Ultracap Bike

[E=IR]

How long does it take to charge up?
How far will it go on a charge?

[JCG]

How long does it take to charge up?
How far will it go on a charge?

Charging is quite fast. Although I have to limit my charging current to around 15 A (the variac starts to hum angrily at me if I try to go faster, and I've long since blown its fuse and replaced it with a sturdy 1/4" diameter stainless steel tubing stub ), I am charged to 48 V from dead (0 V) in under nine minutes. If the variac could handle twice the current, I'd be done in around 4 minutes.

As for range, I would really like to directly test this... but there's a lack of uninterrupted flat roads here in the city! At the moment I'm doing all the wiring for the Kelly controller, but I can make a (probably poor) estimate.

Let's say that the motor wants an average of 20 A of current while the pack discharges from 48 to around 24 V. I could theoretically go down to 18 V before the Kelly went into an undervoltage error, but no sense in pushing it (and the return on investment in terms of energy is not great at low voltages). Let's also say that this corresponds to an average speed of 24 kph (15 mph). Hope I don't make a dumb mistake here.

i = C dV/dt, so dt = C dV/i = (160 F)(24 V)/(20 A) = 192 s = 0.05 h
0.05 h * 24 kph = 1.2 km = 0.75 miles

Hence the use for acceleration/deceleration only, while using the pedals at high speeds to cruise, or more hopefully, to achieve through-the-road regeneration.

[JCG]

Update time... I received my control cable (J2, 14-pin connector) from Kelly on Friday and spent most of the work day Monday wiring things up and making a mess. I mounted the Kelly to the top of the capacitor module; I couldn't resist since the module had frame screws with exactly the right spacings. Here's everything wired up (most of the mess is hidden behind the module):



I generally swiped a bunch of wires and connectors to get it together. All of the controls are mounted on the handlebars (sloppily):



Red circle: Twist brake for proportional braking control.
Orange circle: Brake switch button. This is a hack job in terms of mounting, but it is extremely convenient - it's push to close the circuit, and it opens again when you remove your thumb. No need to flip the brake on and off, and you can easily twist the handle while keeping your thumb in place. Anyway, I liked it.
Yellow circle: Cycle Analyst. I rewired it for the Kelly with a the cap as power supply and to the speed sensor on the front fork. I don't have my high current shunt resistor so it can't track amps or power use yet. It's on the way.
Green circle: Thumb throttle. It's a 1-4 V output type, which called for some fancy reprogramming inside the Kelly's software to make it look like a 0-5 V type. It's used without a switch.

I was still able to go for a spin and get some rough data involving voltage swings... took some trail runs on a mostly flat stretch and went up and down a steep hill. Here's what I found:

1) 0.33 km run on mostly flat street, motor only (no pedal assist), average speed of 10 km/h, capacitor voltage dropped from 33.9 V to 31.9 V.

2) Same distance on same street, pedaling against brake switch resistance (no twist brake), average speed 18 km/h, capacitor voltage rose from 31.9 V to 32.4 V.

3) Same distance on same street, pedaling against stronger brake resistance (switch + slight twist of brake), average speed of 12 km/h, capacitor voltage rose from 32.4 V to 34.0 V.

4) 0.12 km run up a steep hill, motor only (no pedal assist), average speed of 12 km/h, capacitor voltage dropped from 31.9 V to 27.3 V.

5) Braking while coasting down same hill (after riding around for a while), capacitor voltage rose from 19.1 V to 19.6 V.

In short, the regen is working. It wasn't killing me at all to go less than a tenth of a mile and put a full volt back on the caps. I can't wait to start monitoring how much current is coming and going.

[Reid Welch]

Man, I love this thread. This is pure pioneering-inventor spirit.
Never slacken. I am a fanatic for free thinkers.

r.

[paultrafalgar]

+1 Ground-breaking stuff here! Keep it up! Well done!

[JCG]

Many thanks, Reid and Paul. The only two remaining issues in my mind are cleaning up the wire jumble/covering the rear assembly and dealing with some torque arm issues. There's an imperfect fit with the hub's stator and the torque arm - it spun a little bit by chewing off some of the stator's threads. I think I'm going to machine my own when I get back from vacation next week. I'll try to make it super tight, even if I have to file on it for a few hours.

