Introductions
The first serial hybrid bicycle was a success. I can lift it even with two oversized lead acid batteries attached. It takes up half the space in the garage as the long wheel base bike, and is more maneuverable. It’s more fun.
Due to the demise of the first bike, I decided to ease the weight requirements I originally had for this project and work on those in bike #3 or 4. I’ve left this page otherwise intact, however. After some disappointing performance and bad welding, I rebuilt this bike (see the bottom of the page).
Goals
The goals of this build are to improve on the first serial hybrid bicycle. The goals of that bike were simplicity and compact size.
Requirements
- Reduce the weight of the bike
- The first serial hybrid bike probably weighs around 80lbs, my new goal is 45lbs
- Improve the ride of the bike
- Refine the drive system
- Reduce the time to machine and mount to 1.5 hours
- Improve the generator system
- Refine the aesthetics of the bike
Initial Thoughts
Roughly 40lbs of weight is due to the dual 38ah lead acid batteries. Lithium ion batteries could be used instead, or the capacity of the system reduced. The boom for the pedals is made of thick square tubing. Round thin wall tubing would be superior. The weight of the motors cannot be reduced, but the mounting plates could be optimized. Aluminum rims will reduce weight. A standard bike frame might be unnecessary, since the riding position does not match the standard one.
The bike needs a suspension, front and back. A full suspension children’s mountain bike might make a good donor bike.
The drive system on the first serial hybrid still needs quantitative analysis. However, there are already some obvious improvements possible. The motor sprocket is mounted at the end of the motor shaft, which puts stress on the bearings in the motor. This was a compromise for easier mounting. The wheel sprocket needed CNC work to attach to the coaster brake sprocket. Using a standard wheel with a gear set would allow use of standard mounting hardware.
The generator system needs higher gearing that standard bike sprockets can’t provide. No. 25 or no. 35 chain should be used.
A custom frame might reduce weight and could improve aesthetics. Even without that, some sand paper and spray paint would make the bike look much better.
Investigation
I went by Zoran Locki’s Woodside Bike Shop on Woodside Road in Redwood City on Feb 29th, 2008. Zoran’s shop is geared toward BMX bikes, so I figured they would stock small wheels and so on. Zoran recommended sticking with 20″ wheels as the parts are common and cheap. He recommended aluminum wheels, the cheapest aluminum mountain bike crank set with screw in bearings (these don’t require the cylinder holding the cranks to have races welded in place, so I can just use pipe from a metal supplier), and one set of v brakes or, budget allowing, one disc brake. He also said he has light banana seats. For the frame, cro-mo was his choice, which was also my choice several years earlier on the first recumbent bike. He had one frame in particular, but it had no suspension. I’ll have to see how that pans out. I went by Alan Steel, and they only had 1 inch and 1.75″ OD 1/8 and 1/16″ cro-mo for roughly $6/foot. I’ll have to find another source probably.
Necessity
After the first serial hybrid bike was damaged, I decided to ease the weight requirements and get a bike with a full suspension up and running.
The donor bike, a 26″ full suspension cheapo Mongoose
I had a cheapo Mongoose with a broken derailleur sitting around so I decided to work on this bike. I welded a boom to the head tube and removed the cranks and rear brakes. I replaced the seat with the one from the first serial hybrid bicycle. This proved problematic as the geometry of the Mongoose meant there has to be a large gap between the seat and the backrest. I also cut down the steering tube, ground down the sharp bits at the top, and welded it to the Mongoose steering stem. This took roughly 3 hours.
Pushing ahead, I got a one way ride to Ben’s shop with the old bike and the new bike. The lack of space in the rear triangle of the Mongoose was another problem. The Kollmorgen does not fit in between the two tubes that converge at the back wheel, meaning the motor mounting on the first bike couldn’t be copied. Instead, the same motor mounting plate was used and two U channel pieces of steel were attached to the triangle with bolts and then to the motor mounting plate. The motor now sits above and slightly forward of the hub of the rear wheel.
I bought wire mesh battery boxes at Allen Steel, but we ran out of bolts to mount them. Additionally, I forgot to bring the brake cables.
The crank set gearing needs to be increased, so I decided to switch to a no .35 sprocket on the crank and the original 10 tooth sprocket on the generator motor, but there wasn’t time to attach this. In all, this second day took 4 hours of time.
I had to get back to San Mateo, with no battery boxes, no brakes, and no generator. We taped the untested Lithium Ion pack from GreenEMotor to the bike and off I went to the train station. I kept the speed low and made it home. The bike received a good deal of interest on the train.
Desperate times, desperate measures… tape to the rescue
The train conductor asked me if I was bringing a battering ram on the train
Full Suspension
The full suspension totally rocks. After riding the first serial hybrid, I cringe whenever zooming into the driveway. With the full suspension, there is no need.
