The Power Supply that Really Teed Me OffA one year old computer came in with the symptom of rebooting unexpectedly. The computer would run for hours with no problem, other times not even making it through the boot. It seemed to reboot when you put it to work, but at idle or in screensaver mode, it stabilized. The problem wasn't a software glitch, as sometimes the computer wouldn't make it past drive detection on boot. This detail pointed to a power supply problem, but the only way to know for sure was by eliminating all other possibilities first.
Before sending it home with the customer, I had to be sure I'd found the problem and fixed it. Otherwise, the chances were good for a callback, which isn't good for the business. I checked for overheating. I then unplugged the various drives. I swapped out the RAM. It continued to act erratically. I decided it
might be the motherboard itself, so I checked the CPU installation and prepared for another board. At this point, it was worth trying another power supply just to be sure, but my used parts bin was a little sparse. Not having a new one on hand, I found two more bad ones as I plugged them into the system. Both fired up, but had different effects on the motherboard. On the third try, I found a good one, and after another day of testing the system, decided to call it done and hope that was really the fix.
It was, but I'd wasted way too much time, and found myself relying on questionable replacements. It was a crap-shoot from beginning to end. There had to be a better way. Looking for a decent tool, I learned there was next to nothing on the market in the way of a power supply tester. If I'd had a good ATX tester on hand, I would've checked the power supply right away. This is what led me to build the first prototype of what became the GT4.
Putting this particular power supply to the test using the prototype GT4, it very quickly isolated the problem. When the power draw, or load, on any of the voltages changed, the "Power OK" LED blinked, meaning the regulator circuit was dropping out for a split second each time. Most of the time, it didn't shut down the power supply, but at times, as I changed the loads using the Load switches on the tester, I hit a combination or timing that tripped the shutdown circuits. Hooked to a motherboard, this would be seen as a reboot. It took 30 seconds to verify the problem with this power supply. My prototype would have saved me hours upon hours of work.
That, then, is the story of the power supply that teed me off enough to build an entire company based on beating it.
That's how Dick Greenwood, seen above working on the faceplate design for the GT4, came to develop what he hopes will be a line of various pieces of test equipment. With an extensive background in electronics, Greenwood saw an opportunity.
Most computer techs have no real experience with electronics. Even those who use a voltmeter aren't getting the whole picture. I've been in the repair industry for years, and in electronics, power supplies are at the top of the troubleshooting list. It should be this way in computer repair, too. As a computer is running, the power supply is put under a constant variety of loads. It all depends on what you're doing. This constant changing of power demand has to be tightly controlled by the regulator circuits in the power supply. A power supply can look good under ordinary testing, but glitches can show up as the loads change, like happened in the case I mentioned earlier. Just puitting a load on the supply isn't enough. We need to be able to change the load. If this is an easy and quick test, it'll be used a lot more often, and I think techs will catch on that there are a lot more power supply problems than they think.
It was fairly easy to design the piece of test equipment Greenwood imagined (the prototype is under construction in the photo at left). A set of LED indicators and a meter to read voltages was very much like the testers already on the market, but Greenwood felt they didn't go far enough. It was a simple matter to add the rotary selector switch and a jack to feed into a computer running a freeware oscilloscope program. This added to the functionality of the device, especially for technicians with the background to know just how to use the oscilloscope.
The key feature, though, was the ability to put loads on the power supply and change those loads to simulate real-world conditions. This first tester became the first product for Greenwood Technical.
As Greenwood Technical grows, we should be developing a range of power supply testers, and some other ideas as well. Stay tuned . . .