Observed trials bike designing

[bikerpete]

I haven't given up on the belief that a finely tuned controller with enough amps could obsolete a clutch but I am pretty convinced that a clutch is a way easier solution and way more familiar to most riders at this point. Id rather have one and not use it than realize at the end of a build that I want one.

Right on the money I reckon, except that you still might not have dealt with the ability to totally chop drive instantly. For that I think you have to come back to a clutch.
Stopping the rotating masses is not the same as disconnecting them, and even without a discrete flywheel there is still a significant flywheel effect in the motor itself.
There's often times the last thing you want to is to stop the back wheel and so eg. slam the front down off a 2m high step! It's a world of difference between cutting forward drive and actually stopping it.
There's also the consideration of the effect of the reaction torques if you actively rapidly slowed or stopped the motor.

I thought I'd throw this video in here. This is almost a defining maneuver I reckon.
This was filmed at 120 fps I think.
It clearly shows:

• the rise in rpm, storing power and setting the available power.
• The clutch release (4 frames to full release, so 1/30th second)
• His boots are almost off the pegs from his jump
• You can see the tension in his arms as he tries to lift the bike
• Clutch is pulled in 60 frames later (0.5 second), when the rear wheel is fully locked to stop on top of the log.
• Another 0.5 second and clutch back out again to pop off the log (out of this clip).

I think this gives a decent picture of the instant torque that's needed and also the instant cut that is just as critical (if not more so - it's a significant safety valve). Plus how impossible it would be to do all this on just throttle.
https://vimeo.com/manage/videos/657655534

 

[speedmd]

speedmd wrote: ↑Apr 28 2022 12:24pm
Given the much higher torque of the electric motor at low rpms, they should (spin up the flywheel faster) be able to out perform the ICE setup with equal rotating mass.

Therein lies the problem - you don't want it to spin up too fast.
It's one of the problems people have with the EM - it's really hard to use the clutch effectively because the motor/flywheel are too digital. They are either at idle or they are zipping up to full rpm at a rate of knots. Even if you only give it modest throttle, at zero load it just spins right on up.
This is where the throttle needs a relatively slow rpm ramp rate, independent of the current ramp rate.

Good points on the current vs speed throttle setups. Having a motor spin up to max rpm with just a touch of throttle is no good at all. Simply switching from current throttle when clutch pulled should allow for nice infinite rpm control settings as you decide via ramps and your wrist. You could also allow super high (conditions - traction dictating) or low current levels to tame things down when throttle only used and mostly motoring though in current setting. Two independent controls with a clutch lever actuating the switch between the two throttle mapping options may work.

 

[spinningmagnets]

When I wrote "why not both" I only meant to suggest that an "E trials" competition should allows competitors to use either one. Throttle only or throttle plus clutch.

Let the results show which works better.

 

[bikerpete]

When I wrote "why not both" I only meant to suggest that an "E trials" competition should allows competitors to use either one. Throttle only or throttle plus clutch.

Let the results show which works better.

Ah, gotcha. My misunderstanding.

They do & they have.
Clutchless bikes are completely outclassed by the clutched bikes. By a large margin.
I doubt there will be a single clutchless bike in the e-Trial GP this year, but there's nothing in the rules to prevent it.

The current top performing bikes have adopted an "idle" too. The motor never stops spinning, just like ICE.
I believe this is to try to avoid the tiny but nevertheless significant delay in starting the motor from stationary.

 

[bikerpete]

Good points on the current vs speed throttle setups. Having a motor spin up to max rpm with just a touch of throttle is no good at all. Simply switching from current throttle when clutch pulled should allow for nice infinite rpm control settings as you decide via ramps and your wrist. You could also allow super high (conditions - traction dictating) or low current levels to tame things down when throttle only used and mostly motoring though in current setting. Two independent controls with a clutch lever actuating the switch between the two throttle mapping options may work.

