• Howdy! we're looking for donations to finish custom knowledgebase software for this forum. Please see our Funding drive thread

Super Commuter: 6T MAC 48V/120A Kelly in a 26" rim


1 MW
Aug 2, 2011
Some pics of my last built. Did 4 bikes with mentioned motor and controller. Bought them from cell man. Works good.

The whole system is intended to be fool proof. As these bikes are commuters, they have to be reliable. So I chose components that are known to be reliable. The battery is somewhat oversized for the range aimed for (30km). The tires are puncture proof. The connectors can not be plugged in a way that anything would be damaged. Ther is a killswitch/ignition switch directly at the throttle, in case a malfunction of the throttle or other part of the drive system occurs
Fenders, Reflectors, side stands and so on make these vehicles usable and save in workaday life.
The charger is fan-less/noiseless, so the battery can be charged at work without being annoying workmates

A PAS sensor can be used for cruise control if a longer range like 60-80km is desired. The Cycle Analyst (CA) V3 has been setup with an Economy mode: PAS Power is limited to 1000W and PAS speed to 40kph (the throttle still works as usual up to 60kph and 4kW). Not really usable ATM since the brake cutoff switches at the brake levers are still missing. PAS cruise without these switches is a bit dangerous.

Major Parts:
Motor: 6T upgrade MAC from cellman
Controller: Kelly KBS48121X controller from kelly (without 70k e-rpm upgrade, it does >50k e-rpm without issues)
Cells: 13s7p Samsung INR 25R from nkon
BMS: OZ890 Smart BMS from bmsbattery, added 10pcs IRFB4110 MosFETs to make it handle 120A instead of only 20
Display: Cycle Analyst V3 DPS from Grintech
Charger: Meanwell HLG 240H 54 (fanless, 240W, weight is only 1.2kg, a full charge takes about 3.5 hours)
Bikes: 2xKona Stinky (2008 and 2009?), Giant XTC (2001), Centurion (2008)

1st step: Kelly soldering session.

I put on:
- CA-DPS plug: 6-pole JST-SM plug, the yellow wire was used for the thermistor instead of the speed signal from the hall sensors (which is not suitable for geared hubs)
cellman told me to do so, saves me one cable to the handle bar (the rest of the CA-DPS Plug is the same for the CA-DP plug. I had to open the CA V3 DPS and solder the yellow wire of the DP plug to the thermistor pin)
-0.25 m long AWG12 wire as external shunt ( I ran out of AWG10 and AWG8 was too thick/would have been too long for an 1mOhm shunt: AWG8 would be 0.5m @1mOhm) it drifts a bit with temperature ( up to 10%) better go fo Constantan if you want something really precise
- fuse for CA and lights
- killswitch plug: a 2-pole JST-SM plug
-XT90 plugs for the battery, shortened the wires
-4mm bulltes connectors with HXT connector insulation for the 3 motor phases, shortened the wires
-Insulated all remaining unused controller cables with heat shrink
-program the Kelly throttle settings, so it will accept the Cycle Analyst throttle signal 1V...4V (last picture below)


  • P1010204.JPG
    96.1 KB · Views: 14,372
  • P1010207.JPG
    115.6 KB · Views: 14,372
  • P1010212.JPG
    105.7 KB · Views: 14,372
  • P1010216.JPG
    100.3 KB · Views: 14,372
  • P1010222.JPG
    118.1 KB · Views: 14,372
  • P1010333.JPG
    90.5 KB · Views: 14,372
  • P1010345.JPG
    92.2 KB · Views: 14,372
  • !Kelly_CA wiring.jpg
    !Kelly_CA wiring.jpg
    151.8 KB · Views: 14,183
  • Programming.png
    28.5 KB · Views: 14,172
  • Verkabelung Kelly.ods
    10.2 KB · Views: 304
session 2: File down the drop outs and instal the lasercut torque arms
Filed down the dropouts 2mm to correct the position of the rear wheel. Original axle size was 10mm, new axle size is 14mm. Now the wheel sits right where it should and the disc brakes byte where they should :idea:
At the end o´f the day, i installed torque arms on both axle ends and added compression bolts, pics will follow


  • P1010257.JPG
    82.1 KB · Views: 14,425
  • P1010266.JPG
    66.3 KB · Views: 14,425
  • P1010242.JPG
    80.2 KB · Views: 14,425
session 3: Assemble the battery
i glue the cells with FIX ALL TURBO before i start welding the 0.15mm nickel stripes (25mm wide)


  • P1010507.JPG
    107 KB · Views: 14,409
  • P1010420.JPG
    103.1 KB · Views: 14,409
  • P1010432.JPG
    99.8 KB · Views: 14,409
  • P1010529.JPG
    84.7 KB · Views: 14,409
  • P1010551.JPG
    98.5 KB · Views: 14,409
  • P1010516.JPG
    113.8 KB · Views: 14,409
  • P1010518.JPG
    94.5 KB · Views: 14,409
  • P1010552.JPG
    93.7 KB · Views: 14,409
  • P1010554.JPG
    84.8 KB · Views: 14,409
  • P1010532.JPG
    102.9 KB · Views: 14,409
  • P1010555.JPG
    117.2 KB · Views: 14,409
Session 4: Collecting parts


  • P1010248.JPG
    92.2 KB · Views: 14,416
  • P1010236.JPG
    91 KB · Views: 14,416
  • P1010227.JPG
    75.7 KB · Views: 14,416
  • P1010224.JPG
    78.4 KB · Views: 14,416
  • P1010299.JPG
    72.9 KB · Views: 14,416
  • P1010330.JPG
    98 KB · Views: 14,416
Session 5: Insulate throttles (bmsbattery hall throttle). Some thumb, twist and half twist throttles where used


