Global Free Energy Blog

Someone at a forum asked a good question about my “Tiny Orbo Replication” COP measurements” –>

Quote,“I’m pleased to find accurate measurements and calculus here. I perfectly agree with the method you are using to evaluate the COP. Nevertheless I see a possible flaw in the data so I have a question: how do you estimate the coil inductance during pulse?”

Hi,

The inductance was calculated from initial RL curve analysis when the magnet was at TDC. The current pulse followed a typical RL curve. The reason I’m comfortable that this is accurate enough for this measurements is based on a lot of variations in the coil applied voltage. Increasing the battery voltage by varying amounts would obviously increase the current RL curve. The increased voltage did not change the shape of the RL curve, but just the amplitude. For example, increasing the battery voltage by say 4 times would show the same RL curve, except 4 times the amplitude. As you know, during the initial part of an RL curve is mostly reactance, not resistance, and toward the end of the curve it’s more resistance than reactance. This allowed me to see how linear the core was at varying levels of current at these *saturation* levels. I spent a great amount of time analyzing this core on the scope. It has two distinct modes. It’s either incredibly high permeability, or low permeability. So it’s very easy to see on the scope which part of the BH curve the core is in.

Another area to address is how far did the magnet move during this current pulse. As you can see in the blog post, the magnet was rotating at 26.5 revolutions per second, which comes to 0.57 degrees of movement in 100us, which comes to 0.005″ (0.01cm) movement.

So to summarize what was occurring to the core –>

1. The magnet moves to TDC.

2. Voltage is applied to the coil.

3. There is a *brief* period where the core is in ultra high permeability. This was seen in the scope where di/dt was lower then I could detect, and the current was near zero amps. Regardless how much I amplified the signal, it was a flat as a pancake, ~ 0 amps, and understandably so given this cores.

4. All of a sudden the core switches to low permeability as the hits the roof of the square BH curve. This is where nearly *all* of the energy goes into decreasing the cores magnetic attraction toward the magnets. The changes in the cores ultra high permeability have no measurable changes in the core to magnet attraction. So during this phase the core’s permeability was relatively linear relative to current.

5. The current reaches its near max of the RL curve due to electrical resistance. At this point we’re still appreciably less than 100us, and the magnet is still close to TDC.

So that’s why I used the inductance equation of E = 1/2 * L * I^2 because the cores permeability at that current level and cores saturation level was appreciably linear far above 1.26 amps. Furthermore, my COP 1.7 measurement did not even consider how much of that energy could have been recaptured. It is my opinion that at that area of the BH curve (in the saturation area), the core would be relatively linear with or without the magnet. If true, then a large amount of that energy that went into inductance could be recaptured, which would increase the COP measurements.

Here’s the latest circuit, a huge improvement. This low component count design is simple, and is just as efficient as the optocoupler design. Furthermore, the MOSFET  no longer needs to be driven hard. L1 is the toroid.

I am not affiliated with Steorn. Orbo is trademarked by Steorn.

Here’s a much simpler circuit, but not as efficient in holding the toroid current. A 15,000 rpm the pulse needs to be ~ 55us. In 55us the pulse current in the following circuit decays by 11%, while the same version using optocouplers decays by 7%. I understand that Sean from Steorn says the pulse current needs to be constant. IMO that’s untrue. I believe what’s required is the *rising* pulse needs to be extremely fast, a few dozen microseconds at most is preferable.

The 2N2222A transistors should ultimately be replaced with an efficient MOSFET driver chip or circuit. The MBR20100CT could be replaced with a better Schottky diode, but make sure it has a high enough breakdown voltage. Also, adding three diodes in parallel only improves the decay by a fraction of a percent.

The voltage source, V4 (sig), comes from the Hall effect switch.

Obviously you could use a TLC555 instead of a TLC556. The only reason it’s drawn as a 556 is because I’m out of 555′s.

Also, by increasing the battery voltage to 12V you can drop the decay current by ~ another percent.

I am not affiliated with Steorn. Orbo is trademarked by Steorn.

The following circuit is the result of last nights work. It now works good enough in Spice, but I’m still not happy with it because it uses an optocoupler, ugg! Perhaps a tiny toroid could replace the opto-coupler. The goal of the circuit is to get a rapid pulse through the orbo toroid, and then level that current off. The reason for the rapid pulse, according to my old magnetic research by means of conventional physics, is magnetic viscosity found in the NdFeB magnets. NdFeB magnets have extremely high magnetic viscosity, over 100us for an typical test pulse. So the toroid pulse should be a *lot* faster than 100us. Around 10us to 30us is acceptable.

