Observed trials bike designing

[bikerpete]

Virtual Flywheel
I've been considering this idea more and have some ideas that might, possibly work. Not entirely convinced but enough that I've started working on some of the detail.
My plan is a CAN enabled Arduino between throttle and controller.
Controller set to no throttle smoothing.
Micro handles smoothing. Acceleration similar to usual smoothing regimens.
Closing throttle kicks in a routine that calculates the effective 'virtual flywheel' energy at present RPM. That drives a max current limit (throttle value actually) proportional to energy. Note that energy in a flywheel increases by square of rpm.
Also derived from energy is RPM decay rate. This curve is adjustable for experimenting.
The micro reads motor RPM off CANBus.
Now the micro feeds the controller throttle signal to maintain the RPM decay rate. Note that the goal is a constantly decaying RPM regardless of load. This signal is constantly recalculated as long as the throttle is closing or steady.
If the wheel encounters load the micro will increase throttle signal up to the RPM calculated threshold. If load exceeds the ability of that signal to maintain smooth RPM decay the motor torque will be exceeded and the motor will slow more rapidly.
If load is decreased (going airborne say) then the micro will cut throttle signal to try to maintain the rpm decay rate.

My thinking is that this will emulate some of the key features of a physical flywheel. Using a basic energy in a flywheel calculation to drive the throttle is, I think, the key that might make it behave sensibly.
The other key factor is how fast the controller can actually respond. I rather expect that even with all smoothing turned off or set to minimum values there will still be a degree of smoothing to overcome noise, jitter etc. I'll be trying this with a Nucular because that's what I've got that can provide CAN. Maybe another controller like ASI BAC or Silixcon or even CAN enabled Fardriver might give different results.

My Arduino skills are starting from nothing, so who knows if I'll be able to write something that works enough for initial testing. I've started with Simulide to at least see if I can cobble together a few routines into something vaguely promising.
I'm thinking if I make enough progress to see that it might actually eventuate I'll probably start a separate thread for this.
If there's someone who's a whiz with Arduino who wants to help ....

 

[speedmd]

Thinking about it helps also. I went from regularly nearly going over the front to full whiskey "let goes", and after being off the bike for a few weeks to almost getting it to a small degree. Love the effectiveness and the volume of power you can put down when its timed correctly.

Watching the masters, Its the thump delivered by the clutching that is becoming more clear to me. The energy input "thrusting" needs to be more like bouncing a soccer ball off of the foot. They just blow me away as to how in tune they are with the motions.

Noticed a few warming the tire and clutch in setting up for a power move.

 

[bikerpete]

Timing is 90% of trials. Technique is the other 90%.
The clutch doesn't need 'warming', they're just cleaning & warming the tyre for those smooth artificial obstacles. They do it in natural events too just to clean the tread.
Oh that I could feel the difference!

 

[DingusMcGee]

bikerpete,

The Arduino breadboard method (big open space) is also employed in my emulation conception. I have had several CanBus but all were short lived — besides you can get/calculate RPM off the Hall wires with Arduino.

Without a multiplexor board you have some 14 digital terminals posts for input/output wires. Reprogramming ( via Bluetooth or RS232) does not consume these I/O posts.

Inputs: Halls 3 , throttle input 2, clutch pot input 2, throttle signal to controller 2 and common NEG. = 10 pins

Standard aurduino has 6 I/O pins for analog signals. Arduino MEGA 2560 has 16 analog inputs or use multiplexor

See: https://www.amazon.com/ARDUINO-MEGA-2560-REV3-A000067/dp/B0046AMGW0

$48.90

Conveniently Arduino analog I/O operates at 0 - 5V DC which is the same as most controllers.

After finalizing the number of I/O pins and circuitry needed one could eliminate the breadboard and miniaturize the micro circuit box — water proof?

Getting started: Temu | Explore the Latest Clothing, Beauty, Home, Jewelry & More

 

[amberwolf]

I can't offer much advice on the Arduino stuff yet, as I am just beginning to learn it's coding, etc., and experiment with them for my Snuggles project, but I've found there is a huge range of hardware out there for them. Many of them are dirt cheap; I've acquired a handful of different versions, and a few Nanos as they are the "simplest" to deal with and one of hte most commonly used in demos and tutorials (along with the Uno).

Depending on the specific Arduino you use,

Arduino Hardware

Open-source electronic prototyping platform enabling users to create interactive electronic objects.

www.arduino.cc

Arduino Docs | Arduino Documentation

Browse through hundreds of tutorials, datasheets, guides and other technical documentation to learn everything about your Arduino products.

docs.arduino.cc

you can have more or less features, tiny sizes, etc. Once you finish development with something (even a bigger one like Uno, etc),

UNO R4 Minima | Arduino Documentation

The Arduino UNO R4 Minima is the first UNO board featuring a 32-bit microcontroller, the RA4M1 from Renesas . It is faster, has more memory and has a number of built-in features such as a DAC , RTC…

docs.arduino.cc

then depending on the I/O you require, you can move the project to a little one like:

Nano | Arduino Documentation

The Arduino Nano is Arduino's classic breadboard friendly designed board with the smallest dimensions. The Arduino Nano comes with pin headers that allow for an easy attachment onto a breadboard and…

docs.arduino.cc

or one of it's same-size but more powerful MPU, or adding bluetooth or wifi or both, etc

