Reflections on the potential of human power for transportation

Friday, November 23, 2012

The Return of the Recumbent Bicycle

Last September, after allowing it to gather dust in the garage for the last six years, I started riding my Avatar 2000 recumbent on the road again. I had forgotten how much fun it was to cruise down the road sitting with a car-like posture and taking in the scenery instead of being hunched over the handlebars with only a narrow view of the road ahead. I also had forgotten how awkward hill climbing was; dropping into the granny ring and spinning as if my life depended in it. Yes, my relationship with my recumbents was a love and hate thing.
My relationship with recumbent bicycles began three years after I purchased my first derailleur bicycle, a gas-pipe-framed Schwinn 10 speed. I bought the April 1969 issue of Popular Mechanics because it had an article by the do-it-yourselfer extraordinaire, Robert Q. Riley. The article was about the construction of a low-slung bicycle he called the “Ground Hugger”. I had never seen anything like it and started thinking about how I could build one.

As the article said “You’re cradled in a bucket seat and as you lean into a long-banking turn, you have the exhilarating sensation of being on a toboggan with wheels”. To a cyclist, this was an exciting prospect.

Many in the non-cycling community believe that Riley invented the recumbent bicycle. Instead Riley had copied the essential elements of a bicycle designed and built by the ex-airline pilot and bicycle innovator Captain Dan Henry.
Henry, famous for the Dan Henry markers used to guide cyclists on organized rides, wrote an article for the May 1968 issue of Bicycle Magazine describing his recumbent. He designed it based on what he felt were the best characteristics of pre-WW2 recumbent bicycles.

While Riley never admitted copying Henry’s recumbent, the similarities between the designs are numerous, especially the long wheelbase and remote steering. While Henry used a chain and sprocket for the fork-steering connection, Riley used a Cardan universal joint from a socket set. Why neither of them used the much simpler connecting rod between an offset pivot on the handlebars and one on the fork, I do not know. One interesting consequence of the use of a u-joint is the coupling ratio varies with the position of the joint. This could be used to improve steering control by having the steering be least sensitive when the steering is straight and becoming more sensitive with increasing steering lock. (Riley may have seen that the preWW2 Velocar used a Cardan joint to connect the handlebars to the fork.)
(A note here that chopper-style kid’s bicycles like the Schwinn Stingray are technically semi-recumbent bicycles and, for that matter, the children’s Big Wheel tricycle and its variants are recumbents. Here I am only interested in the reemergence of the adult recumbent bicycle.)
Henry was not the first to experiment with recumbent bicycles since WW2. Gunnar Fehlau, in his book “The Recumbent Bicycle”, describes the obscure work of the engineer Paul Rinkowski on short-wheelbase recumbents starting in the late 1940’s and continuing on for four decades.
And then there is this fascinating picture of two riders on very-low Grubb-style recumbents in Popular Mechanics from March of 1952. Notice, in this case, that the remote steering uses cables to connect the under-seat handlebars to the forks.

In addition, Alex Moulton of small-wheeled-bicycle fame experimented with a Grubb recumbent prior to settling on an upright posture for his improved-bicycle design in 1962. Moulton found that recumbent pedaling produced thigh fatigue when pedaled for extended periods and rejected the recumbent approach. It was a lost opportunity for recumbent evolution, given Moulton’s innovative product improvement abilities.   
Henry’s recumbent bicycle may have influenced the east-coast arm of the subsequent human-powered vehicle movement, Prof. David Gordon Wilson of MIT. He must have been aware of Henry’s article, because he wrote an article, “Where Are We Going in Bicycle Design?” published in Bicycling the previous month. He also included a photo of Henry’s recumbent in Bicycling Science, the book he coauthored with Frank Roland Whitt in 1974.
From 1967-1968 Wilson sponsored a design competition for man-powered land transportation. The winner, W.D. Lydiard, designed and actually built an enclosed mid-wheelbase recumbent bicycle. Traditionally recumbent bicycles either had the cranks behind (long wheelbase) or ahead of the front wheel (short wheelbase). Those with a long wheelbase were inconvenient to transport and had a lightly loaded front wheel that could wash out on slippery surfaces. The short wheelbase designs had an overloaded front wheel that consequently resulted limited-front-tire life and in skittish handling. If the cranks are located in the ideal location, above the front wheel, the bottom-bracket height was usually so high that the rider is placed in an uncomfortable posture. I refer to the issues with these approaches as “the recumbent packaging problem”. Refer to “Recumbents and Convergent Evolution”, below.

