Sunday, August 21, 2022

Calibrating a Spoke Tension Gauge and looking at the results

My spoke tension gauge has helped me get a more even tension for the spokes on each side of a wheel than I was able to achieve by plucking the spokes.  But I've never trusted the tension conversion tables provided by companies, even if their gauges look the same as mine.
A slightly different pivot bushing or spring and the kg tables aren't correct
(click on all images to enlarge)
I have the Deckas version, which was affordable, smooth (after a few drops of oil), and quick.  It's worked well.  But last winter in my quest for knowledge, I bought 3 full suspension frames, (because no one I know would let me touch their bikes), and took them apart to study the pivots.

A Stumpjumper, Altitude, and VP-Free that now need wheels 
With 6 wheels to build, it would be really handy if I could use my tension gauge to measure absolute tension instead of just the relative tension between spokes.  I found Dan Burkhart's video about his calibration jig for a spoke tension gauge:
Dan Burkhart's DIY spoke tension meter calibrating device
video on YouTube is a good read

Spoke tension is often recommended to be between 70 to 140 kgf (154 to 308 pounds), so I got a 300 kg digital scale (200 kg would do) that used standard AAA batteries and appeared to have decent accuracy (claiming the OIML R76 standard) and 0.1 kg resolution.  It would be possible to hang the scale and simply wire a spoke to the hook, and then wire a 5 gallon bucket to the bottom of the spoke to fill with metal bits for weight and skip building a test fixture.  However I decided that repeating the test once in a while (for example after dropping my spoke gauge), as well as calibrating different spokes, would both be much easier if I had a calibration fixture.

My spoke tension calibration fixture is easy to build
Looking through my scrap metal pile, I found a nice 2" x 2" x 28" long x 0.094" wall square tube for a backbone for my jig that looked rigid enough to give reliable results. It came off an old exercise machine. People put them out by the road hoping someone will take it after they've realized it's just taking up space, and when I see one that has been sitting outside for a month or two I sometimes pick it up if it might have useful parts. This tube for leg lift weights was the right length and even had 1/2" holes already drilled in one end for slipping a bolt through, it seemed to be waiting to find a new and more useful life.

Mounting the scale to the tube
After sliding a long 1/2" bolt through the tube holes, the spoke had to be 1.25" above the frame for my Deckas spoke gauge to fit, so I used a short piece of automotive rubber coolant hose pushed over the bolt as a spacer.  Another short piece of hose between the scale and the nut provides some give that lets the scale move for alignment so that the load cell inside it doesn't get twisted and give false readings.

Connecting the scale to a spoke
Since the scale was built with a hole for a hook, a stirrup with a plate for the nipple was a natural connector choice.  I kept the U bolt narrow to reduce the spring, but made it long so that I could get my fingers and a nipple inside.  It's 1/4" rod, with a 5/16" thick cross plate that can be changed out for different spoke ends.

Connecting the J bend to the threaded rod
A simple plate as thick as a hub flange with a 2.6mm hole is all that's needed to connect the J bend.  Because I wanted the plate to be changeable to plates for other spoke styles, it's held with a removable 3/16" pin in a slot that I cut into a threaded rod coupling nut with a 1/8" grinder blade.  One problem I can see with Dan's fixture is rotation of the eye hooks as he adjusts the tension.  The red plate in this picture that is sandwiched between the coupling nut and a lock nut is an anti twist shoe that slides along the main tube to prevent this rotation. It also helps hold the threaded rod in position while installing spokes.

The anchor for holding the threaded rod and an alternative mount
I had a short piece of 2" wide x 1/4" flat bar, and after drilling a hole for the threaded rod 1.25" up from the main tube, I welded it on the end.  However if you want to make this fixture without welding, a substitute could be bolting on a short cross piece of ell angle iron.  I used 7/16" threaded rod because I had a short piece, but 5/16" or 3/8" would probably work considering that a spoke is around 2.0mm.  A smaller diameter rod would have a finer pitch which would make adjusting tension more precise, I found 1/8 turn of the nut gave about 10 kg change.  However I put all the test results into a spreadsheet and derived an equation, so it wasn't necessary to aim for exact tension numbers or gauge readings.  Either method should produce a good conversion table, but I'll say that my digital scale took well over 5 seconds to settle and making many small adjustments to reach a particular number could be tedious.