I've thought about that idea and can't quite figure if it would really work.
Picture this.
In the video I put up of the "front-on" onto the log, you loft the front onto the obstacle & lock both brakes as it lands. Once stable, you release the rear brake but press the front into the obstacle by slipping the clutch. You really want a certain amount of pressure on the front wheel to hold it in position and to help balance with, so there's a definite amount of rear wheel drive.
When you pop, the movement of the lever is fairly small - there is a limited range of movement available in your middle knuckle within which you have really good control.
So I can't work out what actually signals the switch between modes.
I think it can't really be lever position because you often adjust the amount of drive with the clutch while balancing, in order to move the front wheel up and down to the correct point, and also to increase or decrease the balance response to turning the front wheel. You'd run the risk of the controller switching erratically between modes which I think would be a complete nightmare.
Similar if you read hydraulic pressure in the clutch line.

So I'm at a bit of loss what you might use to signal the transition between modes?
Perhaps it's actually a torque sensor on the drive output shaft? When there is a sharp rise in torque the controller switches modes?
That could get interesting if say jumping across a gap and a big torque spike as the wheel lands.
Tricky.

I'm looking forward to getting my next shipment of dob soon, these with mechanical clutch and 3kw brushless motor. I plan on modding one with my Nucular which can provide torque, torque & speed, and speed modes, as well as adjustable acceleration and deceleration rates etc. That should let me experiment a bit more effectively with some of this.

 

[DanGT86]

Perhaps the key here would be to have someone do extensive dyno testing of an ice bike data logging as many things as possible including throttle angle.

Then tune your electric motor/controller as closely as possible to match the torque ramp rate and hp profile of the gas bike. Match the rotating mass and clutch as close as possible and go ride.

I do think the hp and rpm are linked to each other in a gas motor in a way they dont really have to be with electric. A lot of what you are used to as clever clutch work is probably you actually catering to that narrow ice powerband.

Earlier you mentioned the wheel coming off the ground and the controller immediately ramping the speed of the wheel to max. This will be remedied somewhat by torque+speed throttle logic. Physics kinda does that on its own in a gas motor and thats what we are used to.

I suspect these E-trials bikes are going to continue to evolve with clutches and very complex throttle logic. Im excited for that.

 

[bikerpete]

Perhaps the key here would be to have someone do extensive dyno testing of an ice bike data logging as many things as possible including throttle angle.

Then tune your electric motor/controller as closely as possible to match the torque ramp rate and hp profile of the gas bike. Match the rotating mass and clutch as close as possible and go ride.

I do think the hp and rpm are linked to each other in a gas motor in a way they dont really have to be with electric. A lot of what you are used to as clever clutch work is probably you actually catering to that narrow ice powerband.

Earlier you mentioned the wheel coming off the ground and the controller immediately ramping the speed of the wheel to max. This will be remedied somewhat by torque+speed throttle logic. Physics kinda does that on its own in a gas motor and thats what we are used to.

I suspect these E-trials bikes are going to continue to evolve with clutches and very complex throttle logic. Im excited for that.

I know a bloke who's been talking to a Uni about just that sort of data logging & analysis - yet to eventuate. I'll keep quietly pestering and offering any help I can to get it to happen.
Dyno testing as such would be pretty pointless I suspect. It needs a bike fitted with a heap of instrumentation being ridden by an expert level rider in various conditions.
Even working out exactly what needs to be measured & how is not insignificant.
Wheel speed, RPM, output torque (top chain line deflection measurement?), throttle position, clutch position, footpeg load (?), supension position front & rear, bike attitude, CoG acceleration vector ...

Certainly everything top riders do now is based on what's been learnt on ICE bikes, so there's undoubtedly compensations happening for their particular quirks & foibles.
There's just starting to be a generation coming through who have always ridden electric, but mostly they cross over to ICE once they get to a certain skill level - there's too much of a handicap with current e-trials.
I certainly believe that when the controllability of e-trials moves up to the level of current ICE bikes (or beyond) then the ability to manipulate torque & RPM somewhat independently will open a whole new world. We're still a long way from there in terms of money spent & research done on it yet.

I'm looking forward to playing with the Nucular torque+speed modes.

I also wonder if part of the eventual solution will be a change in the actual physical controls. Maybe a button embedded in the throttle tube that you push with your thumb to engage certain controller modes? Or an embedded button in the clutch lever - move your finger onto it and the controller reacts differently? Or a foot "gear selector" that changes modes? Maybe the suspension will feed back into the controller?

