The inventors don't tell how it works so all we have are theories, ... many, many theories :-)

Other than the many demonstrations of the working machines, Paul Baumann, the main inventor, gave two demonstrations of basics that serve only as hints of how some of it might work, but even then, not with enough detail to replicate even those demonstrations. They are refered to as the Pinciple Experiment and the Linden Experiment:
 http://rimstar.org/sdenergy/testa/principleexp.htm
 http://rimstar.org/sdenergy/testa/lindenexp.htm
Methernitha's webpage (Methernitha is the name of the group that made the testatikas) also has some details of the workings a little further than half way down:
 http://www.methernitha.com/Mether_2/Free_energie/free_energie.html
-Steve
http://rimstar.org

Here you have some usefull schematics:

(http://www.gewo.cz/ve/Testatika-FED/full-circ.gif)

Hi,

I also found a possible theory:
http://utenti.lycos.it/fischerconsulting/testatpu.html

F_dyne

If, and it's a big if, the Testatika is an electrostatic device working on the principle of electrostatic induction, then I'm sure that plenty of voltage can be obtained at really low rpm. Once the machine has reached an initial charge, it's easy to keep it charged with very slow rotations (as long as the design is clean: no sharp edges to spray corona, good insulation, etc.) The problems are the "down-conversion" problem, as sm0ky2 has outlined, and the problem of where the extra energy must come from (it can't be coming from the rotation per se, or the rotation's power source, this at least we know.) I agree that the key might be in what is happening between the disks.
I've got no problem generating high voltages at low rpm. I'm working on the down-conversion problem, but my best idea so far is kind of like a reverse-Marx bank, and that seems wasteful of energy. And I'm looking between the disks for any possible source of excess energy (assuming all the while that Testatika is real and electrostatic.)
In my sectorless research machines of the Bonetti type, I have often observed an extremely unusual plasma or corona between the disks (the normal and expected corona is on the surface and edges of the discs). This corona occurs in the sector of the disks between the neutral structures and is related to the voltage produced by the machine. In my designs all neutral structures and pickup combs are independently movable, and it is a key feature that is not often seen in other designs. The angle of the "V" formed by the neutral structures on either side of the machine (one "V" facing up, the other down) determines the voltage to which the machine will charge, and it is in this region that the strange plasma occurs. With the angle of the V wide, as in most Bonetti and nearly all Wimshurst machines, the voltage will be relatively low, and this special corona may not be seen. But in my design the angle can be made more acute, and when it is small the machine reaches much higher voltages and the strange corona may be observed. It is paler, and maybe pinker, than the normal edge corona, and is shot through with tiny twinkling sparkles, like a starfield.
You can barely see it in the picture below. Look at the top region of the disks, in the area between the neutral structure supports.
(The left-hand sphere is 5.5 " (140 mm) diameter. The sparks are estimated 200 kilovolts, 200+ mm in length. The exposure is 15 seconds at f/2.8. I counted 13 sparks in that time. The white blur at the right is a pvc rod I use to focus the E-field to initiate the longest sparks.)

Did someone say that when 2 plates of a capacitor  with a charge on them are brought closer together
the voltage drops, but the charge stays the same?
If you charge up a capacitor with a compressible dielectric the plates are drawn together by electrostatic attraction.
If you use a high enough voltage you should get quite a static attraction and thus quite a bit of compression.
This type of capacitor would compress the high voltage static into a lower voltage that could be utilized.
A switching circuit could draw off the compressed voltage to an outside circuit.
Once the static attraction is withdrawn the plates again separate ready for the next charge.
Setting up a number of these caps. could provide a continuous flow of power.

One question I have; Would the plates on the wheels approaching each other work in a similar manner?
The closer the plates get to being directly over each other the more the voltage would drop and the capacitance increase.
Then as soon as the plates are directly over each other a brush makes contact with them and draws off the compressed energy.
Then as the wheels continue to rotate they pick up more static charge and repeat the process.
 
So this presents 2 similar yet different concepts as to how this machine might work.
Maybe it uses both concepts together.

@sm0ky2,
Are you using a Wimshurst machine? And do you mean that in place of the left Leyden jar you put a small tube inside a large tube and the same for the right Leyden jar? How are you observing there's more charge?
-Steve
http://rimstar.org   http://wsminfo.org

charge potential measured by a series of spark-gaps.

i can adjust themese to make longer or shorter sparks, and of course inconsistencies can arrise due to ionization of the surrounding air. some of my spark-gaps are slealed in a glass tube (i.e. broken fuse) which helps eliminate that problem, though those types are not adjustable.


The leyden jars remain attached to the machine.
1 tube of perforated aluminum carries the normal charge from the leyden jar and will produce a given spark with a given number of turns from the machine. This is consistent for the most part,  number of turns on the handle equates to a specific charge value (+/- a little). Though there is a time-constraint on this, as the charge will 'leak-out' at high levels if you turn too slowly or stop turning the handle.

The phenomenon occurs when you bring another piece of perforated aluminum near the original charged piece.
( while its charging or even after it is charged)

this second piece of aluminum will take on an equal and opposite charge from the surrounding environment.

this is completely independend from the equal and opposite charge that is collected in the other leyden jar by the machine.

thus, setting this up on both terminals essentially 'doubles' thge charge potential collected by the machine with a given number of turns on the handle.

adding a 3rd piece of perforated aluminum induces an identicle charge to the first.

so with the same number of turns on the handle you now have 3 charges   - / + / -   all identicle i potential. ( 6 charges of you repeat on the 2nd terminal)

once 3 are charged, you can discharge the original charge in the machine by sparking it to the 2nd terminal-ball
then without turning the handle, a second, 3rd, 4tth, 5th....
spark can be obtained from the terminal connected to first piece of aliminum. This occurs until the charge has dissapated from the other 2 piece of aluminum and is no longer able to induce a charge. ( i assume this is from leakage around the holes and edges of the aluminum)

The second, third, and subsequent plates - even though they are larger, will not induce a charge greater than that of first plate. When the first plate max's out thats the max charge value for all subsequent plates. This differs from the atual designed reported to be used in Testatika, as their 'inner terminal' is a coil inthe center of the charged plates (tubes, cylinders, rings)

i am unable to experiment with more than 3 layers at this time, until i find a solution for the problem of physical attraction, the pieces pull towards each other and ruin the experiment.

i can isolate the charges to 4 or 6 leads and arc them all to a common ground.  i get individual sparks from each one.
so these 'extra' charges are real, i can zap people with them.
The timing on 6 sparks is a little tricky, but i had adjusted the spark gaps at one point and had a 4-phase A/C spark going to a common ground. 

im at a complete loss with the conversion process. automotive coil is getting me nowhere, when it manages to NOT spark over theres almost no current to speak of.

theres a small piece of technology im sure none of you have missed. I speak of a rediculous-looking inductor that is basically a stright wire, with another wire wrapped around it. i couldn't help but toy around with this.

made me go hmm...    try to keep it around the 1-3kv range (tiny sparks) and make sure the inner wire isnt sparking through.
not sure what to think....  it may play an important role when used from both sides of the machine, and crosing over each other they way it is seen in most of Paul's machines.

also, theres sort of 'capacitor' in the middle of the machines, cross-charged from each side and a center terminal comming out of it looks like its charging the spinning disk..

im not exactly sure how ions are created when there is low air-flow such as through a can, 
can charging metal plates actually release ions inside the can?
which can be fed directly to the surface of the disk and carried away?