Is a full suspension necessary? Would a front suspension or rear suffice?
I can take a guess. I’m not sure about the necessity of the rear (a seat post with springs might be enough), but the front fork suspension definitely helps when hitting bumps. Jarring bumps turn into dull thuds.
Ben observed me riding over bumps to see how much the suspension flexes, and the rear does contribute. More evidence: I purchased an Eton Eco electric bike with only a front suspension, and riding over the same bumps in the street is jarring. With the front suspension there is perhaps a roughly 25% better ride than no suspension.
I have a springy seat I will have to mount to the Eton bike.
Battery Testing
I decided to measure the battery performance. After the trip home, I measured each cell of the Lithium pack. All cells were 3.2v except the last, which was 0v. The next day I swapped this cell with a spare. I attached an ammeter to the battery and added brakes to the bike. I also attached the wire battery boxes and did a comparison.
| Pack | V0 | Vf | Imax | Iavg | vmax | vavg | d | Weight |
| LiFePO4 12Ah | 26.85V | 22.90V | 17A | 12.5A | 15.6mph | 11mph | 6 blocks | ~7lbs |
| Gel Cell Pb 38Ah #1 | ? | 24.95V | >30A | 20A | 14.3mph | 13mph | 6 blocks | ~60lbs |
| *Gel Cell Pb 38Ah #2 | 25.80V | 25.25V | >30A | ~17A | 20.4mph | 16mph | 8 blocks | … |
| *Gel Cell Pb 38Ah #1 | ? | 24.50V | 20A | 15A | 16.8mph | 15mph | 6 blocks | … |
* measured on April 10th, with less wind
The lithium pack was totally worn down after this quick test. On the previous ride it took me to the train station and back. But I took it easy on the throttle that time. I guess that shows the value of current limiting circuitry…
I forgot to measure the initial voltage of the lead acid pack. A few things are clear however. The weight of the lead acid batteries is simply staggering in relation to the lithium cells. However, they gave much better acceleration (and for 1/3rd the cost). The maximum discharge rate for the LiFePO4 cells was about 1.5C. The lead acid batteries went off the scale (1C or greater), but somewhere between 1C and 2C seems reasonable. The LiFePO4 cells dropped 4v over the run. Inspection of the pack revealed that two cells (the 5th and the last) had dropped in voltage. I need fresh batteries. Brian Howell, a local electric bike builder, informs me that his LiFePO4 cells can discharge at 3-4C, so my set appear quite past their prime (I returned these to Allen at GreenEMotor at the end of March, thanks Allen!).
Note the difference in top speed from the converted children’s bike. I opened up the throttle for a couple of blocks to see how fast I could go. Interestingly, the speeds were lower than with the previous bike. The frame of the Mongoose is heavier and the bike is larger. Maybe the suspension has some effect as well? The bike is significantly slower (20-25%) than the previous one [now I think this was due to low charge state and the wind].
If portability is not needed, I would definitely opt for slightly smaller capacity gel cells. They are easier to deal with and tolerant to abuse. 25Ah would probably be sufficient at 24v. 12v, 25Ah AGM batteries can be purchased for roughly $50 each. However, the weight of the battery pack would be almost 40lbs.
The LiFePO4 cells are the hands down winner by weight. Assuming I had a pack that was in a better state, the acceleration wouldn’t be as good as the gel cells, but the weight savings make the bike much more manageable, and the performance otherwise is still acceptable. Since I’m aiming to make a lightweight bike, I must go with the lithium ion from now on (but I’d go with the gel cells for the tandem trike I’d like to build) or small SLAs.
With better batteries and more time, I could take some measurements of cycle life.
Batteries and the Serial Hybrid
The gel cells are very tolerant of the 40V or so the generator applies to them (so far). How well do the LiFePO4 cells hold up? I’ll report back once I have the new 60 tooth crank hooked up [actually now its an 80 tooth].
Shortcomings
The motor mounting is still the biggest issue. The lack of 8mm no 35 sprockets means the adapter collar had be to CNC’ed. The freewheel adapters available don’t fit the McMaster sprockets or any others I can find. I will probably return to no.25 chain for the next bike. While the components are more expensive, time and custom machining are far more expensive. That still leaves motor mounting plate machining. It may be sensible to switch to a hub motor. After investigation, the best I could find was the $239 Roadrunner (a rebadged Crystalyte 408, IIRC). This is still a bit heavy and expensive compared to the Kollmorgen.
I want a no welding solution for the steering. I need a steering tube that is far longer than normal. I realized the other night that folding bikes have these, so perhaps I can find a source for those.
