I built a power supply from a computer ATX power supply

Thread Starter

nicklus

Joined Sep 11, 2017
12
I built a power supply from a computer power supply. I used a rotary switch to access the three different voltages [+12, +5, +3.3] to a single output jack. It has a small digital volt meter that I wired in to the output jack to show the selected voltage. The problem is when I switch from one voltage to another [ ex. +12 to +5 or back to +12] the supply kicks off. I have to turn the main switch off, wait a few seconds and turn it back on. It's then ok on that setting till I switch to another and then it kicks off and has to be reset again.
Please keep in mind I'm just an experimenter and love to play with stuff.
 

SamR

Joined Mar 19, 2019
3,240
Instead of the rotary switch changing voltage, use a banana jack for each voltage. They are fixed and will only drop under load. You could switch the meter to measure the voltage at the jacks. The 5V supply needs a low Ω, high watt resistor to switch on. I can't remember the specs for it...
 
Last edited:

Yaakov

Joined Jan 27, 2019
2,894
I built a power supply from a computer power supply. I used a rotary switch to access the three different voltages [+12, +5, +3.3] to a single output jack. It has a small digital volt meter that I wired in to the output jack to show the selected voltage. The problem is when I switch from one voltage to another [ ex. +12 to +5 or back to +12] the supply kicks off. I have to turn the main switch off, wait a few seconds and turn it back on. It's then ok on that setting till I switch to another and then it kicks off and has to be reset again.
Please keep in mind I'm just an experimenter and love to play with stuff.
There are two kinds of switches: break before make and make before break. The former disconnects everything before connecting and the latter connects before disconnecting.

There is a chance that you have the MBB sort and it is shorting the different lines when you swith.

In any case, @SamR’s idea of using separate jacks for each voltage is a good one.
 

djsfantasi

Joined Apr 11, 2010
7,620
What type of contacts does your rotary switch have? If the are make-before-break, you could be creating an internal short and the PS shuts down.
 

Thread Starter

nicklus

Joined Sep 11, 2017
12
There are two kinds of switches: break before make and make before break. The former disconnects everything before connecting and the latter connects before disconnecting.

There is a chance that you have the MBB sort and it is shorting the different lines when you swith.

In any case, @SamR’s idea of using separate jacks for each voltage is a good one.
I think that is what I will have to do. The switch contacts are clearly separated when switched. Not contacting two positions at once.
 

AnalogKid

Joined Aug 1, 2013
9,160
Wafer rotary switches are tricky. Yes, the contacts are widely separated, but the shape of the rotary vane is critical. If it is narrow, you have break-before-make because it cannot span the gap between contacts. But if it is wide, it can reach across the gap for a make-before-break. Can you post a close-up photo of a wafer?

ak
 

AnalogKid

Joined Aug 1, 2013
9,160
I think that should be a 5Ω, 10W resistor.
At least. The rule of thumb for unknown switching supplies is a 10% minimum load on the "main" output. In this case that is the +5.0 V output. Start with the output current rating on the label, divide by 10, use Ohm's Law to calculate the resistance, and then use Watt's Law (voltage) or Joule's Law (current) to calculate the wattage.

Important: double the wattage. A 10 W resistor dissipating 10 W will get extremely hot. For way better long-term reliability (and safer operation), use a resistor rated for at least twice the expected power dissipation.

ak
 

SamR

Joined Mar 19, 2019
3,240
Sounds like a break before make switch to me.
Make before break would be shorting 2 of the voltages and popping the internal breaker which would not reset until powered off or "cooled down". What would the problem be for a break before make? What did I miss here?

Edit: Ahh, I missed a post. Yes the TS said it is break before make.
 

Tonyr1084

Joined Sep 24, 2015
5,856
Instead of the rotary switch changing voltage, use a banana jack for each voltage. They are fixed and will only drop under load. You could switch the meter to measure the voltage at the jacks. The 5V supply needs a low Ω, high watt resistor to switch on. I can't remember the specs for it...
My ATX didn't come with a switch on it. I added one. And I used banana jacks for the various voltages. No need for a meter.
hi nick,
Do you have a fixed load on the +5v output.?
The ATX requires a load on the 5V
E
I don't recall what load I used. I followed someone's YouTube on that part. I think it was the one green wire in my ATX that was connected to something. Either the 5V or ground - I don't remember for sure. Oh, and mine has -12V @ 1A.
1616423219923.png
 

SamR

Joined Mar 19, 2019
3,240
Yup, that gives you +/- 12V for op-amps and 5 or 3.3 for digital. When I was thinking of doing one, I ordered a 10Ω 100W power resistor that was inexpensive. Never got around to doing it...
 

Reloadron

Joined Jan 15, 2015
5,913
Depending on what you read:
"Next we need to provide a small load on the +5V (red wires) output to trick the PSU into thinking its attached to the motherboard and to keep the power supply in the “ON” mode. To do this we have to connect a large resistor of 10 Ohms or less, with a standard power rating of 5W to 10W across the +5V output using just one set of the red and black wires, pins 3 and 4 will do".
The few I have done I used a 10 Ohm 10 Watt which were common at Radio Shack when it existed.
Given a choice I like using banana jacks as mentioned simply because this way you have all the voltages available rather than choosing 1 of 3. I also like the idea of using a LEDs as indicators for the 5 V SB as well as the 12V, 5V, and 3.3V outputs. There is no shortage of tutorials online suggesting ways to go about things.

Ron
 

Tonyr1084

Joined Sep 24, 2015
5,856
5V ÷ 10Ω = 500mA
5V x 500mA = 2500mW (2.5W)
So why a 10 watt resistor? Not that I need an answer, I don't intend to hijack this thread.

Y'all keep this up you're gonna force me to open my ATX just to see what I did way back then.
 
Top