The basic premise of inductive charging is not that complex if you are familiar with the concept of transformers. The transmitter (primary) coil resides on your table/nightstand/charging pad and the receiver (secondary) coil resides on your mobile device. Energy is coupled from transmitter to the receiver magnetically through the air. Most transformers use a good flux concentrator (high relative permiability)such as steel or ferrite to "herd" the flux lines together and direct them to the secondary. In our case, we just have air (mu_r = 1). How good the coupling is depends on the 3D placement of the two coils as well as any intervening medium (such as metal or ferrite).
One smart way people have figured out to increase the power efficiency is to use a concept called resonant coupling. By introducing capacitors on both coils, a resonant circuit is formed between the inductance of the coil and the capacitor. At the resonance frequency only, the reactance of the two (as seen by the driving source) cancel out and you are left with only the parasitic (lossy) effects of finite winding resistance, some "AC resistance losses" to proximity effect and dielectric losses (minimal). So if you have a perfect AC source and drive the resonant circuit you (ideally) have no losses! So you are left in a good position your losses are solely limited by your parasitics.
Two microcontrollers (MCUs) sense and control the action. On the transmitter side we measure the input power (voltage and current), the state of resonance, the temperature of the transmit coil, and also a communication from the receiver (transmitted to the transmitter by the same coil as the power). On the receiver side we measure only the secondary voltage and temperatuer and then drive a communications unit to feedback that information to the transmitter. For example if the voltage on the secondary is too high, the transmitter changes the frequency to pull the system closer to resonance.
Power Control Premise (frequency):
If your primary and secondary coils have resonance at the same frequency and are well coupled,you can modify the frequency of the driving source and change the voltage output of the secondary. One easy way to understand this is by "load line" analysis. By mapping out the load lines at frequencies A-D you can you use this information to decide where to operate your system so that your secondary voltage is just right for your load. For example if your load must operate in the hashed region shown below then you need to operate at frequency greater than C, except at high load, when you need to go towards resonance.
This load line analysis is really useful because it takes into account ALL of your system -- it is inherently an empirical approach.
Wireless charging over a distance has been a dream since Tesla, but the feasibility has yet to be seen. Distances of only 1-10mm, however is completely reasonable and well demonstrated. The idea is that as your primary and secondary coils become misaligned, the flux from the transmitter coil "leaks out" and does not all reach the secondary. The better the two coils are aligned and the smaller distance between, them the coupling (and power transfer efficiency) can be improved.
In this application, where the receiver can be misaligned, the voltage at the receiver can vary widely. With most loads requiring a fixed voltage (such at +5V +/-10%), feedback between the coils is almost essential. Of course, the human can always be the feedback mechanism by fixing the alignment issue or boosting the amplitude of the driving source when their device doesn't charge. Smart people have use the transmit/receive coil themselves for BOTH transmission of the power and of a signal. That way the receiver can say "yep, i'm good, no more amplitude please", or "dude really, I need more voltage over here, give me some more." I have provisioned in my circuit to have the feedback be done all in firmware (by MCU) or by human (debug mode).
Because of the need for a well defined resonance frequency of the primary and secondary circuits, as well as the communication protocol, standards have been reach so that one manufacturer can make a compliant transmitter and another manufacturer can build a compliant receiver and they will both work with each other. This allow the units to be mass produced without coordination between the two manufacturers. The most advanced standard (that I'm aware of) is the "Qi standard" by the Wireless Power Consortium. Many OEMs such as Energizer and electronics ones such as Texas Instruments (See reference) have embraced this standard. The manual ("bible") for the standard is free to download for any engineer. It is relatively easy to read and detailed enough for anyone to get tinkering.
See Attachments at bottom of page for PDF of schematic
On the top left above you can see the DC power supply input of 18V and +5V, +3.6V regulators. The +5V is required for driving the MOSFETs of the power inverter and analog processing opamps; +3.6V is required for the MCU. I separated the digital and analog sections by 1 Ohm resistors to minimize digital noise coupling to the analog sections. The MCU selected is the MSP430F2274. It has the following hardware features:
Two capture/control modules. One is used for generating PWM with adjustable frequency and duty cycle.
I plan on using the second timer module to determine the timing of the communication pulses.
Integrated programmable opamps: I am using one as a programmable gain amplifier for processing the communication.
