RFC: Electronic Solution for Perfecting Friction Drives

[mwkeefer]

Hello all,

I'm not here dissing friction drives, nor am I an expert by any means... just thought I would start a thread to compile some information.

Since I am considering a friction drive but use dd and geared hubs + RC direct link at the moment, I have been following to some extent the work of Kepler and EVTodd on their evolving designs.

Assuming people will want to equip (or can order) hobby type motors with halls (or an addon disc optical sensor type) many might want to look at the < 100.00 infineon controllers (6-12fet stock) as a very viable option to the HV110 / HV140 because they have proven capable of high electrical RPM and they are stable when setup properly...

Even with an RC based setup and controller, there still exists to my knowledge one serious issue - that of potential drive traction loss, the possiblity of tire damage or destruction in such an event (heat, abrasives, etc) and just as important the issue of traction loss actually effects all drive systems so any solution should be applicable elsewheres (ie: front hub motor wheel spin).

I am sure there are other possible concerns for longevity or things those of you using systems would like to see incorporated into an electronic and perhaps programmable black box type solution, so here I am asking for your input...

Anything you can offer for insight would be appreciated!

As of now, I list the following as issues which need addressing (for safety on friction drive but for all ebikes / levs in general)

1.) Traction control - possibly detection of loss and temporary throttle disable for 1/2 second (or programmable interrupt duration) - I have a functional system for detecting slippage on Direct or Geared Hub Motors but... I have no friction drive (which is a fully different beast) to test and see if it will need modifications. It was initially designed and is used now very similar to a car's traction control... it monitors driven wheel (dd motor) and rear wheel RPM and finally current... When it detects the front (driven) wheel moving faster than the rear then sees the current dropping - we have slippage huston... not even about 1/2 second of slippage before it causes the throttle to cut in 1/2 for a second before re-engaging full - lets me know and regain control. Works great for improving the efficiency of DD front wheel drive hub motors.

I see no reason why we cannot monitor 1 hall line (or use one of the 10$ EagleTree BLDC sensorless RPM sensors - connects to 2 of the 3 phase lines and outputs a single hall pulse for each commutation {shit I just realized a properly timed divide by 3 (to generate 3 outputs for each input) would provide hall sensor input for RC motors at a 10.00 cost - maybe} and the rear wheel RPM to very quickly detect and correct for slippage... within perhaps 2 motor roller revolutions, maybe 5 depending on motor speed, I could even see as many as 10.... this would still be better than 5 seconds of spin away right?

2.) True Programmable Throttle Curves - This alone (based on whomevers addition of a High and Low switch) would go a really long way to extending the life of the tire and roller... I see this as being`paramount to proper engagement of a roller drive (see I'm working on it) so that you never apply too much torque in a dry condition which if spin occours - that will smoke up and damage your tire (my only friction drive ever did this, then since it was spring loaded and the throttle was stuck... it destroyed the wheel too.).. in wet conditions it could very likely prevent the slippage (with a good roller) provided proper response curve was programmed into it.

To keep it simple(ish) I would just divide 0-4.7v (throttle range of hall throttle, ball park) by 254 and divide the possible values into as many (254) segments, meaning the same number of steps. I would think on Throttle Change, get the A2D value of the throttle position (a number between 1 and 2047) and store it as target (this is triggered on interrupt of throttle value changing, so it will adapt with dynamic changes to throttle) throttle setting...

Then I would check the previous throttle setting and determine throttle up or down, down should be instant... Up would require... A grab of the ramp interval (time between segment increases) durration, then I would enter a timer loop and each trigger increase the throttle output to controller voltage until it's the same as target throttle voltage.

That would provide us all (I can make it work with PWM type RC throttle signal in and out or Hall or Resistive) with a programmable throttle curve (and no reason we can't have more than one curve given sufficient onboard memory for profile variables - power switches).

Okay so those are my opinion of the remaining issues with Friction Drive control, anyone else (Todd, Kepler?) want to weigh in on other issues you have seen or can anticipate - I realize your issues may be different but I see no reason to create a solution for just one or another type when it's just as easy to do both.

Thanks in advance all, if there is no interest in this... I will simply abandon it as a serious project.

-Mike

 

[Kepler]

Mike,

Mechanically, I think we have these friction drives fairly well refined and one of the beauties of this refinement is the simplicity. I think it is great that you are looking for solutions to make the drive for want of a better word "idiot proof," but we need to be mindful of not over complicating things. That being said, electronics are not my strong suite, however it does appear to yours and as such look forward to any solutions you come up with.

