To the readers of this thread:
As the person on the other end of this technical support case, I can provide a different perspective to this discussion and answer some of the questions raised here. I remember this case because Dave is the only customer who has ever asked to return a motor controller for a refund (as opposed to a replacement).
The motor driver in question is the LB1836 from Sanyo, which has built-in diodes across the driver transistors and is rated for a maximum of 1A per channel. We have shorted the outputs of this driver for many seconds without detecting any damage. Of course, the package is small, and it can get hot, but a thermal shutdown keeps the device from destroying itself.
In response to Ed L.'s questions, the package is rated for up to 800 mW, but on a larger PCB than that of our micro motor controller. This makes it difficult for us to give an exact power rating, and the product is not really intended for users who want to be doing that sort of calculation. However, we are happy to discuss these issues with customers who call or write with any questions.
In that vein, it is worth noting the mistake (or incompleteness) of Ed L.'s analysis: the power calculated (I x V) is the total power consumed by the whole circuit. This is not the power that the motor driver must dissipate, and, ideally, all of this power should be delivered to the motor. The voltage drop in the motor driver at 400 mA is claimed (by Sanyo) to be 0.4V. Therefore, the power that would be dissipated in the motor driver at 250 mA would be under 100 mW, or well below the power rating of the motor driver.
We certainly did not ship back the same motor controller to Dave (the one he returned was visibly altered by the heat), and we made extra sure that the replacement had no problems when we shipped it. These devices can be damaged by improper handling, and there isn't much we can do if someone connects power backwards or unloads a large jolt of static electricity into a tiny semiconductor.
In terms of the support call itself, as even Dave wrote, I tried for a considerable time to help troubleshoot his application. I tried to explain what Gordon McComb wrote, that "it has to be *something*", but Dave was very reluctant to discuss the possible sources of the problem. I eventually found out that he was powering the motor through a linear voltage regulator, and I suggested that Dave at least try with batteries. Even from the limited test done while I was on the phone, the performance was significantly better, but Dave's patience was apparently up. (Powering the motors through a regulator is a bad idea because the varying load of the motor makes it very difficult for the regulator to maintain the proper voltage. We've seen voltage spikes of several tens of volts with just a small motor connected directly to a regulator. Any PWM, of course, makes the matter even worse.)
Of course, the motor controller is not for everyone and every application. It is made for small motors, and it is a component, which means the rest of the system must be designed correctly. At Pololu, our primary goal is to design products that make the system design as simple as possible, and we try our best to make sure that our customers are successful with their projects.
Sincerely, Jan Malasek