10-bit SAR ADC with 15 muxed channels: Used for measurement of input voltage/current, resonance detection and temperature of transmitter coil.
A lot I/O: Useful for debug
Serial (UART) Useful for debug/communication to computer
The input voltage and current to the transmitter inverter & coil is measured so that the system has knowledge of the input power. The current is measured using a high-side current sensor, the LT6105. This is a very useful IC because it has very wide common mode range (0V - ~40V). The output is current, and can be used to create output voltage proportional to references other than ground; this is useful in circuits with split supplies. The gain can be adjusted by a single resistor, R17. An opamp buffers the circuit to drive the ADC. A minimum capacitance is recommended on ADC inputs due to the SAR architecture. The 200-ohm resistor is useful to help stabilize the opamp's output driver when driving capacitive loads. A simple resistor divider (buffered by opamp) is used to measure the input voltage.
I created a resonance monitor by sampling the voltage at the resonance node created by the Tx coil and resonance capacitors. The voltage is divided down and half-wave rectified.The capacitor C22 and resistor R21 form low pass filter which filter the AC to a DC level. The output of this resonance detector is a DC voltage that is proportional to the AC voltage present on the resonant node. I envision this to be used to do some calibration and/or determination of the actual resonant frequency of the tx coil.
The LM60 single IC temperature sensor is used to remotely monitor the temperature of the Tx coil. The IC includes a buffered output stage, so no opamp buffering is warranted.
I added a astable multivibrator circuit (PWM output) created from SN74HC14 Schmitt logic inverters with adjustable frequency and duty cycle controllable by trim potentiometers. The resistors R27 and R25 add limits to the frequency adjustment range. So this circuit just allows for user adjustable frequency and duty cycle for the power inverter useful for debugging. The limits are about 100kHz to 150KHz.
I added clamping diodes for all ADC Inputs to protect the MCU pins from damage. Note that the MCU also implements these, but this adds some more peace of mind for me.
The power inverter consists of a gate driver and a half-bridge leg of two N-channel MOSFETs. The gate driver is the TPS28225 and is used primary for driving logic level (+5V) synchronous half-bridges like this. Please note that logic level gates are not designed to be driven by higher voltage such at +18V!The one really nice thing about this IC is PWM input stage. Only a single PWM signal is required, the complementary signal is internally generated and dead time added between the two gate signals. Another really nice feature is that the input level is automatically detected (adapted to). Therefore a +3.6V PWM signal from the MCU can be directly used to drive the gate driver with no external level shifter required. The high-side gate drive power supply is internally generated by the well known "bootstrap" method. The bootstrap diode is internal to the gate driver.
I used the MOSFET IRLZ44ZSTRLPBF from International Rectifier. I decided to stick to an easy to solder package of the D2-PAK. A lot of these low voltage MOSFETs come in difficult to solder packages with solder tabs not exposed. I chose MOSFETs with Vdss of 55V which is really overkill considering the supply voltage of +18V. I chose this because I like to become familiar with just a few general components and reuse them in other circuits. The current drive of these MOSFETs is unbelievable at Id = 55A. At resonance the current in the transmitter coil is solely limited by parasitic resistances. So if we have +18V and say 0.500 Ohm parasitic resistance the currents can be (18/2Vac)/0.5(Ohm) =18A.
The DC filter capacitors are X7R ceramic multilayer chip. A larger value is not required since the ripple frequency will be 100kHz or higher. Electrolytic capacitors occupy a much larger board space. Since prototype board space is $5/sq inch, clearly anything that reduces board size is always cheaper than a cheaper component. The resonant capacitors are the more temperature stable C0G/NP0 ceramic dielectric and are rated for 100V. This should be augmented with some hot glue, RTV or other kind of glue to prevent arcing on your board from contamination.
The communication detect circuit is a bandpass filtered version of the resonant node at -3db Freq of 600Hz and 70kHz. I can change this based on what the communication frequency will be. A DC level of 1.5V is added to level shift the AC signal half-way between the rail and ground. A Schmitt trigger gate cleans up the slow-moving AC signal into a nice digital one with sharp edges.
