Right. Lucky that I had a T6L knocking about! I have just had a play with the set. Using three Futaba servos and a Viper 10 Marine in ch3. Using a 6v 1600mAh pack to drive everything and an RS380 motor for load.
I can confirm that all channels except ch3 have a 'last valid signal hold' function on loss of signal. So wherever they last were, they stay there!
Ch3 sets its failsafe position on the turn on of the transmitter, the position based on wherever the stick is when the set turns on. The problem is the ratchet, it is not a precise instrument for repeatedly setting a precise failsafe point. So, while testing the system, turning on and off repeatedly, toggling the throttle in between to make sure everything is working, the stick does not return to precisely the same spot every time. So on turn off and failsafe, sometimes the motors creep a bit, as described previously.
If you repeatedly turn off and on without wiggling the sticks, the Viper speed control sits there and does nothing every time. So my thoughts, as mentioned previously, either source a spring return kit for the left side stick on the transmitter (might be a bit of a challenge, all the brainy people at Ripmax are currently furloughed!). Or file a slightly deeper vee in the ratchet inside the set. This needs to be in exactly the right place, dead centre on the throttle range. Best way to mark this is with a bit of tippex on the side of the ratchet, then mark with a pencil where the metal spring arm touches. You could lock the stick off in neutral with a big blob of Blutac on the outside face just to be sure.
Done right this will give you a positive position for your stick every time it turns on. This will set the failsafe in the right place every time you turn on. What it doesn't do is explain why your system keeps going into failsafe in the first place. It would be a good idea to see if you can trace the problem and solve it, rather than rely on the failsafe to get you out of a jam every time.