PDA

View Full Version : AT Power Supply - "Power Good Signal"

TheTekMon
01-20-2001, 09:20 PM
How do I make an AT Power supply produce output without making connection to the motherboard. I think it has something to do with shunting the "Power Good" Signal, but what is it expecting? 5V, 12V 24V -12V ??? A Dummy load???

kenja
01-23-2001, 02:18 AM
Yeah, switching power supplies need a load or they'll either burn up or shut themselves down (usually the latter).

I'm looking at Scott Mueller's "Upgrading and Repairing PCs" book, and he recommends (in most cases) that all positive outputs be loaded by resistors for a PC power supply bench-testing situation. Minimum loads: 0 to 0.3 amps at +3.3v (not applicable to AT supplies); 2.0 to 4.0 amps at +5v; and 0.5 to 1.0 amps at +12v. (He doesn't mention a +24 output.)

Using Ohm's law (Volts = Current in Amps X Resistance in Ohms) to determine the amount of resistance necessary for the midpoint amount of current: 3.3V divided by 0.15A equals 22 ohms; 5V divided by 3.0A equals 1.67 ohms; 12V divided by 0.75A equals 16 ohms.

The amount of power that will be dissipated by the load resistors needs to be calculated (so that they are not literally burnt up; they get hot enough as it is): Watts = Volts X Amps

For the 3.3V output: 22 ohms is a standard resistor value, so 3.3V times 0.15A equals almost 0.5 Watts (these wattage ratings should be considered minimum values).

For the 5V output: 1.8 ohms is an available resistor value, which would give us a current of 2.78A; 5V times 2.78A equals almost 13.9 Watts.

For the 12V output: 15 ohm resistors are common, so the current would be 0.8A; 12V times 0.8A equals almost 10 Watts.

I don't know if there is a commercially made test connector avaliable for AT type supplies (seems like there should be).

If you're interested in making your own, I see partsexpress.com (my favorite source for car stereo hardware) has wire wound resistors that will do the job. They don't stock a 1.8 ohm 15 watter for the 5V load, so I'd get two 3.3 ohm resistors rated at 15 Watts and connect them in parallel (side by side). The effective resistance will equal 1.65 ohms, essentially our original target. (Also, you'll find that a resistor's actual resistance is usually greater than, not less than, its nominal value.)

If you're careful, I suppose this test rig could be constructed using stripped wire and alligator clips. If you don't know how to solder, this would be an ideal opportunity to learn. Wire wound resistors are inexpensive (and difficult to destroy by overheating caused with a soldering iron).

[This message has been edited by kenja (edited 01-23-2001).]