If you find an improvement for the above circuit, then please contact me. [Edit: see my newer design, Tiny orbo replication 2 circuit in progess 2.] The voltage *across* the toroid should be high when the current is rising, and then suddenly drastically drop in order to level off the current. So there might be 80V initially, and then a sudden change to less than a few volts. The few volts is just enough to maintain the current. That makes a more complex circuit requiring multiple voltage sources, but I found an acceptable alternative. That alternative is to simply short the toroid immediately after the rising current pulse. The short will maintain the current long enough for ultra high rpm’s over 10000 rpm. I had my “tiny orbo replication” spinning at nearly 8000 rpm’s no problem and the magnets were not even well glued. This time they’re well glued and will also have a ring of tape around them to keep them from breaking off and also the reduce air flow. So I’m shooting for 15000 rpm. Who knows what will happen. Maybe the ball bearings will be terribly inefficient at such rpm’s.

So the above circuit design is according to my magnetic research, which shows excess energy by means of using conventional physics. I have no idea what Steorn is doing, maybe they’re using a DC to DC converter chip to get the higher voltage.

I am not affiliated with Steorn. Orbo is trademarked by Steorn.

Yesterday I purchased an Enermax Marathon PC fan. Has anyone heard of them? They claim to use a frictionless magnetic levitation ball bearing. Total cost was $8. We’ll see how useful it is for this.

I decided to stay with the 26 Gauge wire for the toroid, but will use a lot more of it to fill up the toroid. Hopefully it will accept 160 turns of 26 Gauge. One reason for using thin wire is due to the skin effect. Another reason is to increase the overall efficiency, as this will allow for higher pulse voltage and less pulse current (as there’s more turns), and hence less losses from the MOSFET.

As a rule of thumb for the “Tiny Orbo Replication,” I use ~ 60uH for 37 turns when the toroid and NdFeB magnet are top dead center. This also comes to ~ 1/6th of an ohm resistance. Anything thinner than 26 Gauge might get into problems of noticeable parallel capacitance.

I am not affiliated with Steorn. Orbo is trademarked by Steorn.

Over the past ~ week I’ve been working on improving my solid-state magnetic designs, but still the losses seem so incredibly high. No doubt they could produce excess energy though.

So today I’ll get back to the “Tiny Orbo Replication” to reduce the losses in hopes of eventually producing a self-runner, now that my measurements show excess energy; e.g., 170% efficient. Here are a few considerations for today –>

For starters, it needs a 2nd toroid to help eliminate the torching on the ball bearings. Right now there’s only one toroid, so every time the NdFeB magnets pass by the toroid there’s a strong tug that twists the ball bearings. Hopefully a 2nd toroid will greatly reduce such friction. Also, the toroid does not have much wiring. So either the coil needs considerably thicker wire or a lot of similar windings all in parallel.

Also, I’ll double the NdFeB magnets to four, thus giving 4 pulses per revolution. The idea is that the extra 2 magnets will not increase the weight of the entire tiny orbo that much, … but it just occurred to me that might not work as well as planned since the magnets are on the outer rim. Hmmm, now I wonder if 4 magnets is better or worse than 2.

Also, I have my eye on another CPU fan in the garage that I’ll put larger magnets on. That will be interesting.     And there’s a PC power supply fan in the garage that’s larger than a CPU fan, which could be interesting.

Another possibility would be the Enermax Marathon PC fans, as they claim to have magnetic levitation bearings, would you believe! If those could somehow work, that would eliminate nearly 100% of the friction.

Another possibility would be to use bonded NdFeB magnets, which have high electrical resistivity, thereby eliminating nearly 100% of the eddy currents from toroid-NdFeB interaction, but that might not work so well if high magnetic viscosity is required for the magnets. I believe the magnets need to have high magnetic viscosity, and IMO bonded NdFeB have less magnetic viscosity than normal NdFeB magnets.

I am not affiliated with Steorn. Orbo is trademarked by Steorn.

[Edit: See Holy grail tiny orbo replication to replicate.]