My more simplistic approach to the issues you have mentioned is to firstly accept the limitations of the system and accept that the drive doesn’t work if it’s wet and that you must be moving at a minimum speed before engaging the drive.

What I have found is that a steel non abrasive roller i.e. an outrunner can, grips incredibly well against a rubber tire in the dry with these variable geometry designs. I program the ESC for a gentle ramp and if accelerating from a rolling start, it doesn’t slip at all with the 1.5kW max I use. However in the wet, its slips even under very light throttle and as such is kind of self protecting.

I personally would like to see a simple device that senses the wheels are actually turning and stops the drive starting until a pre set speed is reached. In my opinion, this would provide nearly all the protection needed.

So Mike, can you build me one of these? All it needs to do is hold open a contact until a safe activation speed is reached.

 

[EVTodd]

I do like the idea of traction control. I was originally thinking about trying something similar with an encoder wheel but I came to my senses and realized I know just enough about electronics to not be able to make it work. After I tried the sliding mount I kind of gave up on the idea. If you can come up with something simple using hall sensors I think it would be very cool. As people try higher powered motors slipping may become an issue again.

Speaking of slipping, what kind of friction setup were you using that destroyed a tire and rim? I can't imagine mine eating through a tire unless it was just sitting in one spot slipping for a long time. Mine also couldn't reach the rim because of the plane it moves on. You must have been outputting some mad power with that thing.

I have another TowerPro 5330 motor here that I keep thinking about mounting sensors to but (knock on wood) I haven't had any problems with my Castle HV85 so far, even from a dead stop. Like Kepler I'm trying to keep things as small and tidy as possible so I dread going back to an e-bike controller. I do have one sitting here though and I'll probably try it some time. Then again, I want to get my hands on a cheap rc esc too because my theory is they will work just fine with the system I'm using.

I also really like the idea of having different throttle maps. I've kind of done that mechanically too. With the throttle I made I can change the gear on the potentiometer which essentially gives me different ramp up speeds. I've also played around with different pots with the same results.

 

[mwkeefer]

Well in friction drive (sorry) the traction loss isn't between ground and tire but roller and tire...

Now with your Jackshafted (sort of) roller... we could simply attach a hall sensor to the mounting platform and you could dimple the shaft to epoxy in 2 little neod. magnets... now for each 1 RPM we get 2 pulses from the hall sensor...

Then a simple spoke mounted magnet on the rear wheel with another tiny hall sensor and we have wheel input RPM

All that remains is a proper calibration of tire diameter (not just wheel diameter, very important in the friction drive perspective) and whalla... we can tell when slippage happens within 1 revolution!

Thing is I love the self engaging stuff but lets be brutally honest... if we solve the hook up traction problem with kurled rollers and using yours and Keplers, let's call it self adjusting drive system (like a timing belt) then what happens when someone wants to push 6kw through it? Well really... with slip detection we can determine the break away point of current for a given wheel speed, meaning we could automatically calibrate for any condition... and recalibrate as needed (each occourance of slip)...

The ability to do adjustable advance curves is a no brainer, and again... considering maximum torque with electric is at 0 RPM... don't you want to optimize hookup through the lowest end accelleration, get as much out of it as possible but without inducing slippage and being able to handle it if slip occours?

I am kinda hoping you guys see other issues, complexities and such that I could assist with solving with such a device.

I will not be selling the basic version but offering instead the PCBs when / if it comes to pass with a BOM... artwork can be used at will for non-commercial purposes and with consent for commercial reproduction.

The idea is to solve several remaining issues...

One thing... I wonder if (I know it would add complexity) adding a latch or pinion against the unit to force it to stay in place could be incorporated and triggered electronically to activate regen braking... again without slippage there would be next to no wear and very much positive braking action.

Also... weather related and road related conditions... what to do if it starts raining (ever try it in the rain, lite pedal assistance from you?) - and when sand gets on the tire (or gravel or whatever) what's the potential towards damage? Finally... maybe just a little sweep brush mounted ahead the roller to get the crud of the tread before passing the roller?

-Mike

 

[EVTodd]

I understand that you mean slipping between the roller and tire.

As for hook up traction loss at startup. I just don't have that problem. I can pretty much jump on the throttle at any time whenever I want and I get very little to no slip from the roller. In fact, it can pop the front wheel up if you get on it too hard. I have hundreds of miles on my rear tire with very little wear from the roller.