The real data will start rolling in once I can monitor currents - the Cycle Analyst's separate recording of standard and regen amps will be very useful. I'm also trying to figure out a way to use my laser pointer and a level to get the slopes of the hills in the immediate area of the engineering building.

Happy Holidays!

[newjeff]

AMAZING WELL DONE. KEPT ME SPELLBOUND

[Miles]

I'm also trying to figure out a way to use my laser pointer and a level to get the slopes of the hills in the immediate area of the engineering building.

I find these gizmos handy for measuring gradients: http://grizzly.com/products/H8131

[TylerDurden]

I found this to be fairly accurate:

HFprotractor1.jpg (8.45 KiB) Viewed 2842 times

http://www.harborfreight.com/manuals/94000-94999/94694.pdf

You won't find them online anymore, but the stores may still have some.

They also have a knockoff of the unit Miles linked.
http://www.harborfreight.com/cpi/ctaf/d ... mber=95998

[JCG]

Miles and Tyler - I'll check into those angle measurers. They look pretty inexpensive! That's always good.

New Year, new update. This has been a productive past couple of days in spite of a few setbacks. I recently connected the high-current shunt resistor (rated at 100 A) in line with the negative capacitor terminal to the Kelly controller:



Now, once I put in its correct resistance value into Cycle Analyst, I will be monitoring accurately-measured amps and power during operation. During another road test climbing hills and such, I had another axle spinout. As some people here already know, the NineContinent BLDC motor (http://www.ebikes.ca/store/store_nc.php) is very "torquey," which I love, as that's the best thing for a start-stop type of bike, but it has a very small axle - a 12 mm threaded bolt with the usual 10 mm separated flats milled into it (http://www.ebikes.ca/store/diagrams/MNCF.PDF). This is a great pain. A torque arm is a must, of course, but my application (throttling and regen braking) causes torquing in two different directions. Here is my junk heap:



A: Torque washer supplied with NineContinent kit. Its tolerances were fairly tight, but it was not up to the task.
B: I had higher hopes for the AmpedBikes arms, which I used together, one to resist pulling and one to resist pushing. Unfortunately, even these had a sub-par fit with the axle's flats. Even a little wiggle room was opening the door to failure.
C: I took a bar of steel (3/16", or 4.7 mm thick) and milled my own slot into it at the angle I required, using a hand file for a perfect slot diameter. The steel wasn't strong enough as you can see.

This last failure really got to me because it yanked the hall sensor wires and I had to disassemble the hub motor to get things right again. I took some time and came up with a plan to use the leftover steel bar to make a compound torque arm. Here's the raw stock:



Now, I was going to essentially bolt two flat edges of this steel, which is springy but not as hard as the axle's steel, and apply so much force that it couldn't possibly escape. I should have taken pictures as the machining proceeded but I was moving fast yesterday. Here's the AutoCad drawing:



The bolts go on either side of the axle, and are 5/16" in diameter (strong). Once the bar was milled down to the right width (0.45" or about 11 mm) and the holes were drilled, I was able to fabricate the longer piece with angled grooves at the top to hold the stainless steel worm clamp. The long arm was first twisted 90 degrees about the Z axis (out of plane) and then bent back to align with the fork. The grooves were aligned with the fork's tube, and I pressed a piece of aluminum as a spacer between the steel and the fork as I tightened the worm clamp. Here's the result:



and a closer view:



It is an extremely tight fit, and I have lots of force applied axially (by the axle nut) and radially (by my twin steel bolts) on the axle. It is not going anywhere. Best of all, the worm clamp holds the arm in place by preventing pulling and pushing (from throttling and braking, respectively).

The day may come when the torque arms are "built-in" to the hub motors, but until then there is always brute force.

Next job: calibrate the current readout and find out regen power and current for downhill coasting and various generator loads while pedaling on a level street. Before the axle issues above, I did pedal around for a while and raise my capacitor voltage from 40 to 48 V. Once I can read currents and get elevation changes, the data will have more meaning.