The first round of testing- how well does it work?
I clamped the jig to a table top with woodworking clamps while making tests, but with it's square main tube it could also be held in a vise.  It turned out to be easy to adjust with one wrench, and had excellent repeatability regardless of tightening or loosening direction, suggesting that these components are very rigid with little spring, and the scale is of reasonable quality.  As time goes on though I expect to compare the digital scale against others to be sure it isn't drifting.

Deckas test results (green and black lines ) show very low scatter and good repeatability
These test results are for 2.0mm (14g) round steel spokes because they are common on my projects.
-the green line has 42 measurements on a used galvanized 307mm long spoke
-the black line (so close that most of it is hidden under the green line) has 54 measurements on a new stainless steel 256mm long spoke
-the blue line has measurements taken from the Park Tools TM-1 tension gauge table
The TM-1 reads lower for a given kg tension than the Deckas, which supports my concern about using tables from other companies.  The adjusting screw on the Deckas looks like it could be backed out one turn which might then match the TM-1 readings, but the ranges and slopes also don't match well.  In the range from 70 to 140 kg:
-Deckas readings go from 26 to 30 (8% of the scale) and change 0.5 per 10 kg
-TM-1 readings go from 20 to 26 (12% of the scale) and change 0.9 per 10 kg
If you want to be very exact with either one you're going to do a lot of squinting because of the small change per kg.  Also neither gauge is useful below a reading of 15 or over 30.

Best fit equation for the 94 measurements from the Deckas
(Warning I've swapped the axes compared to the first graph)
If clicking on the image is still too small to read, save the image to get a jpg that can be enlarged
I combined the measurements of the stainless steel and the galvanized spokes because they were basically the same curves and the greater number of data points would increase accuracy.  The SciDAVis program was used to plot and calculate the best curve to fit the points, and the equation then used to calculate the kg of tension for each reading on the tension gauge.

Here is the equivalent Park Tool TM-1 equation using their table data
The TM-1 table data lines up so well that Park Tool must have also stuffed their data through an equation to smooth it out.  A benefit of using an equation instead of directly measuring a scale reading for calibration is that it removes erratic readings.  This illustrates though that any stickiness while taking a spoke measurement could greatly affect your results.

52% of the spokes on my bikes are between 34 to 60 kg tension.
Looking at the 10 to 20 kg steps between each scale number and knowing that the gauge has still been a big help made me wonder just how exact do I have to be?  Do I need a digital readout spoke tension gauge for better resolution?  So I measured the 5 bikes I've been riding for years.  These wheels are straight and stable, even though I routinely carry 40 to 100 pound cargo loads on dirt roads at ebike speeds.  They failed tension tests miserably, with only 9% in the oft quoted 100 to 120 kg range, and the tensions were all over the map, from 6 kg to 144 kg.  Readings of 28 to 29 on the drive side of a rear wheel often paired with readings of 18 to 19 on the non drive side (a difference of 90 kg), and this doesn't require squinting at the scale to see.  The kg steps between the scale numbers are small in comparison with reality.

Around 120 kg seems to be a common upper tension limit for the rims I use
Note: I assume 1kg = 1kgf = 10N for easy calculations
Now that I have a reliable calibration for my tension gauge, I can see that the spokes on my bikes could be a little tighter.  The average tension of the 102 samples is 50.7 kg and I could aim for 70 to 80.  But some rear drive side spokes are already close to rim limits, so it shouldn't be much tighter.  (Cycling News wrote an instruction page for the Park Tool TM-1 that has a table listing many rim limits: )

Thinking about people on forums saying that their wheels loosen up but that my wheels have been stable, my thought is that the load on a wheel doesn't matter as much as the impact, since I carry heavy loads but I don't jump or bash through rock gardens.  One other consideration for increasing my spoke tension is that many of the nipples on the older wheels are dry and hard to turn even at lower tensions, and only the wheels I've built with grease or antiseize can reliably reach over 100 kg tension.  Given the real world range of spoke tension on my bikes, I don't feel the need to get a digital spoke tester until my eyes get a little older and the gauge scale becomes even harder to read.  I might try making a clearer label for the scale, or even making a new gauge arm with the measuring post closer to the the pivot to increase the pointer displacement.  For how easy it was to make though, this calibration fixture provided a noticeable step up in my wheel building that was equal to getting a spoke gauge, and I think the quality of the last few wheels is getting pretty decent while taking less time.

