While I used a lot of time in designing the Vélassi, in fact three to four time more than time used to build it, I unavoidably made a few mistakes:

Steering geometry

I had the opportunity to buy a fork of Rans Stratus with an offset of 89mm (3’’1/2) and I thought that if I had difficulties with my under-seat steering, using such fork may allow me to change my mind and move the handlebar above seat, so I can have a ‘multi-purpose’ steering.

The geometry and angles I used were approximately the one of the Rans Stratus.

My idea was that if it work with over seat handlebar, it should work either with under seat handlebar.

Short answer is: No, that don’t work like that, as I discovered the hard way.

The bike can be handled without too much difficulties, but low speed is not controllable, my minimum speed being 7~8 km/h. I am also fairly imprecise in tracking.

My front wheel while small (20’’) is fairly heavy (2kg −4,4 pound) due to large tire and heavy hub dynamo.

A bit too late, I tried to think to what can cause the trouble.

I noticed that there is a self-steering effect due to the weight of the wheel, which add to the already large ‘wheel flop’ caused by the high reclining of the steering.

There is no top handlebar which may equilibrate the wheel weight
There is no ‘tiller effect’ which help user auto-centering the steering
I read in this thread that the arm weight also help to center the steering.

So, in short, I have a steering with excessive wheel flop which cannot be counterbalanced by an above seat handlebar and my bike became difficult to control at low speed.

I am very happy with the under-seat steering which is quite confortable and makes minimal loads on my wrist (this was one of the key reasons I am going to use a bent)

I was aware of the Trail recommendations of Bill Patterson and of the Wisil calculator implementing this formula, but I completely missed the ‘wheel flop effect’. Also, the wisil equations doesn’t take into accound low load of the front wheel of a LWB (wheelbase is 1650mm).

I improved the low speed handling with a spring attached on fork which reduced the minimum controlable speed to 5km/h and ease handling when pushing the bike (see details page).

Frame torsional stiffness

While the frame is very stiff in bending, its torsional stiffness is quite low, which gives a bit quirky behaviour at high speed, the gyroscopic effect want the wheel to stay in upright position, so it doesn’t want to follow the frame leaning. This is very noticeable if you try to do small steering movement in alternate.
There is a simple reason for this low stiffness: The low part of the frame is made in thin planks (9mm) and flex sideways, which allow overall torsion. This thin plates were chosen for two reasons:

To not increase the frame width and left room for the chain on right and the steering bar on left while having room for the wheel inside frame when folded.
To increase glued surface, but I could also have done that by cutting a thicker beam in variable height (the bottom beams were cut in height anyway)

That bottom lack of side stiffness could be solved simply for the front part by adding a diagonal. For the rear part you can’t add a diagonal if you want the folded wheel stored within the frame. This is more complex, widening the side beam may conflict with chain and steering bar. A different frame shape with the beam junction rejected behind the chainring might have allowed a thicker beam. As this is very simple, a front diagonal will be added for low weight increase.
A plate in front of the steering may also limit the top/bottom side shift, so reducing the overall torsion. It is needed anyway to protect the bottom steering bearing from rain and mud (yet there is packing tape to do that). And where else could I set my signature if not on a front plate?

Steering wheel clearance is 3D, not 2D.

I looked to the interferences between foot and front wheel while steering, but the front wheel space needed when you steer is not a cylinder but a tore, so you can have the wheel closer to your bottom bracket than you will estimate on side view, especially for small diameter wheels. Not a large gain (~15mm), but this is always benefitting. Also, I was not sure if short cranks were needed and so I studied the clearance with 170mm cranks. Experince show that shorts cranks are definitely a requirement for a recumbent, especially for a small frame as this one, so in the end, the wheel can be 30–35 mm closer than the actual design.
On the other hand, large tire shall be taken into account in the clearance design, including mudguards, I use balloon bike 55mm wide. My revised computer model can show the actual shape of the steering wheel movement (due to rake, this is not a sphere but a tore).

Front wheel does not need to be super strong on a LWB bike

Even without suspension, the load on the front wheel is only 1/3 of the bike weight. On braking, due to bike long wheel base and relatively low center of gravity, there is not much weight transfer. The Balloon tire ‘Big ben’ is quite effective in filtering impacts. So, you don’t need a wheel as strong as a BMX one. I bought my 20” wheel on Taylor wheels shop, it does have 36 spokes and a heavy dyanmo DH-C3000. Total weight is 1.42 kg, which is huge. Taylor wheels also propose a 28 spokes wheel with a lighter 20” dynamo DH-F703 - specific for small wheels -, which weight 1.1 kg for less than 30 Euros more. This will be largely sufficient for expected loads. Far from super-light, but better. As this other wheel is only proposed with quick release shaft, so you shall add an anti-theft nut to the price.

