Bob was given the task of evaluating a wall wart for use in a new product. Rather than design a little power supply like this, it’s more cost effective to buy one that’s usually imported. The problem is that often the specifications are not too clear, and the pedigree of the design unknown. So, one approach is to take a sample of units and test them for a sufficiently long period that the Mean Time Between Failure (MTBF) can be approximated. At the very least, running units under conditions of operation that are similar to how they will be used builds some confidence that the wall wart will not be the weak link in your company’s new product.

The old way of product testing

Before Soft-I/O existed, the process of testing something like a wall wart was kind of complex and involved a lot of hardware and software. In our company, before we had Soft-I/O, to test a wall wart, we would have used a PC-based data acquisition system. It was usually a card-based or USB-based module system. The card or module came with some software that allowed you to set up the data acquisition and then log to a file. Sometimes it was an Excel-compatible file, sometimes not.

One of the biggest problems with the old way was that the data acquisition was not coordinated with the control part, so we would have to set up some kind of I/O card or module that we could interface to a relay. That meant getting a power supply and relay and then wiring it up. Then we wrote some kind of software that would cause the relay to open and close to power up the unit under test.

To correlate the sampled data--temperature, voltage or current--we would usually rely upon the regular period of the on/off relay actuation and then line up the data after the fact. It worked, but it was prone to error. Other ways of product testing.

This is a brief application example, not an in-depth treatise on product testing. In fairness, there are commercial products that are either specifically designed for product testing or can be programmed to do so. They are better integrated than our Do It Yourself solution above. The drawbacks of these approaches tend to be extraordinarily high price coupled with inflexible hardware and unusually complicated software. They are really like with various hardware input and output modules in a backplane structure. Software can be graphical or algorithmic. These systems are way more complicated to use than an oscilloscope, and that was the vision that Bob had.

The new way of product testing with Soft-I/O

We have now set up the problem such that you will probably say, “Foul!” We have pitched ourselves a nice slow one right over the plate. Our solution is so good you will end up thinking that we created the problem just to show off Soft-I/O. After all, Soft-I/O solves the problem faster and simpler while being less prone to error and far less expensive. Here's how Bob made his wall wart tester work very well quickly.

Bob decided that to test the wall wart properly, he would measure the output voltage under various load conditions. He decided to use no load and then 10%, 60%, 100% and 110% load. He also reasoned that he wanted to see how the wall wart handled loads applied after power up and before power up. Bob knew that it's easy to build a power supply. It's much more difficult to make them start and stop and handle changes in load.

Bob built a small panel that held four relays He soldered up three load resistors and put them in a small, ventilated metal box. He then put a 120VAC outlet on the panel and wired it up. One relay switched the 120VAC outlet, so he could insert the wall wart under test into the outlet and this relay would apply 120VAC to it. The other three relays were hooked to the load resistors, so with the four total relays, he could power up the wall wart and apply zero, 10%, 60%, 100% or (10% + 100% = 110%) load. The output voltage of the wall wart, in parallel with the load, was the signal to be measured.

Safety Note:

Our first goal in designing test equipment is safety. Test systems can be unattended, so they must be designed to run in a safe manner and be safe to those who may casually encounter them. For this case study, the goal was to build a safe panel that would meet electrical codes and not constitute a shock or burn hazzard to someone in the laboratory. This test panel is not built to the standards of an industrial control system. The test fixture is a temporary fixture that will be disassembled at the end of the test and the components re-used. Bob employed standard wiring practices, using a three-prong 120VAC electrical cord conneting to a laboratory outlet. The load resistors are placed in a metal enclosure on a metal panel such that temperature rise is on the order of 10C. As always, XiO strongly recommends putting saftey first. Computer controlled input/output sytems must always be fitted with supplimental protection that will protect the user and equipment in the event of a single point failure. Limits and protection must unilaterally act to protect operators, equipment and property.

Bob then wired it all up which was very simple with Soft-I/O. He did not even bother to employ common grounds because he had plenty of Soft-I/O pins. Rather, he hooked up four relays using eight pins. In order to allow Soft-I/O to measure over- and under-voltage, Bob put a voltage divider on the wall wart output such that at 24VDC, it produced a 12VDC signal. He then simply hooked these two wires up to Soft-I/O and configured the pins for Versatile Voltage Input of -12 to 24VDC, so his measuring range of the wall wart voltage was -24 to 48VDC. If the wall wart voltage was outside of this range, it was certainly defective and not meaningful to measure.

At this point, Bob was done with the hardware. Using his browser, he checked his wiring by actuating the four relays one at a time. He could see the wall wart output voltage on his screen while he toggled the four relays. He now wrote a few sequences that defined his cycle. His test had the following steps:

1. Power up the wall wart with no load and measure the output voltage, recording any deviation outside of the specification.
2. Power down and allow the output voltage to decay.
3. Repeat steps 1 and 2 for 10%, 60%, 100% and 110%.
4. Power up the wall wart with no load.
5. Apply a 10% load and note any change in output voltage beyond specification.
6. Increase the load to 60%, 100% and then 110% each time noting any overshoot or undershoot.
7. Increment the number of tests counter.

Bob wrote and verified each test using the Soft-I/O syntax-free sequence engine. He then wired up an LED indicator light that he turned on after each full test. It gave him a nice visual indication in the lab.

With a final flourish, Bob set up his testing sequence such that if a test error occurred, the Soft-I/O module would send him an e-mail reporting the failure. Bob made his module visible over the Internet, added a password screen and had himself the ideal test system. He could start the test and then check on it anytime day, night or weekends. And, if something went wrong, he would know about it via e-mail. When I last talked with Bob, he had run nearly 1,000,000 complete cycles on his test unit. It turns out that his wall warts are very reliable!