Cycling friend Kim, who rides a recumbent bicycle regularly, has written a report about a modification she did to her partner N’s ICE Sprint trike recently.
I thought it was a very good read and would be worth reproducing here, with Kim and N’s permission (which they gave). N’s bike was already previously adjusted for some physical disabilities (gear selection, etc) but is otherwise a reasonably standard model. As you can see from Kim’s report below, she has rather more electronic knowledge and skill than most of the population and has managed to produce a rather excellent result!
– – – – – –
So, for various reasons N hasn’t been getting the miles in. This means less cycling fitness, which has been limiting how far she can ride. Unfortunately, we live a 10km and at least one Bastard Hill round-trip from any pleasant cycling roads, which means we’ve been mostly pootling on the Rea Valley cycle route. Which is nice enough as Sustrans paths go, but gets a bit dull and doesn’t go anywhere – which doesn’t give you an incentive to get the miles in.
This seemed like an excellent job for electric assist – to give a little more speed and make the local hills less of a barrier.
We wanted something that was relatively straightforward to remove, for times when her cycling fitness is better. As the Sprint is a tadpole trike, this pretty much limits the motor options to those that fit in a rear hub. Since the aim of the game is asisted cycling, rather than an electric motorcycle, the BionX system seemed like the obvious choice. Unlike most systems, which simply apply power whenever the pedals are turning, this uses a torque sensor to provide power in proportion to the force exerted through the pedals. As a bonus, this means relatively little button-pushing in normal operation – N already having her hand full with the usual bike controls.
Unfortunately, BionX is currently unobtainum in Europe, and US dealers are reluctant to ship internationally. At this point, a fellow cyclist drew my attention to Falco, a relative new kid on the block. Slightly dodgy website, but the technology is getting good reviews. Their motor supports torque-sensor operation like the BionX, with a higher efficiency, lower rolling resistance when ‘off’ and – even more pleasingly – doesn’t require an expensive proprietary battery pack.
After an email exchange with the lovely Mark at Team Hybrid (the UK importer), I ordered the road legal version of the Hx motor, built into a 406 rear wheel with 11-34 9-speed freewheel (hub motors use freewheels rather than cassettes, presumably due to the larger axles). I’ve fitted larger middle and big chainrings to compensate for the loss of the 10t and 9t sprockets.
The battery was more difficult. We wanted something with lots of capacity, with a view to being able to do the standard 100km/1000m social ride at the lowest level of assistance. There’s nothing off the shelf that fits the rear rack on a Sprint RS (which has a suspension-specific luggage rack), so I decided to molish something. Reading around on e-bike fora suggested that the slightly dubious looking Ping Battery is a good source of LiFePO4 batteries (I decided that LiPo was too scary, and a false economy in the longer term. With EU Pedelec speed/power limiting, the battery won’t be under much stress anyway). I ordered a 48V, 20Ah battery from Ping, along with a 5A charger (this is a simple 60V SMPSU with a really noisy fan – the BMS controls the charge cycle).
So, after a *lot* of tedious research, I decided the simplest approach was to mount this in a durable box on the rear rack. I discovered that Rixen & Kaul make a luggage box that was just about big enough for the battery and some electronics, with a cunning quick-release adaptor plate that fixes to the rear rack (a bit like the Topeak rack bags).
Then it was just a small matter of electronics. Ping batteries include a Battery Management System (which prevents overdischarge and balances the cells properly when charging. I mounted this in a proper enclosure – rather than the mechanically dubious heatshrink wrapping it comes in – taking the opportunity to add a header for a remote shutdown switch. In a second enclosure, I mounted the guts of a Turnigy power meter (a CycleAnalyst seemed like overkill), and a board with DC:DC converters to power the existing dynamo lighting (experimentation showed that a Cyo wants 7.5V +-0.1V to run on DC at decent brightness) and relay logic to enable the BMS when the charger is connected:
The glowing green ring on the outside of the enclosure is the BMS shutdown switch. This avoids the need for a high-current isolator. I’ve used a 4-pole Speakon connector for the motor and lighting power – They’re rated for 40A RMS, the IP54 rating should be adequate protection against rain, and more critically, they’re a latching connector that N can actually undo easily. The charging port is standard 3-pin XLR. I’ve also added plenty of fuses – using 58V automotive fuses on both positive lines to the battery (5A and 30A respectively – the big one is visible front right).