Nano Every | Arduino Documentation

The Nano Every is Arduino's smallest board with dimensions of only 45x18mm and a weight under 5 g. The small footprint and low price, make the Nano Every particularly suited for wearable inventions…

docs.arduino.cc

If you need lots of MPU power or memory you might have to stick with a Mega version but there are small ones of those too

Micro | Arduino Documentation

The Micro runs on an ATmega32u4 processor with native USB communication. Configure via software whether the board is recognised as a standard Arduino, a mouse or keyboard.

docs.arduino.cc

Then there are all the Shields that can be stacked directly on top of them (and each other) to add functionality, connectivity, etc., so you don't always have to come up with your own hardware boards for that stuff.


There are also a bajillion projects out there that people have published code for, that you may be able to borrow code out of to perform functions in your project to save you some coding.

(I am learning the coding partly by breaking such projects apart and modifying the bits I at least somewhat understand to learn about hte bits that I don't).

 

[bikerpete]

bikerpete,

The Arduino breadboard method (big open space) is also employed in my emulation conception. I have had several CanBus but all were short lived — besides you can get/calculate RPM off the Hall wires with Arduino.

Without a multiplexor board you have some 14 digital terminals posts for input/output wires. Reprogramming ( via Bluetooth or RS232) does not consume these I/O posts.

Inputs: Halls 3 , throttle input 2, clutch pot input 2, throttle signal to controller 2 and common NEG. = 10 pins

Standard aurduino has 6 I/O pins for analog signals. Arduino MEGA 2560 has 16 analog inputs or use multiplexor

See: Amazon.com

$48.90

Conveniently Arduino analog I/O operates at 0 - 5V DC which is the same as most controllers.

After finalizing the number of I/O pins and circuitry needed one could eliminate the breadboard and miniaturize the micro circuit box — water proof?

Getting started: Temu | Explore the Latest Clothing, Beauty, Home, Jewelry & More

Thanks.
I've been looking into all this and do tend to keep coming back to one or other of the Arduino boards - the UNO is probably perfectly sufficient but I'm considering going to a SAMD21 or similar based board. I figure my coding is going to be ridiculously poor quality so I can compensate for that to an extent by simply throwing horespower and storage at it (yes, yes, inelegant but perhaps pragmatic). I also figure I might as well start with something with plenty of scope to expand into whatever else I might dream up.

For the cost of them it's totally irrelevant which one to buy, even the "expensive" ones are just stupid cheap in the scheme of things. Although I'm inclined to give some cheap Aliexpress clones a try just because they are so insanely cheap - it would feel miraculous to get something like this running on a board that costs less than a cup of coffee!

I have been a little confused about the 3V3 vs 5V versions. In an ideal world I'd like to feed a hall output straight to the ADC to get RPM in the first instance. I haven't quite worked out if any of the 5V boards accept 5V into the ADC inputs or if that is still limited to 3.3V? Most of the datasheets I've seen talk about the digital IO but don't specifically tell me if the ADC can accept 5V, and others specifically say the ADC is still limited to 3.3V.
No big deal, but it would be that bit simpler and cleaner to just run wiring straight to the board rather than make a little divider board to hang off it somewhere. Or is there a good shield that handles 5 - 3.3V inputs? I haven't looked at shields at all yet.

I've hunted through reams of code and pulled out what looks to be fairly elegant hall to RPM code.
The heart of the program, driving throttle signal as a factor of the difference between target RPM rate of decay and actual rate of decay, is still in the 'pseudocode' or flowcharting stage.
Installed Simulide and Arduino IDE and been simulating parts of the program to get an idea of how things work.
I've also been refining the requirements of the logic and control - it's pretty tricky to consider all the reasonable scenarios and figure how to handle them in a manner that's going to give a vaguely reasonable response.
I've also started setting up a GitHub project for it. Maybe there'll be people interested in contributing who actually know what they're doing

I'm mildly optimistic that this could add a certain degree of flexibility to a basic flywheel. No way it's going to replace the flywheel, but the more I look at it the more I can imagine that it could possibly effectively modify the response of the complete system.

 

[amberwolf]

For the cost of them it's totally irrelevant which one to buy, even the "expensive" ones are just stupid cheap in the scheme of things. Although I'm inclined to give some cheap Aliexpress clones a try just because they are so insanely cheap - it would feel miraculous to get something like this running on a board that costs less than a cup of coffee!

All of mine in the Snuggles thread have come from AE on deep holiday sales (I don't think any of them cost more than a few dollars, most were less than $2, and all were free shipping and arrived in less than two weeks) and so far they've all powered up and talked to the IDE; I haven't used them for much beyond "hello world" sketches, other than the Nanos (that I got a few years ago for some Cycle Analyst things I never learned enough yet to create).

(Before that I also got a couple of the Raspberry Pi "nano" equivelents (same tiny size as the Arduino Nano, but for RPI); haven't used those yet). None of them except the Uno is bigger than a couple inches wide and not even 3" long, most are smaller than that.

I don't think you'll need as much of an MPU as you expect: I think I might have as much computing power in the bunch them as I had in all the desktop and laptop computers I've had in my life except for *maybe* the last three.


