AVR Freaks

The runtime, or flight time i calculated to be about 40 min with 5 min reserve, these specs are if my calculation meets reality about the aircraft being able to maintain level flight with about 5,5Kw.

The pcb-s were completely home made and they are double sided, the power board is with busbars soldered on top and bottom of the pcb, minimizing inductance was a must here, 300A thru 5 cm of bus bar creates a lot of unwanted trouble, the first iteration had about 50nH on power traces (+ and -) and i had to redesign it to reduce inductance to about 16nH to keep the inductive spikes to a decent level.(the red polyester caps on power board were replaced with ceramics , this is a older photo) Command board also had 3 iterations mostly to be able to keep separate analog supply traces from digital ones. On command: board top right power supply, buck converter and linear reg-s; top left motor control, MGD-s, uC ; middle to bottom BMS  ,controller management and analog side.

As mentioned above, the estimated flight time is 40 min with 5 min reserve, if aircraft flies level on 5.5Kw. Battery weight is 28Kg, and has 4,44Kwh 60Ah, it is divided in 2 modules, each 14kg. About motor design and stress testing, i calculated the tension load on the shaft (shaft material is Cr42Mo4 steel) and on the bolts (10.9 grade) of the prop and motor, there are the most stressed parts (and bearings), the rest was done by carefully looking at how others build motors like that, and maintaining some proportions. The winding scheme was calculated (ABC pattern). There was no need to reinvent the wheel, every one that builds them, thy stick to same pattern, so i did so. All in all it came out pretty good. Aircraft on the other hand, i did a lot of load calculation and stress simulations, the aircraft is based on Goat 4  of Mike Sandlin design http://m-sandlin.info/goat/goat.htm ,and has some modifications to the wing size in order to acomadate propulsion system weight and maintain same wing loading and more important the same slow flying speed of about 43km/h. Controller i design it completely, i took some inspiration from here and there, but it did require lots of math to be done like thermal design, high current path design ect.        

Yes indeed, i have put some thought in to that, the command board is mounted semi rigid inside the unit, all the connections to it are with wires, so that there is less transmitted vibration to the board and no mechanical stress on pcb connections, the power board is mounted rigid to the unit heat sink, all the connections that go out have a snug fit thru the plates of the sides including bulk caps. There are a few things to improve, one of them is to find a way to attach those 35sq mm cables to the board by NOT soldering them, initially i wanted to use crimped lugs but they are bulky and i could not get that compact design with them. The hole unit attaches to airframe by rubber vibration isolators, that is a must to ensure that no heavy vibrations is passed to the unit. Now about mechanical shock, this system is powered by a Li-Po battery with pouch cell type, these cell types are easily deformed if subjected to mechanical stress. The cells are mounted in a aluminum box with 1,5mm thick walls, witch offers good mechanical rigidity and protection, so in case of light shock is ok, but something like crashing a PPG from 3-4 meters, will almost certainly damage the battery and possibly cause it to catch fire. So the electronics surviving mechanical shock is limited by the battery ability to survive mechanical shock. The controller unit is very rigid and tough and if dropped from 3-4 meters it has a great chance of still working, or at least greater than the battery. Pcb-s manufactured by a professional company will certainly improve mechanical safety aspect, so there is room to improve.   

Thank you for suggestion. The bus bar that i used was kind-a of the shelf, it comes in 1meter length 11mm wide 1,5mm thick. A neat solution would be CNC or laser cut plates to match the pcb trace area, and then it can be used thinner pate, something like 1mm, witch is easier to solder and give better performance plus will look top notch.

Yep, thank you for advice.

Nope, it has about half the energy density per mass. Li-po battery's are seafe, if you treat them with care. In 15 years of working with them, and i had a few smashed to the ground hard, they bent but i only had one catching fire.

I'm a bit worried that it will warp the bord becouse copper and fibreglas(fr4) do not expand and contract at same rate, it warped a bit the board that i already made and the bus bars were soldered with hot air one at a time.

I cut with water jet the copper busbars, they are awesome, i have to de-burr therm a bit, and they fit perfect.

Yes, the caps and case+heatsink are grounded, they fit very snug in those holes. They are out for cooling purpose, but also the case can be smaller and lighter, plus it look cool :D . 

For circuit breaker, i did not use any  , i assumed that the battery is capable of delivering 1200A continously and 1800A in bursts and that can vaporize anything from the controller in its path, and i went in to the full power tests all or nothing. On earlier test i did blew a few mosfets NOT all 18 at once (lucky me  ) , and i don't know how the hek do they fit so much smoke inside them . For flight i have a handle that disconects the positive power cable in case something goes PUF, but the thing is contained in the case so only smoke can come out, it is kinda anticlimactic .

The mosfets are bolted directly to the heatsink (see photo, this is v1.0 but v1.3 is the same except copper plates busbars), TO247 package would be pointless without direct contact with heatsink , and they do have separate gate resistors (the white isolators in the photo), and i dit a looooot of fiddling to get rid of the ringing, there is a topic around here about that subject; https://www.avrfreaks.net/forum/... , hopefuly the new fets with better Ciss/Crss ratio wont give so much head bashing.  Cable clamps, yes still on to do list, now they are installed on the motor/controller mount/pole so the hole thing gets secured in place, a small amount flex allowed to reduce vibartion transfer, but that i might change it, i am not compleately satisfied with that arrangement, i still have to figure a way of securing those cables better, i might have a ideea to clamp them inside the case.

Thank you, will do!

No RTV, only colophonium rosin thined with isopropylic alcohol.

Yes, but that is overkill and is masive, it wood take to much space, for a 50kw or 100kw that wood be the solution but not at 20kw, it is way to big

I understand you, i did consider these options, but thy wew ruled out because of weight and dimensions beeing to big

Hot glue, glue gun, but on v1.3 only on command module, on power module there is no need any more.

well, on command module i use it to get some rigidity to all those wires, board to wire and wire to connector, i'v noticed that it does not bond very well unless i encapsulate what i'm gluing, i'l try some "special" RTV. The  colophonium rosin thined with isopropylic alcohol, was brushed on the boards to protect copper from corrosion.

it is time for a late update, V1.3 is done (it was done 2 months ago) ...