Fallbrooks Harmony and hack potential

[Oran]

Got my hands on a load of Harmony stuff at a great price on Ebay, about 10% of what I could have payed for all new. 2 Harmony hub interfaces one of them still in the packet. An N360 built into a wheel, 2 Harmony controllers and a few extras, manual shifter and cable hub interface.

 
 
I’m interested to have a go with Fallbrooks version of automatic shifting. Find out how they’ve done it and see if there’s any potential to control the hub interface with an Arduino and my code. 

So far I’ve only had a quick look inside a hub interface. It’s as I was expecting, basically a custom made servo and the control circuit. Already showing potential for replacing the control circuit as an Arduino Nano would fit with careful trimming and even the USB socket would be accessible externally.

 

[Oran]

Soon had the hub and Harmony installed on my old mountain bike. Just using the middle ring a 38t and the 20t sprocket that came with the hub, no derailleurs or tensioner. To power it I’m using old 5 cell, 18V li-ion drill battery. 

Initial impression of it was not good the cadence seemed to be random at times. It also kept asking to be recalibrated frequently during the ride. This made me suspect there had to be something wrong. The most obvious reason for this had to be the position feedback. 

The most obvious weak point that I noticed when I first opened a Harmony hub interface was the position feedback. At the time I assumed it was a potentiometer but later discovered it’s an encoder. The problem is that its connected to the main output gear. Both pots and encoders are susceptible to damage from mechanical forces and ingress of oil.

Opening the used hub interface it was apparent how much more grease was used on the gears compared to the new one. There is the potential that oil has leaked in from a leaking hub. However its just as likely to have come from the grease, a case of "dab will do you, big gob will screw you".

Using an Arduino as an oscilloscope it became obvious the encoder was the source of the problem. 

Old encoder
 
New
 

I ordered a close replacement off Ebay and made it work by reshaping the shaft. The number on the original encoder is either 3315C-102-216M or 3315C-102-006L I now think it might be the one ending 006L meaning its 6 pulses per rev not 16 that I got, number  3315C-111-016L. Should have counted manually with multi meter first. Weirdly it seems to work so need to investigate further when I get time.