If I had plenty of money to spend I'd probably buy everything I could from places like Adafruit, as they've taught me quite a bit of stuff over the years just by existing and publishing their projects, and whatnot.

I have been a little confused about the 3V3 vs 5V versions. In an ideal world I'd like to feed a hall output straight to the ADC to get RPM in the first instance. I haven't quite worked out if any of the 5V boards accept 5V into the ADC inputs or if that is still limited to 3.3V? Most of the datasheets I've seen talk about the digital IO but don't specifically tell me if the ADC can accept 5V, and others specifically say the ADC is still limited to 3.3V.

AFAIUI, 3.3v versions are limited to 3.3v I/O of all types, but you can use various "shields" with level converters, etc., or buy similar boards separately and wire them up, including analog "level shifters" to rescale I/O voltages as you need them. Or buy op-amps and design the circuit yourself if you need something specific that's not already out there. Or just use an external ADC/DAC that talks via whatever serial flavor you prefer to the MCU.
A quick google finds a few things that would probably work but I don't know anything about them; here's a couple:

Extended ADC Shield | Mayhew Labs

The Extended ADC Shield gives your Arduino the ability to measure extended voltage … Continue reading →

mayhewlabs.com

Oak Level Shifter Shield Kit - Digistump

Digistump.com

digistump.com

See my Snuggles thread for some tiny amount of info about a couple of the digital shifters scattered somewhere in the posts; I know I will have to use some at some point.

Robotics Project: Snuggles, The Wolf

Technical discussion and idea-workout thread for Snuggles, The Wolf, a robot intended for emotional support interaction. Discussion of the project itself is welcome; ridicule or other negative feedback about the idea itself, etc is not. ;) The project concept is pretty complex, so reading all...

endless-sphere.com

For the coding it sounds like you already know more than I do by far, so I don't know if I can help much with that. You should link the Github here. (I haven't started one for the Snuggles project yet, as I have nothing to put there besides ideas).

 

[bikerpete]

Thanks @amberwolf lots of good info there.

I fully expect you're right about processor power - really what I'm doing will be very far from taxing even the most basic Arduino type micro, but there doesn't seem to be much downside to throwing brute horsepower at it so why not? I'll probably buy a few different micros from AE just so they're there if I want them - they're so stupidly small & cheap it's no big deal to throw them in a box.

I had a look at the Uno R4 Minima you linked to - that seems a pretty decent option - 5V, powerful, CAN bus (with a transceiver board), usual Arduino user base to get help ... From AE they're AU$10 so not exactly going to break the bank! I also found a guide that specifically fed 5V to the ADC so golden there.

Once I've got the GitHub with something useful in it I'll certainly link it. I'll probably start a new thread for the e-flywheel project and link it here too.

Have you seen Simulide? I've already found it very helpful to throw something together and see what happens without needing any hardware at all. Can even add external components to the micro. I've fed a UNO ADC pin a voltage signal and watched the serial data of my Tacho code - fiddled with the software, rinse repeat. Change the micro to something else - easy. Had an initial fiddle with some code for the RPM decay rate calcs.

 

[bikerpete]

A couple of days ago EM released some info and pics of their prototype "Factor-E".
I feel vindicated in much of what I've been banging on about

• Small high KV motor - 17,000 rpm max
• 'normal' trials clutch
• 4 speed gearbox
• 19kW peak
• 1400 N.M torque at the back wheel

Now we're talking! A real e-trials bike at last.
I knew it had to come one day, but it seemed to be taking forever for someone to figure it out.

I'm expecting it'll have some good flywheel inertia even if it's only got a physically small flywheel - energy increases by square of RPM, so 17,000 rpm gives even a puny flywheel some serious energy!

It has a VW logo on the side which is very interesting. No hints if VW have provided the motor, or money or manufacturing facilities or what. I'm kind of suspecting VW might have provided the motor - that kind of motor isn't exactly commonplace! 19kW air cooled at 17,000 rpm and in a production vehicle that needs to give a largely trouble free warranty period out in the wilds at a vaguely affordable cost - that seems a fairly rare beast. If I could find one at a vaguely affordable price (I'd consider a couple of thousand dollars plus affordable for something like that) I'd probably jump at it.

The only video I've seen purporting to be that bike in action didn't really look exactly like the prototype on show. Hopefully there'll be some more coming out soon.

It really seems to border on stupidity that e-trials manufacturers all decided to forget almost everything trials designers & riders have learnt about trials drive systems over the past 5 decades or so and gone off on a tangent informed largely by road going vehicles it seems. To me that is really pretty dumb.
Start with what works then work out what you can change, rather than what's happened which is start from scratch and gradually work your back to what was already proven!
Weird.

 

[DingusMcGee]

Bikerpete,

The hyped Factor-E edirtbike with 4 speeds likely has a clutch for shifting gears? But this design is simply like an ordinary motored motorcycle but powered by an electric motor instead of an ICE. Easy to imitate — just substitute a BLDC motor into some gas motorized dirtbike so as to retain the OEM gears and clutch/pressure plate/flywheel. Already been done several times and reported in ES.

The trend of this thread has been to question how one might add a micro processed flywheel /clutch emulator module to the circuits of BLDC motored [one speed] edirtbike to get similar control as one gets from having a real flywheel/clutch. This task is still an unsolved problem.

