To meet my requirements for speed and power I need a peak of 20kw power and a wide range of speed. I evaluated a large number of both hub motor and mid mount options. Hub motor was appealing as it would be simple, packaging the battery etc would be easy. Mid mount offers superior efficiency, lighter weight with the downsides of chain maintenance and noise.
The challenge to reach my goals was not making the power, it was cost - many of the motors capable of this 20kw figure required elaborate, expensive and very physically large controllers. John in CR's Hubmonster stood out for affordability, compatibility with relatively cheap/compact controllers (despite requiring two) but primarily it was attractive for out and out performance. Due to the high KV it's a challenge to use this motor in its ideal efficiency range in wheel, requiring quite a small rolling stock. By pulling it out of the wheel it allows me to volt up and gear down, delivering substantially more thrust at increased frequency. Plus I can vary this gearing with a relatively simple sprocket change.
With an effective phase to phase resistance of 0.016ohm, low no-load losses and general high quality construction this motor should actually be capable of substantially higher peaks than my 20kw goal, but I'm going to take this one step at a time
The plan is to mount a conventional sprocket using the 3 bolt disk brake flange on the motor. From my brief mockups so far the chain line actually ends up superior to the original Madass due to the increased front sprocket size giving additional chain clearance to the swing arm. Motor positioning is going to be approximately in the original location of the 50cc's crank case, as tight to the top tube and as far rearward as feasible. The motor will be mounted in a tubular fabricated subframe that picks up on the original motor/swingarm mounts on the main tube. Motor axle will be fully captured in 10mm+ thick torque plates. To the bottom of this structure I will mount the foot pegs and kick stand as well - these actually bolt to the bottom of the engine on the stock bike. It looks like I will be able to get these in the original location/height within a few millimeters.
I have purchased a pair of Sabvoton 100v100a capable units from zombiess. These are cheap, generally regarded to be reliable and should be capable of the ~10kw each I require to hit my performance goals. They are NOT well supported by the manufacturer and I fully expect a lot of pissing about with frustrating chinese software before I get these working as expected. These should run up to 120v and apparently deliver 100a and up to 350a phase current.
Long term I will likely replace these with something capable of higher current and better support, but only after proving the rest of the package.
I'll be using a CA v3 with external shunt for a Dashboard.
To make the most of the volt up, gear down train of thought while using the full capability of the Sabvoton controllers (without pushing them too hard) I decided on 28s or ~102v nominal, 117v HOC.
To get my range target I require about 3.5kwh of energy. That's applying a 40% modifier to my long term around town average on my other bike of 25wh/km. Depending on the situation, the new bike may actually be MORE efficient rather than less, but it will certainly weigh more, has higher rolling resistance tires and will be putting out a lot more power which I will definitely be using, so I think 40% more juice by distance is appropriate for estimation.
To get the peak power I need without excessive sag I need a pretty low ir pack. Based on my experiences with other bikes I don't like the reality of living with a large lipo pack of unknown quality/longevity. I have had very good experiences with 18650's and have built the tools I need to assemble reliable packs. Unfortunately I can't get enough power out of an entirely 18650 pack without ending up with weight and capacity well above the demands of my range requirements. This has flow on effects for power to weight and desired performance, requiring more power and in turn an even larger pack to supply it.
One potential solution is to build a hybrid pack, assembled from 28s12p of Samsung 30q 18650 cells in parallel with a single string of high power lipo cells. Based on recent product releases and testing it seems that the Turnigy Graphene 65c cells are the go to for consistent high C performance with surprisingly long cycle life. I have done some small scale testing with cells of varying levels of power density and the behavior of the current draw follows a proportional split according to pack IR.
So tentative pack specifications are
28s12p Samsung 30q cells - 15.5kg - 65mOhm - 3.6kwh
28s1p 6ah Turnigy Graphene 65c - 6.5kg - 35mOhm - 0.6kwh
All up 22kg, 22mOhm and 4.2kwh. Enough for my range requirement + 20% and enough punch to deliver 30kw efficiently.