To reduce the pedal height, yet maintain their location over the front wheel, Lydiard used a squashed pedal path generated by a crank-slider-type mechanism. Lydiard felt that this approach could be refined to eliminate the problem of interference between the feet and pedals when putting the feet on the ground.

Wilson was apparently quite captivated by Lydiard’s mid-wheelbase, squashed-pedal-path approach. In a private communication, Wilson shared 17 permutations on this design approach, the latest being dated 1977. To my knowledge, none were actually constructed. Based on one of Wilson’s drawings, I built up a crank-rocker or treadle mechanism for my EcoVia 2.2, but power-production limitations caused me to abandon the design. See “Transcending the Pedicar, Part 2”, below.
In 1972, while Wilson was designing linear-drive recumbents, H. Fredrick Willkie, being inspired by Wilson’s design completion, contacted him for a sketch for the design of an advanced bicycle. Interestingly, the recumbent that resulted from that sketch didn’t utilize a linear drive. Ultimately there were five iterations in this design exercise, culminating in the Avatar 2000. Willkie did two designs which he christened “Green Planet Specials”. The first version had handlebars in front of the rider’s chest connected directly to the fork and a high bottom bracket. The design was very similar to the preWW2 Rivat recumbent. Willkie found the compressed body posture uncomfortable and at Wilson’s suggestion, the second version had a lowered bottom bracket and a more leaned-back seat angle. In addition it had direct steering with the handlebars mounted beneath the seat. Wilson bought the GPSII from Willkie and continued to modify the design as the Wilson-Willkie. The seat back was made more vertical and the weight on the front wheel was reduced from about 70% to about 65%. The Avatar 1000 followed and the front-wheel loading was reduced to about 62%. Finally, with the Avatar 2000, the radical step was taken to move the front tire ahead of the bottom bracket, and the front wheel loading dropped to about 31%. Under Wilson’s guidance, two Boston-area bicycle builders, Richard Forrestal and Harald Maciejewski began manufacturing the Avatar 2000 in 1979, making it the first production recumbent since WW2. (A more comprehensive description of the evolution of the Avatar recumbent, including photos of the five vehicles above, can be found in Wilson’s article, “Evolution of Recumbent Bicycles and the Design of the Avatar Bluebell” in the proceedings of the Second International Human Powered Vehicle Scientific Symposium.    

The Avatar 2000 recumbent was very similar to the preWW2 British Grubb recumbent. Both use indirect steering with the handlebars coupled to the fork by a connecting rod. Differences were the Avatar’s seat was higher and more upright and the Avatar had a shorter wheelbase due to the use of a 16” front wheel.  The Avatar’s wheelbase is 63”.
So there is a circuitous linkage between the Ground Hugger and the Avatar 2000.
In 1984, after becoming somewhat bored with upright bicycles, I purchased an Avatar 2000 from Angle Lake Cycle in Seattle. It was serial number 085 and it had been sitting in the store window for several years.
I was disappointed with both the on-the-level speed and, even more so, the hill-climbing speed of the Avatar. On the positive side, the extreme comfort riding the Avatar on the level terrain somewhat compensated for the reduced speed by eliminating fatigue from secondary effects like a sore seat, sore back and numb hands.
I did make one component change that significantly improved its hill-climbing ability. I replaced the conventional cranks with a Power-Cam crankset which I had purchased several years before.