 

[DingusMcGee]

Bikepete,

There are 2 methodologies to approach building an emulator for the clutch/flywheel.

1. Employ the physics that accounts for the motions. Mainly here we have the change in momentum equals the Impulse which is Force x the time the force acts. There is torque version of the impulse/momentum equation and this describes the physics of what happens across the clutch/pressure plate. Think: collisions.

2. Heuristic Method— a set of calculation rules (procedures) that are frequently logic tested to see when another set of behavior rules come into play. If a certain set of changes happen then the new set of rules come into play.

If energy calculations vs. momentum calculations are used to determine motion you will likely need heuristics because some energy that might seemly produce motion simply vanishes from the system for from doing such motion - entropy.

Example — 2 box cars on rails of the same mass and speed collide and have working couplers. What happens to the energy?

Remember: momentum is always conserved.

Heuristics, employing very little physics, likely can be used to create a usable emulator because if it could respond better you simply add more logic tests and rules of procedure .

 

[thepronghorn]

@bikerpete
I haven't read the entire thread in detail, but it seems like in some posts you want a virtual flywheel and in other posts, it seems like you think a physical flywheel is needed. Are you interested in trying out both methods?

How much continuous power and top speed is needed for a trials bike? How much peak torque is needed?

 

[speedmd]

Clutch is needed not for shifting but more for impulse control IMO. Like the idea of the synthetic "electronic" flywheel very much. Some concerns I have on clutch lever feel - behavior to work through. Reasonably Sure it can be adapted.

Not sure if a single reduction gearing setup can match the torque dynamics - options a gear box allows. Gut feel is that we should be able to easily eliminate a few of the gear selection options and have fully capable bike with two or three speed (gear reduction) options with a suitably capable motor.

Motor does not need to be very powerful (possibly 15-20kw peak), but needs to have smooth and high torque -power delivery. Peaky - fast spin up is not ideal. Most have top gears suitable for roadway - faster path speeds to help travel between event segments which can be some distance apart. When riding with my friends on their enduro bikes, I am in 4th or 5th gear on most all of the trails. Only drop to the lower three when I do some segment exploring when regrouping or down to third on some near impossible trails for the enduro bikes.

 

[amberwolf]

Have you seen Simulide? I've already found it very helpful to throw something together and see what happens without needing any hardware at all. Can even add external components to the micro. I've fed a UNO ADC pin a voltage signal and watched the serial data of my Tacho code - fiddled with the software, rinse repeat. Change the micro to something else - easy. Had an initial fiddle with some code for the RPM decay rate calcs.

I have seen references to it but have not yet looked at it. It sounds like a good way to do some of my prototyping. I suspect it's electronics capabilities aren't sufficient to simulate the most important things I would need for the robotics stuff (IMUs and other such sensors), but I'll have to try it out to find out for sure.

 

[bikerpete]

Bikerpete,

The hyped Factor-E edirtbike with 4 speeds likely has a clutch for shifting gears? But this design is simply like an ordinary motored motorcycle but powered by an electric motor instead of an ICE. Easy to imitate — just substitute a BLDC motor into some gas motorized dirtbike so as to retain the OEM gears and clutch/pressure plate/flywheel. Already been done several times and reported in ES.

The trend of this thread has been to question how one might add a micro processed flywheel /clutch emulator module to the circuits of BLDC motored [one speed] edirtbike to get similar control as one gets from having a real flywheel/clutch. This task is still an unsolved problem.

Forgive me if I'm wrong, but I really don't get the feeling you have much experience on trials bikes. ?
Calling the Factor-E "like an ordinary motored motorcycle" is technically correct, it is. But so is a MotoE GP bike and I'd suggest it's going to end badly trying to do trials on one of them! No trials bike worth the name is an "ordinary motorbike", they are extremely specialised machines for a specific niche. Yes, they've got two wheels, an engine, handlebars and footpegs, all in familiar relationship to each other. But it sort of ends there. You only need to watch someone who's never ridden one (even if they've lots of experience on other bikes) hop on to see that they are very, very different beasts.

The Factor-E is a purpose built trials bike coming out of the factory that has more experience building e-trials bikes than anyone else in the world by a very long margin. I'm generally pretty sceptical of EM and haven't thought much of their meandering development path, but I'd be a fool to ignore the fact they've accrued an enormous amount of knowledge over the years they've been building bikes. It is rare for a manufacturer to produce a new product that is worse than the previous one, and I kind of doubt EM have done so with this bike. It looks to me like they've finally pulled together the parts of an ICE bike that work well, the parts of electric that work well and found a way to unite those aspects effectively. Time will tell, but it looks pretty exciting to me. Is it likely to be the last word in e-trials? No way. But it looks a great step forward and I hope it performs significantly better than the previous model.

I've spent many, many hours trawling through ES looking at trials-relevant info and I've found very little that is seriously applicable to trials at more than a beginner level.
Every geared moto I've seen on ES is a trail bike at absolute best, or a road bike. When someone takes one and rides a trial in mid-level (Expert, Trial 3 etc) competition and comes in with results equivalent to their results on a reasonably current ICE bike I'll take serious notice of what they've done.