As for riding on dirty roads, or sand, or gravel. I've had zero issues with this. In fact, I ride dirt trails around my area all the time. Again, when they're dry. The mount gets dirty and you would want to clean the bearings and sliding surface from time to time but there has been no issues with sand or other objects getting caught in the roller.

As I said though, I would still be interested in a simple traction control for the roller especially as I move up to higher powered motors. I'm still not convinced it's needed yet though. I could simply machine longer grooves on top of my mount so it slides back further thus putting more pressure on the tire as you accelerate.

Honestly, now that I think about it... Traction control might be a problem with my setup anyway. As you accelerate, the mount starts to slide back to tighten against the tire. If a traction control system sensed slipping it would cut the motor out for a second which would stop the mount from sliding back. I could see it defeating the purpose of the sliding mount.

 

[mwkeefer]

Todd,

The motor mount sliding back on accelleration would not be detected as slippage becase the minimum trigger would be a 1/2 turn or 1 full turn, your not actually slipping there but rather motor is climbing in effect.

-Mike

 

[EVTodd]

Todd,

The motor mount sliding back on accelleration would not be detected as slippage becase the minimum trigger would be a 1/2 turn or 1 full turn, your not actually slipping there but rather motor is climbing in effect.

-Mike

I think I worded myself wrong. What I'm trying to say is if the roller slipped on the tire during the process of the mount sliding back a traction control system would stop the roller for a split second causing the mount to pop forward again. I don't think it would be productive to use both systems at the same time. In my mind I don't think they would work together at all.

 

[mwkeefer]

I do see what you mean now, that would / could be handled by instead of using an interrupt which would cause it to fall back down a bit... just a throttle back (not entirely, perhaps 5% each cycle) until traction returns which it will then all is good again?

-Mike

 

[mwkeefer]

Kepler,

Long as your cool with adding a normal bike computer magnet to a spoke on the driven wheel, sure enough ... that's actually just a pulse counter trigger, solid state or MicroController?

-Mike

 

[Kepler]

That’s the sort of setup I am looking for, just like a bike Speedo pickup setup. Solid state or micro processor make no difference to me. Just needs to be very compact, cheap to make, and easily repeatable. Also some way of tuning the start speed is needed. If its just a simple solid state device, maybe you could just point me to a simple circuit diagram or quickly draw me something up. My electronic skills amount to making a servo tester using a 555 IC and I can change out fets on my hub motor controllers without too many problems. Hopefully that level of electronics’ skill is enough for me cobble something together from a circuit diagram.

I like the path you are going down in regards to traction control and can see this could take contact roller drives to a whole new level of sophistication.

 

[EVTodd]

I don't know... Maybe it's time to try more powerful motors to see what the limit of mechanical friction drive really is. I really really don't mean to dismiss anyone's ideas because I do think traction control is valid... But, I just don't have the the problems that are being addressed by it.

Maybe it's time for me to try an Astro or a large Turnigy motor and document everything. So far people only have anecdotal evidence from Kepler's and my build. It really does work incredibly well if done right. I think people need proof now. Then again I'm not sure it matters, like I've said before, people want exotic drives and they don't see friction drive that way.

 

[deecanio]

not sure i agree with you there Todd, i'd be up for a friction drive in a flash, but have no concept where to start on one, imo friction drives can be the simplist way to get what you want but as you say we only have yours and keplers thus far to go on, its not something thats easy to do whereas there are plenty of rc setups to follow for noobs.
show the way and ppl will follow

D

 

[Kepler]

I can see high powered friction drives as in 3kW plus being more than achievable with agressive contact rollers, progessive pressure geometry, and perhaps electronic traction control. Not sure if I see the point though. Once you go to all that trouble you then need big batteries, super ridgid mounting systems and then your back to the weight and complexities of a reduction drive or high powered hub motor. Thats not to say I wouldnt like to see someone put together a high powered contact roller system but I cant see myself going down that path.

For me, a contact roller drive is all about not even realizing its attached to the bike most of the time. Now I am not anti power and love hooning around on my 100V 5kW hub motor setup but what I like even more is riding my efficient and light weight hybrid with an ocasional 1kW boost.

I ride some serious kms with a group of strong conventional riders. I can hold my own most of the time but these blokes are strong riders and leave me for dead up some of these hills if I don't use the drive. However with a mere 80whrs on an average 50km ride, I am right with them at the end. We have a great laugh as me with my chicken legs power past them up the hills. They have nick named me the Terminator.

