[msandford]

My guess is that this is why they have so many problems. I'd guess that there's probably an accelerometer somewhere to give them an idea of what the swipe velocity is, so they can know how fast to play everything back.

But ultimately the way they've chosen to do this seems to be very difficult, especially if those are two very long single coils. It feels like it would be much easier from the software side if they had made them addressable somehow so that they could control every bit individually by either energizing or not energizing a particular coil. Then you'd have eliminated the swipe speed variable from the equation and a lot of complexity would drop right out.

Honestly looking at this teardown I'm surprised it works at all.

[joezydeco]

Doesn't seem to be any accelerometers anywhere. There are two hidden membrane switches to either side of the coils. The card is probably using the press and release timing of that button as it travels through the reader to guess at the speed of the physical card.

There are patents for dynamic magnetic stripes that have individual coils for each bit. That is probably way too costly for this design, or perhaps too expensive to license it.