The Power Cam was invented by Dr. Lawrence Brown of IPD (International Patent Development). The design involved chainrings that could float relative to the cranks. A cam follower attached to the cranks rode on a cam attached to the bottom bracket and drove the chainrings through a gear sector. The inertia of the bicycle kept the chainrings rotating at a near constant speed and the cam mechanism caused the cranks to speed up and slow down relative to the chainrings as a function of pedal position. The net result was the mechanical power pulses during pedaling became shorter and larger than with a conventional crankset using round chainrings. As a consequence, the rest periods between the pulses (two per cycle) became longer. The longer rest periods are physiologically more efficient and, for a given oxygen consumption, the aerobic power output was increased. (More on factors influencing power production in a future post.)
Using a Power Cam on a recumbent was suggested by Edward P. Stevenson in his book “The High-tech Bicycle”. The Power Cam did not work well when standing up on the pedals and forced you to climb seated. On a recumbent you couldn’t stand on the pedals so a Power Cam on a recumbent made a kind of sense. Stevenson was correct and it improved hill climbing. The Power Cam had only two chainrings, however, so an adaptor plate was machined to allow the mounting of a third granny chainring.
The increased comfort of the recumbent posture allowed me to take longer rides than on my upright and I logged a lot of miles over the 22 year period between 1984 and 2006. I eventually added an arm-power attachment to the Avatar to scavenge some of the power I was loosing through the inefficiencies of my leg pedaling. See “Arm Power and the Avatar Recumbent” below.
In 2006 I started mountain biking regularly and put the Avatar in moth balls. I began with a hardtail and later purchased a full suspension bike. My hardtail acquired road tires and became my road bike. I didn’t ride on the road much, however, because of numb hands and tired back from sitting in one position for extended periods. I didn’t have this problem when riding on dirt because of a continuingly changing body position and the reduction of road vibration due to the full suspension. The level of comfort riding an upright bicycle off-road was acceptable.
It was on a car trip up to Snoqualmie Fall for breakfast on Labor Day that I was reminded of the number of time I had ridden the scenic hill climb up to the falls on the Avatar. I have never done that ride on my converted mountain bike because of the associated lack of comfort. Realizing what I had been missing the last six years, I decided I would ride the recumbent on the road and the upright on dirt, thus having the best of both worlds.
The first few miles back on the Avatar after the six year hiatus were a bit shaky and the long hill climb back home was a painful grind. But when riding on the flats what I was left with was the feeling of being in a touring car where I could comfortably view the scenery while exercising at the same time. For me, the recumbent bicycle has returned.

Saturday, September 29, 2012

Transcending the Pedicar: The EcoVia, Part 2

Human Powered Commuter Vehicle Criteria

1.       Weather Protection
2.       Statically Stable
3.       Reasonable Cruise Speed
4.       Cargo Carrying Capacity
5.       No Wider than a Bicycle
6.       Same Height as an Auto
7.       Comfortable posture and ease of entry
8.       Two-wheel drive
9.       Car-type Wheels
10.   Electric Assist for Hills

Part 1 of the EcoVia design post can be found below.

The EcoVia 2.1 was well on the way to becoming a vehicle that satisfied the HPCV criteria. Unfortunately I did not feel it satisfied #7. I did not consider the riding posture as comfortable as it needed to be. The bottom bracket was higher than the seat and it needed to be below the seat, at least four inches below if possible.
EcoVia 2.2
I had fallen victim to “the recumbent design problem” where the pedals wanted to be where the front wheel was. (Refer to “Recumbents and Convergent Evolution” below.)

For reasons of leaning functionality the wheel layout was fixed at one front and two rear wheels, with the front wheel steering. I wanted a compact package, so having the cranks behind the front wheel was rejected. So too was putting the cranks before the front wheel because of the skittish handling that came along with that location.
My first thought was to squash the pedal path, thus lowering the upper position of the pedal without causing interference with the wheel in the lower position. In his article “The Development of Modern Recumbent Bicycles”, Dave Wilson presents three sketches of mechanisms that produce elongated pedal paths.
 The simplest of these is known as a crank-rocker (actually crank, connecting rod and rocker) mechanism in kinematics and historically as a treadle mechanism. It can be made very rigid, since there is only one link between the pedals and the frame. It also has a fixed range of motion with built in decelerations and accelerations at the ends of the stroke. The literature seems to suggest that this type of reciprocating motion produces more power than levers pulling chains over one-way clutches. Recall the latter approach was used in the Pedicar. (Refer to “Pedicar Technology”, below.)

After rotary cranks, the treadle drive may be the most often used method of converting leg motion into rotary motion.  It predates cranks being used to propel velocipedes. It was used in the first attempts to build a safety bicycle and it is used in most children’s pedal cars. My favorite period picture of a treadle driven bicycle is Oscar Egg’s recumbent form the 1930’s. Egg had set the hour record several times on a conventional bicycle and he competed with Francois Faure in recumbent record attempts.

