We discussed (guessed?) the keep alive issue with a LM3 in a past post. The LM3 has a ground pin on the connector, not present on LM1 or LM2. The ground pin is what facilitates attaching a conventional keep alive. However, there is a caveat - a keep alive connected across the ground pin and the blue positive wire places a power source, the keep alive, "downstream" from the current overload protection within the LM. This means if there is a short on an output pin and the keep alive is charged then there will be no current protection for the output transistor when the keep alive discharges through the short. Output transistor goes poof, output no longer works.
We speculated Tim may be working on a current regulated/limited version of a keep alive. This approach would spare the output transistors in the event of a short but would also limit drive current to the motor to <100mA (output pin limit). Likely 100mA is more than sufficient for motor drive current in all but a few cases like single locomotive under heavy load.
I'm guessing this current protection dilemma may be the reason it is taking Tim a long time to bring a product to market. If the LM had pins that connect to the output of the internal rectifier (upstream of current protection) then all would be well. That likely entails more change than it sounds like.