I Wanted a Rowing Machine. I Built an eBike Instead.

I bought the bike in early 2019, and not for any considered reason. A coworker rode it into the office one day and I liked the look of it. When he moved away not long after, he asked if I wanted it, and I said yes.

Then I barely rode it. Commuting on it never even occurred to me at the time. At some point we moved house and the bike stayed behind at my parents’, where I mostly forgot it existed. I brought it back in mid-2023, gave it a clean, and then promptly forgot about it again.

I should say clearly, at this point, that I was a complete novice when it came to bikes. Everything from here on I learned by doing it, usually the slow way.

The current spec lives on its own page, My eBike. This is the story around it.

The rowing machine that didn’t fit

What actually got the bike moving again had nothing to do with commuting.

I wanted to exercise more. My first instinct was a rowing machine, but there was nowhere in the apartment to put one. Looking around for alternatives I came across Zwift and the Wahoo smart trainer setups people build around it, and the idea clicked: instead of buying a dedicated indoor bike, I could use the one that was already sitting there doing nothing. So I pulled the Winora apart and kept only what it needed to sit on a trainer and talk to Zwift.

I rode it like that for about four months. I liked it, and still do, but I don’t see much point doing it indoors once the weather outside is decent. Along the way I did all sorts of routes, and one day, more out of curiosity than anything, I loaded up a route that matched my actual commute to the office and rode it on the trainer.

That’s the moment the idea of doing it for real took hold, though the eBike part hadn’t arrived yet. It planted something more basic: that I might actually want to cycle to work outdoors, not just simulate it indoors.

Once I was seriously thinking about the real commute, an eBike started to look like the obvious answer, and my first instinct was just to buy one. A proper pre-built eBike, off the shelf, done. The problem is that they’re expensive, and I’d be spending that money on a guess: I had no idea yet whether I’d actually stick with commuting by bike, or whether it’d be another thing I was enthusiastic about for a month.

So I looked into hiring one for a few weeks instead, to try the commute properly before committing. That turned out to cost far more than I expected. Enough that it stopped making any sense, because I’d be spending real money to answer a question and still own nothing at the end of it.

That dead end is what set me off. If buying was too expensive to risk and renting was too expensive to justify, was there a cheaper way in? Which is roughly when the thought turned up: I wonder if there’s a way to convert a bike into an eBike. I already had a perfectly good bike sitting in the apartment, after all.

I was probably close to trying the commute unassisted first, on the bike as it was. It’s not a small climb, about 400 metres over the ride. But before I’d settled on a route or waited for the right weather, the conversion idea got there first, and I knew straight away that was the way I wanted to do it.

The only kit I’d heard of

What I found was Swytch: a hub motor with the battery clipped to the handlebars, simple to fit, no need to touch the drivetrain at all. They do a front and a rear version, and I weighed both up without ever getting very exercised about it. I was close to ordering one.

I should be accurate about how little I knew at this point, because it explains most of what follows. I wasn’t comparing options and choosing the best one. As far as I was aware, Swytch essentially was the option. There were other hub motor kits out there if you went looking, but they struck me as messy and cheap, a bag of loose parts and exposed wiring bolted onto a bike as an afterthought. None of them appealed at all. Swytch looked like a designed product, something a company had actually thought about, and that mattered to me more than I’d probably have admitted at the time.

It had not occurred to me that mid-drive kits existed.

What made it feel obvious was the price. The kit cost roughly what a single month of hiring an eBike would have, which was exactly the trial period I’d been considering. Same money, except at the end of the month I’d still have it, and it’d be bolted to a bike I already owned. Renting suddenly looked like the silly option.

Two things stopped me. Their stock was backordered for a while, which gave me time to think. And the more I thought about the climb, the more I suspected the standard battery wouldn’t give me the range I actually wanted for the round trip, which meant stepping up to a bigger pack, which meant a noticeably more expensive kit than the one I’d been looking at. Range anxiety, basically, before I’d ridden a single metre of it.

That sent me further into research, and that’s where I found Johnny Nerdout’s channel on YouTube, and specifically a video of his making the case for mid-drive motors over hub motors. Two things from it stuck: reliability, and the fact that a hub motor has no idea your bike has gears.