Sunday, January 30, 2022

The Breeze, the Stratus battery (S1E4)

Over the years I've used several types of batteries.  My first bikes had huge LiFePO4 bricks with A123 type cells, which could take a lot of my beginner's abuse but were very heavy and actually needed a cargo bike to carry them.  I liked them a lot because it's a safe chemistry- there were too many reports of people burning their houses down with the other main choice, which was hobbyist model airplane radio control (RC) packs.  I eventually moved on to Lithium NMC 18650 cells as they became affordable and Battery Management Systems (BMS) became better, because they were 2/3's the size and weight.  I've stayed with 48 Volts, because I think it gives the best performance for the smallest wire and component size, while staying below most safety standards for high voltage and not needing more expensive components.

I prefer to use as large a battery as possible for 3 reasons:
1. range
I figure about 15 Wh per mile, (depending on the rider it can range from 10 to 20 Wh/m), so on average 840 Wh is good for 56 miles which covers most situations.  I've told everyone I've built bikes for that they still have to pedal.
2. surge capability for climbing hills
Larger batteries won't have as large a voltage sag under load, and my riding is one long hill after another.
3. longevity
Lithium cells don't like to be fully charged or fully discharged, and they last longer if kept between 20% to 80% full.  The first two batteries I owned lasted for only a couple of years, because I used to put them on the charger right after I got home to find out how many Watt hours I'd used on the ride.  But after installing a meter on the bikes (Grintech Cycle Analyst) I stopped the frequent charging and took more rides between charges, and the batteries lasted twice as long.  Then I started trying to charge the batteries only right before a big ride, and the last battery was a few months into it's 7th year before it did not have power anymore.  Interestingly Stanford University researchers recently found that by charging a cell and then immediately briefly discharging it they could move plated out lithium back towards the cell electrodes, in effect rejuvenating the cell.  I think charging just before a big ride has a similar effect and may help explain the 7 year life span   Clean Technica wrote a summary of the research here:
To take more rides between charges, you'll need a larger battery. 

I've found the down tube packs to have the lightest cases and the best bike balance,  I also like the smooth nonconductive plastic with rounded corners so that I don't hurt myself as much.  After I'd installed half a dozen different styles of down tube cases I settled on the Reention DP-6 case as my standard choice.  It will fit 65 cells in a 13 serial by 5 parallel pattern (13s5p), which if  assembled with 3500 mAh cells adds up to 840 Watt hours (48V x 17.5Ah).  This is as good as I've been able to fit in the front triangle of most bikes I've converted.  (Reention does make a DP-9 case but it's usually a bit too big and if more Wh are truly needed, I end up using a 1 kWh (13s6p, 48v x 21Ah) rear rack battery.)  I often have a DP-6 on the shelf because I'm  converting a bike.  Having a standard battery came in handy for the Upper Valley Ebike Lending Library last summer when they dropped a battery.  I was able to substitute one of mine while I glued the broken case back together.  (PVC pipe plumbing glue from the hardware store works well.)  Because of this background, the Stratus got a Reention DP-6, 48 Volt, 17.5 Ah, 840 Wh battery without a second thought.

An empty Reention DP-6 case ready to be stuffed with cells.  When purchasing an already built battery, I look for one with name brand cells and a BMS that has at least a few certifications.  I prefer this case because the base has these features:
A. it has 4 retention tabs down each side instead of 2
B. a full length aluminum spine instead of 1/2 way 
C. 5 large spade terminals that can be doubled up
D. no large cavity for a controller that I don't need  

For comparison most of the Hailong cases available have only 2 retaining tabs per side plus the empty cavity for a controller with a shorter spine, and use bullet terminals that can fill with water and dirt when the bike is transported with the battery removed.

However the DP-6 case could still be improved by adding these 3 features:
1. an IP-65 rating (to match the BBS02) and a more waterproof switch.  These cases are OK in the rain, (I've ridden in downpours without a problem), but they can't take a garden hose pointed directly at the switch and connectors.  In many ways they remind me of the electrical switches on the motorcycles of my childhood, which had bare switch contacts screwed inside a plastic shell on the handlebars that were very exposed to the water coming in.  There were millions of these switches made for decades and they worked, but we can do better now.
2. a Vee channel molded up the center under the base to fit on down tubes more securely.  Numerous companies make water bottle and accessory mounting rails with this groove, and Grintech's Bottle Bobs also have this groove, here's an example:
3. an XT60 socket instead of wires.  Here is an example that I made for a mountain ebike conversion:
The terminal base of the DP-6 fitted with an XT60 plug.  Normally the power wires would exit here.
The outer 2 spade terminals are paralled together for each pole.  The short wiring is almost stiff enough to hold the XT60 in place by itself, but a RC connector hold down strap was also added.  Note: live power receptacles should have recessed female sockets, similar to household wall receptacles.  Using exposed male terminals that could easily be shorted is not good practice.

