Update 5/8/2015
Labeled as Dead End since no progress has been made in two years.

Project starts here

Many people are tinkering around with RC controls and robotics and trying to install independent sources of power into their projects.  Testing devices with these independent power sources often presents some problems for the lab, and especially, the sensitive bench instruments on which we all rely.  Ground loops are often a surprise event, and can be costly, in terms of loss of the project device, or the test instruments themselves.  Here is my answer to this, a fully isolated and floating regulated power supply, with the capability to charge a lithium cell either through USB, a wall wart, and even wirelessly.

This project is in three parts. First, the BLiVIT. This is a small (60 mm x 37 mm) board that contains an integrated single Li+ cell charger and management circuit, plus a fully isolated buck-boost DC-DC converter to set the board power output at either 3.3V or 5V. The board can be powered one of three ways: 1. The 2.1 mm DC "wall wart". 2. The micro-USB-B connector, which is limited to either 100 mA or 500 mA constant current load, in accordance with the USB 1.0 specification. 3. The optional mezzanine board "POQiRX". The prototype battery I am using is a 950 mAh cell that is approx 15 x 20 x 3 mm, but others can be used. The POQiRX is a WPC Qi-compliant wireless power receiver that uses a bqTesla power manager from TI to accept and control transfer of up to 5W wireless power, which is directed to the BLiVIT for regulation and isolation, and charging of the Li+ cell. The board dimensions are 25.4 x 22.6 mm The third component of this project is the POQiTX, which, as one might expect, is the complementary transmitter for the POQiRX. This device is a (60 x 60 mm) low-power-driven transmitter that can be powered either from the 2.1 mm DC jack or USB connector. It is my belief that since I am using the lowest-power configuration of the bqTesla devices the POQiTX might achieve EnergyStar certification at some point. Possible uses include RC models, field-charging portable devices, small robots, security devices. Really anything where you would want a wireless charging capability, or other portable device power capability. The full isolation is a significant safety enhancement over typical boost/buck dc-dc converter implementations.

The power coils are different for the receiver and transmitter.  The receiver is a 32 mm oval coil from Vishay that is being used for a lot of cell phone applications these days.  The transmitter is necessarily more beefy, so I am using a thicker 50 mm round coil from Wurth.

There are a lot of ways the POQiTX/POQiRX coils can be brought into proximity, so simple is probably best here.

The enclosure uses a laser-cut acrylic interlocking sheet technique (thanks to Ponoko).  A 3D rendering is included here.  The sides and bottom of the enclosure are assembled with acrylic welding cement, but the top uses 3mm screws through to the bottom to secure the enclosure.  The exposed screw heads on the bottom plate are also used to center the receiver coil over the transmitter.

To provide an on/off switch for the target load power the Powah-1,  a small 0.5" x 0.5" module, is used.  The mounting pins of the Powah-1 elevates the module so that the tact switch is flush with the case top before soldering the long pins to the pads on the BLiVIT.  I have included photos and a schematic for the Powah-1.

Update for 8 July 2013:

I have been thinking about future improvements to this design.  I think the electrolytic caps C9 and C10 should be either 1. moved so that there is more "real-estate" around each, so that it is easier to assemble, 2. respecified as 50V is probably overkill on the rating.  3. changed to use tantalum parts.  That would have benefits such as much lower ESR, smaller footprint, and importantly, no longer dependent on liquid electrolyte which improves shelf-life and service life.  The only downside I see for tantalums are higher cost and less availability.

For another factor I am needing to either 1. substitute a 1.3mm jack instead of the 2.1mm jack for DC power, as I have done for now; 2. Maybe use a right-angle KK type header and use a shrink-fitted adapter to bring in the power.

I have noticed that either way I will have to respin the enclosure design (The Powah switch hole needs to move a little, and if I make it larger the resulting round plug created in the laser-cutting process can be reused by gluing it to the tact switch with cyanoacrylate).  It would also be vastly more useful if Ponoko can use 1.5mm acrylic, but that may be hoping too much.

Update for 12 July 2013:

It appears that there are better values for R11, R12, R14.  The datasheet for the ADP2503 boost/buck regulator has a formula Vout = (R11 + R12 (or R14)) * Vref / R12 (or R14).  Vref is typically 500mV, so the better value for R11 is 300K, R12 is 53.6K, and R14 is 33.2K.  This results for jumper-selecting R12 sets Vout=3.299V.  Jumper-selecting R14 sets Vout=5.018V (nominal values).

A spreadsheet calculator is enclosed here so that other values can be substituted for other outputs.

The updated Mouser numbers are:

R11 300K 71-CRCW0603300KJNEA

R12 53.6K 71-CRCW060353K6FKEA

R14 33.21K 71-CRCW060333K2FKEA

Update for 3 Sep 2013:

I have been giving some thought to the output interconnect story.  If the BLiVIT is going to be used to the fullest degree then some sort of quick connect would be needed instead of the screw-terminal design currently used.  When looking at laptop computer batteries, for example, a great deal of thought seems to been put on the mechanical interconnects of the batery terminals.  If there was some way to have a simple device that contains a pair of pogo pins, for example, then the screw terminals can be replaced with a pair of contact pads.  Then the BLiVIT can be pulled out of the application device and placed on the charge plate of the POQiTX, all without needing any sort of extraction tool.  That would be really nice.  I will look into this.