I’ve recently completed a number of printed circuit boards to help control the rides in carnival. After a few turns on the PCB due to operator error and inevitable “betterness” over time, I arrived at my final design and fully populated 4 PCBs will all the parts necessary to control the rides.

After getting all the PCBs built and mounted, I connected a few rides and tested the software. It all worked perfectly for a while, then some of the rides would randomly restart. Hmmm, clearly a software problem. Many hours ensued, many small issues resolved, but still restarting over time.

Then I noticed that two or three of the rides would reset at the same time. Clearly not a software problem, because the rides all have different sequences and different designs. Hmmm, it started looking like a hardware problem. Using my voltmeter, I determined the power supply was cutting out. I assumed with all 4 boards connected, I was approaching the limit of the regulator chip and it was shutting down. I ordered a commercial switching power supply rated at 5V, 2A. That should be good.

Not so much. Although the restarting was less frequent, it still happened. Further debugging showed it happened with only one board connected, though much less frequently. Thus, one board could not be approaching 2A.

Without access to an oscilloscope, debugging the problem much further became very problematic. A quick search on the Internet yielded tons of new and used oscilloscopes – for hundreds of dollars. Hmmm – again. I don’t really want to spend that much money, so I sought alternatives. I eventually found a low cost, PC-based oscilloscope that got good reviews in some magazines. I ordered it and was ready for detailed debugged.

Once I got the oscope, I connected it to the circuit board and was a little shocked at what I found:

Though an extreme case of considerable harmonics, the amplitude and duration of the noise was pretty consistent every time the relay released the contacts. Note the peak-to-peak voltage of 23V on a 5V circuit. Wow!

It took some debugging, but I determined the issue was not the relay de-energizing, but rather, the switching of the AC voltage for the ride motor. It seems the rapid removal of the voltage was coupling to the signals and ground plane of the PCB and causing some pretty deleterious effects.

I’m in the process of adding a snubbing circuit to the motor to see if I can quiet the noise. More when the problem is resolved or mitigated.

Update – 3.28.09

After adding a snubber to both the incoming AC and the outgoing signal, I was able to eliminate nearly all the transient noise. Here’s a picture of the noise after adding to the outgoing signal. I was not able to get the oscope to trigger at 4.8mV after adding a snubber to both in and out. I hope that means the noise is gone.

Noise Reduction

Noise Reduction

Seems like this “project” is now complete. Next step is to get all the boards wired, mounted, and we’ll be done.