[JCG]

Whoa, an update already. I had to take advantage of the mostly empty streets today and cruise around. I have some neat data now that my shunt resistor is a known and I can now record accurate amps and watts.

First off, I was interested in determining a trend with the bike's speed and the regen amps that would be returned to the capacitor as a function of that speed. To do this, I went down a very, very slight hill (nearly flat) and pedaled to just above the speed of interest with the brake switch on (very slight generator resistance), and used the hand brakes as necessary to get the speed constant. Then, all I had to do was read the displayed current. All tests were with the capacitor module at about 43.6 V.



Everyone loves a straight line. Then, I did a few cycling circuits around a block near the building. It is essentially a rectangle - two blocks wide (west to east) and one block long (north to south). There is a stop sign at each intersection. Here the layout:

Leg 1: Sixth Street south. Moderate slope downhill. 0.110 km length.
Leg 2: Bryant Street east. Gentle slope uphill. 0.234 km length.
Leg 3: Fourth Street north. Moderate slope uphill. 0.110 km length.
Leg 4: College Street west to return to start. Gentle slope downhill. 0.234 km length.
Total distance: 0.688 km.

My riding guidelines were:
1) Stay below 20 kph.
2) Use regen braking to slow down as much as possible before each intersection.
3) Use throttle to accelerate to at least 15 kph out of each intersection.
4) Leg 1: Use strong regen braking to keep speed below 20 kph.
5) Leg 2: Bike at speed up gentle slope, using very minor regen braking while keeping speed up by pedaling.
6) Leg 3: Allow motor to do most of the work, pedaling with only a little effort.
7) Leg 4: Pedal easily with moderate regen.

Power usage:
Leg 1: Average 150 W regen (-)
Leg 2: Average 25 W regen (-)
Leg 3: Average 200 W motoring (+)
Leg 4: Average 100 W regen (-)

Circuit data (all from Cycle Analyst):
Initial Voltage: 42.0 V
Final Voltage: 42.3 V
Forward energy: 37.9 mAh
Regen energy: 52.7 mAh
Maximum Speed: 19.8 kph
Average Speed: 14.9 kph
Trip time: 2 min, 42 s
Maximum regen current: 5.45 A
Maximum forward current: 10.58 A
Minimum module voltage: 41.6 V

The torque arm performed like a champ. Riding around all afternoon like this wouldn't wear me out either. I think that as long as the trip doesn't involve a huge net elevation change, the capacitor can be adequately charged by the person riding the bike.

[JCG]

More torque arm fun... the custom-made torque arm I described earlier failed by bending out of shape along the 1/2" thick stem, so I made another, wider (1") one... which failed later on by spinning out. My stator bolt has taken a lot of punishment during the various spinouts, getting rounded off and losing threads in the process. Since you can't replace the bolt, I was looking at needing to get a whole new motor. Today I tried a long shot idea to fix it up: I drilled a hole from one side to the other of the stator bolt (the side without the wires coming out, from one flat side to the other). I had to take the motor apart to get things lined up right in the drill press, but here's the result:



The hole is big enough to let a threaded quarter inch bolt through (~0.243").



I reassembled the hub motor:



I had also drilled a matching hole in each side of the external torque brace (from Mark II):



Here is everything bolted on, without the spacers (to keep the first failure from re-occurring):



And here's the whole setup, including spacers and the worm clamp on the fork.



The test drive was a success, and if this thing manages to fail again, I quit! Long live Torque Arm Mark III.

[TPA]

I love the super capacitor avatar

[JCG]

I love the super capacitor avatar

Behold the power of Microsoft Paint!

[paultrafalgar]

CJG, this is all very exciting! What we all want to know is what it feels like: how hard were you pedalling? In a sense, you got more than 100% regen. Of course, that's because you are pedalling to add energy. Are you left with the impression that this is either a) a practical ebike with out a battery or b) would make a worthwhile addition to an ebike with a battery? If either of these are true, what is the possibility of others building one of these ebikes - in other words, can one buy the ultracapacitor and is it prohibitively expensive compared with say LiFePO4 cells?
Can you think of experiments which take your pedalling strength and weight out of the assessment? Are you planning any test with a battery as well as ultracapacitor? Any longer trips planned? With video (this will give an impression of your effort)? How about a test against another ebike - one circuit with you on your bike and someone else on a known bike, followed by another circuit where you swop bikes?

