Hi
as I modified my first defective one (with the big hole in the sideplates which let water in and therefore destroyed it) with different components I did the following (after I had to completely disassemble it, thanks to good guidance with the foruim here) to catch up with state-of-the-art design:
a) I let a vacuum dip coating of the stator wiring done and a final shaving of the pol heads
b) change both bearings with energy efficient bearings from NTN which reduces loss from (at 50kg load) from 13W to less than 6W per Bearing, unfortunately they costed a lot due to my "I only need one" so each costed me about 30EUR instead of about 6 or 8 when I did as volume buyer; the difference is only measurable under heavy load. The endurance of the bearings should be more than 200mio turns, so enough for me ;-)
c) I coated the SMD upside+downside (not the FETs) with PU to have them shielded from moisture. Because it's a low adhesive version it's easily detachable by just pulling harder on it.
d) I did upgrade the PCB with a thick coating of soldering tin with a pure copper cable on top on all current switching lanes
e) I replaced all external cables as I need only PAS, SerCom, PowerBrakes, ground, 5Vout and Powersupply. The main issue was about cable length and thickness. The original may have a square of max 2mm^2, while the new one is a 4mm^2 with pure copper and for the other 5 signals I used 0,15mm highly flexible signal cable. So I kept the original mechanical brakeout, reducing my external "cable mess" to a minimum
f) I sealed (with a highly capable to creep PU mixture) both sides of the ext. cable so no water/moisture will come in. That's only to keep corrosion away from the cables. As Germany has become a bit of a wet country (maybe not as rainy as the UK) I wanted to conserve the cabling for a long life span.
g) I let the 14mm wheel be conserved by covering it with some "magic liquid" (from a friend who pimps 1950's cars as hobby) which coates the iron skin with a light blue shimmering; so far it hasn't corroded more (old rust was cleaned before)
What I missed:
a) strengthening the alu stator
b) exchanging the quality limited in-hub controller (which I see as an advantage to have it included in the motor) with an external
c) I did not (what I planned before) exchange the FETs to lower RDSon versions (better n) or higher Voltage versions (which I do not need)
d) neither did I change the standard condensators to solid capacitors with lowESR; it'd have taken to much time (investigation where to get and the right values) & money ( shipping costs mainly) to replace all of 'em
e) I planed to replace the 5V regulator with a high-eff. switching controller. TI and National.Sem offer a few with a huge range (up to 90V in) and n of about 90% with only five ext. components. I need them for my lights and (yes) iphone navigation, which consumes a bit more than the allowed 500mA. As I drive always with lights on, charging (iphone) is not possible. I'd still do it as the replacement would have been always external.
So I think this one would be the best MP3 I could imagine: the efficiency must be (pure estimation!) about 5-10% better, by just adding 50EUR to the component mix excl. the dip coating (must be lower for wholesale). The water resistance is still the same as original, but this was already in acceptable ranges: how the cables are (U) bound is the important point.
Conclusion: The MP2/3 is an exceptional hub motor unluckily hit&reduced by cost management. I own an Bionx as well and I find the MP3 more interesting (apart from the weight) by it's construction. When my second MP3 will have a maintenance needed, I will surely upgrade it as well!
Cheers and a good and fortunate 2014
Sam