CleanHawk 250 Quadcopter Power Distribution Board

Got around to testing some stuff, and it's actually better then I expected with one minor surprise

First the bad: The TPS54331 regulator I'm using appears to have a drop out voltage of around 0.5V with light load (i.e. just LEDs) and around 1.2V @ 1.2A. Agrh, looks like the 10V regulator just became a 9V regulator. I expect people to crash their quads by running the batteries in to the ground at 3.3V / lipo cell which translates to 10V @ the Vbatt inputs. There is a schottky diode protecting against reverse voltage protection at the input of the regulator and may help explain the dropout voltages above the regulator output voltage. When crashing their quads I want it to be due to the motors stalling rather then the 10V regulator wimping out and adding noise to the FPV gear.

The observed affect at dropout voltage was significant noise on the power rail and audible ringing from inductors and caps. Not good, probably alot of stress on the components.

I'm going to stick with the same resistors for the LEDs and let them get a little dimmer. LEDs aren't bright enough to be considered useful in high sunlight... and I already 3 reels of 5000 resistors each.

tl;dr 10V -> 9V regulator for safety. The LEDs in series still work at 9V, so that was easy.

The Good: The TPS54331 + inductor + schottky diode selection is ultra efficient, better then design calculations.

5V regulator:

• Vin=12V + load=1.17A = 92% efficient
• Vin=12V + load=1.5A = 90% efficient
• Vin=15V + load=2.9A = 88% efficient

10V regulator (will be converted to 9V regulator):

• Vin=12V + load=1.4A = 94% efficient
• Vin=12V + load=1.6A = 95% efficient
• Vin=15V + load=1.6A = 95% efficient

My bench supply doesn't go beyond sourcing 1.5A, so it held be back a little bit for higher power numbers. I don't think there is too much else to learn with numbers this great. I was hoping for 80% efficiency + at max load, and this clearly does it.

I tested the reverse voltage protection and short circuit protection both intentionally and a few accidental times. No problems.

Only casualty of testing may have been some KISS ESCs that I may have overvolted around Vbatt=18V. Goes to show you that the KISS ESCs have almost no design margin. Haven't confirmed that they are dead, but they were definitely making bad noises. I'm hoping it was the LDOs on the KISS ESCs whining.

This is not necessarily the fault of the KISS ESCs, but I wouldn't ship something that close to the edge. In all reality my board could almost be rated for 6s if similar margins were applied to the input limit due to input caps (25V) and TPS54331 (28V).

Now, I need to hide these numbers from the "cost" trolls on RCGroups who will scream bloody murder for extra cost and blame it on over engineering.... but we all know that part cost is not the significant cost in building this board.

Board # 1 was assembled. No major deal breakers, so that's good. Quality isn't quite what I had hoped for so hopefully the second board house delivers a little better.

First set of 10 PCBs showed up after an epic international next day DHL delivery:

My current task is to find a local assembly house to put these together for me. Can anyone recommend a small volume assembly shop in the SF Bay Area?

Ordered (10) PCBs from 2 different manufacturers to evaluate quality. I have one in mind for large volume runs but wanted a back-up plan so I ordered both in parallel. Hopefully they'll arrive next week or the following week.

All the components are ordered for prototyping and the first run. this is my first time with this much invested in inventory. Reels of expensive components are... well... expensive! Almost all the parts are coming from US distributors, so that gives me confidence in both time and performance of the components. Re-orders of components should be easy if this were to take off. Boards will have a longer lead time.