Design problem in the MEG circuitry
By Arend Lammertink, MScEE. July, 2024.
Contact: lamare at gmail dot com
I have recently been looking into the Motionless Energy Generator or MEG for short. The working principle is based on the principle also used in the electropermanent magnet:
https://en.wikipedia.org/wiki/Electropermanent_magnet
In this video, the ideas behind the MEG are explained by the inventors of the device:
http://www.youtube.com/watch?v=no50_5iSr2Y
In this video, it is said that higher frequencies would be required, so the first thing that comes to mind is to use ferrite cores, as has also been done by Jean Naudin in his version 2, judging by the text on his diagram:
As well as a photograph of the device:
This version had a COP of 1.75, quite an accomplishment given the design flaw in his circuitry, which we will discuss below:
Freewheeling diodes
Let's take a look at Naudin's v1 schematic:
Notice the freewheeling diodes D1 and D2?
A freewheeling diode is used in inductive circuits to provide a path for the current when the driving voltage is suddenly reduced or removed. It prevents voltage spikes that can occur due to the inductive kickback, protecting the circuit components.
However, in this case this is not a good idea, because this way the coils get shortcutted and all the energy that is put in to energize the coil gets burned up by the diode and series resistance of the coil wire.
In the v2 schematic, these are still there, but integrated in the BUZ11 mosfets:
On the v2 page we read:
The new MEG control board v2.0 is now fully in line with the original Bearden's MEG comparing to my previous version.
So, apparently both Bearden's team as well as Naudin made this same mistake.
However, an elegant solution is to use an asymmetric bridge converter as described in my earlier article (optionally switching a number of BEMF recovery capacitors between series and parallel), with this picture showing a triple bridge converter:
https://en.wikipedia.org/wiki/Switched_reluctance_motor#Power_circuitry
Replication plan
Now that we found a mistake in the control circuitry, we can think about how to test a new control circuit, which requires a core. I made a drawing of how this could be done:
The idea is to use 4 126 x 28 x 20 I-bars for the main construction and up to 20 38 x 4 x 25 I-bars for the inner part with the magnet.
This has a big advantage, namely that this way we can play with the magnets and the core in the same was as shown in the video above. This way, it is possible to determine how many magnets would be needed to obtain the best results by varying the number of magnets and see how the core reacts.
Once that is determined, we can think about adding coils and a control circuit.
A little demo with the core (English subtitles):
http://www.youtube.com/watch?v=nWHmHftx3ao
In this demo, I used a neodymium magnet with a diameter of 17 mm. Using ChatGPT I now also have a method to estimate the diameter of magnet to use, which can be found in this Jupyter notebook:
https://github.com/l4m4re/notebooks/blob/main/MEG_stuff.ipynb
ChatGPT's conclusions:
Summary
- The flux density of 0.405 Tesla in the larger bars is within the allowable limit of 0.49 Tesla for N87 ferrite material.
- Your observation with a 17 mm diameter magnet confirms that the flux switching works without causing saturation.
You are on the right track with your experimentation. If you want to ensure even more margin, slightly reducing the magnet size or adjusting the configuration to distribute the flux more can help.
Controller design
Am thinking about designing a pcb for controlling the MEG and to include the capacitor switching trick I described earlier as well, so various circuits can be driven with one pcb (design). Would be nice to also include the possibility to drive brushless DC motors (as Ron Brandt worked with), so this project might be a good starting point to base the design on:
https://github.com/pgrady3/EasyController3
We will have to replace the driver sections of this design with an asymmetric bridge and I have been thinking about using IGBT's, because of their ruggedness. Mosfets don't seem to appreciate high voltage kickbacks, like in Bedini's circuits.
It appears the DGD2304 gate drivers are capable of up to 600V and can drive IGBT's as well:
"the DGD2304 is a high voltage / high speed gate driver capable of driving N-channel MOSFETs and IGBTs in a half bridge configuration. High voltage processing techniques enable the DGD2304’s high side to switch to 600V in a bootstrap operation."
It seems very doable to adapt this design, see the asymmetric bridge and the EasyController3 schematic next to one another:
As an extra bonus, I'm thinking about implementing the BEMF recovery circuit described before as well, so we can experiment with this circuit as well.
In this picture the capacitor switching trick I have in mind next to the power stage of the EasyController3:
We would have replace the power stage with three of these capacitor switching banks, one for each bridge.
For the transistors, I've chosen the C3M0280090D, which can handle 11.5A up to 900V, after the design of Master Ivo who uses this series of mosfets in his experiments on his YouTube channel.
This way, we would obtain a pcb with the possibility of driving BLDC motors, the MEG as well as experimenting with the BEMF recovery circuit.
The current version of the schematic is available in KiCad format on my github, with the pdf version here:
https://github.com/l4m4re/BEMF_Controller/blob/main/kicad/BEMF_controller.pdf
Further updates may become available at:
http://www.tuks.nl/wiki/index.php/Main/DesignProblemMEGCircuitry