A 4-speed gearbox on a road or trail bike makes very little sense, it has it's uses in limited cases but generally a single speed provides ample torque, speed and control for what people need. That is so far from observed trials use-case as to be almost laughable.
If you drop an electric motor into a trials bike chassis and transmission, you also need to retain the flywheel. If the motor spins significantly slower than the original ICE, then you'll need to increase the diameter/mass of the flywheel to achieve the same energy capacity. If you don't do that then you've completely missed the point of the past 50-70 years of trials bike drive train development. And that's just a starting point. You've still got to address issues around spin-up that occurs on loss of traction due to the flat torque curve and characteristics of an electric motor, plus all sorts of other details.
Trials ICE engines are nothing like say an Enduro engine, they have tiny carburetors to get sensitive low end response and also to deliberately choke down high end response and power. They have virtually non-tuned exhausts (you can easily double the power out of a 2-stroke with a tuned exhaust). Their internal porting is worlds away from most high performance 2-strokes. These things aren't accidents, they've taken decades to refine and they work incredibly well. It seems pretty silly to me to ignore that knowledge base.

Re. the comment

Already been done several times and reported in ES.

As I said above, for anyone to think they can pull out the engine of a trials bike, replace it with an electric motor with totally different characteristics, remove one of the most significant components of the drivetrain and then say, "I just tested that and it doesn't work. You don't need gears or a clutch." reflects an immense naivety and ignorance about trials.
It may work perfectly well for their use, but it's not a shadow of a trials bike. If they can show me video of real-world modern trials technique, or results as mentioned above then I'll happily change my tune. But I've yet to see either on such a machine.
As a reference, the current world electric trials champion is an incredible rider on an EM, but I've never seen him do some of the moderately basic maneuvers the ICE riders do routinely. So even what is arguably the best e-trials bike in the world still can't even do some of the basics, let alone do them as well as an ICE bike. For the record, Gael (the rider) has the ability to ride very competitvely at World level on an ICE bike - I doubt he'd be anywhere amongst the leaders, but there's probably less than 20 people in the world who could even complete all the GP sections! The best riders in Australia and USA can't even make it through all the Trial 2 level sections, let alone the GP sections! And these guys are pretty amazing to watch.
And recall that EM unquestionably have the most experience building e-trials bikes.

The purpose of this thread is to pull together some ideas for how best to put together a real fit-for-purpose trials bike.
One of those ideas is the flywheel emulator which I feel might, perhaps, if we're very lucky, add some flexibility to a physical flywheel.
Another idea is a virtual clutch. Personally I find that a fairly pointless path to head down. I can accept it might be possible to do although I really struggle seeing how it can be implemented in a manner that doesn't introduce so many disadvantages as to be useless in practice. That's just me though.

Your input has definitely contributed to me delving into the e-flywheel project - it's got me intrigued. Your input has challenged some pre-conceptions. I'm guessing that part of the reason you succesfully challenge my thinking is because you have good knowledge in areas I don't.
But I suspect when you start moving into discussion about what a trials bike needs as contrasted to what electronics can and can't do, then you're moving into territory that you're really not familiar with.
I could be wrong, you could be riding pro-level trials for all I know, but I suspect not.

 

[bikerpete]

Clutch is needed not for shifting but more for impulse control IMO. Like the idea of the synthetic "electronic" flywheel very much. Some concerns I have on clutch lever feel - behavior to work through. Reasonably Sure it can be adapted.

Not sure if a single reduction gearing setup can match the torque dynamics - options a gear box allows. Gut feel is that we should be able to easily eliminate a few of the gear selection options and have fully capable bike with two or three speed (gear reduction) options with a suitably capable motor.

Motor does not need to be very powerful (possibly 15-20kw peak), but needs to have smooth and high torque -power delivery. Peaky - fast spin up is not ideal. Most have top gears suitable for roadway - faster path speeds to help travel between event segments which can be some distance apart. When riding with my friends on their enduro bikes, I am in 4th or 5th gear on most all of the trails. Only drop to the lower three when I do some segment exploring when regrouping or down to third on some near impossible trails for the enduro bikes.

I think you & I are thinking along similar lines.
Except I'm not interested in moving away from a physical wet multi-plate clutch. That's just a bridge too far for me.

To be clear, I also don't think it's realistic to replace the physical flywheel with an electronic one. I think the e-flywheel will simply add some flexibility to the overall flywheel response.
Maybe the e-flywheel will exceed my expectations, but I completely expect that anything I can put together will be little more than an indicative test-bed. It'll take some serious time and money to really develop it into a viable replacement for a hunk of steel. Assuming it shows any promise at all.

I don't have any friends, because I can't be bothered riding with people on enduro bikes.
If I want a social occasion I'd rather not combine it with 2-stroke engines ripping along on stuff that's either technically uninteresting or just a slog-fest.
I rarely get out of first or second gear except to move from one set of obstacles to another.

 

[bikerpete]

Bikepete,

There are 2 methodologies to approach building an emulator for the clutch/flywheel.

1. Employ the physics that accounts for the motions. Mainly here we have the change in momentum equals the Impulse which is Force x the time the force acts. There is torque version of the impulse/momentum equation and this describes the physics of what happens across the clutch/pressure plate. Think: collisions.

