Commuter E-Bike from Novara Hardtail MTB

[Alan B]

Have been reading about the Magic Pie hubmotor. Looks like a nice choice for my type of hilly commuting. Probably better at climbing and shedding heat than the 9C (though the 9C is very good). Just get the model for an external controller. A dual Magic Pie (one in each wheel) would really be amazing. A number of folks are doing that.

http://goldenmotor.com/SMF/index.php?board=15.0

[dogman]

OMG, a 2810 on 72v isnt climbing good enough for you? What are you doing? riding up a ski area road?

[Alan B]

Hi dogman. The 9C is doing well. On the slacker's peak run in Marin it got really hot. On my run to work it gets very warm. It is doing the job.

At some point I'll put a temperature sensor in there and get real data. It seems warmer than I'd like, but perhaps it is not as bad as it seems. Slacker's hill was probably near damage level, though. It was really hot for a very long time. Took hours to cool off.

I was thinking we should fund a Magic Pie for you to test. That would be interesting.

[wineboyrider]

Hi dogman. The 9C is doing well. On the slacker's peak run in Marin it got really hot. On my run to work it gets very warm. It is doing the job.

At some point I'll put a temperature sensor in there and get real data. It seems warmer than I'd like, but perhaps it is not as bad as it seems. Slacker's hill was probably near damage level, though. It was really hot for a very long time. Took hours to cool off.

I was thinking we should fund a Magic Pie for you to test. That would be interesting.

Yes! So I can film DogMan climbing up to the Sierra Blanca with smoke coming out of the motors!

[dogman]

Yeah, really steep grades like slackers would be stalling it enough to get real hot. That's why I went to the 2812 on my dirt bike. It does pretty good. I did 70 wh/mi for 30 min continous the other day, and only got the motor up to about 150F. Maybe most of it was less steep than slackers, but I managed to ride an ebike up a certain slickrock part of the trail for the first time ever. About 30 degrees for about 50 yards.

Once hot, it does take hours to cool off a hub though. Big hunk a copper.

If you put a thermometer on it, you'll be able to tell if you are getting hotter and hotter, or have reached equilibrium while still in the safe, under 200F zone.

[Alan B]

Charging setup at work now with balancing charger, power supply, and bulk charger



Upgraded XTR brake lever with stiffer design and adjustable pivot

[Alan B]

The new brake lever went on easily and has great feel.

I've been looking at the Zippy 6S 8AH packs from HobbyKing for a long time. My main pack is 18S2P of 5AH Turnigy, and I have a few more for the 18S3P upgrade, so the Zippy's don't fit into the primary battery plan. But I'm thinking about making a secondary pack for those long weekend excursions that need more range. Three of the Zippies weigh about 8.5 pounds. So not a lot of weight to add. Should be possible as long as it is removable.

So let's see what that would do:

The Stock Setup Now:

18S2P is about 70 volts at midpoint, 10 AH so 0.7 kilowatt hours. At 35 watt hours per mile this is 20 miles. That is running pretty hard.

The Next setup with 3 more Turnigy packs added to the triangle (coming soon):

18S3P is going to raise this by 50% to 30 miles.

And then with the extra range pack tacked on:

Adding an extra pack at 70V 8AH would be 23 amp hours and at 70 volts this is 1.6 kilowatt hours which would go 46 miles at 35 watt hours per mile.

At a more gentle 25 watt hours per mile this 1600 wh pack would go 64 miles.

If you wanted to go 100 miles you'd have to pedal to the 16 watt hour per mile level to make it. At the efficient level of 12 watt hours per mile this setup would go 133 miles.

So I would call this range extended setup good to about 50 moderately agressive miles, possibly a lot more if riding efficiently and pedaling a lot.

Or I could swap things around and make the Zippy pack be the main pack in the triangle, and then have 8AH for the commute to work (just enough), but drop the bike weight by almost four pounds.

So with my battery set I can make 5, 8, 10, 15, 18 or 23 amp hours if the brands can be mixed.

Lots of choices!

I could also leave the main pack at 10AH land make up a 5AH and 8AH auxiliary packs.

Have to give this more thought. But I'm leaning toward 15AH in the triangle and 8AH auxiliary.

Any thoughts out there?

[Alan B]

Awfully quiet out there.

I received a pile of Anderson connectors for the battery upgrade. Plus more for future needs.