 

[EVTodd]

I think we all agree for the most part. I'm with you Terminator. For myself I really wanted a simple, lightweight drive mainly to get up a couple of the huge hills we have around here (especially after a long ride). But I think if I build a crazy powerful drive it will show that the limitations of a system like are mostly caused by poor designs or just don't exist in many cases.

Then again, I don't plan on making these drives for a living so I don't really have that much to prove to the world. :lol: To be really honest I kind of want to build something crazy just for fun. Come on, you gotta have a couple projects going all the time!

 

[Kepler]

Not trying to discourage anyone from building a high powered drive and I recon your'e are the man for the job. I agree, we all need on going projects. I have got my hands very full at the moment, believe me, and spend way too much time in the workshop. Lucky my wife is reasonably understanding.

 

[plutonet]

That’s the sort of setup I am looking for, just like a bike Speedo pickup setup. Solid state or micro processor make no difference to me. Just needs to be very compact, cheap to make, and easily repeatable. Also some way of tuning the start speed is needed. If its just a simple solid state device, maybe you could just point me to a simple circuit diagram or quickly draw me something up. My electronic skills amount to making a servo tester using a 555 IC and I can change out fets on my hub motor controllers without too many problems. Hopefully that level of electronics’ skill is enough for me cobble something together from a circuit diagram.

I like the path you are going down in regards to traction control and can see this could take contact roller drives to a whole new level of sophistication.

Mike:
I just found a few things that might help on designing that rpm limiter. Both could use the speed sensor from a bike speedometer. One is a discrete circuit that could provide both a low and a high rpm cutoff. The other is an Arduino. There is a DC clone from adafruit.com for about $15 that may work. I don't know if we would need any kind of bounce detection, i.e. in reed switches.. maybe hall will work. One advantage of the Arduino approach is it could be expanded to measure volts, amps, temp, etc. (AC version). even have Bluetooth to connect to an iPhone etc to show battery levels etc. (future stuff). The programming is pretty straight forward. Since it is an open platform, may attract more ideas. The other is a discrete circuit I found at http://www.seekic.com/forum/22_circuit_diagram /25990_BIKE_SPEEDOMETER_CIRCUIT.html could just use the tach chip and trigger the limiter based on voltage out. Don't need the IC4/5. Ideas??
Joe

 

[plutonet]

One more thing on that low rpm limiter for Kepler. I was reading the data sheet on that LM2907 http://www.national.com/an/AN/AN-162.pdf On pages 8-9, there is a very simple example that looks like it might do it with the one tach chip with its variable latch out. I am not sure about the lower limit of rpms of the wheel we need, but it looks like it works below 10hz. Kepler, do you know about what the motor rpm should be to start?
Joe

 

[recumpence]

I really like the idea of friction drives. I have considered it myself. Also, I am all for 1kw drives. The only reason I build big power systems is because I can. But, my next build is a relatively modest power system that is also quite simple in execution.

It does seem however, that the amount of work involved in buiding these one-off friction drives is very similar to a chain drive system. So, to that end, there is not a huge difference. The weight is the lowest attainable, though.

Lastly, as far as the high powered friction drives, there will always be nay-sayers until someone builds one to test. I got the same thing with my recumbent. I finally videoed it just to prove it does run and run well.

Friction drives will always have their place. I especially like the idea of using the can of an outrunner as the roller. That is perfect!

Matt

 

[EVTodd]

It does seem however, that the amount of work involved in buiding these one-off friction drives is very similar to a chain drive system. So, to that end, there is not a huge difference.

I think it only took me an afternoon to build my mount and another when I switched it over to the sliding system. Waiting for bearings was the biggest issue for me at the time. For some reason you can't find anything here locally and I live in a fairly big area.

You can keep friction drive real simple if you want. I mentioned it before but there are only around 20 parts on my drive and that includes hardware. The one thing that helped me is that I purchased a few ev warrior rollers when they were still for sale. They would have taken a little extra time if I had to make them from scratch. I also just kinda threw mine together to see if it would work. I never took the time to make it look nice.

I'm honestly hoping someone eventually has an e-bike race in the Midwest. If that ever happens I will probably follow through with my plans to make a crazy fast version.

I think you should build a friction drive Matt. I would love to see what you come up with. That's my vote for your next build.

 

[deVries]

.I have another TowerPro 5330 motor here that I keep thinking about mounting sensors to but (knock on wood) I haven't had any problems with my Castle HV85 so far, even from a dead stop. Like Kepler I'm trying to keep things as small and tidy as possible so I dread going back to an e-bike controller. I do have one sitting here though and I'll probably try it some time. Then again, I want to get my hands on a cheap rc esc too because my theory is they will work just fine with the system I'm using..