This is the most complex part (and probably critical) part of this whole project. The overall strategy is termed power backscatter amplitude modulation. The secondary impedance is modulated at a frequency other than the power transfer frequency and this is seen as a change in the reflected impedance seen by the transmitter coil. So, the communication can be a modulated signal of 1kHz while the power transfer occurs at 100kHz. The way the receiver impedance is modulated is by switching in the capacitors C7,C8 at the communications frequency. Another way to think of it is that the receiver is being periodically detuned from the resonance frequency and the transmitter detects this. What is seen is an amplitude variation of the resonance voltage at the communication voltage. The voltage variation depends on the system operating frequency (how close near to resonance), the Q factor (how steep the resonance is) and a number of other factors that I don't even understand yet.
My simple communications strategy will be encode information in the communication frequency. For example:
By measuring the frequency (timing) on the transmitter side, the transmitter can decide if the system frequency needs to change to accommodate the receiver (load). This will definitely not fly for FCC or a commercial product because there is no safeguards against environmental noise being interpreted as a valid communication signal. What the commercial products do is encode the communication into packets with CRC bytes/bits attached so that the transmitter can verify that the packets being received are indeed valid. But this is too complicated for now -- I will put that implementation on the back burner.
The big issue is that the amplitude of the communication signal is not well defined and depends on the coupling, system operating point (freq), and Q. Anyways here is my thought: sample the resonance node through a bandpass filter to filter out only communications signal and reject the power signal. Additional scaling and adding a DC shift the signal to make it amenable for a single supply processing. The next step is to convert the AC signal (sine wave) into a pulse waveform with sharp edges for measurement of frequency (period). The frequency contains the information.
My first option is to use a fixed gain and feed the signal into a discrete Schmitt trigger logic gate with clamping diodes at the input. If the amplitude is sufficient to exceed the positive going/negative going thresholds, then the output will be a digital signal whose frequency is the same as that being transmitted by the receiver..
The second option is to use a programmable gain amplifier (PGA) to adjust the gain to ensure that if the input amplitude is too low/high we still can read its frequency content.
An MCU timer will be used to measure this frequency (period) and then decode this into a communications command. The MCU will have to do some validation to make sure that noise is not processed into a command
This is getting complicated and I don't fully understand all the details at this time. The important thing is not to get bogged down in details but provision options on the hardware (PCB) to test out different things. These two options, fixed gain and programmable gain will allow me full flexibility. "Let's make some mistakes already!"
Design of a PCB Tx Coil:
Winding a coil is always a pain. Winding a planar coil is another level of pain where you are trying to hold down this springy wire that flies into the air as soon as you look at it wrong. It's like having a cowlick on your hair, no matter how much gel you use that dang hair will stick up in the air. To also complicate matters, at 100kHz the skin depth is non-negligible and so litz wire (bundles of electrically isolated) wires is recommended. When working with multiple strands of wire, it can be difficult to strip each strand at the end of the enamel and make a good solder connection.
My approach is to design the coil on the PCB with well defined analytical expressions and hope for the best. To maximize the use of the PCB, a 3D structure is used whereby half of the coil is on the top side of the PCB and the other half of the coil is on the backside. Vias on the board (and a jumper wire) is used to connect the two coils.
I specified that the solder mask be exposed over the traces so that I can either 1) electroplate the copper or 2) flow some solder on the traces in order to decrease the total resistance of the coil. The voltages in this project can be quite high, on the order of 100-200V at resonance, so at the end I will pour some hot glue or RTV to protect myself.
Analytical calculation of the self-indutance is based on Modified Wheeler Formula as presented in . When you have two inductors in close proximity (top and bottom of PCB), the total inductance increases to L1s+L2s+ 2M, where L1s is the self-indcutance of 1st coil , L2s is the self-inductance of the 2nd coil and M is the effect of the mututal inductance. See the article "A new calculation..." by J. Zhao in Pulse magazine (in attachments at bott. of page) for a simple to read expressions for calculating all of the above.
PCB Service was Laen's PCB order which can be found in the Dorkbot PDX website at : http://dorkbotpdx.org/wiki/pcb_order. Technology is 2-layer FR4 with 6mil resolution. Cost is $5/sq inch for 3 boards with no limits on min. size or order. I highly recommend this for anyone who is developing their own PCBs.
Receiver coil is Vishay IWAS4832FFEB9R7J50 available from Digikey.
I plan to use the ez430-Launchpad which only costs $5 for a debugger.
 S. S. Mohan et al. Simple Accurate Expressions for Planar Spiral Inductances. IEEE Journal of Solid-State Circuits. Vol 34. No. 10. Oct 1999.