http://www.ebay.com/itm/L293D-Motor-Drive-Shield-Expansion-Board-For-Arduino-Duemilanove-Mega-UNO-/221541254774?pt=LH_DefaultDomain_0&hash=item3394e34676 Chris Purola  could this be made to work? Anyone else have an opinion?

http://www.ebay.com/itm/L293D-Motor-Drive-Shield-Expansion-Board-For-Arduino-Duemilanove-Mega-UNO-/221541254774?pt=LH_DefaultDomain_0&hash=item3394e34676
@Chris_Purola_Chorca could this be made to work? Anyone else have an opinion?

The H-bridges on that board are fairly low-current, only 0.6A apiece. It would work for a smaller motor. The biggest hurdle for me would be the software and doing PID. The video you posted awhile back had someone who had used the arduino PID library to handle that along with some hardware interrupts for high speed encoder processing. At that point it’s just how fast you can run the motors and how well the PID tuning works.
The H-bridges can be scaled up using MOSFETS, which it appears is all that Gecko does, there is a set of TO220 packages on the rear heatsink.

Have you seen any open or cheap alternatives to the gecko?

Could there be a PID autotune algorithm, similar to tuning a hot end or heated bed, that uses the encoder feedback in the same way thermistors give feedback ?

That would be AMAZING and awesome. From what I’ve read, some high-end commercial drives have that feature. See: https://www.youtube.com/watch?v=EeyWN5bhexA

Like in the hotend auto-tuning, you have a closed feedback loop here due to the enconder, so once you set up your system (belts and carriages are added, you’ll need to set the PID values for each axis too. I do not see why it could not be automated (something like M303 XYZE)