Additionally, creating a braze on or bolt on solution for the front boom would also be helpful.
Full suspension parts aren’t cheap– usually. I did find 20″ front forks for $40-$100, but did not find rear shocks aftermarket for reasonable prices, so I will have to investigate how well spring seats work.
Second Incarnation of this Bike
I tried a Tidal Force 750 bike with a 1hp hub motor and NiMH batteries; it had a springy seat and a front fork, and it totally rocked. So I think the spring seat might be good enough.
The 26″ wheel on the front made steering… strange. Against better judgement, I took an old battered fork and cut the last 3 inches off with the drop outs. I then welded these onto the 26″ suspension fork. Now, this is a weld you don’t want to get wrong. A failure at high speed would not be pleasant. This is why I welded 3″ of steel. I also cleaned the parts and did the best welding I could muster. I tested it afterwards by beating on the drop outs with a hammer, but the weld held without any visible cracking or chipping.
The suspension has rubber bits inside, so I kept pouring water on the fork while welding, but it wasn’t enough and I’m pretty sure I harmed it. A better solution would have been to submerge the fork in water. I used a no.2 tip and turned the torch way up so I could weld as quickly as possible, thus limiting the heat affected zone.
Once that was done, I had trouble lining up the generator and crank set on the front again. One the previous bike, this was fixed by many washers. Additionally, the voltage was not high enough. To raise the generated voltage, I had to increase the gearing, but this wasn’t possible with bicycle chain. TNCScooters sold me the generator motor with a “no.35″ sprocket on it, so I decided to go to no.35 chain for the generator too. So I purchased a 60 tooth sprocket (no.35) from McMaster…
I pulled this nice aluminum crank set off a parts bike. The sprockets are held on by bolts and nuts (allen key head)
I drilled 3 holes and then dremeled them out since they were a little off…
While I was at it, I decided that the stickers were ugly, so I used a wire wheel to remove them
To use the new crank set, however, I couldn’t use the old bottom bracket, because the new one is 3 piece. Since the welds on the boom weren’t very good I knocked it off with a hammer. I cut the bottom bracket and a couple of feet of tubing off the donor frame. I used my reciprocating saw to cut a triangular mounting plate for the generator, then cleaned it up with the dremel.
The generator motor mounting. Three bolts to hold it to the boom, 3 bolts holding the motor down. Any less, and parts will flex.
I cleaned the ends and welded it on. Wait for the metal on both pieces to turn liquid, then slowly insert the welding rod, and work a continuous bead…
Time to paint! Resist the urge to throw on a thick coat the first time; it will just run and take more coats to hide the imperfections. I did about nine coats of Krylon yellow and black.
Painting the bike
At this point I attempted to hook up the generator and… the “no.35″ sprocket is actually for 1/2″x1/8″ bicycle chain. Oh great. I measured the shaft and it was either 7/16″ or 11mm… McMaster stocks 3/8″ and 1/2″ sprockets. The 1/2″ was loose, and 3/8″ was definitely too small. Grr. Is this a common size in China? I tried putting the 1/2″ sprocket on anyway, but I couldn’t get it to stay on straight. I searched through my box of parts and pulled out a 10mm D bore flat no.25 sprocket. I used a very worn down dremel cut off wheel to remove the flat from the D and the sprocket fit over the threaded portion of the generator motor shaft. I tightened down the nut and it seemed to stay. I placed some no.25 chain on it and it appears to fit fine. I didn’t have any no.25 sprockets for the cranks, so I went back to McMaster and ordered a 72 tooth sprocket.
I decided I would run the bike without the generator just to see how well it works. I spent half a day creating a nifty battery box to hang between the front wheel and the frame that held the batteries in line to reduce frontal area and drag, but I realized after finishing that it didn’t fit when the suspension was fully compressed. Oops. So I took my wire mesh boxes and bolted them to the diagonal part of the frame with 3″ long 1/4″-20 grade 5 zinc plated bolts and some pieces of scrap metal that I used as washers.
The bike reassembled
Not happy with the positioning of the seat but I didn’t have any suitable metal to make another mounting.
Test ride was positive. The bike reached speeds reminiscent of its predecessor. I went on full throttle for a few blocks but could not reach 21mph like the previous bike had. But 20.6 was pretty close, and the ride was far superior.
I held the throttle still at around a 17amp current draw. The bike accelerated from 12mph to 16mph, and then the acceleration slowed dramatically The last mph went by pretty slowly. If the voltage was 24v, then it takes this bike roughly 400 watts of energy to go 16mph. This is much higher than my calculations for the long wheel base recumbent. I suspect my ammeter is innacurate and the voltage may have sagged below 24v.
Hopefully the new sprocket will arrive soon.