There's absolutely no question in my mind, e-trials is going to come to dominate the sport in a relatively short time. It's just still in the "backyard tinkerers" realm at present - EM are building a nice enough bike, but basically they build a frame and bolt in a bunch of more or less off-the-shelf general purpose components. It's not even vaguely close to what a company like Honda or KTM could bring to the table. Now KTM own GasGas maybe they'll put a bit of money into e-trials - they'd have to be willfully blind not to see that's where it's going to go.

Unfortunately, or fortunately depending on your perspective (I tend toward the latter), trials bikes don't sell like Hardley Rideables so it could be a while before the development gets done.

 

[DanGT86]

I agree that the dyno and datalogging would be no small task. I guess I was thinking the primary purpose would be to figure out what the unloaded rpm acceleration rate of the ice motor is so you can copy it. I see that being the #1 challenge to getting the feel right. If a rider has the clutch pulled completely and wants to rev to say 3000rpm they need to have some idea how much throttle angle that is. With no noise and a torque only throttle they would surely rev to max rpm quickly. A fake noise generator or perhaps some bright colored LEDs across the bars to indicate general rpm would be a big help for training ones brain to the electric behavior.

This all has me thinking about my younger days when I used to tune fuel injection maps for cars. With an ICE engine the volumetric efficiency is different at every different combination of RPM and vacuum (engine load). So the torque of the motor at 3000rpm is completely different at 10% than it is at 80% throttle. The engine will require a different amount of fuel at all of those loads. Based on that amount of fuel it will find an equilibrium and stop accelerating.

On the attached graph Full throttle at max rpm is in the back and closed throttle minimum RPM is the front center corner. The height of each intersection point is the VE percentage which roughly corresponds to the torque output at that point and the required fuel.

Lets say you hold the ICE bike at 3000 with the clutch pulled in. As you release that clutch the engine vacuum drops (MAP increasing on the graph). This causes more fuel to be delivered and the torque output increases without the RPM changing. The carb or EFI is in the background increasing the torque for you based on the load. I don't really see an analogue to this in the electric realm. The closest thing would be a very well tuned speed+torque throttle.

One big takeaway is that the ICE profile is still a ramp everywhere. At lower RPM your torque is limited so you have way more resolution with your throttle angle and an unloaded motor. Its a more coarse adjustment to rev an unloaded gas motor without instantly overspeeding it vs an electric motor. Opening the gas throttle all the way and dumping in all the fuel at 2000rpm is never going to make the same torque as 5000rpm. The electric motor by contrast would have no problem providing you nearly all the torque at 2000rpm.

If this graph below was for an electric motor it would look like a flat plateau rather than a slope. The trick is going to be mimicking an ICE curve even at the expense of power to get it in line with what human riders are capable of controlling.

 

[bikerpete]

I agree that the dyno and datalogging would be no small task. I guess I was thinking the primary purpose would be to figure out what the unloaded rpm acceleration rate of the ice motor is so you can copy it. I see that being the #1 challenge to getting the feel right. If a rider has the clutch pulled completely and wants to rev to say 3000rpm they need to have some idea how much throttle angle that is. With no noise and a torque only throttle they would surely rev to max rpm quickly. A fake noise generator or perhaps some bright colored LEDs across the bars to indicate general rpm would be a big help for training ones brain to the electric behavior.

This all has me thinking about my younger days when I used to tune fuel injection maps for cars. With an ICE engine the volumetric efficiency is different at every different combination of RPM and vacuum (engine load). So the torque of the motor at 3000rpm is completely different at 10% than it is at 80% throttle. The engine will require a different amount of fuel at all of those loads. Based on that amount of fuel it will find an equilibrium and stop accelerating.

One big takeaway is that the ICE profile is still a ramp everywhere. At lower RPM your torque is limited so you have way more resolution with your throttle angle and an unloaded motor. Its a more coarse adjustment to rev an unloaded gas motor without instantly overspeeding it vs an electric motor. Opening the gas throttle all the way and dumping in all the fuel at 2000rpm is never going to make the same torque as 5000rpm. The electric motor by contrast would have no problem providing you nearly all the torque at 2000rpm.

If this graph below was for an electric motor it would look like a flat plateau rather than a slope. The trick is going to be mimicking an ICE curve even at the expense of power to get it in line with what human riders are capable of controlling.