2. Heuristic Method— a set of calculation rules (procedures) that are frequently logic tested to see when another set of behavior rules come into play. If a certain set of changes happen then the new set of rules come into play.

If energy calculations vs. momentum calculations are used to determine motion you will likely need heuristics because some energy that might seemly produce motion simply vanishes from the system for from doing such motion - entropy.

Example — 2 box cars on rails of the same mass and speed collide and have working couplers. What happens to the energy?

Remember: momentum is always conserved.

Heuristics, employing very little physics, likely can be used to create a usable emulator because if it could respond better you simply add more logic tests and rules of procedure .

I'm definitely heading down the Heuristics path.
If throttle is closing Then switch to Flywheel routine. Else use Throttle smoothing routine.

In Flywheel routine it's basically:
Read RPM & calculate equivalent virtual flywheel energy to set max torque (throttle) level. (High RPM = high max torque response).
Calculate Delta-RPM of motor and compare to pre-configured Delta-RPM target. If DRPMm > DRPMt then apply proportional throttle correction up to current max.
Loop

There'll need to be a fair bit of other stuff going on, but that's the crux of it as I currently see it.
I don't really care if the correction would overshoot the target because in reality it's going to be a pretty slow analog response (tyre spin, real flywheel inertia etc.) so I think the corrections will be calculated so frequently compared to the response that it'll become more or less self correcting. So I'm not going to implement any PI or PID.

If I think it through various trials maneuvers it feels like it should work.
At low RPM, say just giving little pulses on clutch to hop the front wheel, then the throttle will be limited fairly low so things don't get too violent.
At high RPM, say a splatter onto a face, then the throttle can go much higher when the wheel impacts and almost stalls (there's always slip, so the controller has time to respond & ramp up). The limit is there to emulate the way a flywheel rapidly loses energy & rpm as it drives. Effectively the limit allows the motor to decelerate more quickly than the preset Delta by simply overpowering the correction signal. So as it dumps energy into drive it also reduces the amount of correction signal it can apply in the next correction (RPM feedback). A decaying feedback loop that is dependent on energy consumed in driving the bike.
For starters the power limit will be a fairly simple parabolic curve (driven by the power calculation, which allows the virtual mass of the flywheel to be altered with a corresponding change to the correction curve), but it might need to become more sophisticated than that to work nicely in the real world. We'll see. Start simple.

 

[speedmd]

I'm not interested in moving away from a physical wet multi-plate clutch. That's just a bridge too far for me.

For starters, agree, smaller-fewer changes will most times be best.

To be clear, I also don't think it's realistic to replace the physical flywheel with an electronic one. I think the e-flywheel will simply add some flexibility to the overall flywheel response.

Much depends on the mass of the motors rotor and rpm its running for a drop in replacement. The larger diameter axial flux designs may be able to punch as well as a ice-flywheel setup but I have to agree, it may be wishful thinking. Regardless, adding a flywheel plate is easy enough to plan into most any motor setup and should be included.

BTW, Agree on the cow trailing with folks that think they are dirt bike riders. The last outing was hard to get out of. The teasing stopped at the first washed out steep trail that none of them could get through while I made multiple passes up and down stopping to help rescue them getting unstuck in rocky sections and big wet erosion ditches. Heavy bikes suck! They all want a Trials bike now.

 

[DanGT86]

based on everything so far if I had to build this theoretical bike tomorrow I would do the following.

1. Use a clutch
2. Use a flywheel equivalent to ICE
3. Have a gear set between E motor such that the flywheel spins at the same rpm relative to ground speed in my most used gear
4. Incorporate a microcontroller between throttle and controller such that my torque ramp and hp at all rpms are as close to the gas bike as possible.

Once that is all dialed i'd add a simple 2 speed gearbox to have a gofast gear to get me around from section to section.

With enough engineering you could cut some of these corners like maybe calculating the energy from the flywheel and adjusting accordingly to eliminate the need for the motor to flywheel gearing.

Im also assuming most technical trials riding occurring in only one gear.

But I think going down your priorities list its clear that you are really after the feel of a clutch and the behavior of a flywheel. Id start with a design that leaves those as close to native ICE environment then slowly refine from there.

 

[lnanek]

Personally, on my commuter ebike, I use feathering my brake lever like a clutch. Pressing it slightly stops the electric motor without actually triggering the brake. Guess it's true you can't clutch walk the bike around a parking lot in the clutch's friction zone like a motorcycle, though.

One of Grin's new analog ebrake levers connected only to regen might help, or a two way twist throttle, forward for throttle and backward for regen would be my next step if I wanted more impulse control, though. Don't think I'd ever want a physical clutch on an ebike.

It's not like a spinning, unpowered electric motor is going to drive you forward anyway, it's an electric generator at that point and immediately becomes engine braking.

 

[DingusMcGee]

Ohbse,

The Nucular 24f controller may be fantastic and for a price $$$?

In all of nature (nuclear explosions included) and man’s contrivances nothing has an infinite rise rate — a true step function for what you may think of here as a change of 0.0 amps to 500 amp in zero time does not exist and cannot be produced. This statement applies to the any trials bike freewheel/pressure plate momentum transfer rate also. Drop the clutch and you do not get instantaneous angular momentum transfer.