[JCG]

Paul, thanks for putting this into words; I have been wondering on and off about all of these things but I haven't taken the time to think in enough detail up to now.

What we all want to know is what it feels like: how hard were you pedalling? In a sense, you got more than 100% regen. Of course, that's because you are pedalling to add energy.

The propotional control of the regen brake is giving me a lot of options here. Some background might be necessary since we're talking about effort... I've been biking through city traffic for distances of around 5 miles (round trip) on a regular bike for about four and a half years now. That kind of training is really more akin to strength training than endurance, with all the stop and go. I also do weight training a couple of nights each week with traditional olympic lifts, incuding squats (I'm not even bronze medal material though ). That gives you kind of an idea of my general fitness level. I feel that I can comfortably (feeling as if I was riding a wide-tired mountain bike) cruise at 20 kph while sending 75-100 W back to the capacitor when I'm on a mostly flat street. For my usual voltage range, this is a bit less than 2 A of regen. Honestly, I don't feel any discomfort or unusual effort. Coming to a stop is capturing things less efficiently, due to phases shorting together (or something like that, I'm still reading about it). I feel I really make my best headway pedaling against low resistance at cruising speed.

Are you left with the impression that this is either a) a practical ebike with out a battery or b) would make a worthwhile addition to an ebike with a battery? If either of these are true, what is the possibility of others building one of these ebikes - in other words, can one buy the ultracapacitor and is it prohibitively expensive compared with say LiFePO4 cells?

I wouldn't have known how to best answer this without some real comparison with the battery types out there; I had no idea that some 48 V battery packs were as heavy as they are. Not counting the hub motor, I've added about 35 lbs total to the bike's weight. I think that compares pretty well with other ebikes (please correct me if I'm mistaken). I would say that both of these [(a) and (b)] are true, but probably I would suggest 48 V for a battery-free ebike (like my prototype) and a 36 V capacitor module matched in parallel with a 36 V battery pack for the mixed setup you suggest, or 24 V of each to save a bit of weight. With this kind of regen, city biking with this bike will mean that I won't ever need to charge from the wall as long as I don't leave the bike sitting around for a month. What I'd really like is a rear-front wheel linked dyno so that I could charge the cap while sitting still before a trip (to top it off). For now I have to stick to through-the-road regen, which means I'll ride around in circles.

The immediate concern closest to my heart is the way things bounce around behind me when I hit a bump. It's all locked down pretty solidly, but it still scares me sometimes.

*Edit: I just realized I didn't answer that last question. A 3000 F, 2.7 V boostcap cell costs about 115 USD from Maxwell. I have 18 of them in series to make my module, which is a pretty hefty price tag (2070 USD) if it was purchased new (mine were free, donated to my hybrid research by the good doctors at Maxwell Tech). Making a 36 V stack wouldn't be as bad, and 24 V would be quite reasonable (at around 1000 USD). The thing is, they will last PRACTICALLY forever (that's for you, Toorbough ) and I expect that the prices on ultracaps will continue to come down as more companies get into the mix. As I mentioned before, they've already come down a factor of 10 (dollars per stored joule) over the past 5 years. I really like the outlook.

Can you think of experiments which take your pedalling strength and weight out of the assessment? Are you planning any test with a battery as well as ultracapacitor?

It's going to be hard to take those values out without a propulsion model. I need one of those to be refined for use in the paper I hope to get out of this, so I will see what I can do. It will probably be a data fit approach; I can add my mass to that of the bike, but quantifying pedaling power is going to take some creativity. I realized that we have one perfectly flat area nearby - there's a running track at the high school just to our west. I am going to sneak out there with some students and get some zero-grade data. I think that a battery-cap mix would be great. I'll take any suggestions of a good, long-life, light 36 V pack that someone could recommend.

Any longer trips planned? With video (this will give an impression of your effort)? How about a test against another ebike - one circuit with you on your bike and someone else on a known bike, followed by another circuit where you swop bikes?