I'm considering using the PowerPole 75's to do the paralleling of the 3 packs in each section. Four #12 wires will fit into the PP75 pins designed for #6 wire. So there is enough room for the 3 pack wires (using 4mm polarized bullet plugs) plus a fourth wire to the PP30 that is used for 6S balance charging. This solution will not require a terminal strip, all connections are handled by the crimping in the pins. Mechanically this is very nice. I have different colors of PP75's so the color coding will be used to avoid mistakes in connecting the PP75's. These connections will not be disconnected frequently as charging will be done by either bulk charging the whole 75V stack, or by balance charging each 6S bank with all connections in place. The only time these will be disconnected is when servicing the pack or mounting system.

[Alan B]

Rough Cost Breakdown

Made a rough cost breakdown for this project:

https://docs.google.com/spreadsheet/pub ... utput=html

[GCinDC]

nice rap sheet. i've gotta do one too! might be nice to link the items to source!

not sure how i've missed all this! looks great. now i gotta go back and read it all.

[Alan B]

Thanks for the comments! You've been in this thread before, so you haven't missed the whole thing. It has been quiet for awhile, too.

It is kind of scary adding things up. I will put in more links, did most of this from memory, so it will improve. It is easy to update it in Google spreadsheets.

Received the hi-visibility jacket today. Wow! Wife says it is over the top. Looks great!! Well made. Probably a bit warm for the really hot days, about perfect most of the time.

[Alan B]

CellPro 10S Charger

Last evening I used the new JST-XH adapter on the Cellpro 10S charger to charge the batteries on my son's ebike. (On this first solo ebike trip he used 90% of his 10S 5AH pack with no meter, no LVC, etc! Really Lucky!!) That reminded me how nice the Cellpro 10S charger is. Fewer button pushes to get things done, and it measures cell impedance which is interesting. Not the lowest cost unit, but very good quality. It is field firmware upgradeable unlike most imports. It will do 250 watts output on 15V input so just under 10A on 6S, and very close to 6 amps on 10S. It is also capable of logging though I haven't tried that part.

I have a large screen Cycle Analyst to put on my ebike, and then my regular screen CA will go to his bike. But he was not patient and went out without consulting me. Luckily the batteries were at 3.7V/cell and balanced when he returned. The charger indicated 10% at the start and put back 4.5 amp hours. His next pack will also have 10S2P (double the capacity) when we complete it. But he got his first taste of solo ebiking (we went out together for a short trip once before). He really likes it. Another ebike grin success. Since he doesn't drive ebikes may be a major part of his life from now on.

His ebike is made from all the upgraded parts from my ebike (combined with my old first mountain bike, actually). There's a separate thread on it somewhere around here, but it is not complete. Is any ebike ever complete? Here is the thread:

viewtopic.php?f=3&t=27751

If we make up the right cabling the CellPro 10S can balance charge his whole pack at once at nearly 6 amps. That might be the way to go for him. Hard to argue with that simplicity. A db25 for balance wires and a powerpole for charging current. Simple. Use the CA to do low voltage cutoff and watch the integrated current (amp-hours) on the CA for a gas gauge. Can't get a lot simpler and more reliable than that!

[Alan B]

Upgraded the Cycle Analyst today to the large screen model. Very nice!

The shunt setting from the old CA didn't work by about a factor of two. Recalibrated, need to re-test. My usual ammeters aren't helping here due to the different connectors and high voltage. So I'm calibrating to the recharge amp-hours on the Cellpro 10S. Not the best, but good for now.

Next task may be to set the controller parameters and get regen going.

Thinking about setting the battery and phase current limits in the controller. I don't subscribe to the "2.5 times" or whatever factor rule. There are too many potential issues here that can make those factors wrong, even dangerous. Ignoring this can lead to overheating and smoke.

Factors affecting Safe Current Limits:

Battery Current Limit Factors

* Cell Current (factoring for parallel)(Turnigy 25C 10AH, 250A optimistically)
* Cell cables and connectors (2x #12, 2x 4mm bullets in parallel)
* Battery Cables and Connectors (#6, SB50)
* Controller cables and traces and shunt (#12)

I suspect the battery current limit for this setup is the controller wiring and traces. A #12 wire, type TW, single wire in open air, 86 degrees ambient, wire no more than 140 degrees F, with single layer plastic insulation, is rated at 30 amps.

