Miniature wireless power demonstrator

Marko, Mon Aug 10 2009, 11:03PM

20. 04. 2010 - update:

I added little FAQ summarizing the questions I've been most frequently asked about the project. I may update it with time and feel free to post a question you think might be suitable for me to add.

**FAQ**

- Are there kits or complete models available for sale?

Not yet as I'm currently unable to build any. I might make some this summer if I find increased interest, although I'm not promising anything as I have a number of hurdles to overcome regarding sales.

- I can't find the WIMA FKP capacitors like you used. Can I use ceramic/MKP/other type of capacitors available in my store?

No. Other types of capacitors, if they work at all are most likely going to be very lossy and overheat and melt in minutes.
Capacitors don't really even need to be WIMA brand, pretty much any polypropylene capacitors should work well.
This includes CDE942 caps generally used by tesla coilers here, as well as various brands of 'MKP' and similar caps. MKP is somewhat more lossy than FKP but worked fine in my first designs.

- I want to use the circuit to charge a mobile phone. Can I have schematics/plans?

I just don't have time to write about it now, and if you built the circuit I think it's straightforward enough for you to figure it out yourself.
Hint: a rectifier of suitable high speed diodes and a DC/DC converter chip.

- What is the diameter of transmitting and receiving loops?

Not sure why does everyone need to know this exactly, since I don't really remember the diameter I used myself. A larger loop is going to help transmit the power somewhat farther at cost of devices being about proportionally larger. Larger loop with same capacitor will result in lower resonant frequency.

- I can't find the copper tube and wire you used for the loops, what can I do?

Choice of the conductor used is really not critical as long as there is enough surface area to keep conductivity high and avoid overheating. A 6mm copper tube is OK for the transmitter and it should be easy to find in air conditioner stores. Receiver doesn't need to use wire, copper tube is fine too. If you have a large amount of thinner wire you could make a litz conductor by twisting several of them in parallel. 15-20mm width of copper strip should also make excellent conductor for the purpose.
Having equal loops on both transmitter and receiver will make the tuning easier.

- The transmitter circuit does not oscillate, instead it shorts the power supply and one mosfet and inductor heat up rapidly, what to do?

Firstly, if you are using a version without the relay, this is a common problem, and is caused by power supply voltage rising too slowly on powerup. To fix make sure you use a switch on low voltage side, that is immediately between your power supply and the circuit, to turn the circuit ON.

If it'+s still happening, make sure:
1. that the mosfet that suffered the condition is still working
2. check for connections of your circuit, misplaced components, directions of the diodes...
3. The circuit won't run without the loop attached!

Hope this helps.

- The transmitter oscillates but I get very little power on the secondary side. What can I do to improve?

There are two problems (or two parts of a same problem, more precisely); tuning and load impedance match.

Firstly, we want both LC circuits to resonate at about the same stand-alone f0 - the best starting point is to make them both with identical loops and capacitances.

But, to achieve maximum power throughput, we will need to fine-tune the system, prefferably during runtime. This can be done by increasing or decreasing the stand-alone resonant frequency of either the receiver or the transmitter. Some of ways to acheive that are -

1. Changing the diameter of loops - may be difficult to impossible depending on the construction, probably easier on receiver. Should not be done on the transmitter while it's operating, and it's loop should be soldered down anyway.
2. Changing the tank capacitance.
3. Much more convenient to do while system is running - is to insert a large ferrite core (AM radio ferrite rod, or a TV flyback transformer core) into a loop which we want to decrease f0.
4. Alternatively, we can bring a copper or aluminium plate behind the targeted loop to do the opposite.

Secondly we want to match the load impedance to the best possible way to the ''transmission line'' we created with coupled LC circuits. Playing with the circuit you will notice that the ''tuning'' methods described above also work to match the system to different loads! There are some reasonable limits we need to follow with the load if we want best power throughput. We can't use unreasonably low (like a car headlight) nor unreasonably high load resistances (25W 230V incadescent bulb?).
I found a 24V 5W bulb to be a decent load, although I suspected 3x 12V/2W bulbs in series might have performed better.

- I want to supersize the circuit/increase the input voltage/transmit the power over a few meters?

None of those are really practical with this design. The best shot for increasing the power throughput would be to use a ferrite transformer between the active section and LC tank - I tried it and it works but at cost of even greater mosfet heatage... if you are already at this point then you probably already understand that this circuit sucks and have better ideas than it anyway.
Just increasing the voltage to the circuit as it is beyond 18V is most likely to cause it to blow up. And using higher voltage mosfets is actually just going to make the problem worse.

Transmitting power this way on a scale of meters with sub-metre sized devices at any reasonable efficiency is pretty much science fiction as of today.

- Have you been developing any new ideas on wireless power? When may we see updates?

Yes, but I'm keeping it top secret as of now. It's not going to be simple and I don't have time to explain it anyway. ANd builds are unlikely to start before this summer.

- I built the circuit, but I measured a different frequency at the transmitter than your 1.5Mhz. What is wrong with my circuit?