The USB port is there primarily because the Falco console (slightly pointlessly, in this case) uses Ant+ wireless to communicate with the motor (there’s a dongle under the seat), and recharges its internal battery via USB. It also gives the option of using the pack as a 1550000mAh backup battery for your iThing!
I extended the cable between the console and the assistance level control buttons to put them somewhere sensible for under-seat steering:
The console is the weak point in the system, IMHO. While the motor is reassuringly solid, it feels cheap and plasticy, and the display has poor contrast. Advanced functions are accessed via cryptic combinations of presses on *spit* membrane switches. On the other hand, the wireless link to the motor seems surprisingly robust, and it can display the heartrate from an ANT+ strap. There’s a possibility that future firmware might allow a device like a Garmin Edge to communicate with the console to record torque sensor data.
The remaining piece of the puzzle is a crank rotation sensor:
Theoretically this shouldn’t be needed, but it provides a backup interlock to the torque sensing to ensure that the motor stops properly when it should. This consists of a unit containing a pair of Hall-effect sensors and a ring of magnets – unlike a standard cadence sensor, this can tell which direction the cranks are turning, and responds more quickly. It’s designed to fit inboard of the chainrings, with the sensor on a collar that’s secured by the bottom bracket. Unfortunately, ICE use a fairly narrow bottom bracket to keep the Q factor down and there simply wasn’t room. Mark suggested that I could fit it to the drive-side idler pulley instead, which seemed feasible with a bit of bracket fettling, but I opted for inboard of the the non-drive side crank instead, on the basis that it would be clear of skog, twigs and so on. As the sensor is directional, this meant mounting it upside-down. Hose clips and self-amalgamating tape to the rescue!
Beyond that, it’s cable ties all the way down. Lots of wiring for power, lighting (I removed the existing bottle dynamo, and provided a connector for re-fitting it in future), and the various data lines. All bundled up with the existing lighting and computer wiring, and the usual proliferation of brake and gear cables.
I took it for a proper road test today, and the motor performed as expected. It takes a bit of getting used to – the feeling is something like a strong tailwind that kicks in as you reach about 3mph and stops when you exceed 15mph. It took a bit of time to learn how to use the gears effectively to find the optimum balance between human and motor effort. Yes, there’s a lot of mass on the rear rack (though well within its load rating), and you do feel it’s there in corners. It’ll do, but if I were building this with a motor programmed for *cough* ‘off-road’ use, I’d want the centre of mass lower down. On the other hand, having lots of weight on the back greatly improves the rear wheel traction on mud. There’s also some rattling going on somewhere that needs attention – probably the box’s QR bracket moving in its clip.
Bastard Hills are where it really makes a difference. With the legal power limit of 250W, it’s not going to conquer a chevron without a bit of work from the rider, but you can maintain a much more respectable pace without stressing your lungs too badly. You certainly don’t notice that you’re lugging an additional ~15kg of battery and motor!
In the interests of SCIENCE, I just took it for a 30km ride, including the silly slalom hill in Cannon Hill Park, Primrose Hill and Weatheroak Hill, with the lights on (not that they use much power in the scheme of things – I estimate they would discharge the battery in a little over two weeks) and being less than sparing with the assistance level. When I got home, N took it out for a quick lap of Canon Hill Park. 35km in total, and some 300m of climbing. Ambient temperature was mostly around 9C, though the battery and electronics felt slightly warmer in use. The power meter claimed it had used 4.5Ah (234Wh) in that time. If those readings and the battery rating are accurate, that bodes well for a range in excess of 100km…
– – – – – –
and then an update:
And we’ve just ridden our 36km loop (which N hasn’t managed since last summer) out to the reservoir and back. Average speed of 17.8kph, which would probably have been higher if it hadn’t been for N and her motor having to wait for me to catch up on hills.
Rattling greatly reduced by the addition of a ratchet strap around the box, to stop it bouncing upwards.
I’ll check the exact figures later, but electron use was in the region of 240Wh again.