Shorting coil gives back more power

[gyulasun]

Hi Romero,

You wrote:  "I have got about 1.7 COP with the mechanical version. I have tested the solid state version too but with simillar results."
Would like to ask when you tested the solid state version what was your load?  An incandescent lamp?  I mean how did you estimate the COP of 1.6-1.7?  Not 'nitpicking' here, just wondering on a possible looping with a correct AC/DC or DC/DC converter with a 85-90% efficiency still seems feasable. 
So if you could give some more details on the solid state version, that would be great, including the coil heat issue too. With a COP of 1.6-1.7, can it work for a longer time without getting too hot?

Thanks
Gyula

[romerouk]

Hi Romero,

You wrote:  "I have got about 1.7 COP with the mechanical version. I have tested the solid state version too but with simillar results."
Would like to ask when you tested the solid state version what was your load?  An incandescent lamp?  I mean how did you estimate the COP of 1.6-1.7?  Not 'nitpicking' here, just wondering on a possible looping with a correct AC/DC or DC/DC converter with a 85-90% efficiency still seems feasable. 
So if you could give some more details on the solid state version, that would be great, including the coil heat issue too. With a COP of 1.6-1.7, can it work for a longer time without getting too hot?

Thanks
Gyula

Hi Gyula,
I was using a bulb as the load in both versions but not connected directly to the bridge rectifier.
I have modified a computer PSU and the output from the bridge rectifier is applied to the PSU.
This way worked best for me.I have used this trick to modify a computer PSU in other experiments and I got best results.
The coil gets very hot in about 20-25 minutes in the solid state version but the mechanical version works for about 40-50 minutes before starting to get burn smell, maybe it takes longer before is too hot because of the air flow from the rotating wheel that cools it down a bit.
No selfloop at this moment.
Another problem I found is the diodes and bridge rectifier.Most of them are not capable to forward all power resulting in at least 20-30% loses. One of the picture posted by Richard shown a board with many bridge rectifiers connected in parallel,... why is that?
I have used different diode types and the last experiments I have used BY255 with the best results so far. Every time I add another diode in parallel with an existing once I get a bit more power output without affecting the input. In my test I used 5 diodes in parallel in all setup and I runout, no more diodes Maybe other diodes will do better , can you suggest something?
I am not sure that dimensions of the coil are the best.In my case I get the best results at about 58-60KHz, depending of the coil used.
Driver of the coil is not standard too. Switching is not sharp enough from any electronic circuits used then I decided to use a reed activated by a coil(I got one small coil with reed inside) and that reed drives a SS DC Relay.
If I drive the relay from the same source directly, without the reed I get much less COP and the coil gets hot much faster.
I hope people understand what I am talking about.
The Magnacoaster patent has 2 diodes connected wrong, the ones comming from the coil to the output rectifier.

All the best,
Romero

[popolibero]

Hi Romerouk,

a diode tipically has a 1V voltage drop, if you put more in parallel the drop gets smaller, a bit like if you put resistors in parallel. Diodes mean losses, especially if working at a low voltage like 12V! In fact it would be way more efficient using a smaller gauge wire for the coil that does the same magnetic field but with less amps and more voltage...

I'll write more later when I have time.

Mario

[gyulasun]

....
Another problem I found is the diodes and bridge rectifier.Most of them are not capable to forward all power resulting in at least 20-30% loses. One of the picture posted by Richard shown a board with many bridge rectifiers connected in parallel,... why is that?
I have used different diode types and the last experiments I have used BY255 with the best results so far. Every time I add another diode in parallel with an existing once I get a bit more power output without affecting the input. In my test I used 5 diodes in parallel in all setup and I runout, no more diodes Maybe other diodes will do better , can you suggest something?

Hi Romero,

Fast or ultrafast type diodes are better at the frequencies you mention than the BY255 (which is a normal 50/60Hz type rectifier with t_rr=3usec reverse recovery time  http://maxdiode.com/Uploadpdf/BY250%20THRU%20BY255.pdf )
The reason you found better output when paralleling them is that their forward voltage drop get reduced, as Mario mentioned,  the reducement can be anything from say 60mV-80mV per diode, this is also nonlinear, in case of 5 paralleled diodes you may have got 0.7V loss instead of a single diode of say 1V.
At Farnell you may find the BY500-600 fast diode type (600V/5A, t_rr=200nsec price 0.191Ł one piece, order#1651065), this is much faster and has 1.35V loss at 5A (your BY255 has 1V loss at 3A) so at 3A this diode can be better if you also parallel 5 of them.

Best solution would be to use syncronous rectifiers: these are controlled MOSFET switches, see Fig. 3 in this link: http://www.national.com/vcm/national3/en_US/resources/power_designer/national_power_designer112.pdf

Will return to this later.

rgds,  Gyula

[yssuraxu_697]

popolibero:
"In fact it would be way more efficient using a smaller gauge wire for the coil that does the same magnetic field but with less amps and more voltage..."

Strongly disagree. What is actually needed is:

ultra low resistance across *all* the components
ultra low inductance
ultra low HF losses

This means tech along the lines of RF class
cores, litz wire windings, physical switching and so on.

This is from hands-on experience with the system.

The perfect system would consist of:
- superconductive winding that is totally compensated with according capacitance
- zero hystersis core
- 10ns class switching

gyulasun:
"Fast or ultrafast type diodes"

UF4007 rocks. And you can parallel them also...