My implantation of the treadle mechanism for the EV2.2 was a bit different from previous embodiments. I fabricated a pedal that would allow connections to both sides of the pedal’s axle. The second connection was formed by brazing a bolt to the pedal’s steel dustcap. The pedal mounting was reversed with the normal thread being mounted outboard of the pedal. The rocker link was connected to this thread. The other end of the rocker was supported by a pair of Igus bushings, which, in turn, were held by a tube that was brazed to a rectangular frame that surrounded the front wheel. This frame did dual duty, since it also contained the pivots that would hold the front tilting faring. As result of the outboard location of the rocker links, they did not interfere with the front-wheel's ability to turn. The other, inboard end of the pedal was attached to a bent connecting rod whose other end was attached to a conventional crankset with the arms shortened. The connecting rod was bent to clear the top of the tire. The entire mechanism was very compact and it positioned the rider’s feet in a low and very comfortable position.
As successful as the packaging turned out to be, the performance was anything but. You see, I had overlooked one critical fact. Almost all the historical uses of treadle drives were on bicycles with fixed gears. The motion of the vehicle carried the pedals over the dead spots at the ends of travel. Problems with the dead spots became painfully obvious when attempting to climb even the most gradual of hills. As a bicycle moves progressively slower when the rider climbs increasing grades, the rider must apply the propulsive torque over a greater portion of the pedal stroke. (An explanation of this will come in a future post on why bicycles climb steep hills so poorly.) With the treadle mechanism, the rider can only apply a torque through the middle of the stroke. Pedal forces required to produce torque near the end of stroke become enormous and forward motion is lost.
I had an old Bullseye elliptical sprocket with a 1.56:1 ratio. I thought I could use it to modify the sinusoidal pedal velocity to improve hill climbing. Mounted in one orientation it would slow the pedals in the middle of the stroke and speed them up near the end.  This is the orientation I expected to improve performance but to no avail. Rotating the elliptical sprocket 90deg also showed no improvement, and I concluded that the round sprocket gave the best performance, which was still woefully inadequate. Oddly enough, however, the cable-clutch system used on the Pedicar, while lacking in power output due to limited pedaling cadence, did work well on steep hills because the output torque was constant except at the extreme ends of pedal travel.  While this approach would have solved the hill climbing problem, its limited power output would have restricted the speed of the vehicle. So much for unconventional drives…

EcoVia 2.3
I realized that if the bottom bracket was almost touching the tire, I could drop the bottom bracket height from 26” to 20”. If I kept the bottom bracket close to the steering axis and spaced the crank arm as far apart as I could, I might have enough front wheel lock to have a viable solution. I bought the widest triple-crank bottom-bracket axle I could find. I coupled this crank to an intermediate crank (left over from the treadle drive) on the left side and tensioned that chain with a spring-loaded chain tensioner. The right side of the intermediate-crank carried triple chainrings.
The acid test was riding the two miles to the library. The overlapped-crank drive passed with flying colors, the only crank/tire interference occurring during very tight, slow turns.
 In addition to lowering the bottom bracket, I changed the lower mounting position of the shock to raise the seat and I moved the seat back forward by adding an extra bend near the top. The result, while not being as ideal as the Avatar 2000, was a great improvement, enough to make the EV2.3 comfortable to ride.
All of the EcoVias used caliper brakes on all three wheels. Per tradition, the left lever actuates the front brake. The right lever actuates both rear brakes through a linkage that balances the forces applied to each rear brake. The travel of the right lever was increased to produce the greater cable motion required to actuate two brakes.
The middle cable from the brake lever is attached to a toggle linkage. The ends of the toggle linkage are attached to brake links that pull cables attached to the individual brakes. As the toggle flattens out, the forces on the brake links increase. If one caliper clamps the rim first, its brake link stops moving and the toggle motion continues to move the other brake link until its caliper clamps the rim. This balancing mechanism would be unnecessary if hydraulic disk brakes were used. Unfortunately there is no room for the disk rotor on the cantilevered hubs.
Now, actuation of the rear brakes had an unexpected consequence.  During leaning, each wheel rotates slightly with respect to its wheel beam. Since braking prevents this rotation, the act of braking the rear wheels inhibits leaning.  So during an emergency braking maneuver, when leaning is undesirable, it is automatically prevented. 

Below is a picture of the suspension-shock attachment locations, the wheel-beam-reverser linkage and the disk brake sector used as the lean-lock mechanism. The vehicle is in its upright position.

Below are two pictures of the vehicle leaned over.