There is an easy and partly safe way to test your controller and others (junk controllers) on an ebike/system as to their potential fitness for quick high torque deliverance to the rear wheel.

As you suggest, you will need LiPo batteries or a large bunch of others in parallel to produce the amperage and quickly.

The CycleAnalyst can be set to go WOT when some set voltage threshold has been twisted up by the throttle.

The other CycleAnalyst step to do at the test beginning is to set the max amperage quite low and then do some easy WOT runs and proceed by incrementing the amperage. You then may get a feel for what 70 amps WOT is like and you could trying timing that rise time/distance? If you also want to play with phase amp you can adjust/limit max phase amps on any Fardriver controller.

The point here is that any controller/battery/motor system that can do a good short WOT wheelie can be programmed with an Arduino throttle signal intercept circuit to actually get any kind of ramp function less steep than what you got on the WOT test. Likely some junk controllers for far less $$$ could work?

Beside in trials, you need to control the delivery rate of momentum change and do not want an “ instantaneous” buck you off wheelie — a very steep ramp function.

 

[bikerpete]

Ohbse,

The Nucular 24f controller may be fantastic and for a price $$$?

In all of nature (nuclear explosions included) and man’s contrivances nothing has an infinite rise rate — a true step function for what you may think of here as a change of 0.0 amps to 500 amp in zero time does not exist and cannot be produced. This statement applies to the any trials bike freewheel/pressure plate momentum transfer rate also. Drop the clutch and you do not get instantaneous angular momentum transfer.

There is an easy and partly safe way to test your controller and others (junk controllers) on an ebike/system as to their potential fitness for quick high torque deliverance to the rear wheel.

As you suggest, you will need LiPo batteries or a large bunch of others in parallel to produce the amperage and quickly.

The CycleAnalyst can be set to go WOT when some set voltage threshold has been twisted up by the throttle.

The other CycleAnalyst step to do at the test beginning is to set the max amperage quite low and then do some easy WOT runs and proceed by incrementing the amperage. You then may get a feel for what 70 amps WOT is like and you could trying timing that rise time/distance? If you also want to play with phase amp you can adjust/limit max phase amps on any Fardriver controller.

The point here is that any controller/battery/motor system that can do a good short WOT wheelie can be programmed with an Arduino throttle signal intercept circuit to actually get any kind of ramp function less steep than what you got on the WOT test. Likely some junk controllers for far less $$$ could work?

Beside in trials, you need to control the delivery rate of momentum change and do not want an “ instantaneous” buck you off wheelie — a very steep ramp function.

I've got a 12F here that although somewhat underpowered on the QS138 it's actually perfectly useable. There is only a very small difference in power between it and the Fardriver on my DOB.
While the the NUC has some really nice characteristics (I really liked using the speed & torque mode) it can't be set to idle properly. I've put that in as a feature request, but realistically it must be so far down their priorities that I doubt it will ever eventuate. No idle is a showstopper.

You are absolutely correct on the 'infinite rise rate', and in fact that is one of the reasons a physical clutch is not really negotiable to me at present. Trying to develop an electronic analogue for the control-ability of a really good wet clutch is going to be massive I think. All the issues around rapid torque increase, decaying RPM, consistent drive during varying throttle inputs etc it becomes SEP (Someone Else's Problem).

I'm still very skeptical about the worth of trying to go down the path of having a motor with sufficient instant torque to do dynamic trials moves. The bottom line is I don't believe it's possible to produce the sort of torque that's needed without a bigger weight penalty than a flywheel gives. So why bother?
A further problem arises when you start talking snap from stationary or even rolling backwards (it's not unknown to be rocking the bike slightly backwards for balance or clearance to the obstacle right at the moment you want to Go.), given the way all e-trials bikes have moved to having idle modes I get the feeling everyone has come upon the limitations of known controllers to provide that really clean pickup from zero or even reverse rpm.
And that's without addressing the stuff we're looking at in the characteristics of the power delivery in the virtual flywheel concept.

that can do a good short WOT wheelie

Throttle wheelies have virtually no correlation to trials back wheel moves. Just because a bike can flip you backwards in an instant when you twist the throttle does not make it in any way relevant to trials. Please, trust me on this.

My little e-trials with the QS138 has an odd power curve (regardless if it has Fardriver or Nucular. Nuc is better, but doesn't solve it) that as it gets to about 1/3 rpm it hits a significant 'power band'. Very odd. Also very, very easy to flip the bike over backwards. It's crazy how fast the bike rotates if you're not on the ball when you wind on full throttle from low RPM - it hits that power band and for a short time the acceleration goes way up. I've looped it several times. But that level of acceleration is just nowhere near enough to do what I want to do on obstacles.
I've tried locking both brakes hard, winding on full throttle against the slipping clutch (keeping motor RPM fairly low so I'm not activating the flywheel effect, and staying down in the peak torque area of the motor) to get past the current ramping in the controller, then releasing brakes and popping clutch as close to simultaneously as I can. It tends to go badly pear shaped because it's too hard to get the throttle closed after the bike has responded, but during the actual 'launch' phase it's anaemic. Just nowhere near the snap I can get out of winding up a decent flywheel. You don't feel a 'snap' of instant acceleration, you still feel it build as the bike starts to move, and that's useless. See note 1 below
That suggests to me that I'd need a much bigger motor to get the response I want, which doesn't make sense because the QS138 is already a big chunk of steel & copper that provides more than enough power for everything I want to do on that bike. Increasing it's mass with more iron & copper by even a relatively small percentage would cost more weight than the flywheel I threw on and which works an absolute treat. Maybe it's a controller limitation, but then I'd also need to considerably up-spec the battery to be able to deliver considerably more current - a vicious circle.
I would then need to configure the controller to restrict the power during most riding (unlike many motorsports, 'enough' power is best, more is not better!).

