Sherlock Ohms and the Case of the Reluctant Robot

Tension rises when a robot being used ingeniously as a test fixture crashes repeatedly in front of the suits

By William Ketel, Contributing Writer

This was an interesting case of a most unusual kind for our company. We were approached by an organization that had been awarded a contract to build two endurance testing machines for a new turn signal switch, which was a new part for a major truck manufacturer.

The manufacturer was Yazaki, and because the switch was a new product for a new customer, the testing would be very extensive. The part was an all-in-one package, so in addition to the turn signal function it also had a twist switch for wiper control, a push-the-lever-in function for the washer operation, and a longitudal axis motion for high beam-low beam switching. In addition the customer wanted to verify the correct calibration for the trip points for the turn signal lever unlatch-on-completion function.

The company that contacted us saw that the fixturing to provide all of these motions would be very complex, and thus very expensive. It also posed quite a challenge to produce two that were identical. My company proposed that the best and least expensive way to do the test would be a small Motoman robot.

This turned out to be a very good choice, since using a robot would save a lot of fixture design time, and a great deal of fixture build time and expense. The cost break-even point was reached before the second tester was completed.

Using a robot as the test fixture made everybody involved look good. The custom fixture was a simple aluminum block that allowed us to mount the switch at the correct angle. A bracket also supported the motor to turn the simulated steering shaft that we used to check the cancel function. It turned out that the smallest motor we could order with the robot was a whole 1 horsepower unit.

Both systems went together very nicely on our production floor, and we were able to deliver them on time to our customer. That is where things got interesting. The convenient power on our production floor was 240 volts, 3 phase, so that is what we used for developing the robot programs and doing the system checkout. During this time everything worked flawlessly.

When we installed the robots at our customer site, the power available was 208 volts, 3 phase, so we set the taps on the internal transformer accordingly. With this change made, I started the initial checkout, which went well. All of the functions worked, the polished and buffed fixtures were still in the correct positions, and the run-through of the preliminary test program ran perfectly.

Then I selected the actual test program to run. During the initialization part of the program the robot suddenly announced an overload and shut down. I first assumed that it had been caused by a power spike, so I reset the robot by cycling the main power switch, and tried again. After quite a few more tries, I went back to the checkout program and it ran perfectly again.

Of course, by this time I had quite a few customer managers watching, since this was a “showcase” project. So there was a bit of pressure. I went over all of my power connections again and verified that they were indeed all correct. I put a meter on the incoming power leads and monitored the voltage as I ran the robot again.

There was just a hint of dip, less than a volt, as the robot went into its failure mode. I did this check between each of the phase pairs, and had similar results. The only real difference was that I was using the 208 volt connection instead of the 240 volt connection. That was the correct choice, since my customer’s building had 208 volt service.

The explanation wound up being at once simple and complex: The problem was that the transformer was sized EXACTLY for the load of a standard robot. When we added the extra axis motor to drive the “cancel cam” on the simulated steering column, it increased the load beyond what the transformer could handle. That caused the voltage input to the rectifiers to collapse, giving us the error. The reason that it had performed correctly on our floor was that the rectifiers fed from the 240 volt taps, so that the only load on the transformer was a low powered winding used for powering the external I/O and some safety circuits.

The solution and repair was to install external transformers with greater capacity, which we did.The result was that the robots functioned correctly, and the customers were pleased that I had been able to find the problem so quickly.

Contributing Writer William Ketel is a hands-on electrical engineer who enjoys troubleshooting and diagnostics and Ham radio,(extra class). His industrial machine projects range fro, an evaporator valve calibrator to a brake drum inspection machine to crash sled controls, and a package to calibrate developmental crash sensors.