Today I used Kapton (okay, Koptan) tape for one of its common purposes for the first time, I used it to stop insulation melting and disappearing while I was tinning leads. I got the leads out of an old parallel cable, and they probably weren't meant to be soldered at all. Old computer cables are bar none the cheapest source of highly multi-colored, small-gauge finely-stranded wire I know of. If you pick them up from thrift store warehouses or flea markets you can get them for less than a dollar, and get yourself a meter of 20+ different colors of wire.
What I was doing with the specifically-colored wire was making a fresh new cable for the GPS on my dead cat drone. I had hacked something together from stuff I had lying around, and it worked okay but it was lumpy because it had an extra splice in it. Now I have "correctly" color coded wires running all the way from the GPS to the flight controller. While I was in the GPS' case I also verified that the compass chip is a Honeywell (or clone) HMC5883L by googling its markings, but I knew that already because I had specified it in Cleanflight (
mag_hardware=2) and it worked. But today I finally got the bright idea to wave a magnet near both the FC and the GPS and I got a reaction near the GPS, so I win. The SP Racing F3 Evo carries an onboard InvenSense MPU-9250, which in turn includes the AK8963 compass chip. The HMC5883L actually seems like a good, low-noise compass chip which is what you really need for a noisy environment like a quadcopter. The onboard 9250 is connected to the processor via SPI, which gives it a higher update rate than typical IIC-connected designs. The HMC mag sensor chip is connected via IIC, which it has to share with the upcoming OLED display as well as my planned external barometer. This means it's going to have a substantially lower update rate than the onboard equipment, but it's still well worth it, mostly because it's on the stalk with the GPS (and its antenna) and thus should be subjected to substantially less noise from the motors. The GPS is marked "BD" and it contains a counterfeit uBlox Neo-8M, the HMC chip, and a 3.3 volt regulator to provide it power. The compass produces a 3V IIC signal regardless of whether you supply it with 3.3 or 5 volt input, which is convenient for connection to the F3 Evo — it only has 3.3 volt IIC.
This particular use of the Koptan tape (a Chinese knockoff of Kapton with less heat resistance, but otherwise similar properties) was not the most interesting one of the last week, however. I recently used it to separate layers of copper foil tape with which I built a PDB (Power Distribution Board) for the upcoming RGB LEDs on the upper board. The tape can only handle about half an amp, but that is sufficient for the four LEDs which I am attaching. Over each layer of foil tape, I placed Koptan tape as an insulator and protective layer. Exposed contact points were touched briefly with the iron and standard rosin-core solder, which made very nice solder pads. I ran ground and power traces in a horseshoe shape, and a broken data wire to the four corners. Power will come from one of the ESCs, each of which has a pair of 78M05 linear regulators, which should provide clean if inefficient power.
The motor and ESC leads turned out to be long enough to bring the ESCs inboard, which I've always wanted to do, both to protect them and to improve handling by reducing the moment of inertia. These are absolute-cheapest 30A ESCs running SimonK firmware, and lucking into four of them is what led to this build. This meant that there would not be a lot of ESC lead to manage, but it remained an issue. My solution was to order a sheet of copper from China for a couple of dollars, and cut it into 6mm strips using shears. I then folded it over at the proper angle (just a little more than ninety degrees, here) and soldered the fold point liberally, creating the feed point. I double-insulated it with color-coded heat shrink tubing, and tinned the ends for connection of the ESC leads. This meant that only two leads had to be joined to each pad, which makes the soldering job a lot simpler than trying to group all four leads together and solder them. This also takes up less space than a PDB, and I didn't want one in the center of my design. The official "Defibrillator" PDB for this design is not only expensive, it's also reported to frequently be in poor condition with traces connected together internally, or broken. The dual-voltage power supply is also often reported to be of poor quality; I have a cheap one already.
Since I am using very little of the ESC BEC output, I am not very concerned about efficiency. (I have a switching mini PDB with dual BEC for efficiently producing power for video output, and anticipate installing LC filters fore and aft for 5V and 12V respectively.) Motor #1's BEC is currently supplying the FC, and #3 feeds the RX. It's an OrangeRX R820X v2 which I purchased as a reasonably-priced Spektrum-compatible unit which could bind satellites, and it has telemetry and requires too much power to run from the voltage regulator which supplies power during USB connection. Motors #2 and 4 are mounted at the front of the airframe, where the center of the LED power bus sits — the logical place to make connection.