http://en.wikipedia.org/wiki/Race_Condition

sorry not so great news in the call, not what you wanted to hear.

here are all the scans.

to recap
for 8" to 15" base freq ~ 1.4khz - 3.2khz
freq 2 = base x 2.1 +/-
freq 3 = base x 3.2 +/-

driver2.jpg was my preliminary interpretation

some comments.
important to have the collector outputs around 1-2" separated, not together.
wind a / b / c with same # of turns on top of each other
lowest wind (closest to collector) uses slowest freq
the control coils should be wound over top and into middle (winding direction) clockwise
the control coils should meet in the middle between the collector outputs, and be brought into the middle (not come outside)
all the electronics should be housed inside the torroid (not outside)
use tinned stranded insulated wire
fine tune 2nd, for max output, then 3rd for max output then repeat (for max output)
more collector turns for more voltage

sometimes a build would not work, so start again
found older wire (some corrosion) would work better

use a outer circumference wind to stablize electronics ground (notice the 2 grounds mentioned in diagrams) float gnd
2 grounds (9v battery) plus the outer wind ground

lots of buffers everywhere.

too many stories to tell (12 or so meetings over 12 months, sometimes 4 months of no contact)

I figured out what the second set of windings in the drawing is for. If you remember what "S" said, part of the output from the collector was wrapped around all the control coils together as a magnetic bias that helped to intensify the effect - the "furnace that feeds itself". I also noticed that one of the collector rings didn't appear to be attached to anything but the VCC source. Perhaps it's functioning as some sort of potential bias?

Also, I studied the pictures containing the big ICs with the resistors and stuff on them. I looked up the 4009 and it is a Hex Inverter IC. If you look above it in the picture where he has all those resistors tied to inputs and outputs, it looks like the things on the left are actually electrolytic capacitors. He is using the RC values probably to male an oscillator to generate one frequency for each channel. It looks like he has three of those ICs there which create pulses for the three channels (you can see them labeled A, B, and C). So by controlling all of the RC constants, you can effectively make both a digital oscillator for the frequency and a delay line setup to control the pulse width. I haven't studied the circuit in as much detail as I will later but those are my initial impressions from the drawings he gave.

That's my thoughts for now. This is making more sense by the day :-). Also, I checked my rat shack coil spools (the pack you can get that has a red, green, and black set of wire). They are definitely tin plated!

Here ya go pretty coils straight from the bedside: go to downloads and look for gk1-gk27
http://www.overunity.com/index.php?action=downloads;sa=downfile&id=406
You still have to work.

The big chips are 'break before make' when the address lines change. Try a JK flipflop with the !Q tied to CLR then clock it. Very high speed. That is the big secret. Floating ground, floating fets(IRF840), low current, controls all the way around(snake biting its tail).

The fet driver chip is a p605. The clock chip could be a 4 bit binary out. Look at the truth table. 1 line then 2 lines are on at the same time then 1 line.
There is also a funny looking switch in the circuit. I have used a fet. It is in the feedback loop internal to the ckt itself.
All cmos for 9v bat operation in center.
Notice no caps on the coils? That is where we come in and what about the magamp? Iron centers. Color picture posted previous. I have this build 80% done. I prototype now.

Rat shack arent plated anymore.
I got mine from http://stores.sacelec.com/sacramentoelectronics/ search for wh22-

Look at the TPU this way:
All who know are under NDAs. How do they get it out? Recruit somebody who has a clue, who can be shown with out mentioning vital terms and then the recruit takes it from there. Might the recruit be you?
Could be my last build. I am not saying anybody has to follow. I just give

You should not believe everything people say.  You should perform simple experiments before building full-blown devices.  Otherwise, you waste a lot of time and money, and get no where.

GK, Thanks!!!   I think I am getting it.  I think you mean De-Mux's...

Interesting that this goes way back to the "Original" circuits I used for my first try's, though mine were TTL and my "Driver" section stunk.  This IS a great way to line up and sync multiple outputs, very cleanly.  I really wonder what this would/could do, built in CMOS and allowed to float up via non-grounded battery power.   Hmmmmm......   

Rats, now you have got that little curious part of my mind working, that won't leave me alone till I try.  I'll probably go 3 to 8 line and see what happens, as digital logic and counter response WAS my original job description, before LSI became popular.  This will take a while.  (Note: old school shortcut...  If the entire "Bus" in connected, then aligning the chips on the drawing doesn't require drawing all the lines.  Just assume they are all in use.  If not all are in use, the some draw just the "Last" one, and the assumption is beginning to drawn.  Just a comment.)

Wind up a big solenoid coil with 2000 feet or more. HT mag wire.  Insulation looks like bronze.  28 awg is good.  Doesn't really matter.  Use a hollow insulator material for the core or no core if you pot it somehow.

Hit this coil with a few KVDC (at least 2kv), through a spark gap pulse circuit.  This is just a test, so you don't need a full blown build.

Set this up and you are at step one.