I almost forgot this one although that resulted in the biggest loss of all similar events. Lean back, this is going to be a long story, but the weekend is approaching anyway, so maybe I can afford it...

Some 10 years ago we used hoppers (coin payout mechanisms) of a renown English manufacturer. They had a spectacularly solved mechanism (as most of the english products do) capable of handling coins of different thickness and diameter (almost all of the competition have coin-size-dependend parts, so you have to stock a bunch of different hoppers or at least their parts, if your machines operate in different countries, or if various types and sizes of coins and tokens are in use even in one country). It had also a relatively big coin box (that's important for the operators, as they don't want to refill it very often) and a nice provision to prevent the coins from "arching" (if you have many coins, especially if they are dirty and a bit sticky, which they usually are; they like to get stuck leaning against each other in various funny ways; and any shape of the box can really prevent this).

The hopper had a relatively complex electronics, capable of detecting the motor current and "shaking" the motor (moving forth and back) if it got the impression that the motor is stuck; plus it shaped the output of two optical sensors sensing the coin output. It was controlled by a 4-bit microcontroller, I don't remember the brand.

It was very expensive, the most expensive in fact of all similar products at that moment, but we liked its ruggedness and reliability and had installed it in relatively large quantity.

One day, coins started to be missing.

We knew payout happened (we had an electromechanical counter attached to one of the the coin-output-signaling lines) and that the machine's control board did not pull the control line ("run") of the hopper (that information was stored in a log in NVRAM of the board). And when it happened, it was always completely empty, which is rare in normal operation.

Initially we suspected that somebody broke into the machine's enclosure and grounded the payout line manually. But we've seen no evidence of force (the police was investigating it, too, they've brought the specialists with those powders and brushes and UV lamps and stuff) and also it would require certain knowledge and skills (I know the audience here finds it trivial, but I believe most of the people couldn't sort it out too easily without leaving a trace). And it started to happen at various places, first in the city, then in the smaller places nearby.

Finally, somebody noticed that one of the "drained" machines was moved farther from the wall, and this together with the fact that the power switch was on the back of the machine, gave us the first clue. I don't remember who came with the idea but soon half of the company was playing with the power switch, and indeed, sometimes the hopper started out of no reason and dumped itself dry.

It required a relatively quick flip of the switch off and on again. Of course this depended on the power supply itself and on the capacitors, but the point was to put a short down-up glitch on the VCC of the 4-bit microcontroller. It took some experimenting to get the movement on the switch right, but very soon we all managed to achieve almost 100% success.

Of course, the microcontroller had an RC reset. When it went into the "dump" mode, it did not react to the input signal, nor the stuck/overload condition anymore; only a repeated power cycle (a long one, to provide the required reset pulse) stopped the "dumping".

We tried to reach the manufacturer, but I don't remember if they even responded - we needed to act really fast, the plague was already spreading. We added a standard reset (in a crude way - the original product was a fine SMD board, and we did not master SMD technology that time, so we hooked up a nasty big subboard with a 7705 and what's needed for it), then we subjected it to thorough testing - all the staff tried to throw it to the "dump" mode, then we attached a slow-running sweeped pulse generator for a couple of nights. Not a single coin was dropped anymore.

This was so successful, that after we "repaired" all our hoppers, the local *competition* started to bring in theirs (of course, the culprits, finding out that the trick does not work anymore on our machines, moved on to the competitions'... :-) ), so, from the profit of what was already a routine operation, we were able to partially compensate for the loss...

I hope you all enjoyed the story... :-)

JW