Another point to recall is that a flywheel adds a very useful gyroscope. Balancing a trials bike is much easier engine running than engine off because of that spinning flywheel. That's not to be sneezed at. One of the most common reasons I fail tricky obstacles in sections is because I take off before I'm fully balanced and settled, and that's with the gyro helping me. Without it my failure rate would go up even before I made the attempt on the obstacle itself.

Note 1.
In 'punch' or 'Zap' type moves you place the front wheel as low as you dare on the face of the obstacle and lock the front brake. You then drive the bike into the locked front wheel to compress the front forks fully before releasing the front brake. As you compress the forks you are driving your hips forward to get your body moving toward your destination, then extending up to get the vertical momentum, kind of an L shape action. Part way through the Up action you fully release the clutch and the front brake so the compressed forks can extend and help rotate the bike around the back axle.
If the front wheel has moved forward even 25mm (inch) then you lose much of the impulse out of the forks and your vertical lift is dramatically reduced. If the obstacle is rounded like a log it's even worse, as now the front wheel is simply rolling forwards rather than driving up.
So it's unacceptable for the bike to have to move forward even a small amount in order to build torque. It has to do it while virtually stationary. In reality the back wheel rotates a few degrees as it drives in under the suspension, but it's literally just a few degrees.
This is where the response of the dob off throttle and locked brakes feels rubbish - you can feel the bike accelerate over those first inches, and then it's all over, you've lost it, and the rest of the action is somewhat pointless. It's completely different from the agggressive snap you get in the first moment from a flywheel/clutch, which then rapidly fades into something more manageable.
Doing this sort of move from stationary onto a decent sized obstacle is my "acid test" of how well a bike performs in these sorts of moves, it really sorts the goats from the sheep as the saying goes.

 

[bikerpete]

BTW, Agree on the cow trailing with folks that think they are dirt bike riders. The last outing was hard to get out of. The teasing stopped at the first washed out steep trail that none of them could get through while I made multiple passes up and down stopping to help rescue them getting unstuck in rocky sections and big wet erosion ditches. Heavy bikes suck! They all want a Trials bike now.

Yep. They take you to the "hard singletrack" at the end of which you're often left wondering where the hard part starts.
Otherwise they take you to some ludicrously long, loose hill that if you fail you end up spending the next half hour manhandling your bike around the place and end up so exhausted you're ready to stop riding. Fun factor - zero.
And yes, heavy bikes are awful, beats me why so many people ride them.

 

[bikerpete]

based on everything so far if I had to build this theoretical bike tomorrow I would do the following.

1. Use a clutch
2. Use a flywheel equivalent to ICE
3. Have a gear set between E motor such that the flywheel spins at the same rpm relative to ground speed in my most used gear
4. Incorporate a microcontroller between throttle and controller such that my torque ramp and hp at all rpms are as close to the gas bike as possible.

Once that is all dialed i'd add a simple 2 speed gearbox to have a gofast gear to get me around from section to section.

With enough engineering you could cut some of these corners like maybe calculating the energy from the flywheel and adjusting accordingly to eliminate the need for the motor to flywheel gearing.

Im also assuming most technical trials riding occurring in only one gear.

But I think going down your priorities list its clear that you are really after the feel of a clutch and the behavior of a flywheel. Id start with a design that leaves those as close to native ICE environment then slowly refine from there.

3. is an interesting one. I read that as you're suggesting putting a gear between the motor & flywheel to speed up the flywheel?
Bit tricky space & cost wise maybe but an interesting idea. I've been increasingly leaning toward a higher RPM motor as it probably makes the clutch gearing easier (smaller dia motor) and lends itself to a lighter motor.

4. I'm not too sure about. I kind of think that there's no need to emulate an ICE torque/power curve. As long as you've got what the clutch and flywheel offer the flat e motor curves are a bonus.

Yes, most ICE trials riding is in the bottom 3 gears which are all very close to each other. With an electric flat curve one gear is likely enough to cover all three. Maybe 3 gears would be good, but I'd try 2 first.

I spent quite a bit of time trying to get a huge rear sprocket made for my e-trials so I could prove out the low gear theory, but it kept getting derailed for one reason and another. I'm now waiting on delivery of a 2-speed box. Hooray.

 

[speedmd]

The motor (+flywheel mass spin up) and clutch engagement dynamic is a interesting one. Need to think on it a bit to get a better visual on all the factors at play. Your certainly looking for a good solid kick for many of the moves. No matter how strong a soccer player you are, you get very little distance kicking the ball if your foot starts right against the ball and then accelerates vs winding up a bit and hitting it with the foot and your mass moving.