hi Jesus
i've been thinking more about your circuit
while it's true that connecting your 1N4007 causes a direct DC path from 12V to the 1.5V rail of the battery and LT device, i don't think this is the direct cause of the damage
the 12V point is the output of the LT device and the 1.5V point is the input of the LT device
so i believe that the direct cause for damage is that connecting the output to the input via the 1N4007 is causing the LT device to go into external oscillation - and this oscillation, if it occurs for long enough, is what damages the device
there is some evidence which supports this idea: you mentioned that connecting the 1N4007 causes a high-pitched whine from the coil, and the neon gets a lot brighter - both are signs that the circuit has started generating strong oscillations
manufacturers of devices like the LTxxx advise users to avoid such feedback and oscillation, but i'm interested by the details you've reported
so why not let it run automatically, in a controlled way - as you first intended - and tell us what you find!
my previous circuit is aimed at blocking the damaging feedback to allow the circuit to operate continuously with the coil-collapse energy feeding back to the battery
however, as i've mentioned, i think you'll find that when the feedback is only the coil-collapse energy on its own (ie. the usual method used in JTs & Bedini, etc) that although the efficiency of your circuit should increase - you probably won't see that dramatic recharge of the battery
so i suggest you try my previous suggested circuit first and try to get a measure for any increase of time running from, say, a AAA or AA battery (smaller is better, if it can drive the motor, because it won't take so long to wait for the battery to discharge)
you could then add a circuit, similar to the one below, which DOES cause the 'unwanted' oscillation across the LT device - but only for a fraction of a second, say, every few seconds
then, repeat the discharge test with the new circuit added and see if this gives better or worse efficiency for the 1.5V battery
the new circuit, which allows occasional DC feedback from the 12V point to the 1.5V battery, is controlled by a similar switching circuit to the one you've posted recently
once VR1 & VR2 are adjusted to give a short pulse every few seconds (like your manual 'tap' of the wire onto the battery), then the pulse will switch on Q4, via the opto-isolator, to connect diode D4 to the battery
you would use this new circuit in addition to the previous DC blocking feedback circuit - on this new diagram i've shown where C1 & L2 from the previous circuit get connected (still in the same place as before)
it would be best, if you decide to try this circuit also, to use a separate battery for the new switching arrangement at first, just to see if the new circuit improves the efficiency of your system - it's decoupled by the opto-isolator, so it won't add or draw any energy from your basic system
the new battery could probably be 2 or 3 AAAs or AAs in series
let me know if you'd like some guidance with component values
hope this helps
all the best
np
http://docsfreelunch.blogspot.com