destruction of a few major components that will over time cost you more in components than purchasing a ready made unit.
You should pick and chose your project time carefully. There are off the shellf units that you can buy for slim terminals, (check out new egg that have some decent mini machines) That or single board computers are a good choice as well, they can be incredibly small (less than a 6 inch ^2 foot print.)

__________________

 

Well, I see ... I can, of cause, go and spend couple of hundreds/thousands and buy a nice diskless setup (server + number of ws), but the idea is to build everything myself, not just go and buy readymade things.
Would I be asking for an 'where to buy' advice here?

The thing is that I'm interested in building it, not in having readymade.

 

The rest of it is probably worth building yourself, just not the power supply.

If you really want to build the PSU yourself, then look at the Linear Technology, Maxim semiconductor and National semiconductor websites for ICs which you can use to make DC-DC converters.

__________________
I also post at the following sites:
http://www.stop-microsoft.org http://www.heated-debates.com
Screen name: Aloone_Jonez
And http://www.silicontronics.com, same screen name as here.

 

Well, the PSU is perhaps the hardest part, of cause. But the thing is that the only trouble with it is producing stable 3.3V line(s), the rest seems to be obvious to some extent. Shouldn't be that hard, cause diskless system shouldn't require much power or am I wrong about this?

 

Yes. Any other voltage is not that problematic and they use MosFETs as synchronous detectors to reduces losses at 3.3V. Loop control also , at 3.3V poses issues.
One can do it but time taking and with three or four iterations on transformer design. I think there are application notes from TI and others.

__________________
Regards,
Sarma.

 

just in case anybody is interested ...
This project is actually possible to implement. Just been browsing dc/dc converter IC section of maxim-ic.com ... 3.3V is possible with quite a cheap ICs (perhaps even sampling is OK). So, it is not as bad as it seems ... The only thing to do is to calculate required max. wattage for each line. (I seriously believe that 2-3Amps or even less should do for diskless system). The implementation of the rest of the lines (5, -5, 12, -12) is obvious.

 

lexter, the 3.3v line isn't used for disks, it's used for the logic on the motherboard processor and memory.. Drives usually use 12 volts. Even laptops use 30-40 watts of power a desktop system will likley use significantly more.
The 3.3v line on my generic ATX supply is rated at 14 amps. I would strongly suggest you test the system you're going to build before you build the supply, power supplies are a rats next of wires so just find all the 3.3 volt wires (they only go to the motherboard connector) and test them on a clamp meter so you know what kind of power your system really needs. I'm guessing it's going to be under 10 amps so you could probably hook it up to a multimeter for more precise current usage. A clamp meter will give you a clue. I'm guessing even an older slower processor is going to be a minimum of 20 watts that's going to be a baseline of 7 amps, and I'm going to hazard a guess the actual power is going to be higher. What are the specs of the system you're building so far?

__________________

 

well, obviously, 3.3 line is not used for disks. thanx for commenting about it The specs are not known yet, and the wattage is to be checked. I was just saying that if there is no fancy gpu, fd, cdrom, etc it would consume much smaller ammount of power.
Actually, processor power requirements can be checked from the datasheet. (same goes for m-board)

 

Right. So what are the specs? I checked Wikipedia they have a nice entry on CPU power dissipation. The original Pentium/P2/P3 are in the range of 10-50 watts it only goes up from there. So you're looking at a current draw minimum just from a very slow processor of 3amps, the higher end is 15amps. Looks like any modern processor (aside from mobile versions) is going to be in the 40-80 watt range.

__________________

 

Well, 15A * 3.p3V = 49.5W ... Not that bad though ... in the worst case, can put some small sized PSU (was it called µATX?) in test versions ... Just was saying that if I would want to buy the readymade product, I would just go and buy the whole setup, not just a PSU. (And wouldn't, of cause, ask for advice on electronics forum.)

 

If you don't need the speed, you could underclock it to save power.

__________________
I also post at the following sites:
http://www.stop-microsoft.org http://www.heated-debates.com
Screen name: Aloone_Jonez
And http://www.silicontronics.com, same screen name as here.

 

hmm ... nice idea! The speed is not of any importance in such a project. I will take that into consideration.