I want to ride the bike back and forth from home soon. I am thinking about getting one of those helmet-mounted cameras that Nashbar offers (http://www.nashbar.com/profile.cfm?sku=21385). There is a guy in EE here that has an older NiCad, external motor-powered bike, we even might be able to race... Thanks as always for the interest & support.

[mace1934]

JCG wrote, about the torque arm struggle:

"if this thing manages to fail again, I quit! "


PLEASE don't quit! We have so much to learn from you and your project.

It's so sad, with all you've accomplished on electric matters, that this mechanical thing is giving you so much frustration. Maybe you need to go with a different motor?

My front Crystalyte 408, without a torque arm, running at 66V on A123s, over hundreds of km, has had no such problems - and I push it quite a bit coming up the hill to my home (but I never attempt a quick acceleration from a stopped position).

Justin, at ebikes.ca, went all the way across Canada on an old mtn bike/extracycle using a front Crystalyte 5304 and stock torque arm - secured by what seems to be a rag (judging from pics) to the fork arm.

Maybe others using Nine Continent motors will chime in with their experiences; seems like the NC shaft is too weak for the torque the motor generates?

(I am currently considering torque arm options for my Big Dummy with a front CL 5304, and I'm grateful that you shared your torque arm experiences. Thanks!)

Larry

[JCG]

Don't worry Larry, I'm slightly more than cautiously optimistic that this time it'll hold! In fact, I'll be going out for a spin here in a few minutes. I just hope no one else has to do that many weird things to their axle...

I wouldn't really say that the NC motors have a weak axle, in fact it may be the strength of that axle that ironically spun through my steel but also means I can still use it... it spun out, and in sequence vented its wrath on my fork droupouts, defeated the torque washer, rolled through the Ampedbikes torque arms (simultaneously), my custom torque bar, bent the heck out of Torque Arm Mk I, pried open Mk II and in the process all it lost was some threads. I think the real complaint is that with a 12 mm nominal bolt diameter, cutting flats to get 10 mm across is just too small of a "flat" area for any passive (non-clamped) torque arm to handle. Compared to the 12 mm axle, the 14 mm axles increase the flat width from 6.6 mm up to 9.8. that extra area can really help prevent spinout! Plus, it's a lot harder to spin a 14 mm rounded edge through a ~10 mm wide slot than a 12 mm rounded edge, especially after the threads are gone. Combine all that with the extreme torque the NC motor can put out, and in both directions thanks to regen, and I was asking for trouble.

I would love to hear other NC users share their stories, though. Good luck with your torque arms, Larry! I hope you have success in Round 1 (not 6, like me )

[E=IR]

Do you know how they make the caps? Are there layers of plastic and foil?

[paultrafalgar]

...

Can you think of experiments which take your pedalling strength and weight out of the assessment? Are you planning any test with a battery as well as ultracapacitor?

It's going to be hard to take those values out without a propulsion model. I need one of those to be refined for use in the paper I hope to get out of this, so I will see what I can do. It will probably be a data fit approach; I can add my mass to that of the bike, but quantifying pedaling power is going to take some creativity. I realized that we have one perfectly flat area nearby - there's a running track at the high school just to our west. I am going to sneak out there with some students and get some zero-grade data. I think that a battery-cap mix would be great. I'll take any suggestions of a good, long-life, light 36 V pack that someone could recommend.

One thought: If you had access to a rolling road you might be able to take out those parameters. That rolling road would have to measure the forces it was applying to your ebike, so you would be able to quantify your inputs and outputs. Some maths for the students to do?
Edit: maybe you could fabricate a rolling road with some rolling pins and use an anchored spring balance to measure the force you were pedalling with?

[JCG]

Do you know how they make the caps? Are there layers of plastic and foil?

Word on the street is that they have foil current collectors with a bunch of acetonitrile-soaked, pourus, activated carbon on top as each electrode. Those layers are separated by something with good dielectric strength (probably a plastic or a crosslinked dielectric polymer, like you suggest). You can find a few diagrams online, but the stuff people are really working on at the moment is likely a trade secret.

Paul: Do they make bike-sized treadmills?!

[TylerDurden]

Paul: Do they make bike-sized treadmills?!