So perhaps 30 battery amps is a good current limit unless the controller's input wiring is upgraded:

* #10 would be 40 amps (NEC ratings)
* #8 would be 60 amps
* #6 would be 80 amps


Other Battery Current Limit Factors (not equipment protection)

* Enforcing a power input limit for the motor
* Enforcing a power limit for range extension

It is better to use the Cycle Analyst for this since you can easily reprogram it in the field. On the other hand, for some applications like a power limited bike for kids or newbies, perhaps the controller is the right place to do it. In my case 20A for 1500W is about right.


Phase Current Limit Factors

* FETs, accounting for parallelism (2 each 4110's in this 12 FET)
* Traces
* Controller phase wires, traces and connectors (#12) (#12, PowerPole 30's)
* Motor wires cables, coil wires (#14?)
* Adjustment for 3 conductors (1.5x)
* Motor magnetic saturation

The motor saturation is hopefully the limit here. But we also have to worry about the #14 wires. To the first order I'll ignore the traces, hopefully they aren't the weak link yet. If the motor wires are really #14 the rating from the NEC, for single conductor open air, ambient 86F, wire less than 140F as before, is 25 phase amps. May need to beef up those phase wires!

The 30 amp PowerPole connectors in the phase leads, with #14 wires will have a temperature rise of 40C at 29 amps. So they are just better than the #14 wire itself.

Since three wires are used for phase current, and only two at a time carry current so the heating is reduced. This allows a 1.5x improvement over the wire's rating for current. So this would be 25*1.5 or 38 amps.

Magnetic Saturation

Justin's measurements for the torque vs current knee are here:

http://endless-sphere.com/forums/viewto ... =2&t=14494

It shows a drop in the torque to current ratio of 40% above this point:

70 n-m torque
50 amps
9C 7x9 (9 turn) motor
50*9 = 450 amp-turns at the knee

With my 10 turn motor that would be 45 amps. Going above this phase current is quite inefficient, so we don't want to do that.

So the wiring is the limit at just under 40 phase amps. I'll have to measure the wires and confirm that.

If I upgrade the phase wiring and connectors the next limit is the motor saturation at just under 50 phase amps.

Regarding the FET limits, we have a 12 FET so there are two FETs in parallel in each phase. The TO220 package limit is generally considered to be 75 amps, and the paralleling derating factor is 50% of the additional FET for balance, so we only credit half of the second FET and 75 * 1.5 amps. Now we also have the 1.5 factor from using two phase wires at a time, which is another 1.5 so 75 * 1.5 * 1.5. Then half the time we are using the alternate set of hi side vs low side fets, so that gives another factor of 2 in heat dissipation so 75 * 1.5 * 1.5 * 2 which is about 300 amps IF the one FET can really handle 75 amps. If we want to run the FET conservatively we would take 80% of that, or 240 amps. If we want to be ultra conservative it would be half of that or 150 amps. A close inspection of the FET data sheet is indicated to see if further derating is needed. But since we already have a 50 amp saturation limit we can assume the FETs are not the problem here.

So to be absolutely reliable use 40 amps. To take minor risks with the connectors and cable use 50 amps. Going beyond 50 amps is a significant waste of energy and will cause a great deal of motor heating. If this is not enough torque, look into dual hubmotors or driving through reduction.

On the Regen, I need to set that for 75V max. Seem to recall there is a problem with high voltage regen, not sure where that happens. Have to do more searching. Thanks for any tips here!

[gtadmin]

Hi Alan, higher voltage regen with Infineon controller is here

[Alan B]

Hi Alan, higher voltage regen with Infineon controller is here

Thanks! Looks like over 60V requires an R12 mod.

[Alan B]

Here is a sketch of my plans for the battery pack upgrade. It takes advantage of the PowerPole 75 connectors with #6 wire capable crimp pins. This will accept four #12 conductors so we can parallel three batteries and still have a connector for parallel charging of the 6S bank.

The wiring modules will parallel up to three batteries, and then the PowerPole 75's are used to series these blocks up into the pack. This flexible approach will work for 6S or 5S or other size packs (just don't mix them in the same bank).

[Alan B]

I just noticed in the schematic above I showed the controller main power connector as PP75 whereas I'm using an SB50. They use the same pins, but the SB50 is a bit smaller and fixed at 2 pins whereas the PP75 has single pin bodies with dovetails. Either will work but I'll stick with the SB50 since I've already used that on the controller.