[2010/2/11 update: See "Tiny orbo replication 2" major update]

Today I took more detailed measurements to obtain the efficiency of my “tiny Orbo replication.”  Once again it was well over 100% efficient. My “tiny Orbo replication” has changed since the last measurements. The toroid was rotated 180 degrees around so that there are no wires directly facing the magnets. Also, the voltage was low, just 0.36 volts. Furthermore, I moved the toroid closer the magnets, so the inductance is different. The following math and results excludes joule heating from electrical wire resistance, as the goal is to see if there is any excess energy:

General info
Revolutions per second: 26.5 rps (1590 rpm)

Pulses per second: 53.0 pulses/s
Peak voltage pulse across coil: 0.36 V
Peak current pulse through coil: 1.26 A
Inductance of coil during pulse: 60 uH
Power input into inductance: 0.5 * 60 uH * 1.26² A * 53.0 pulses/s = 2.52 mW. Note, inductance was linear in this range. [2010/2/11 update: Nearly all of this inductance energy can be captured back. See "Tiny orbo replication 2" major update]

Deceleration control test:
Time duration: 0.33 s
Starting rps: 27.6 rps
Ending rps: 24.7 rps
Deceleration rate: (27.6 rps – 24.7 rps) / 0.33 s = 8.79 rps/s
Calculated rps from 26.5 rps in 0.1 s = (26.5 rps – 8.79 rps/s * 0.1 s) = 25.6 rps

NdFeB magnets
Quantity: 2
Density: 7400 kg/m³
Length: 0.0033 m
Radius: 0.0028 m
Total volume: [(pi * 0.0028² m) * 0.0033 m] * 2 = 1.63e-7 m³
Total mass: 7400 kg/m3 * [(pi * 0.0028² m) * 0.0033 m] * 2 = 1.20e-3 kg
Average distance from center of rotation: 0.016 m
Average velocity @ 26.5 rps: (2 * pi * 0.016 m) * 26.5 Hz = 2.66 m/s
Kinetic Energy @ 26.5 Hz: 0.5 * 1.20e-3 kg * 2.66² m/s = 4.25 mJ
Average velocity @ 25.6 Hz: (2 * pi * 0.016 m) * 25.6 Hz = 2.57 m/s
Kinetic Energy @ 25.6 Hz: 0.5 * 1.2e-3 kg * 2.⁵⁷² m/s = 3.96 mJ
Deceleration energy loss in 0.1 seconds: 4.25 mJ – 3.96 mJ = 0.290 mJ
Power output from friction @ 26.5 rps: 0.290 mJ / 0.100 s = 2.90 mW

Black plastic outer rim
Quantity: 1
Density: 1000 kg/m³
Height: 0.00635 m
Thickness: 0.000889 m
Distance from center of rotation: 0.0146 m
Volume: (2 * pi * 0.0146 m) * 0.00635 m * 0.000889 m = 5.18e-7 m
Mass: 1000 kg/m³ * [(2 * pi * 0.0146 m) * 0.00635 m * 0.000889 m] = 0.000518 kg
Velocity @ 26.5 Hz: (2 * pi * 0.0146 m) * 26.5 Hz = 2.43 m/s
Kinetic energy at 26.5 Hz: 0.5 * 0.000518 kg * 2.43² m/s = 1.53 mJ
Velocity@ 25.6 Hz: (2 * pi * 0.0146 m) * 26.5 Hz = 2.35 m/s
Kinetic energy at 25.6 Hz: 0.5 * 0.000518 kg * 2.35² m/s = 1.43 mJ
Deceleration energy loss in 0.1 seconds: 1.53 mJ – 1.43 mJ = 0.100 mJ
Power output from friction @ 26.5 rps: 0.100 mJ / 0.100 s = 1.00 mW

Black plastic flat top disc
Quantity: 1
Density: 1000 kg/m³
Radius: 0.0146 m
Thickness: 0.000686 m
[extra info: Volume: 9.19e-7 m^3]
[extra info: Mass: 0.000919 kg]
Integrated Kinetic energy at 26.5 Hz: 0.679 mJ
Integrated Kinetic energy at 25.6 Hz: 0.633 mJ
Deceleration energy loss in 0.1 seconds: 0.679 mJ – 0.633 mJ = 0.0460 mJ
Power output from friction @ 26.5 rps: 0.0460 mJ / 0.100 s = 0.460 mW

Total friction heating from both NdFeB & black plastic outer rim: 2.90 mW + 1.00 mW + 0.460 mW = 4.36 mW

Efficiency: 4.36 mW / 2.52 mW = 173%

Friction heating was calculated by doing a control experiment. During the control experiment, the “tiny orbo replication” was rotating over 28 rps. The deceleration rate was then recorded after turning the “tiny orbo replication” off. This showed how fast the machine decelerated between 27.6 rps and 24.7 rps. The deceleration rate along with mass and velocity of the pieces provides the loss in kinetic energy over time, which provides the friction heating output of the device at the operating speed.

This the circuit I’m using.

Click to enlarge image.

I am not affiliated with Steorn. Orbo is trademarked by Steorn