Below is a detail of the seat. The lever on the rider’s left engages the motor. The lever on the rider’s right engages the lean lock.
Electric Hill Assist
The electric hill assist uses a geared Astroflight cobalt motor that drives one of the rear tires through a 2”dia. aluminum puck. The puck is brought into contact with the tire by pulling upon the motor engagement lever, above. The gear ratio of the motor is 2.7:1. At 24v, the motor can put out 675W at 3430rpm at the motor rotor. The results in a vehicle speed of 7.6mph for a vehicle payload of 300# climbing a 15% grade. The speed should top out about 15mph on flat terrain. The batteries and motor controller are not incorporated at this point.

The faring frame is made up of a ½”dia. tubular steel loop and ½ x1/8”aluminum strips bolted to the loop and pop riveted to each other. ¼ x 1/8 ” aluminum strips are pop riveted to the wider strips. A polycarb windscreen is pop riveted to the ½” strips and a foam nosecone made up of laminated insulating foam is bolted to the front of the loop. At this point the faring frame remains uncovered. I am soliciting suggestions for a heat-shrinkable material to cover the frame. 

 The faring tips forward to allow for the rider to get into the vehicle.

The rider sits at about the same height as the driver of a typical minivan.

When the faring frame is covered, the total weight of the EcoVia will be about 100#. 60# is for the tricycle proper, 20# is for the motor and batteries and 20# is for the faring and mounting framework.
I consider the EcoVia 2.3 a proof-of-concept vehicle. All the features that comprise the design work, but features were added piecewise instead of being integrated as in a production vehicle, or even a prototype. So, before discussing how the EcoVia transcends the Pedicar, I would like to paint a word picture of what a productized version of the EcoVia would be like.
·         11-36 tooth, 10-speed cassettes are readily available. (The EV2.3 uses a 12-32 tooth, eight speed cassette.) Attaching one of these to the central driveshaft and using a single chainring would make an intermediate bottom bracket and secondary crankset unnecessary for flat terrain and would greatly simplify the drivetrain. For hills, the motor drive would provide the extra climbing ability. (Sram now makes a 10-42t, 11 cog cassette, but at a cost of almost eight times the 11-36 model, it is not a cost-effective alternative.)  Instead of using bar-end-shift levers at the ends of the handlebars, a twist grip or a trigger shifter could be used. Elimination of the bar ends would allow the width of the faring to be decreased at least by three inches.

·         Recall that the frame is made up of two pieces. The upper frame is a seat tube which is supported by a pivot at the fork end and a shock absorber at the rear. The lower frame is a 2”dia. tube that supports the fork, pedals, driveshaft, wheel beams and banking linkage. Much of the lower frame was carried over from the rear-steering EV1. If the lower frame is redesigned, it can be made significantly lighter. The weight of the rider is supported by the seat-tube pivot in front and the shock in the rear. The seat pivot loads go almost directly into the front tire through the fork. The shock loads go almost directly to the rear wheels through the wheel-beam-reverser linkage. So the lower frame does not carry these loads. The lower frame holds the head tube stationary, reacts the chain forces between the cranks and the driveshaft and fixes the lower shock mount and the reverser –pivot with respect to the drive-shaft tube. These loads are small enough that a space frame made up of thin-wall ½”dia. tubing can be substituted for the 2”dia. tube in the current version. This should significantly reduce the weight, since three, ½”dia-.028wall steel tubes weigh .45#/ft, while one, 2”dia-.065 wall tube weighs 1.4#/ft. The seat  tube could be dipped in the middle to aid in step over when getting onto vehicle.
Refer to the sketch below.

The second major change to the lower frame is to support the wheel beams directly on the drive shaft using ball bearings instead of using an intermediate tube and bushings. The over all goal is to try and reduce the weight of the tricycle proper to about 50#.