[Alan B]

It is amazing. Everyone asks about regen. Even if they know next to nothing about electric bikes, EVERYONE asks a question about it in some form. I suppose it is just so novel that electric vehicles can "make fuel". I'm very interested in getting regen going for a couple of reasons. The main one is to save my brakes. I have no illusions about getting a lot of free charging or massively extending range. Regen just doesn't make that much energy. But I do want to save my brakes, and brake more quietly on the long downhills I have on my commute.

The first question is how to set this up. My controller has this feature. It has a pair of white wires with one pin connectors that plug together to "enable" regen. Then the ebrake connector which has 2 pins needs to go to the brake switch that actually fires off the regen and disables the power to the wheel at the same time.

One question is what voltage this controller will regen to. Many will only go to about 60 volts. To go higher requires modifications to hardware. So I need to check on that.

Good news! - My Controller is rated for regen to 75V with NO R12 mod which is exactly enough. I need to try that soon then! I probably do need to load new parameters into the controller but I have the gear to do that, which is software and a special serial cable from a PC.

One question is how to handle the regen control (ebrake input). Put a switch on the brake handle, or the rear Vbrake itself, or mount a pushbutton or control lever to actuate it.

My favorite is probably to mount something on the rear Vbrake itself. The question then is how to do it clean, simple and reliable.



Before I do anything fancy on the Vbrake I will set up a quick test and try it out.

One other concern is torque arms. I have two, plus NordLock washers, but I do need to watch this and make sure it is not working back and forth. Regen causes forces on the torque arms opposite the direction of motor power, and forces in both directions can work things loose and that is a major problem. There are two levels of regen selected at programming time. I will keep the regen at the lower level and verify things are not getting loose or working back and forth.

For the initial test some kind of simple thumb button near the throttle would be good. This would allow the left hand to be free to work the front brake. Ideally the control would be below the right hand bar and in a position where the right hand fingers are still on the rear brake handle. Then one can control all three at once. Have to see what I can cobble together.

One other thought. Should the kill switch actuate the regen? Or just kill power to the controller?? That will be on the left hand side when I get to it. More later..

So many choices...

[Racer_X]

hej Alan B, how about using one of Ebike.ca's new Half Grip Throttle with Regen Button?
T-HTwist_B_Upgrade

[kold kanuck]

Glad to see your still getting on with this.I would possibly lean more towards the brake activated regen with a little free play in the cable. Initial pull would activate regen without apply brakes themselves and if you need more whoa just pull more. If this makes sense.

[GCinDC]

Do you coast much? If not, set regen to activate when throttle off. (It may even be programmed that way by default - can't remember setting.) I love it that way, cause the way I ride, I'm either WOT or braking... And I didn't think I'd like it before, but I do. Works just as well in the rain too, unlike most vbrake pads. And it kicks in faster than reaching for the brake, like wearing a small parachute or something. When I want to coast, I turn the throttle 1/10th of a turn... Oh yeah, below 15 mph, regen doesn't kick in, so it's not annoying...

You got a BMS on there? Might be an issue.

[Alan B]

Great comments folks. I took a little break from working on this project, but I'm getting back to it. Have two ebikes in our stable now, and my son is probably going to use his a lot more than I use mine, and so that is driving some of this. Also I have done a few commutes now and am keen on making improvements for that. The rains have started, but our rainy season is very variable so we'll still have plenty of good weather for ebike use.

I like that throttle with the brake button built in. Really nice integrated solution. Have not seen that before. I also like my Magura throttle, wonder if I can do something similar with that. Having it out of sight is also appealing, mounted on the Vbrake arms somehow. Another thought is to take a thumb throttle and use the mechanism for this, mounted inboard of the Magura throttle. Would really be nice if it was variable regen. Someday I'll do my own controller and will provide for that...

I do coast quite a bit, and if you are peddling you might want the ebrake off. Of course a switch somewhere could change that mode, might be interesting. I think there is a software setting and a jumper that talk about the slip regen/throttle regen feature. OK, you say at normal pedal speed regen doesn't do much anyway? I thought it did something down to 5 mph or so, but we'll have to try it on my setup to see exactly where it stops. Perhaps on these high voltage setups it just doesn't work to as low a speed. With my higher KV motor it might regen to a somewhat lower speed. It would be fun to try that throttle regen control, did you have to put a jumper into the controller as well as programming the feature? So far I have not taken the controller apart. But if I can try it with only software changes that is more appealing.