Training rollers?
http://en.wikipedia.org/wiki/Bicycle_rollers

[nomad85]

I would love to hear other NC users share their stories, though. Good luck with your torque arms, Larry! I hope you have success in Round 1 (not 6, like me )

I have a NC front hub as well. I took it off in favor of the rear hub for higher speed, but when I used it my front forks would shake whenever I accelerated without a torque arm. I tried to use the ampedbikes torque arm, but it didn't fit... I ended up using a torque arm from hi-powercycles.com and it made the shakiness go away.I had no problems for over 200 miles @ 48v until I got tired of going so slow and put on the rear hub instead. I think they are both the same speed now though from ampedbikes(slow)

[JCG]

Well I'd say it 's time for an update. I mentioned in an earlier comment that I was distrubed by the diving board-type of effect I was getting with my rear rack mounting of the capacitor and controller, not to mention that it wasn't the most secure setup if I parked the bike and walked away from it. After trying to make a secure sheet metal shell around the rack that was lockable and had rear access (which also scared me since the sheet metal was close to the cap leads ), I finally took Dennis' advice from his first comment in this thread and mounted everything in a trailer! I decided on a two-tire trailer for a bit more side-to-side stability. Here's a picture that shows the bike against a beautiful backdrop of junk in the lab where I've been working all this time. It was clean back in December...



Here's a better look at the front end. All of the controls are still in place, brake twist throttle, brake button switch, CA, thumb throttle from L to R. I cleaned up the wiring a good bit and waterproofed what I could.



And here's the back end. I put a large black plasitc tube to cover the trailer arm, and hid the three wires under it (motor phase power (3 conductors, AWG14), hall sensors (5 conductors, AWG18), and control wire (12 conductors, AWG22)].



Now for a look under the trailer cover. I have a large waterproof "Fibox" (made of tough plastic) with a lockable cover. It holds most of the wire mess and the cap, controller, shunt resistor, and a terminal block. I can see through the cover when it's locked to check the controller for any error signals (LED-based) without unlocking it. The Fibox is mounted directly onto the plastic trailer bottom.



A clearer view inside the Fibox with the cover unlocked and raised. You can see the gasketing edge around the outside of the lid's groove and also the switch I can use to power both the controller and Cycle Analyst on and off. With both of them off, the capacitor stays nice and charged - holds its voltage level very well overnight. Also in this configuration, I have easy access to the programming port (serial) on the Kelly Controller.



The terminal block is hidden behind the capacitor on the left. Here's a peek at it:



It's kind of a monster, 20 terminals with 0.44" spacing. I hope to not be playing around with it much at all. It took too long to figure out where to wire everything in the first place!



I guess the best part is that everything is detachable. At the place where the trailer connects to the rear hub hitch, I have plugs for all 5 of my device sets: brake wiring assembly (switch and twist brake), hall sensors, motor phases, throttle, and Cycle Analyst; I can unplug them all and take off the trailer and just ride the bike like I used to.

There IS room in this trailer for more stuff if I don't use the Fibox. Connecting three 48 V caps in parallel is not out of the question, nor is adding a 48 V battery for parallel connection with a single cap module. Riding with the trailer is taking some getting used to; you can kind of feel it push and pull at you when you are accelerating or decelerating, or taking a curve.

The weather has got to warm up soon, then we can take to the streets with some students and get some better data. I hope I'll have some Cycle Analyst-based data acquisition by then... logging V, A, and speed is going to be a must for my reports. Thanks to all here for helping this project along - especially Dennis for pointing me to the NineContinent motor (which has been great despite spinout issues), Paul and Miles for discussions on regen, and methods for some very critical assistance with the Kelly Controller.

[JCG]

"The Long Ride Home"

Join me if you will for an evening ride through the busy streets of the US capital. The camera's mic picked up a strong signal from the motor! It's about all you can hear for some reason. It sounds like a damn motorcycle.

Initial cap voltage: 36.7 V. Final: 38.6 V. The big downhill stretch starts at 2 minutes in to part 2.

Note: you may see some idiot drivers, pedestrians, and myself "bending" the traffic laws. Please do not attempt. Professional driver on closed course.


Part 1

Part 2

More details coming soon.