·         Currently the motor drives only one rear wheel through a puck. To take advantage of the dual-wheel drive, the motor must be coupled directly to the driveshaft. This can be accomplished by mounting a large diameter gear on the driveshaft and driving it with a small gear on the motor. The motor can be pivoted out of the way to disengage the gears. A ratchet could be used on the motor gear so the motor could stay in continuous engagement, but then the ability to drive the vehicle in reverse using the motor would be lost.
·         Currently, the faring pivots on a frame mounted ahead of the front wheel. If the faring was to pivot at the back of the vehicle, above the storage rack, the front frame could be eliminated and the weight of the faring reduced. The goal is to reduce the faring weight to 15#. This approach also opens up the area around the front wheel to make swapping out the wheel in the event of a flat tire easy.
·         The overall weight goal for the tricycle, the faring and the motor drive is #85. Since much of the weight for the motor drive is in the batteries, a weight reduction in this feature is unlikely
·         The EcoVia would be sold as a base tricycle, with the motor drive and the faring as individual options. This would allow the cost of the base tricycle to be minimized.
So how would a production EcoVia compare with the Pedicar?
1.       The pedal drive would be more efficient.
2.       Because of the compact and streamlined faring , the EV could attain higher speeds.
3.       Because of the ability to lean into turns, the EV could corner at these higher speeds without the danger of tipping over.
4.       The EV would weigh about 2/3 of the Pedicar.
5.       The rider posture of the EV would not be as comfortable as the Pedicar, nor would the entry and exit be as easy.
Improvements in four out of five categories makes the EcoVia a clear improvement over the Pedicar.
So the concept is proven. What needs to happen now is to find an agency with the resources and interest to take the EcoVia into production. Possibly a reader of this blog?

Part 3 can be found below.

Sunday, August 12, 2012

Transending the Pedicar: the EcoVia, Part 1

For those devoted students of Bob Bundschuh's Pedicar, the title of this post may seem presumptous. After all, the Pedicar is a high-water mark in the production of a practical human-powered commuter vehicle. However,we know a lot more about human power than in 1973, when the top speed of a rider sprinting on a bicycle was about 45mph and the first edition of "Bicycling Science" was a year in the future. So the Pedicar's linear transmission was based on a lack of understanding of human-power generation and the speed potential of this type of vehicle was underestimated by about half. There is significant potential to improve a Pedicar-type vehicle's performance and still meet the design criteria below.

Human Powered Commuter Vehicle Criteria
1.       Weather Protection
2.       Statically Stable
3.       Reasonable Cruise Speed
4.       Cargo Carrying Capacity
5.       No Wider than a Bicycle
6.       Same Height as an Auto
7.       Comfortable posture and ease of entry
8.       Two-wheel drive
9.       Car-type Wheels
10.   Electric Assist for Hills

It should be no surprise that when I compiled the HPCV list above, I had quite a bit of time invested in building a proof-of-concept vehicle that I felt would satisfy these criteria. Early on, the vehicle was christened the EcoVia.
The broad approach was a banking-recumbent tricycle with two-wheel drive.
Four functions had to be distributed among the wheels. These are steering, driving, banking and braking. Each wheel had its own caliper brake, so braking did not affect the distribution of the other functions. Driving and steering the same wheels required the complexity of some type of constant-velocity coupling, so steering and driving were separated. From this decision, since two wheels were driven, the single wheel did the steering. The two driven wheels did the banking. To eliminate the interference between the cranks and the steered wheel the steered wheel was placed in the rear, resulting in a rear-steering tricycle.

EcoVia 1.4
Below are several sketches of the expected body shape for the enclosed vehicle. 