BBS02 motors come with a 20" lead, the battery cases have a 12" lead, and usually there is a 10" long adapter cable for a total power cable length of 42 inches.  This is a bit long since the distance from battery to BBS02 is around 9" on most bikes.  I cut a lot of this out and leave only one pair of connectors.  However the Stratus motor is a foot forward of the battery so I kept the wires and hid an XT60 connector under the battery base instead of fitting an XT60 terminal into the base. 

Connectors I have loved or left
(First loves on left, current connections on right)

AC-The first battery connectors I used were AC power cord plugs such as on a computer printer- big and heavy, just like the LiFePO4 batteries they came on.

APP-I used to use Anderson 45 amp Power Pole connectors, but found that the jiggling wires twisted the contacts apart slightly during riding, and then the contacts started sparking and pitting.  This would melt the housing, and then all hope of getting home was lost. (Although hard to see, in this picture the red housings are melted, and the contact is pitted.)  After the second meltdown, I moved on.

EC5  (blue)- I liked these, they're tough and stayed together, but they are completely open to water
HST (red)- a little smaller than the EC5, but the spring contact collar squished I was while playing with them on the bench, so I never put them on a bike.  These are the same size as banana plugs though, so I'd consider using them for a bench testing setup.

BF- These are Bafang bullet terminals, which are not the same diameter as common 1/8" bullet terminals.  I liked these a lot because they have most excellent silicone rubber boots.  But they have a very weak neck between the wire crimp and the socket, and after breaking 3 of them off I now only use them for the connection inside the BBS02 between the controller and the windings.

WP- Delphi Packard Weatherpack automotive connectors (similar to AMP/Tyco/Molex) did not rate a trial, they are either too low in amps for a suitable size (20A on left), or too big for a usable amp rating  (46A on right) due to their stamped steel terminals.

XT90- The early motor controllers I used had large capacitors in the input circuit, and there would be a very bad spark when plugging the battery in.  The XT90S version has a resistor that moderates this spark and was a step up for my earlier bikes.  But they are big and heavy, and my batteries have switches now.  Normally this connector would have a solid yellow exterior casing but I sat down one night with a hacksaw and an Xacto knife and cut a connector in half to see the resistor, shown in the photo.  .

XT60- My current fave.  There are several versions shown in this photo.  Machined brass terminals give a 60A rating in the smallest package, (my rides consistently have 12 to 18 amp climbs, with some momentary spikes to 30A, so this is a good safety margin). Because the standard yellow version on the left has a hooded design it can be reasonably water resistant if you put heatshrink and silicon caulk around the outside wires, and then smear the terminals and inside hood with silicone grease before plugging them together.  I'm using the gray version on the right which has snap on outside covers.  I still put heat shrink tubing over the wire solder joints to provide some bending strain relief for the wires, then I fill the area with silicone caulk and snap the cover on.  After drying it's a pretty robust setup.  I've tried to improve this by putting a boot over the connector, however the available boots (in blue) are too short.  A substitute is a piece of heat shrink that is an inch longer than the connector with some caulk sealing the stationary end.  (I place a wood shim in the tubing while shrinking it so that afterwards when I remove the shim there will be some slack.)   However I think the best protection so far is putting the XT60 inside the battery base.  The black aluminum U shaped RC model hold down strap for the connector is shown near the top.  Note: the yellow bulkhead connector at the bottom with exposed male terminals is OK for an appliance that draws power, but shouldn't be used on a battery that is a power source.

All of the large forces on a battery case act to press the case and the downtube together, and the 2 small bottle mount screws are sufficient for most riding.  On standard diamond frame bikes I prefer to install a third screw for additional sideways strength when the bike is ridden with the frame tilted.  (The Vee groove battery base improvement would help here.)  I use an M5 popnut that matches existing bottle mount hardware.
A popnut gun with M5 inserts.  A bolt with nuts and oiled washers (center) can also be used to compress the popnut with some effort because it will try to unscrew while compressing

An easy way to find the center of the downtube is to clamp a stick on each side, and then put a mark on a piece of masking tape halfway between the sticks.

The Stratus didn't have any bottle mounts on the tube for the battery, and I decided not to drill holes in the frame.  I switched to tee nuts held on by hose clamps instead.

