View Single Post
Old
  (#69)
BrianG
RC-Monster Admin
 
BrianG's Avatar
 
Offline
Posts: 14,609
Join Date: Nov 2005
Location: Des Moines, IA
08.01.2008, 01:52 PM

I had an old broken Antec TruePower 330w PS laying around in my parts bin. The motherboard connector had melted at the 5v pins, so it was either throw it away, repair it, or do something else. I chose "something else". It is rated 17A @ 12v, so should make a decent 12v supply.

This supply is a bit different than most supplies I've seen. This one has seperate feedback lines for each of the main supply rails. So, on each of the main 3.3v, 5v, and 12v wires, there is a smaller gauge wire of the same color that is connected to the main wire at the connector. Presumably, this feedback wire reads the voltage at the connector. During heavy current draw, there could be voltage drop along the wires, so the PS can compensate for this by boosting it a little.

Loading the 3.3v and 5v lines did nothing to change the 12v line. This PS uses seperate feedback circuits for each line. Nice. This is the way all supplies should be built IMO.

Also, there was no fine-tuning potentiometer to boost the voltage.

So, I got to thinking about those feedback lines; what if I use a voltage divider to trick the 12v circuit into thinking it's only putting out ~10v? The circuit would then boost the output so the feedback wire is at 12v. A quick test confirmed this. After some playing around with different resistor values, I found that anytime I tried to go over 13.5v, some other protection circuit would engage when powered up. I was hoping for ~14.5v, but that's ok. Also, it seems I needed to draw at least 1A for the feedback circuit to act according to my calculations. 1A at 13.5v is too much wasted power IMO. The resistors would heat up for nothing, and I'd have that much less 12v current available.

So, I decided to use a couple series connected diodes with a resistor to draw ~250mA. The combined diode voltage drop of ~1.5v at the sense wire will boost the 12v to 13.5v. I used two 100ohm 10w resistors in parallel to get the 50 ohms I needed. Then, to make sure the diodes wouldn't overheat, I used 3A barrel diodes because they have larger packages to dissipate any heat. Below is a picture of the circuit.



No on to testing. The output of the supply was actually at a default 11.89v with the sense wire attached to the power wire (no v gain). So, after the circuit was added, the actual output voltage rose to 13.35v. What surprised me is that even at over 16A, there was only ~0.1v of voltage drop! Wow, talk about good regulation! I don't think I've ever seen a supply (linear or switching) that doesn't sag more than that when run at so close to the max value. This supply just became my new field PS!

Other mods I did: It comes with two fans. I removed the rear fan and kept the one that blows directly on the heatsinks/circuits. Where the rear fan was, I attached a piece of plastic, and on that is where the banana jacks go. The resistor/diode package was mounted to a piece of channel aluminum (for protection and heat dissipation) on the rear of the PS. All extra wires were completely removed for neatness and I used Deans 12GA wire from the PCB to the banana jacks.

Last edited by BrianG; 08.01.2008 at 01:54 PM.
  Send a message via Yahoo to BrianG Send a message via MSN to BrianG  
Reply With Quote