I did order some of Methods new LVC/HVC pre-production boards. Very nice design, epoxy coated, very low power consumption. Also some 4110 FET's for the keyswitch board. Still adjusting my design. Really want to get to one switch for everything. Working on a new keyswitch version that has three positions similar to a car ignition: OFF - ACC - ON. In the ACC position the Controller, Cycle Analyst and accessories (such as lights) would come on, but the throttle would be clamped off. Then in the ON position the throttle would be active. This would allow turning on the CA, lights, etc without enabling the throttle which is handy and safer than making the bike 'live' when that is not needed. I've got the keyswitch, just need to make the electronics...

[Alan B]

I dug up the thread that Justin announced the results of his torque testing on the Nine Continents hubmotors.

http://endless-sphere.com/forums/viewto ... =2&t=14494

It shows a drop in the torque to current ratio of 40% above this point:

70 n-m torque
50 amps
9C 7x9 (9 turn) motor
50*9 = 450 amp-turns at the knee

So the place to operate is at or below 450 amp turns, which on a 10 turn motor is 45 amps. Going above this the torque increases 40% less per amp than below it, and the extra power goes into extra heat. We have enough heat already.

I updated the post on page 31 about 2/3 the way down where the limits were analyzed.

[Alan B]

Future Upgrades

It is always interesting to play with the motor simulators. The upgraded simulator at ebikes.ca allows easy comparison between two motors. I compared the Nine Continents to the new Crystalyte light motor, HT3525. They are not too different. I would consider the Crystalyte if I was starting from scratch, but the difference may not be sufficient to change for. Probably the same goes for the Magic Pie, though I have not simulated that one.

I keep coming back to the trip in Marin where I had insufficient torque and overheated my motor. How to improve that? Either a major upgrade to the rear motor, which is not simple at all, or 2 wheel drive.

The most appealing thought is to add a front motor. This can be removed when not needed, since I don't do this kind of riding every day. If this front motor were small and light and had a freewheel I might be able to leave it even when it wasn't being used. So how about a gearmotor like an EZEE or BMC torque? Use a six fet 4110 controller set to limit the current so it will produce a lot of low end torque, but not consume a lot of power to keep from overheating it or overpowering the front end??

This would DOUBLE the available torque which would really help out on the steep stuff, and keep the 9C from overheating on those steep dirt hills.

[Alan B]

Did more simulations at http://ebike.ca/simulator/

To compare the 6x10 9C with the HT3525 requires some fiddling since the simulator is missing the 6x10 wind. So you have to scale the voltage and use the 9x7 2807. That is 28mm magnet and stator width and 7 turns. The 3525 is 35mm magnets and 25 kph at 36V.

You would expect roughly a 35/28 improvement in torque per amp, which is 25% if all else was equal which it never is. There are many other factors including turns, magnetic gap, magnet strength, etc.

Conditions of the comparison:

70V 70 milliohm 10ah battery on 3525
49V 50 milliohm 14.3ah battery on 2807 (adjusted to behave like 2810 on 70V)(adjusted to same watt hours as the 70V pack)
40A 4110 controller
26" wheel mountain bike 125kg

I wish Justin would put the 2810 into the simulator. Would make this so much easier and avoid mistakes.

How does top speed compare?

26.3 mph for the 9C vs 31 for the 3525.
Range at top speed is 20 vs 16 miles.

How does efficiency at 20mph compare?

Adjust both throttles to 20 mph. Range is 32 miles for the 9C and 29-34 miles for the Clyte. I call it a draw with the 9C just leaning ahead.

How does climbing effectiveness compare?

Set grade to 15%.

9C shows 9.4 mph, 3.4 miles range, 3.1 minutes to overheat
vs 17.8 mph, 4.4 mile range, 4.4 minutes to overheat for the Clyte.

A little extra torque really helps here. This grade is tough for the 9C.

How about the HS3540 on this test?

16.4 mph, 4.1 miles range, 4.1 minutes to overheat. Not great. So much for fast winds.

The HT3525 is well suited to this task, for a few minutes. It may be worth considering an upgrade at some point. The HT3525 performance matches very well to my goals. It provides more climbing torque and also provides slightly higher top speed, which is not excessive at just over 30 mph. Too bad nobody stocks them. Hopefully that will change.