A metal hole punch is used on the hose clamps.  The 3 sizes of tee nuts I use are at the bottom, along with a piece of heat shrink tubing for covering the clamp.  The tee nut in the red circle has been bent to fit the frame tubing.

Furniture Tee nuts come in many styles, I use the type with 3 small holes.  (Metric tee nuts are not available here and I've been using M5 Weld nuts instead.)  The 8-32 size will slip into most battery base mounting slots but it's a bit light duty.  I often use the 10-24 size but the base slots need to be filed slightly wider.  The M5 size requires a lot of filing for use with a battery base, but once in a while this thread size is necessary for mounting an accessory.

I start by tightening the hose clamp on the frame in the desired position and marking the location for the hole in the clamp band for the tee nut with a Sharpie marker.  A metal hole punch works better than trying to drill a hose clamp.  Drilling is possible but start undersize and then use a chainsaw file to smooth out the hole and move it back towards where it was supposed to be.

Using a pipe, vise and punch to bend a tee nut to fit the tube.
Sometimes the tee nut has to be trimmed, but most often the shim washers and hose clamp take up any excess height.  I bend the tee nut to a radius that matches the frame tube, slip it into the hose clamp, and then fit a piece of heat shrink tubing over the clamp band (with cutouts to fit the tee nut and screw housing) to protect the paint on the bike.
A hose clamp with a tee nut fitted, then I cover this assembly with heat shrink.

The Stratus hose clamp mounts installed.  The power cable to the motor runs under the base.

For the Stratus I tried to hide the screw heads up under the base, but they could be pointed down for easier access.  The heads were useful in this position for holding the wiring and XT60 connector in place.  I also painted the screw housings black to blend in.  Hose clamps and tee nuts provide a very secure mounting, it's just that they aren't as clean looking or light as a pop nut.

I'll write one more short post to finish up this series about The Breeze - the Stratus,  with a couple of the other small details that make it a nice bike.

Friday, January 7, 2022

The Breeze, the Stratus drivetrain (S1E3)

From the start the Stratus was going to have an electric assist installed.  I'm in my fifth year without a car, in a rural area, in the middle of a transportation system that was 99.9% designed for cars.  The electric assist helps me compensate for the poor road design, the distances, and the nonexistent bus and train network.  Also, recumbents aren't as easy to peddle uphill, and there are plenty of hills here.  So I took most of the original drivetrain parts and stored them in a plastic zip bag in case the bike needs to be restored later on.

Because they can be shifted to different gearing ratios, I've found Bottom Bracket motors are more versatile than the hub motors I've used.  I knew the Stratus was going to be fast, but it also had to climb up my dirt road.  Because my state recently enacted the (questionable) 3 class ebike legislation and I wanted to stick with 750 Watts, I decided to use a Bafang BBS02.  It's more powerful than the 500 Watt Tongsheng TSDZ2, and not as heavy as the Bafang BBSHD,

left to right: TSDZ2, BBS02, BBSHD (click on all pictures to enlarge)
Because these are all the same voltage (48V) and about the same motor speed, the size of the motor case is a pretty good indicator of their amps and power capacity.  I find the BBS02 will climb steep roads around 8 to 12 mph, but the TSDZ2 has to be shifted down one or two gears and is slower.  The BBSHD is a good choice for heavily loaded cargo bikes because it's built like a truck motor and can deal with the force and heat much better.

  I'd installed a dozen BBS02 which had been strong and reliable motors, but then the 13th motor had a manufacturing defect:

There were two dry solder joints on the Hall sensor printed circuit board of the 13th BBS02.  This was easy to diagnose because a 08H error code showed up on the display, and after removing the board the dry joints were visible and an ohm meter showed bad connections.  I fixed this by resoldering the joints because the vendor was slow in sending a replacement.  I consider this a significant fault, but otherwise the rest of the motor looked very well designed and solid inside, and I was willing to give the BBS02 another try.

Testing the Hall sensors to be sure it was the solder joint and not the sensors

The sensors themselves could have been bad, so I tested them before reassembling the motor.  With a voltmeter connected between the ground and the output of each sensor, the output should toggle between a high and a low voltage when a magnet (a magnetic screw holder in this picture) is passed by the sensor.  If you are trouble shooting this part and not familiar with doing this, please unplug the Hall board from the circuit and use a separate power supply.  Note:  these appear to be latching Hall sensors, and the magnet has to be flipped to the other side to change the polarity.  I had expected the vendor to send a replacement Hall sensor board so that it could be swapped easily, but they cheaped out and sent one single Hall sensor.  I don't think this would have helped most bicycle shops.  