Hi Todd, are you using batteries that can't do a high C rate, enough, to blow the ESC, so you are protected by the limited battery power? What's your battery set-up and volts now?

If one does have a high C rate battery, don't the ESC have a programmable current limiting feature that would protect itself from a blow-out? Or, is the solution done with a CA, fuse, -or- how about a 150amp+ ESC? Would a high-amp ESC be protected by virtue of its high amp rating, so one could do a full throttle from stop (ideally using Mike's non-slip device) and not burn things up during initial acceleration?

Also Mike, can an RC ESC be programmed (with its programmable feature set) to speed-up the acceleration using this method you described below or maybe the Infineon? ...

I've played with it extensively...

The purpose in the eyes of eBike designers is quite simple... Although some hub motors are really torquey, most just arent... to that end, you need the most torque when launching from a dead stop and so...

An engineer decides what wattage and determines the motor's torque output based on 1355 / No Load KV * Amps... Then depending on the requirements, for instance a max load weight of 300 lbs the engineer needs to find a way to get that mass (the 300lbs of you bike batteries and motor) moving along quickly... though accelleration takes more power, it only needs that power for a short time... hence Block time is supposed to be the delay in seconds before the limiting kicks in and brings you back to the programmed primary current.

Another bit you may or may not have realized is the purpose of Phase Current and why it is different than Primary or Battery Current...

At lower speeds the controller can multiply the current at the cost of voltage which is only needed (the higher voltage) once the motor is trying to gain speed.... That's why people with 9C usually recommend 2.5 X the Primary Current for this setting and it's the current your phase windings are slammed with on startup... the block time allows for a momentary surge of Primary Current to enable the output which is current amplified (the FETs do this) which in a 26" DD 9x7 Rear (Loaded kV of 10.10 @ 48v - load rating 100KG / 220lbs) - these motors don't have enough torque at 45A to get you moving and thus reduce duty cycle and power handling of the FETs, phases and every other component... then when you approach speed... you will see your current taper down because "An object in motion stays in motion unless acted upon by an outside force".

By comparison - I have found that using Phase Current of lower multipliers of Primary Current like 1.5 X will result in higher top end speeds because the voltage sag is less in the current amplification process.

If your using geared hubs... these are the most fun, here I begin at 1:1 with block time at about 5 (I do run 69A limit on a 9FET shunt soldered and reprogrammed infineon with stock fets and caps) - I can't keep the nose down and my top speed reaches 30+ mph where as when I had the Phase Current at 2.5 the max speed was 27mph... Next I increased the multiplier until my top speed on a flat without wind dropped measurably... This gives me the absolute best combo of slamming accelleration, higher top speeds and better efficiency since the motor spends less time at low inefficient speeds.

.

 

[EVTodd]

deVries,

I'm still using 3 Bosch Fatpacks. I'm not sure what the c rating is on them. I know they can do high discharge which is why I grabbed them when they were cheap. Maybe we can get one of the battery gurus to chime in on the discharge capability of these batteries.

I keep saying it but I HAVE to get my hands on a watts up meter so I have some real numbers.

 

[Kepler]

I really like the idea of friction drives. I have considered it myself. Also, I am all for 1kw drives. The only reason I build big power systems is because I can. But, my next build is a relatively modest power system that is also quite simple in execution.

It does seem however, that the amount of work involved in buiding these one-off friction drives is very similar to a chain drive system. So, to that end, there is not a huge difference. The weight is the lowest attainable, though.

Lastly, as far as the high powered friction drives, there will always be nay-sayers until someone builds one to test. I got the same thing with my recumbent. I finally videoed it just to prove it does run and run well.

Friction drives will always have their place. I especially like the idea of using the can of an outrunner as the roller. That is perfect!

Matt

I agree that a one off build of a drive like my design is probably as much work as a chain drive. for me though it was always about minimum impact on the bike and was well worth the time involved.

 

[recumpence]

Personally, I like the radical light weight of these setups. My recumbent weighs 47 pounds with pack. That makes it still pedalable. But, these things ahve got to be more like 37 pounds. Besides, if you are only looking for aid up a hill, you do not want a heavy system slowing you down when you pedal.

Matt

 

[robert.smu]

It's great idea, I think about it as ABS in car - it keeps tire on road without skidding. It has one big advantage: static friction is higher then kinetic friction: http://ffden-2.phys.uaf.edu/211_fall2002.web.dir/ben_townsend/frictiongraph.JPG