Some geeky stuff, for those who might be interested.
A signal from the transmitter to the speed controller via the receiver is detected as a pulse of current which lasts for a variable amount of time. This time is dictated by the position of the transmitter stick and is typically somewhere in the range between 1mS and 2mS. The pulses for each channel are sent in a sequence (called a frame) followed by a longer timing pulse, and then the series is repeated. This happens about 50 times a second. The exact length of each pulse determines where the servo should be (or the ESC, which works similarly to a servo in this respect). It's logical, therefore, that the central position of the servo – or the neutral position for the speed controller – should be around 1.5mS, or halfway. This allows the same amount of travel in each direction. i.e. 0.5mS. Any signal received which is outside the limits (1-2mS) is rejected by the speed controller as invalid, and the thing usually goes into failsafe and stops the motor.
Things being what they are, the exact value of the neutral signal can vary between one transmitter and another (or one manufacturer and another). If the speed controller didn't make any allowance for this then the motor would probably run slowly in one direction or the other when you switched the set on, and it would need a small amount of trim to stop it. This is annoying and so many speed controllers require that you set them up with the neutral, full ahead and full astern values initially. So-called Plug-and-Play or Autoset speed controllers do away with this. They are programmed to "listen" to the first signal they hear when the set is switched on and, once that signal pulse has been received say 20 times without changing its value, settle on that value as neutral irrespective of what that value is. Thus you can have a set whose neutral signal is quite a long way from 1.5mS and the motor will still remain at dead stop if the transmitter stick isn't moved for a fraction of a second on switching on. OK so far?
NOW if you set the transmitter stick at fully back when you switch the radio on, the throttle signal will be at some extreme value like 1.05mS and that will be identified by the ESC as neutral. The speed controller is programmed to cope with signals 0.5mS either side of neutral because it knows those values correspond to its full speed ahead and astern. Those limits in our hypothetical case are now 1.55mS and 0.55mS. However, once you push the stick more than halfway up, the signal value exceeds the 1.55mS which the speed controller can accept as valid – so it shuts the motor down.
Push it further forward and all sorts of strange and unpredictable things can happen. Return the stick to any position below about halfway and the motor will respond again. Pulling the stick all the way back should stop the motor, but you won't be able to get the motor to reverse because you can't pull the stick back any further than all the way.
Now compare this with the behaviour of Peter's ESC and you'll see why I suspected that this is what he'd done. All we need now is for Peter to reply!
Dave M