This 14th motor has been working fine, it's smooth, quiet, and pulls nicely.  I reprogram the BBS02 motors mainly to turn the power down in the first and second assist levels for easy riding.  With these settings, the bikes ride a lot like the Specialized Turbo Como in our ebike library.  I also change a few other settings, such as reducing the starting current to prevent a large starting load from burning out the motor, but I don't do anything complicated.  There is a ton of BBS02 programming information available on the web so I'm not going to repeat it here.  (I use Penoff, Endless Sphere, EM3EV, Karl Gesslein, Lectric Cycles, and Lunacycle suggestions.)

The new rear wheel setup, with a 126mm freehub, 9 speed 12-36T cassette, and a rear derailleur capable of a 36T granny gear.  (The stick is because I hadn't installed the kickstand yet.)

The Stratus has a 27 x 1 1/4" (32-630) rear tire which is one of the larger diameter wheels, and the original gears were a two speed chainring (52/42T) and a six speed freewheel (13/14/17/21/26/32T).  I needed to keep the 52T chainring for speed, but wanted to add a larger granny gear to make up for the single large chainring.  Almost all Shimano rear derailleurs in my price range have a maximum granny gear capacity of 34T.  I had experimented with a short hanger extension on my Marin MTB conversion, but I had to drill and insert a peg to correct it's angle to Shimano specs.  It worked but it wasn't easy or elegant.  The Shimano Deore M592 rear derailleur has a 36T maximum, so I decided to try it without a hanger extension and see if the motor made up for having less than a huge granny gear.  It will climb a 16% grade OK, but it's a couple mph too fast in rough, bumpy spots.  If this setup lasts, I might experiment with a 10 speed setup to get a 40T or 44T low gear, but there are a few considerations with doing that:

-Many people talk about the torque of a mid motor ebike wearing out chains quickly.  I've been using wider 6/7/8 speed chain, and the wear has been so minimal that I decided to try out 9 speeds for this bike.  But 10 speed chain would be thinner still.
 -Six to nine speed parts are somewhat compatible and I can mix and match some parts while building, but I wouldn't be able to use any of my spare parts for a 10 speed.  The cable pull and the parallelogram angle are too different.
-I've been buying quality regular chains for about $14 each, but have found a comparable ebike chain was $42.  Since the Stratus uses 2.1 normal length bike chains (246 links) and The Breeze will be similar, there is some financial motivation to design for a low wear drivetrain that can use a regular chain 

No one ever talks about using larger gears to reduce the stress on a bike chain like the industrial drive people do for machinery.  At first glance having only a 52T chainring seems like a bad idea, but if it's paired with large rear cogs then it's actually beneficial.  Besides less stress on the chain, I use the lower gears more often than with a 44T or 48T chainring which spreads the wear around.  I'm also becoming less and less of a fan of 11T top cogs because I've noticed chains jumping on them more often, so they seem to be not a great idea for a strong mid motor ebike and 12T or 13T high gears are better.  In general I'm leaning towards larger diameter gears, while still trying to keep a very wide gear range on the cassette.

The original Suntour Mountech rear derailleur has an extra spring loaded pivot that is concentric with the upper idler pulley.  This was famous for packing full of dirt, and then jamming the derailleur into the spokes.  It was called "The wheel builders best friend" by one blogger, and was responsible for much of Suntour's financial problems.  It went into the storage bag with the other original parts.

The Stratus rear dropouts have a 125 mm spacing.  I wanted to upgrade to a stronger freehub with more speeds, but all my MTB parts hubs had a 135 mm OLD dimension.  Since the seat and chainstays are very long and slender on the Stratus and the frame alignment measured true,  I didn't want to cold set the stays and take a chance of it becoming off center.  I found a wheel with a 130 mm Shimano FH-RS300 hub, and it even had a 4 mm spacer on the left side!  After removing the spacer and recentering the rim 2 mm over to the right, I had a 126 mm wheel that fit into the dropouts almost nicely with just a light nudge.  Unfortunately the chain did not fit.

The inside of the right rear dropout had to be modified to clear the chain.  The seat stay tube (top) stuck out too far and had to be cut back and then rewelded closed.  The tube had grooves already worn in it by the old chain, so this probably was a problem even when new.

The finished seat stay with decent chain clearance in high gear.  Originally the stay was a tube all the way around.

