Rewiring a psu....how to get 3.3v?

Monkey34

Monkey34

Supreme [H]ardness
2FA
Joined
Apr 11, 2003
Messages
5,140
I've got a dell psu that I want to try and convert to standard ATX specs. I already know that the plug looks like atx, but isnt....the wiring is different, with different color codes. I'm PAST that issue, and have all the wires switched to the usuall spots EXCEPT that there are no 3.3 volt lines( i need three - pins 1,2, and 11). After swapping the others, I'm left with (2) +5v, and (1) -5v.:mad:

I know if you wire a line as 12vdc for positive, and 5vdc for the negative, you can run things(fans) at 7vdc ( the difference). What would be involved in trying to wire for 3.3vdc? would i need a voltage regulator? could i use a potetiometer to "dial it down" from 12vdc? Wont adding a resistor to the line change it?
 
You could open up the PSU and look for a 3.3V spot silkscreened on the PCB.
 
not worth the time and effort. just buy a regular ATX PSU. I have an old 250 Watt Sparkle I'll give you for $10
 
Chances are it doesn't even provide 3.3v, if it doesn't have an aux connector either. Only way to know is to look.
 
Buy a new powersupply. It's not worth the effort making a 3.3V line. Seriously, a good Sparkle or Fortron PSU is like $25...
 
Format_C - I just might take you up on that psu offer, but now i'm just curious about this. I found out a linear potentiometer will do it, but I'm not 100% sure on the ratings(also max amps is an issue). I'm going to dig some more. A refresher on Ohms law might do it.
 
The current requirements, voltage flucuation, and noise requirements of modern systems are enough to make it unfeasable to add 3.3v output. Question though: Have you checked inside about the 3.3v pad yet anyway?
 
You cannot use a resistor (ie potentiometer) to set voltage. A resistor's job is to oppose the flow of current, so as the current requirements increase and decrease, the voltage is going to change. If you must use this power supply (and you sure seem dead set on it) look towards a voltage regulator. It is a simple 3 legged IC with in, out and ground as the legs. Personally, I would just replace it as others have said. Good effort though
 
If you want to use a potentiometer, you'll have to use a voltage divider. To do this, you'll acutally need two potentiometers. You want to connect them in series on the 5V line. Then you adjust the pots until the first one drops 1.7V. That means that the second one drops 3.3V. Now you can hook the output up in parallel with the second pot and you will have around a 3.3V supply.

Of course, this is a very bad design. As soon as you hook the output to the second pot, the resistance will change because of the input resistance of the load. You can minimize this by picking really small values for the two pots. If you have a really small resistance (~3Ohms) in parallel with a moderate resistance (~250Ohms) then you effectively get around a 3 Ohm resitance back, so your voltage wont change that much.

The problem with this is that you are dropping the same voltage (3.3V) but at a much higher current. This increases your power consumption a lot. Basically, voltage regulators are a lot better method, as mentioned above. It sounded like you were just curious, so I thought I'd help out.
 
Yea, I really dont think I want to do the voltage divider/pots set-up. Ive seen a voltage regulator online that does +5vdc to +3.3vdc at nice current ratings, and I'm going to look around locally for it.( I'm trying not to spent forever doing this though).

Seems too easy just buying an atx psu ...... it IS [H]arder doing it myself:D Maybe I'll do a window/led mod while I'm at it;)

Oh btw: I cracked it open to look for a 3.3v spot, but untill I can "can-o-air" it, and move some wiring over, I cant see well enough yet. I'll look more tomorrow after work(time for bed).
 
Exactly what kind of supply are you planning on doing this to? There are ways to do it but without a board to mount stuff too and tie things down it could get a bit messy.

Of course, this is a very bad design. As soon as you hook the output to the second pot, the resistance will change because of the input resistance of the load. You can minimize this by picking really small values for the two pots. If you have a really small resistance (~3Ohms) in parallel with a moderate resistance (~250Ohms) then you effectively get around a 3 Ohm resitance back, so your voltage wont change that much.
Click to expand...

A better design would probably be a bridge circuit. Think of 2 different series circuits with 2 resistors in each. You'll want the voltage in the middle to be around 10V on one side and 6.7V on the other or something like that. The 3.3V difference is what you want. You could use a 5V or 12V line to get the values you want. Put the load in between them and you have a 3.3 V supply. Replacing one of the 4 resistors with a pot would probably be easiest otherwise controlling your voltage would be a pain in the ass. Do a little math and use a multimeter to adjust close to the 3.3V and it should work. This should let you keep the voltage pretty stable and prevent a ton of current from flowing through the circuit.

Assuming you use relatively high resistance values you won't have a ton of current flowing through the circuit and still have a 3.3V supply.
 
Of course it would be best to leave the fine points of fixed voltage supply design to something like this (pdf)

a SIL15C-05SADJ-V or 2 should give you plenty of power. Though really, for another $10, you can get a decent PSU so.....
 
There's no 3.3 spot on the psu's board, I looked good. I do have a suggestion from someone that may work good...........depending on availability of the pieces. A 3.3 volt ziener diode with a resistor in series between that and ground. Whataya think?
 
Again, having proper regulation and enough current handling is going to be a problem
 
Think this'll work good, if I can get the right values on the individual pieces.
volts.JPG
 
For a 250watt psu it is( max 6A PER PIN). I'd have to do 3 of these=1 for each pin.
 
Hmm, I suppose so, except the problem with that is that you have a high possibility of having slightly different voltage between the three pins, which are all connected on the mobo. That's bad :(
 
You must log in or register to reply here.
Back
Top