A hub motor drives the wheel directly. Whatever gear you’re in doesn’t matter to it, it always turns at the same ratio. Shift down for a climb and your legs get an easier time, but the motor doesn’t, it’s still working exactly as hard, just more slowly, which is worse for it rather than better. A mid-drive motor turns the chainring instead, so it gets the same benefit from your gears that your legs do. Change down and the motor changes down too.

Once that clicked, something I’d half-noticed while shopping made sense retroactively: cheap pre-built eBikes are almost always hub motors, and the properly built, more expensive ones are almost always mid-drive. I’d assumed that was just a price thing, more money buying a better version of the same idea. It isn’t. It’s a design decision about whether the motor is allowed to use the bike’s gears at all.

Worth being clear that this is where the penny-pinching stopped. Once I was comparing the Swytch against a mid-drive kit, there wasn’t much in it on price, so cost simply wasn’t the deciding factor any more. What I wanted was the more performant option, the more reliable one, and frankly the smarter piece of engineering. All the thrift in this project went into the decision to convert a bike at all. Having got there, I wasn’t going to pick the worse motor to save a little.

Which also, belatedly, justified a judgement I’d made on no evidence whatsoever. I’d dismissed most of the hub kits early on simply because they looked messy and cheap to me. That was pure aesthetics, and it turned out to be tracking something real. A mid-drive genuinely does look cleaner, tucked down at the bottom bracket instead of bulging out of a wheel, and it also happens to be the better engineering. Not a reliable way to make decisions, but I’ll take it.

It also made the front-versus-rear question I’d been idly turning over completely irrelevant. That had felt like the decision, and it wasn’t one. Both put the motor in a wheel, and a motor in a wheel can’t use your gears no matter which end of the bike it’s at. The actual line runs somewhere else entirely.

I landed on the Bafang BBS02, a mid-drive kit that bolts into the bottom bracket shell. It’s also sold in a more powerful 48V 750W version, and that’s where a Swiss rule mattered: a converted bike here only counts as an ordinary bicycle, no registration and no plate, if the motor is 500W or under and assistance cuts out at 25 km/h. Go over either line and you need a cantonal inspection, or you’re restricted to private land. I went with the 36V 500W version and stayed under that line on purpose, because I wanted a bike I could just ride, not something I had to think about administratively every time I left the house.

The tools you don’t know you need

The kit arrived in early June 2025, and I had the motor turning within a couple of days. That sounds fast, and in a sense it was, but “running” and “finished” turned out to be very different states, and I’d be working on the gap between them for a long time afterwards.

Part of the delay was tools. I own tools, plenty of ordinary ones, but bike work turns out to want oddly specific kit that has exactly one purpose, and I had none of that. Before I could really get started I had to order a crank puller, a bottom bracket tool, and the particular lockring tool the Bafang needs.

Worth mentioning, because it’s the part that gets left out of the cheap-DIY story: you don’t only buy the kit, you buy the tools to fit it as well. In this case it barely registered. The tools were a small fraction of what I’d saved by not buying a pre-built eBike, and unlike the kit they don’t get used up. They’re still in the drawer.

The hardest parts, once everything arrived, were getting proper torque on that lockring and the cabling. The lockring needed a lot more force than I expected. The cabling is the longer-running story: I hate messy cables, and keeping them tidy on this bike has genuinely been an ongoing pursuit ever since, not a one-time job I finished.

Getting it working was quick. Getting it right took a lot longer than I expected. But I never really doubted I’d get there, mostly because of how much research I’d already put in by that point.

The first ride, kept deliberately cautious

The first time the motor actually worked, I just rode around the block. The bike was barely finished: no lights, no fenders, about as minimal as it could be and still function. It was fun, and I knew immediately it had been worth doing, even though there was still a lot of work and a lot to learn ahead of me.

I don’t remember much about the first real commute, which I take as a good sign. What I do remember clearly is how conservative I was: assist only on the climb, throttle never touched at all, and still nervous about range, even though I’d already solved that exact worry by choosing a mid-drive with a bigger battery in the first place. Old habits.

Learning what it actually needed

Everything since that first ride has been a response to something specific going wrong or feeling off, rather than a plan I had from the start.

