Agentgates´s TPU setup with strange wavehill hump

[Spider]

Step 1.......


Spider

[Magluvin]

Hi Tony

I made that coil in haste. By the few differences in the pics, I went with the angle you see on the primary.
But now it is 45 deg. I understand NOW that it is still being tested and the theory is not fully complete.

It just came to me as to what might be going on, a vision of sorts and came to some fair conclusions according to these principals you laid out,
1 The secondary need not be in the middle of the primary, or wrapped by the primary, but proximity is necessary.
2 the new and hopefully final deg angle is 45, and it probably should be. we will see. It also makes for more inner and outer primary "crossings"
3 Secondary should be less turns than the primary, or at least not more.

Considering the thin wall tube, the idea of the inside and outside primary crossings need to be close to each other, and apparently the secondary need not be inside the primary.
The star on the wood maybe should be shown much later and concentrate on the tube, but it shows something useful.
So probably the best way to go about it all for the tube style is to use pvc tube with pins on the outside and just zig zag the primary winding outside the tube. This way the primary winding "crossings" are close in proximity also to each other, thus when a secondary is wrapped to it, the inner and outer primary crossings are equally in proximity to the secondary. 

I say this due to the fact that it makes no sense to have the inner primary windings any further away from the secondary than the outer primary, IF what you state is the case. IF my way is tried and the effect decreases, then I would go with a very thick wall tube, wrap the primary around that, at your 45 deg, then do the secondary to the outside, where the secondary is more influenced by just the outer most angles of the primary.

I am picking up some more wire today and will post pics later.

Magluvin

[agentgates]

An attempt to visualize the setup a bit more, from my views after reading this thread.
Below circuit is my understanding, please indicate whether it is correct or not.
Maybe we can help in improving.

Yes, thank you for your kind offer, we will certainly make it more compact.

There are a few fundamental issues that seems not easy to solve.
It also seems to me that with peak voltages on the output only, this system is not yet OU. Is there someone that understand the reason for the large dissipation in the FETs?

Yes I already answered this important question somewhere. The huge heat dissipation is because the gained energy moves electrons but there is not equal hole from the opposite direction. At ordinary devices the electric current moves as follows:

1st You create holes on the positive side.
2nd Electrons say: "hey guys, there are some holes to fill" so they will start moving to the positive pole, while the holes are travelling toward the electron flow.

This is normal electric current. The more holes you create, the more electrons will try to struggle to comply the holes "demand".

But in this device not this is what happens. Due to the dot-like crossing pattern of winding and the large spaces between the turns you create vortices of electrons inside the copper wire WITHOUT holes. So how do the gained electrons know that which way to flow in the wire if there are no holes?

Two possible ways:
- They will try to find holes wherever they can. There are plenty on the PN transition layer of the IGBTs.
- They will try to find other ways, like the secondary wire if the electron "traffic" is not to "busy".

Conclusion
- Increasing the diameter of the collector will let more electron to flow inside of it.
- Decreasing the diameter of the primary will let less electron to flow back toward the FETs.
- Increasing the numbers of channels will demand more and much smaller spikes and WILL NOT let the already fast moving electrons in the secondary wire to slow down. (when I am done with this we can remove the IGBTs, presently I am using them temporarily for prototyping)

In other words, we don't need mass circuitries to force the electrons moving that way we want, instead, we already know that how they want to move, so make them their way and let them moving easier.

My personal view is that using square wave pulses, the coils will only transform the energy during the waveform transitions, while the FETs consume power during the whole 'on' time of the positive input period.

Yes but this is the TEMPORARY case. Right now I am working on the multi channel version to throw out the brutal IGBTs and don't let the electrons spin down. I will only ticle them to preserve their speed.

Not using a fly-back diode will force at least one of the three FETs to consume significant power for a short period of time, since not all three FETs will switch off at the same time, due to small differences between the three.

Yes but presently if I use one IGBT that has smaller PN surface than 3xIGBTs. So again they are TEMPORARY solution to protect my low voltage power supply from the spikes and they will be removed in any hour. I oversized them by the parallel fashion to make sure they will not follow the dead one.

I don't want to kill the spikes ---> I want to harvest them in the seconday wire and make usable electricity with them.

And an interesting observation: there is a small electrostatic field on the heatsink which proves my theory of concept and the lack of holes.

On your drawing:

Two things are missing. A 4700uF 35V cap between the low voltage PSU and a 56 ohm resistor right after the mosfet driver as I DONT want to turn them on so fast and with a 9A current. The mosfet driver is unimportant but I had to use something to reach the ~10V with my PG and turn the IGBTs on.

[teslaalset]

I see the 'hole' effect that you explain a bit different, maybe we are both talking about the same thing, but let me explain my understanding:

If you charge a coil with current and switch off that current, electrons still keep on traveling to the side of the coil that was initially connected to the positive power source connection. This is caused by the magnetic field that is now present in the coil and leads to a very high amount of 'holes' at the other side of the coil, resulting in the well known high voltage peak.
So, nothing special occurring so far.
In that situation the coil is in a a-stable situation, which is not caused by mysterious effects that add energy into the circuit.

Are you saying that there are even more electrons moved besides the normal electron flow occurring right after switching off a coil?

[agentgates]

1. from your previous posts the 45 degree angle seems to imporve the operation correct?

I don't know whether it improved anything or not as I did some other critical modifications and altogether it performs far better than the first operating device. But naturally if somebody interested may play with the angle.

I will also look into it but there are other priorities right now.

2. The secondary should be close to the primary windings?

Yes and they should touch each other's surface on a dot manner.

3. The secondary should be wound with spacing between turns to reduce interaction/capacitance between the wire?

Yes. Both the primary and secondary should be wound with spaces so PRIMARY/PRIMARY wires are far, SECONDARY/SECONDARY wires are far and PRIMERY/SECONDARY wires touch each others surface. (use a thin plastic tape, like clear acrylic to make sure they dont rub each other and damage the enamel)

4. Have you used a sweep generator or dip meter to determine the resonance of the primary yet?

No but you can see it clearly on the scope as you varying the input frq. Resonance is not critical here.

I ask since knowing the coils resonance may help match the drive frequency and using some capacitance with the primary may help reduce the driving current and possible heating.

When you have the 3 channels done and driving them after each other the current flow in the secondary will look like charging a cap and watching it on the scope. Your TPU is operating at the best performance when you get an almost perfectly straight line DC out of it. This is what I got with my first 2 operating device but at that time I didn't really understand the nature of this device so there were some huge spikes and about 40-50VDC straight line out. Therefore I had to give it more juice to light the bulb, and use another stage of TPU to boost the performance.

But the new design is refined. It gives a beautiful rising and falling dome. All I will do to make more and overlapp them and I will get something like stairs until I reached the voltage I want. When I did, I will simply decrease the height of each dome.

Another idea that comes to mind is to use a 1:1 transformer to isolate the driver FET's or transistors from the primary this might help in resolving the driver heating.

Thanks we will look into this one if the multi channel thing wont bring the expected results. For now I am focusing onto decrease the "correcting" circuitry and use as small amount of parts as possible.