Jfet joule thief

I have made npn joule thief's and ran into a reference to an old patent that showed a JFET joule thief, but I can't get it to simulate, works good for npn though.
Can any one tell me whats wrong, I have tried several variations.

You have tried the simple one, reversing the feedback winding connections?
lack of positive feedback is probably going to be the reason, or lack of resonance.
Have you seen the schematic for the 'thief that just puts a fet in parallel with a bipolr transistor?, I've tried it, it works.

With your sim:
put a cap (e.g.10n) across L1, set V4=1V, check the 'Start external DC supply voltages at 0V' and 'Skip initial operating point solution' boxes, sit back and enjoy.

I tried your suggestion and there was just no way I could get the sim to oscillate, did you? can you post the .asc
Thanks,
Kinarfi

The other change I made to your sim was to shift the ground symbol. Don't know why that made all the difference.

Thanks Alec, Don't know how the ground symbol got turned on it's side, but it effectively cuts the wire when it's like that. I changed the inductance to 5uH and it failed until I upped it to 500uH, I just guessed at what the inductance of the theif was, and I doubt this would ever drive an LED like the thief, just a 1K load kills the oscillation, fun little exercise, thanks for playing.
Kinarfi