In the mid 90s someone posted a question in an online forum asking what kind of network cable it would be good to install in a house with the idea of it being an addition that future owners of the house could use. Someone replied that wireless networking would replace cables in home networks. I wrote that this claim was optimistic. To get the throughput needed, you’d need a high symbol rate, which would make it vulnerable to inter-symbol interference. Or you could have a large alphabet size, but that would require a high SNR. 
Moreover you’d need collision avoidance that works well. I knew it worked in MATLAB, but I didn’t foresee it being adopted widely in the home.
Two years ago, when I bought my home, I found there were Ethernet cables connected to an each room. In some of the rooms they weren’t terminated, and the easiest way to deal with them was to pull all the slack and cut them. It would be a pain if I ever wanted to have those cables in those rooms, but I was completely confident I wouldn’t. As I did it, I felt like the universe was ribbing me about my forum post 15 years earlier.
So I try to be open-minded about the idea of eliminating power cables. It’s easy to couple power magnetically. The efficiency drops rapidly, though, when the distance between the coils exceeds one coil diameter.
I tried testing some Qi-complaint TI eval boards. The transmitter was bq500210EVM-689. The receiver was bq51013EVM-725. You provide 19V DC to the transmitter. The receiver provides 5V DC and provides a maximum of 1A.
I observed the efficiency decrease when I placed paper between the the coils, but not by all the much.
| Coil Separation | Output Load | Input Power | Output Power | Efficiency |
|---|---|---|---|---|
| No Link | NA | 0.17 W | 0 | |
| Direct Contact of Eval Modules | Open | 0.80 W | 0 | |
| 170 mils [4.3mm] of paper | Open | 1.00 W | 0 | |
| Direct Contact of Eval Modules | 50 ohms | 1.69 W | 0.5 W | 30% |
| 170 mils [4.3mm] of paper | 50 ohms | 2.02 W | 0.5 W | 25% |
With no link, the "analog ping" looking for a load drew an irregular current for 35ms every 400ms. (Below I measure it with a 10-ohm current sense resistor.)
I couldn’t go more than about a quarter inch [6mm] before it would shut off. I suspect this is because it needs communication. The Qi standard is not designed to operate at distances where efficiency would be very low.
I suspect non-Qi-compliant products will appear that receive a tiny fraction of transmitted power. If someone works out a way, say, to harvest 5% of the power coming from a 100W transmitter over a distance of a meter, applications will crop up everywhere. Coffee shops could put them under the tables so computer users wouldn’t have to plug in. The efficiency is terrible, but people will plug in those 100W transmitters with the same lack of thought people used to give to turning on a 100W incandescent bulb.