1/2 inch black poly tubing instead of idler pulleys (shown with the chain on the top sprocket)

There were a couple more small modifications.  One was that I needed to extend the speed sensor cable.  The bike was almost ready to ride and I didn't want to wait for another order, so I cut 28 inches out of a USB cord and spliced it into the cable.  It looked the same, plus it was shielded.  I filled the splices with silicon sealant before putting the last piece of heat shrink tubing over them.  The other mod was the chain stay protector.  I'd planned on making an adapter in front of the rear wheel that could mount both the kickstand and chain idler pulleys.  However because of the 52T chainring, chain rub on the stay was not a problem.  The kickstand was also more stable when located further forward.  So I took a 6 inch piece of black poly 1/2 inch plumbing tubing and cut a slit in one side, and then snapped it on the chainstay just to protect against rubbing when the chain was bouncing around.  It's light, and I've never heard it make any noise.

This bike is a treat to ride.  It takes a little effort to get it up to 25 mph, (there's some weight, and I should probably revisit the motor programming), but then it will stay cruising at 25 to 30 mph with not much trouble.  The one problem I've had with it is getting started with 60 pounds of groceries on the rear rack.  I don't have a hand throttle on it, (there wasn't any room left on the handlebars!), and I've found that if you use the walk assist mode to get going, the motor will not stay running when you start pedaling.  I need to make a start button to turn the motor on until I get balanced and pedaling.  Overall it's exciting to ride, and it's probably the closest I'm going to get to a high performance sports car with 9 speeds with paddle shifters on the back of the steering wheel.

I'll write one more post that will cover the battery and remaining details that make this bike work.

Sunday, January 2, 2022

The Breeze, the Stratus frame (S1E2)

I'm only moderately familiar with recumbent design, and my first step was to figure out what the Stratus was supposed to be like originally, before years of different owners had worked on it.  I didn't find any original documentation from 1985, but there was a 1992 review in the Recumbent Cyclist News archives ( , Vol 6 No 12).   Amazingly my bike was pretty much whole, with the original Shimano 600 group set.  I was going to have to change some of that out because I didn't think a 35 year old freewheel would be reliable enough for my use, but I'd be changing the front chainring anyway with the addition of a motor.  There was also no way I'd use the shifter and brake on the joystick. But overall the frame was straight, and mainly needed a new coat of paint.  It had a sticker on it from Dana Point Cycle and Sport, and I imagined the original owner cruising up and down the California beach front listening to "Hey now, Hey now, Don't Dream it's Over".

The RANS Stratus A as received

Out of the many bikes with a lower seat height, I was lucky to receive this Stratus.  This model and the Easy Racer Tour Easy ( appear to be 2 classic bikes that are so popular that one blog writer claimed they are the bikes that have crossed the US the most times.  It's a very good place to start from.

An Easy Racer Tour Easy with an early electric motor drive kit added under the seat.
Photo credit: I don't know the original source, I found this photo on back in 2014, and that site now appears to be dead.  I've seen this picture on a few other sites since then.

The Stratus has a bottom bracket height of 14 inches off the ground and a seat height of 21 inches, while the Tour Easy has a slightly more upright position at 13" BB and 22" seat.  I had been planning on a seat height of 18" for The Breeze frame to match many household chairs I have and also the seat height in my old Saab 900, but I found while riding the Stratus I liked a 1" pad, which gave me a 22" seat height that matches the Tour Easy.  However the Stratus BB height is better at 14", as the heel of my shoe was often less than an inch from the ground.

The frame after welding the rear dropout and painting, with the new wheels
(It's mounted in an old exercise stand to prop it up.)

Evaluating the frame design while cleaning it up was lot of fun!  There were a few things I wouldn't have done, such as the plastic cap closure for the seat tube, and the termination of the stays on the rear dropouts (which I'll say more about in the drivetrain post).  Also I was wondering how the BB to seat to rear dropout triangle would handle the torque of a motor.  Otherwise the frame was thought provoking!