Frame shall allow a better rear chainline

I made a small mistake by having part of the frame near the articulation preventing the chain to be perfectly aligned with the sprocket - remember I use a gear hub. This frame part does not have influence on strength but is used for the articulation of the rear suspension plate. This could be solved easily on a new frame, but it is a bit too late for this one. New frame will have a real air shock, anyway, so no need for this articulation.

Hub gear shifter installation

I did face some troubles with my initial rotary shifter installation. To have the housing not contacting hand and wrist, I gave the rotary shifter an orientation with the cable output quite to the outside of the handlebar so the down part of the housing will naturally be well placed between the wrist and torso. That had exposed the shifter to rough contacts either with the ground or with other bikes when parked. This have moved the shield and displaced the cable outside its plastic guide, creating very serious indexation problems, which can be very dangerous for a geared hub. So I installed a stronger screw to better maintain the shield and rotated the shifter to have the cable output less exposed (see photo with bike parked in its shelter).
So an aluminium strip was installed to firmy guide the housing. See details page.

Building

Do not pre-impregnate parts

I did a pre-impregnation of my wood parts, which was a big mistake, as further gluing is of lower strength. Epoxy does not bond so well on itslef, even after mechanical scraping/sanding. You shall glue direct and be careful to not impregante areas which will be glued later. The full epoxy impregnations will be done after the whole frame assembly. Sure, there will be many recesses to be taken care of, but that will be structurally better.

Serrated glued areas on hard wood

Hard wood epoxy gluing requires surfaces to be serrated (striated ?) for good structural gluing. Hardwood are difficult to be well glued with epoxy as they lack porosity. I had read that it was a requirement for structural gluing of hard wood to serrate surfaces (with a metal saw blade). But I forget it… My memory came back later and I hopefully did this serrating for the rear frame which is the most critical part.
Also, it is reminded that oak shall be treated with aceton to remove tannin (I did not forget that…).

Hole size matters

The bottom of a 1”1/8 (28.6 mm) steerer tube is larger than the top (30mm), so the frame shall be drilled accordingly, as I discovered the hard way because my steering was stuck up in the bottom hole with very high friction, grinding the epoxy coating (this is the epoxy powder exit which helped me to find the cause of my steering problems). I did my very first recumbent runs with this half-locked steering!

Safety

You shall use nitrile gloves, not latex one.

I initially used latex gloves when doing the epoxy work, but latex does not resist to epoxy which end up on your skin, creating a high risk of skin allergy and sensitization.
Having already experienced skin allergy with epoxy, I should have known better.

Lighting

Don’t expose cables to suspension movements

I poorly routed the wire to the rear light and it got cut by the suspension movement.

Rear light connection

The rear light is connected to the front light, as is usual to benefit from the on/off switch of the front light. Though, my rear light does have a ‘brake’ function (increase lighting when it detects dynamo current decrease).
This wiring is wrong for two reasons:

It seems that the brake system don’t work. This is based on an inertia system, not dynamo current detection as the Bush and Muller rear light, so most probably a faulty light.
There is no point shutting down the rear light, especially one equipped with brake function. LED does not fear road impacts as incandescent bulbs were, so there is no mechanical reason to shut down modern lighting (I ride with light always on). In addition, I installed an USB charger for navigation with my phone which got difficulties to charge simultaneously with the front light on, so I have now a reason to shut down front light (2.5W) but I am not willing to abandon the rear light which does have a much lower consumption (0.5W). So connecting the rear light directly on dynamo is better and that reduce wiring as the USB charger (installed under the seat) is already supplied directly from dynamo. Also, when climbing, I shut down the front light to limit the muscle power required by dynamo. Gain is fairly limited (~4W) but I have difficulties to climb. As my muscle power is approximately 100W (Yes, that is my problem…), in slopes 4W gain may be equivalent to lower the bike weight by 2~3kg.

Dynamo plug connection

I initially unproperly wired the dynamo plug till I understood hown it shall be done. See details page.

Conclusion

The big deal here is the steering geometry which have driven to low speed handling difficulties. Being of known geometry, this frame may work a bit better with an over seat handlebar but I like my under-seat handlebar so I am decided to pursue in this direction. Low speed stability of over seat handlebar LWB is a known problem, which made some renowned designers recommending extremely low trail, which improve it, at the cost of high speed stability. This is a tradeoff.
That is one of the reasons the geometry of the second version was completely different, allowed by the indirect steering.

All other stuff is minor and can be corrected without difficulties (or already have been).

The Velassi

Mistakes