  • P1010387.JPG
    77.6 KB · Views: 14,406
  • P1010378.JPG
    84.8 KB · Views: 14,406
  • P1010390.JPG
    67.9 KB · Views: 14,406
Putting it all together


  • P1010571.JPG
    161 KB · Views: 14,173
  • P1010569.JPG
    97.9 KB · Views: 14,173
  • P1010570.JPG
    104.6 KB · Views: 14,173
  • P1010586.JPG
    95 KB · Views: 14,173
  • P1010583.JPG
    127.1 KB · Views: 14,173
  • P1010572.JPG
    113 KB · Views: 14,173
  • P1010567.JPG
    110.1 KB · Views: 14,173
  • P1010562.JPG
    118.4 KB · Views: 14,173
  • P1010580.JPG
    99.3 KB · Views: 14,172
Hella power! exciting battery build! :D
thanks, i post this so others can learn and for me so others can criticize things that could fail/ be improved

swapping the blue and yellow cables (both Halls and Phase wires) makes the kelly 100% MAC compatible. CA V3 compatibility need sadly same of soldering

So why not an Infineon?
1. throttle control of the Infineon is speed mode. It's un-ride-able at this power level. Not controllable. The kelly has really smooth torque throttle
2. The Infineon will burn if too hot
3. the infineon will burn if you connect the halls wrong
4. the infineon is large. it wont fit into the saddle bag
5. the infineon is weak. A 6 fet infineon would fit into the saddle bag, but it would not handle 120A like a 12Fet Kelly of almost the same size
6. you loose warranty if you reprogram the infineon
7. the infineon is noisy, it does not do sine wave commutation

surely there are more reasons against the infineon that just dont come to my mind ATM. But there is really no reason to use such crap

The oil cooling is no experiment. It was planed in the first place, but I wanted to test how much difference this does, so I drove around without oil for some days. Shell Diala S2 oil was exactly designed for such purposes. And cellman does not sell his motors with shaft seals just for fun ;)
at low speed, the infineon throttle is just not controllable. That is why there are throttle tamers. The CA V3 can work as a throttle tamer, but that induces throttle lag. I want instant, but controllable throttle, so there was no way around a controller like the kelly. The price was lower as an infineon, too. An infineon that does 120A stock would have cost me more than the $150 for the kelly.

sure the kelly can be programmed. like all controllers. Thing is, you get the software right on their homepage. For others, like the x8m06c you have too search a lot, but at the end of the day, nothing really works the way you want
Great work, Crossbreak :D A photo of torque arms from each side holding the motor shaft, please :)
The main reason for the CA V3 is the motor temperature rollback feature?
thanks David :) Jep, that is exactly what makes this power level possible.

I could as well have connected the thermistor to the kelly. This would make live safe, too. But the Kelly has no roll back feature, it simply shuts down the power stage completely if the motor has reached the maximum temperature. This could be dangerous in traffic.

The rollback feature is worth a lot. It makes live easy. You can still hit full throttle all of the time without wondering about anything. You can't burn your motor and still dont risk a power stage shutdown.
it's stainless, 104mm BCD, 52T, fits 6-8 speed chains

another pic of the CA and Kelly marriage which may give some insight in their relationship


So I assume that you are happy with Kelly.
How to estimate max battery current based on peak motor current?

As far as I understand KBS72121X will give 130A phase on 100% settings, but how is it related to battery current.
(basic Infineon rule is simple: phase current = 2.5 x battery current)
So far, i have not experienced any battery current limit, even if running more than 100A for more than 30 seconds. So the maximum Battery current is equal to the maximum phase current at the point of maximum power. This is at 40kph for this bike. As a result, at all other speeds it is less.

these controllers are actually about 15% underrated. A 120A model will actually output ~150A. This is why i programmed the Kelly to 80% motor current, otherwise it would be 150A phase current for a speed range ~10 kph to ~40 kph. 150A is a bit too much for this motor. Feels like it saturates. I just explained that further there: http://endless-sphere.com/forums/viewtopic.php?f=2&t=63578#p1022721

Sadly the Kellys dont output full phase current at low speeds. This is why many say that a weaker Infineon still pulls faster off the line.

Just measured KV for the 6T/kelly combo: It is 655rpm/52.4V = 12.5kV (62.5kV @ rotor))
Hi crossbreak,

I was wondering if you could maybe shed some more light on that programmer for the Smart BMS ? Do i need to buy there "proprietary" programmer or can i just use a generic one ?

Nice Build :D

thx. Some succeed in programming with a generic one, but they could not use the vendor software. Most easy way is to use the proprietary one.

I did both. We made a GUI software to use our own arduino based programmer. But i did not write all of the code myself... so sadly i cannot give it away.

ATM i am again using the proprietary programmer, even if it has some minor bugs. It does the job and is easy to use. For example, i still i did not manage to use external 10k thermistors. they are always ~40°C off. They show 60°C if it is only 20°C. Did not investigate further here.

Another issue is to get clean shunt readings. There is a lot of drift in it. I have a 0.5mOhm shunt, but if i calibrate, i have to program a value roughly half of that: 0.26mOhm
Guess this is due the strange shunt circuit on these boards. This circuit makes it possible to use the MosFETs to be used as a shunt (which does not make so much sense as FET resistance depends highly on temperature) But i did not investigate further here as well yet.

there are some threads about this topic: