Hey Guys -

As I mentioned a couple of weeks ago, I'm building a new bench supply. I'm using an ATX as the base. I've attached my schematic (please forgive the shoddy work, I had to use MS Word). I plan to have five outputs: 3.3V, 5.0V, 12.0V,-12.0V, and 1.5V-9V adjustable. Red "diodes" are LEDs that indicate "power good." The black diode above the LM 317 is (I'm assuming?) a switching diode to prevent backflow while providing feedback for the voltage regulator. Capacitors are added to the variable output portion before and after the LM 317 to smooth things out.

I'm going to use a six pole rotary switch to select a given output and route it to a volt meter (this is something I wanted specifically for the variable output, but hey, why not make it available to all of them).

Any comments or suggestions for improvement?

 

FYI - the resistor at the bottom of the LM 317 is a 10k pot. Also, the meter switch leads will be reversed for the -12V output since the voltmeter is analog and not able to indicate negative voltage.

 

Many ATX power can be adjust the voltage, and it just need to adding a Pot in the IC of the PSU, if you can open the case of PSU and find out the IC number, then maybe you can get a great current and no need to use LM317 if you like to do this.

 

I read about that, too, Scott, and decided against it. I won't be doing any projects that require a lot of power.

 

If you still want to using LM317, then the R2(10K) is too big, you can in parallel a 2K with VR10K on the 1,3 pin, it will be 1.667K.

Then calculating the Vout:
Vout = 1.25V(1+(R2/R1))
= 1.25V(1+(1.667K/220Ω))
= 1.25V(1+7.58)
= 1.25V*8.58
= 10.73V.

According to the calculation formular that you can reducing 2K to make the Vout to close to 9V if you wish to do that.

 

dont forget to put a load on the 5 volt output, some supplies will not work if no load. 1/2 amp is enough.

 

dont forget to put a load on the 5 volt output, some supplies will not work if no load. 1/2 amp is enough.

Yeah, I have a 10 ohm 10 watt resistor for that. Thanks.

 

The few PC power supplies that I have attempted to repair have usually had the +12v & -12v etc scaled off of the 5v supply, IOW, once the 5v is produced, the others are tracked off of it.
The only down side to a Bench supply made from a P.C. supply, there are times when you want circuits fed with isolation to each other, these have a separate common terminal which can be linked to other commons when necessary.
Max.

 

If you still want to using LM317, then the R2(10K) is too big, you can in parallel a 2K with VR10K on the 1,3 pin, it will be 1.667K.

Then calculating the Vout:
Vout = 1.25V(1+(R2/R1))
= 1.25V(1+(1.667K/220Ω))
= 1.25V(1+7.58)
= 1.25V*8.58
= 10.73V.

According to the calculation formular that you can reducing 2K to make the Vout to close to 9V if you wish to do that.

Many specs I see have a 5k pot. I need to go back and look at my notes again. Thanks Scott.

 

The few PC power supplies that I have attempted to repair have usually had the +12v & -12v etc scaled off of the 5v supply, IOW, once the 5v is produced, the others are tracked off of it.
The only down side to a Bench supply made from a P.C. supply, there are times when you want circuits fed with isolation to each other, these have a separate common terminal which can be linked to other commons when necessary.
Max.

I plan on making a couple of isolation boxes from DC/DC converters. That should solve that problem if I encounter any circuits that require isolated power supplies. I think. I hope?

 

I plan on making a couple of isolation boxes from DC/DC converters. That should solve that problem if I encounter any circuits that require isolated power supplies. I think. I hope?

Also a thing to keep in mind is that most PC P.S. common's are connected/referenced to earth ground, you may be able to isolated this by detecting the grounding point and remove it if necessary.
Max.

 

Many specs I see have a 5k pot. I need to go back and look at my notes again. Thanks Scott.

Yes, I knew that, and it will be over half range of VR5K is the same voltage when you adjust the VR5k, if you using 1.67K comparing to 5K, it will be over 3K is useless.