Now, rear-steering tricycles have inherent stability issues. (Refer to “Bucky and the Urbee”, below). These issues were exacerbated by the requirement that the vehicle lean into the turns.
Recall that vehicle lean in a stable turn is an inverse function of the steering angle and a direct function of the square of the speed. (Refer to “The Drymer and Varna Lean Forward”, below). Now for a HPCV, without any electronic banking control, there are three approaches to control leaning. The best and simplest approach, for tricycles that are inherently stable, is free leaning, where the vehicle balances like a bicycle and leans accordingly. Rear-steering trikes are not inherently stable so this approach was not attempted. One of the remaining approaches is to control steering and leaning separately. This can be ergonomically challenging in addition to requiring twice the typical steering hardware.  The remaining approach is to link steering and leaning to one control. This, while simpler than the former, usually results in too much lean at higher speeds and not enough lean at lower speeds. The approach I used was a combination of these methods. More specific details a bit later…
The heart of the dual-drive leaning design is the use of a dual-function central driveshaft made up of three concentric tube assemblies.
A chain from the cranks turns the drive shaft proper, which is made from a thick-walled, 1” dia. tube. The free-hub body is brazed directly to the drive shaft without a ratchet. On each outboard end of the drive shaft, a 16t freewheel is attached. The left side used a convention right-handed freewheel and the right side used a 16t Southpaw freewheel. This freewheel had a reversed ratchet and a left-handed thread to prevent it from unscrewing during driving. Second and third chains connect the freewheels to 14t fixed gears on the drive wheels. The fixed gears were glued to the hubs using red Loctite, again to prevent the cog on the right wheel from unscrewing during driving. All three wheels, and the spare, have cogs glued on them so the wheels are interchangeable. The wheels are built up from Bullseye hubs running on 12mm Allen bolts as axles and use 20” dia. rims. I am using 1.5” wide tires.
Driving the paired wheels through a ratchet for each wheel allows for differential rotation of the wheels during turning while providing a posi-traction effect which preserves traction on slippery surfaces. The Pedicar used a similar approach, and, as a result, was able to climb steep grades on snow covered roads.
The drive shaft is supported by four ball-ball bearings that are in turn supported by two segments of 1.5” dia. tubing. These segments are attached to the 2” diameter tube that makes up the spine of the frame.
Two 1.75” dia. tubes fit over the 1.5” tubes and are supported by four IGUS bushings. Each 1.75” tube has a 13” long, 1.5” dia. tube brazed to it at an angle slightly less than 90 deg. On the other end of this tube a small tube with a 12mm bore is brazed parallel to the 1.75” tube. The small tube accepts the axle for the hub and a nut secures the axle in place.
These tubular wheel beams can rotate around the drive shaft raising and lowering the wheels with respect to the frame. Having one wheel raise and the other one lower causes the vehicle to lean. When leaning, all three wheels remain parallel, and, as a result, during a balanced turn, the loads on all three wheels are radial as they are for a bicycle. A three-link reverser linkage using spherical bearings connects the wheel beams together so they move in opposition.

Below are pictures of the EcoVia 1.1. The seat has yet to be constructed and the cranks are absent.

Below is a photo of the EcoVia 1.4 before it was painted and I began road testing. The numbering indicates it was the fourth iteration of the lean-control mechanism. Comparing the pictures above with the picture below gives an indication of all the modifications that were necessary to make the vehicle rideable. I will skip many details concerning minor aspects of the vehicle's construction for reasons that will become clear later. However, I will discuss the details of the lean-control mechanism because, at the time, I thought it was an innovative approach.
The steering handles were located on either side of the rider. They were capable of two motions. They could be moved forward and back and in so doing rotate the wheel beams down and up respectively, resulting in the vehicle leaning. They could also be moved from side to side for steering. What was novel was that the leaning motion of the handles was coupled to the steering motion through a device I called a Push-Me Pull-You, PMPY. Essentially it was a linkage using two springs that would exert a spring force when it was compressed or extended from a neutral position. So, for a given lean position of the levers, the interconnect linkage would move the steering to a specific location. Through changing the connection point on a coupler linkage, this location of the steering could be adjusted.  This steering location corresponded to the neutral position of the PMPY. But this steering position could be overridden by moving the levers laterally, and the PMPY would return them to its neutral position when they were released. As a result, the spring-neutral position for the steering was not always straight ahead but would move according with the lean angle. 
The approach when riding the EV1.4 was to adjust vehicle lean to initiate the vehicle to turn. At low speeds, correction to the steering was usually minor or unnecessary. I was happy with the performance and took several four-mile-round trips to the library parking lot to practice turns on Sunday mornings. Low speed testing being considered a success, I began moderate speed testing, say faster than 10mph. While riding along straight road about 15mph, one of the front wheels hit a pot hole. The next thing I knew I was upside down in the ditch. Since the steering and banking were interconnected, the wheel drop associated with the pot hole caused a steering perturbation. The steering went into an unstable oscillation and the vehicle flipped. I pushed the vehicle back home. One advantage of rear-wheel steering is the vehicle is easy to push and steer at the same time. It was little comfort, and I avowed to never ride the EV1.4 again. I felt lucky I didn’t break anything on my person.  My son, Kyle, had crashed an earlier iteration of the EV1. At the time I wrote it off to his lack of familiarity with the riding technique. Instead it was merely an early warning that the rear-steering approach was seriously flawed.
In “Bucky and the Urbee” I discuss the difficulty in defending against accusations that the inherent instability of the rear-steerer was responsible for the accident that the claimant just had. I also refer indirectly to my experience with the EV1.4.  So, in retrospect, the abandonment of this approach was inevitable. 