I rewired the front and rear lights, which had run off the front dynamo hub, to run off the Bafang’s battery instead. One less thing to think about.

I put the pannier rack back on, which sounds minor but changed more than I expected. Without a bag on my back I can actually turn and check over my shoulder properly, which feels like a real safety improvement, on top of being far less sweaty and lowering the bike’s centre of gravity a bit.

I added a small mirror to the left end of the handlebar, which has quietly become one of my favourite parts of the whole build. Most eBike mirrors are big and ugly. Mine is small enough to fold flat out of the way when I don’t want it there.

About a month in, I swapped the stock Bafang display for an EggRider V2. Partly because I hated the stock one, but mostly for a reason I hadn’t appreciated when I bought the kit: the BBS02 is far more configurable than it first appears. Current limits, assist levels, how much the motor actually gives you at each one, all of it can be changed. The catch is that the standard way to change any of it involves plugging a proprietary programming cable into the motor and sitting there with a laptop.

The EggRider does the same job over Bluetooth from a phone app. That turns tuning from a thing you do once, at a desk, into something you can change whenever you want to. Given how much of this project has been finding out what I actually wanted only after riding it, that mattered more to me than the display itself did.

That, along with the mirror and the tidy cabling, is really one decision wearing different clothes: I don’t want the bike to announce itself as an eBike. You can still see the motor and the battery, obviously, but I’d rather it read as a normal bike that happens to have a bit of help, not something trying to look electric. A bit of a sleeper eBike, if that’s a real category.

The bigger changes came out of a single ride, a couple of months after getting the Bafang running. I was descending a steep gravel hill nearby and the rear of the bike started stepping out on me under braking. Unsettling enough on its own, but it also made me notice how upright my riding position was, which wasn’t helping. That one ride is what put two changes on the list, though I got to them in my own time rather than straight away: I swapped the stock Marathon Plus tyres for Schwalbe G-One Allrounds, and I lowered the bars so I’m leaning forward now instead of sitting bolt upright. Both were about wanting the bike to feel more like a mountain bike, or more specifically, more like a gravel bike.

The expected trade with a grippier tyre is that it wears faster than a Marathon Plus and picks up more punctures. Half a year in, I haven’t really noticed either. I assume the wear will catch up with me eventually, but the tyres are still absolutely fine, and the grip has already paid for itself several times over.

The most recent fix is a bit embarrassing. When I first set the bike up I chose a 48-tooth chainring for the Bafang, on the reasoning that bigger would be better for climbing. That’s backwards, a bigger chainring is a harder gear, not an easier one. I only worked this out very recently, on that same gravel hill, riding with the assist off for the exercise. I was in my lowest gear, working hard, and noticed the mountain bikers around me were spinning a noticeably higher cadence at the same speed. I looked at their chainrings, saw how much smaller they were than mine, and it finally clicked. I’ve since swapped to a 42-tooth ring.

Where it’s at now

These days I commute on it once or twice a week and take it out just for fun about once a week on top of that, along with errands around town.

I’ve settled into using the throttle for exactly one thing: getting the bike moving from a dead stop, mostly at traffic lights. It’s less about laziness and more about balance and safety, getting myself moving and settled on the bike, and being able to get a bit ahead of the cars beside me the moment the light turns green.

There’s more planned. I want drop handlebars at some point, and slightly wider tyres, to push it further toward feeling like a proper gravel bike. It has genuinely become a tinkering hobby along the way, one I hope is mostly behind me now, though I’ve said that before.

Looking back at the whole thing, there’s a pattern I didn’t notice while I was in it. At every single point where there was an obvious thing to buy, I ended up adapting something I already had instead. No room for a rowing machine, so use the bike that was gathering dust. Pre-built eBike too expensive to gamble on, hiring one worse value still, so convert the bike. The dynamo lights that were already on the bike, rewired to run off the Bafang. A rack I already owned, put back on. None of that was a principle I set out with. It’s just what kept turning out to be the better answer.

So I still occasionally wonder whether I should replace the frame outright and start again with something actually built for how I use the bike now. Maybe I will, someday. But so far, every time I’ve hit something the bike wasn’t quite right for, I’ve found a way to wrangle it into doing what I needed instead. That’s held up for a year of riding it now, and it’s hard to argue with.