A 59 degree head angle with a centering spring added to counteract flop

The 59 degree head angle confirmed something I've learned since I wrote my blog posts about steering geometry back in 2014- the basic function of the head angle is to put the handlebars comfortably within the reach of the rider, and handling is secondary.  Here's an example that demonstrates the important aspect is trail, not head angle:

Tony Foale rides his modified BMW with a 15 degree rake (motorcycle terminology, or 75 degree head angle in bicycle terminology).  Tony modified his motorcyle for testing with rake angles of 15 and 0 degrees, which under some conditions performed better than the stock 27 degree angle which could oscillate. 
Credit: Motorcycle Chassis Design: the theory and practice, Tony Foale and Vic Willoughby, 1984, Osprey Publishing Ltd, page 62

The Stratus head bearings were notchy, and I found a current version that looked the same but now had seals!  The front wheel wasn't original, it had nuts and 90 mm Over Locknut Dimension (OLD) while the fork had 100 mm spacing.  I dug out an old Shimano 100 mm quick release front hub that looked period correct and built a new wheel using an aero shape rim.  A major side effect of the 59 degree head angle is a tremendous amount of annoying wheel flop, and after a couple of weeks of having the front fork turn sideways whenever I let go of the handlebars, I made a centering spring.  It's just 4 inches cut off of an old spring for a wood screen door (the kind that always slammed shut with a bang), and is attached to the fork with a little ell bracket on the brake caliper bolt, and onto the frame by hooking it on to a hose clamp.  I still have to figure out a front fender though, because RANS cut it a little too tight with less than 1/8 inch between the tire and the fork crown.

While I did remove the central joystick, I wanted to keep the front fairing + storage cubby.  I added bar extensions to the hoop handlebars instead of the usual update of  "ape hanger" bars.  This bike is steered by leaning, not by turning the handlebars, and it took me a while to learn to steer with my seat.  I had to not lean on the backrest for about 20 miles before it started to become more natural.  There's a small amount of tiller effect during slow, sharp corners, but at speed it's not noticeable at all.  The handlebars aren't very solid and mostly just a place to put your hands, but the fairing does block the wind, and the cubby is really handy! 
The cockpit with a pair of Dia-Compe SS-4 two finger brake levers and a Shimano Mega9 shifter sandwiched in between them.  The BBS02 keypad is on the left handlebar extenion, and it's LCD display is on a custom bracket that mounts to the fairing in front of it.  (The handlebar extensions are cut off in this photo.)  All cables and wires fit under the bar tape for a comfortable grip, although it is a pretty close fit.   The brake switches are the inline cable type and hidden inside the fairing.

I've adjusted the hoop so that my knees have just an inch of clearance to the underside, and realized that while a fixed solar panel could be mounted over my knees on a bike frame that has steering with a remote linkage, there wouldn't be much space for a front trunk underneath it.  Also a solar panel over my head would be much taller than I had planned, so the next step is to reconsider the seat height again.

The rear wheel is also new.  I had planned on upgrading the antique freewheel hub to a freehub to better handle the motor torque, but I found a complete wheel that I was able to modify, so I simply replaced the whole thing.  I'll cover this part a bit more in the next post about the drivetrain.

The seat cover had shrunk and I had to make a new one.  Also the original kickstand simply spun around on the chainstay, so I took a standard Greenfield kickstand and lengthened it by epoxying on a 4 inch piece of a salvaged seat stay that I had hammered into a profile that slipped on the leg nicely.  (This took three tries:  on the first try I hammered too hard and snapped the leg off, the second time I tried using a torch to soften the tube and melted the aluminum kickstand.  The third time I applied a lot more patience.  It's good that these kickstands are common and inexpensive.)
The extended Greenfield kickstand
The chainstays are very slender and could be bent, so I made top and bottom mounting plates out of 1/2" CDX plywood with grooves routered in them that cupped the tubes to spread out the force of the kickstand   Then after mounting the kickstand I scribed the end of the leg to the correct angle and welded on a circular foot.  I had originally planned to combine the kickstand mount with a pair of chain idler pulleys, but the pulleys weren't necessary.  I'll write more about this in the drivetrain post.

The almost finished Stratus project, (it still needs a front fender and another layer of foam on the seat)

I finished the bike in September and rode it for a couple hundred miles before snow started and then I parked it to keep the road salt off of it.  It's very fast- easily able to keep up with cars in 25 mph zones, and under most situations it's very comfortable.  It's definitely in it's element on a paved road, but it's been fine on all the dirt roads here except for the worst ones where the town has dumped a lot of 3 inch stone.  For The Breeze I'll probably go one size wider in tire width, from 1 1/4" (32 mm) to 1 1/2" because of the stone, but I don't wish to go wider because I also need to cover distance quickly.  I've found it hard to choose between cruising along on this bike or the more upright posture of my converted Marin eMTB and I split my time half and half between the two.

For the next post I'll take a closer look at the drivetrain.