EcoVia 2.1 
The EV1.4 was dead and there was nothing to do but take the hacksaw to it and make it a front-wheel steer -two-rear-wheel-drive trike. There were three factors that made me optimistic about this approach. I could reuse all of the wheel-drive hardware and much of the frame. I would need a new reversed seat and need to move the bottom bracket above the steering fork. (It didn’t turn out to be quite that simple as I had hoped.) With front-wheel steering the vehicle would be as stable as a typical recumbent bicycle, so I could use free leaning and not actively have to control the leaning.  And I had seen a video of the Munzo Tilting Trike which had a similar configuration of rear trailing wheel beams. It seemed to free lean just fine.
 To insure that the EV2.1 would lean as much like a bicycle as possible, I kept the wheel beams horizontal when the vehicle was upright and the links for the reverser linkage were vertical and horizontal. I felt this would insurt that the vehicle would lean about a point between the two wheels at ground level. This approach appears to have paid off. On the first coast down my slightly-sloping driveway, I thought the friction in the linkages was preventing the vehicle from leaning. I was wrong! The EV2.1's leaning was so natural that without looking behind me I couldn’t tell that I wasn’t on a bike.

The EV1.4 had spring suspension on all three wheels and for the drive wheels the springs were in series with the lean control levers. With the EV2.1, I wanted to insure that any suspension effects did not influence the leaning. To accomplish this, I made the frame structure two pieces. The main frame supported the drive-shaft wheel-beam-pivot assembly, the support for the wheel-beam-reverser linkage and the front fork. Since it no longer needed to be suspended, I went back to a two bladed fork. However, to maintain the interchangeability of the wheels, I used a through-axle approach, similar to that used on mountain bikes. One blade of the fork had a 12mm clearance hole and the other blade had a 12mm nut.  The seat, the handlebars (mounted below the seat) and the rear rack were clamped to a horizontal tube. They could be slid along that tube to adjust for rider-leg length. The horizontal tube could pivot about a point near the bottom bracket, similar to a Softride suspension. However, on the Softride suspension mounted on an upright bicycle, the pedal forces are in the same direction as the beam motion, so the pedaling forces excite the suspension. On the EV2.1, the pedal forces pass close to the pivot and these forces are orthogonal to the beam’s motion. The spring supporting the beam is located behind the seat and it extends from the top of the seat to the main frame, attaching near the support for the reverser linkage. The spring is quite compliant . The static deflection at the seat is about 4in. The natural frequency of the suspension can be determined by the formula Fn=(10/SD), with SD being the static deflection at the center of gravity for the suspended mass. The  calculated natural frequency of the suspension is about 1.6Hz., or 47rpm pedaling cadence. If the rider pedals above about 60 rpm, the suspension is not excited. 

Below is a pic of the suspension-beam pivot near the bottom bracket and steering head.

The EV2.1
There was one expected and one unexpected advantage to the front-steering approach. The expected advantage was that the free-leaning drastically reduced the mechanical hardware needed to steer and lean. The unexpected advantage was that the vehicle package became much more compact. The beam holding the crank and the beam holding the fork in the EV1.4 stick out in opposite directions. With the EV2.1, they became one beam sharing the crank-support and fork-support functions, thus shortening the overall vehicle length substantially.
Since the handling was everything I had hoped for, I set about to make some minor improvements.
I added a lean lock that consisted of a section of a disk brake rotor bolted to the horizontal arm of the wheel-beam-reverser linkage. I got this idea from the Munzo trike. To activate the caliper, I used a down-tube shift lever with a tube slipped over it to extend its length. This was attached to the side of the seat. I originally expected to use the lean-lock only for parking purposes. (The front brake lever has a locking feature as well to act as a parking brake.) However, at low speeds it became very convenient to lock the leaning. One forgets how much effort can be associated with balancing a recumbent when going slowly.

I enlarged the rear-storage rack so it would hold a plastic milk crate.
I added fenders and I added a frame around the front wheel to support a faring that would tilt forward to enter and exit the vehicle. I got the whole thing painted so I could ride it to church for the annual Earth Day observence.  I began to work on the faring.

 I rode it to the library several times to drop off books and to the local Safeway, but I realized that I didn’t like how it felt pedaling it. Yes I was out of practice riding recumbents, but even so, the bottom bracket felt too high, resulting in my knees coming  too close to my chest when pedaling and the seat was leaned back farther than I expected. I was getting more suspension compression than I had designed the seat’s back angle for.
I put the faring on hold and pulled out the hack saw again.