Nothing may be wrong, you might just have used a larger loop diameter or more capacitance than me. If parts are really closely matching mine, thenthere might be a problem. Some things to try:

1.Just try powering something if you have the receiver - if not build one from a piece of wire and a light bulb, capacitor isn't even required for proof of concept.
If you get any incadescence on the bulb then your circuit is working already! Just proceed to tune up the receiver then.

2. Look for the current drawn from the power supply. If it's less than 0.5A or more than 1A without load then something is wrong. It's recommended to use a current limited supply for initial tests.

3. Try measuring the LC tank voltage with an oscilloscope, not a frequency counter - this also has to show the peak voltage which has to be about pi*supply voltage (37V for 12V supply). Some precautions are required while doing this - the power supply must not have grounded " - " (like a PC power supply does) because placing an oscilloscope ground clip to one of 'hot' ends of LC tank will cause short circuit. In that case we need to measure with scope input set to DC input and measure between ground and one ''hot end''. This will yield a ''halfwave rectified'' waveform but the peak voltage value should remain the same.



....that's about it for now.




Update 16. 9. 2009.
Eagle files of the new PCB


]wirelesspower.zip[/file]

New update 24. 8. 2009. :

I designed a new PCB with goal of ease of replication by newbies. Including schematic with explanations. Preliminary board, may be subject to change.


]through_hole_wireless.pdf[/file]
]through_hole_wireless-schematic.pdf[/file]
]through_hole_wireless-components.pdf[/file]


I also modified the old, prototype PCB (without the relay) a bit to make it more practical to use. All components are intended to be surface mounted on the copper side so there is no drilling involved.

]pcb_old.pdf[/file]
]pcb_old_parts.pdf[/file]


A number of people has expressed interest into this project, and I decided it would be best to post it up here as a project thread.

The idea behind the project was to create a small tabletop demonstrator of magnetically coupled wireless power transfer, resembling a miniature version of the MIT 'witricity' device. The goal was to keep the circuit simple with easily obtainable parts, and to keep voltage and power levels low so the device is safe for handling and doesn't require special methods of cooling.

The basic idea is to feed a parallel LC tank circuit from an AC voltage source at it's resonant frequency, which allows large reactive current to circulate in the circuit while only real power is being drawn from the source. This sets up a large alternating magnetic field in the inductor, which is designed as a single conductive loop in this case.
Now, another LC tank with load attached is brought in proximity to the excited LC circuit, significant amounts of power can be transferred via weak magnetic coupling between them. This is because AC current itself in the transmitting loop is very large, and inductive reactance of the receiver loop is canceled out by the capacitor.

For a practical device, the AC voltage source had to be substituted with an appropriate oscillator, which would take feedback from the tank circuit itself and hence always drive it at it's resonant frequency.

The circuit of choice was a slightly modified royer oscillator, such as popularly used in CCFL inverters and for flyback drivers.
Input voltage was limited to 15V for safety and because the circuit tends to become unstable at higher voltages.

The idea of the prototype circuit is rather simple.






Mosfets I used were IRFZ44, but any similar ones will do. A small piece of Aluminum for a heatsink is recommended, although in prototype I just soldered the mosfets onto the PCB.

Rg were at first 50, but later increased to 100 ohms which is enough and wastes less power. Resistors need to be rated 1 watt at least and they get quite hot.

Radio frequency chokes were 100uH, at first iron powder cores, but later switched to ferrite which produced much better results.
Powdered iron cores tended to heat up from magnetic flux they picked up from the transmitting loopas well.

Diodes - 1N4148's. Similar small high-speed diodes are ok.

The LC tank circuit is the part where heavy current circulates, and is required to be sturdy. The copper pipe used as conductor heats up significantly under ~20A it's passing continuously. To handle the current while keeping losses tolerable, capacitor consists of 6 paralleled 6.8nF 1000V wima FKP1 capacitors. It's important that capacitors are polypropylene dielectric and foil or foil+film based - other types will heat up and melt in this application.
The transmitter still oscillated at relatively high frequency and had to be tuned by insertion of a ferrite core into the loop, as shown on the picture. This lowered the frequency to about 1.5Mhz without load.
Alternatively a copper or aluminum plate can be brought near the loop to increase frequency, by decreasing inductance.
Number of capacitors was later increased to 8, removing the need of additional tuning.

On the receiver side only a single capacitor and a loop of 3mm solid copper wire was used. The wire heats up significantly, though.
You can notice I used a small matching inductor in series with the load, which is a 24V 5W. It's choice was guessed at 6uH and it improved performance somewhat at larger distances.

This prototype, though, had one problem - if the supply voltage rose too slowly, such as while DC filter capacitance is charging, it tended to fail to oscillate and just keep shorting the power supply with one mosfet ON. In the final design this was solved with a relay, which acts as undervoltage lockout of sorts, applying Ugg power rapidly after supply voltage rose high enough.

Jumpers in the schematic are to allow connection of a step up autotransformer between the mosfet amplifiers and the LC tank circuit.

I provided the schematic of the finished circuit, but not the PCB since it's a fairly odd design with odd components (SMT inductors) which I thought might not be too useful for most people. PM me if the PCB layout is desired.


PCB layout and part layout for the final circuit shown in the bottom picture, along with schematic:

]parts_top.pdf[/file]
]pcb_bottom.pdf[/file]






Cheers,

Marko