That's fascinating. I've never looked at that sort of graph with any particular comprehension, just gone "Whatever, that's for people chasing power".

I think what you're saying about rider feedback and RPM rise rate in the first paragraph is a trivial issue really. We humans are amazingly good at detecting and responding to change if it's at a rate we can deal with. An electric motor has a quite audible noise, and if one really was too quiet to hear, a noise generator would be trivial to add. The actual rise rate is also trivial to determine - competent riders can move from a trials bike to an enduro bike with very different motor response and adapt in minutes to the new RPM ramp. You could just sit an electric next to an ICE and listen - if it sounded like a similar ramp I'm sure that would be close enough.
Oh, and visual isn't going to work I reckon - your eyes are already busy scanning your path, helping plan your strategy and feeding into your balance - glancing at a visual indicator is just going to be a distraction, interference & far too slow.
You'd need the RPM ramp to be totally separated from the torque or current ramp though so that all those moments when momentary or short high torque is demanded it is supplied. I'm looking forward to playing with those two ramps when I get my Nucular into a bike - it wont be ideal but it should give me a better picture of how it all works together.

The two issues you point out are creating a "map" that riders can predictably and accurately control, and getting the motor/controller response to automatically do some of the torque, load and RPM integration that ICE does.

This raises a couple of questions immediately for me, and I'll probably come up with more as I process it.
1. What do you think an "ideal" map would look like? Is it an inclined plane? A ridge running up and back?
2. How would a controller achieve the varying automatic load response of an ICE? Is it via a complex "3D" speed/torque map instead of a 2D type formula?

Thanks for sharing that perspective, it'll give me hours of contemplation I think.

 

[DanGT86]

This raises a couple of questions immediately for me, and I'll probably come up with more as I process it.
1. What do you think an "ideal" map would look like? Is it an inclined plane? A ridge running up and back?
2. How would a controller achieve the varying automatic load response of an ICE? Is it via a complex "3D" speed/torque map instead of a 2D type formula?
Thanks for sharing that perspective, it'll give me hours of contemplation I think.

Sorry I don't have super definitive answers to these questions. I'm far from an expert. I just enjoy thinking out loud through these engineering problems with you guys.

I wish I knew what an ideal map would look like. If we are considering an ICE bike to be the gold standard of rideability then an ideal map would look like the map of the gas bike. This is why I mentioned the Dyno testing and datalogging of the gas motor. You need to know what your target curve looks like.

I think the torque might have to be limited and based on rpm to create the most "natural" feel. This is again assuming ICE feel is your goal. Its a bit of a shame to de-tune the nice torque of the electric motor but the goal is precision control not max power.

As for how the controller achieves a varying load response, that might kinda be the result of an RPM based torque curve. Seems to me if the delivered current were tied to the RPM then a clutch dump from 6000 rpm would be violent and a clutch dump from 2000rpm would be milder. This would feel like what we are used to. The ramp up RPM would be controllable at low rpm and the power wouldn't let you down at high rpm. If you needed to get to the power quicker you could slip the clutch a little bit to get up into the power. This sounds a lot like the experience of riding a gas bike to me.

The current+speed control logic is what will allow you to hold an RPM with the clutch in without the motor just continuing to spin up. Without that almost any throttle will cause the RPM to continue rising. Same with the wheel not running away in the air. I believe you will need to get the speed data from the motor output because its motor speed you are concerned with not wheel speed. I don't recall if the Nuc current+speed control is speed or rpm.

If it were me building this I would just put it together with a flywheel and clutch and a Nuc with current+speed control and see what happens. It might just be great. If not then you are probably going to have to build a device that looks at rpm data and buffers the throttle signal between your hand and the controller to "trick" it into doing what you want.

 

[speedmd]

So I can't work out what actually signals the switch between modes.

Thinking, when clutch level is out (fully engaged), you would be in current throttle mode. As soon as you pull the lever, it would default to speed throttle. When pulled, you are effectively limiting torque (current) drive force with the clutch finger. You are then somewhat more liberated to adjust motor RPM- stored flywheel energy for the task at hand.

 

[DanGT86]

Using the clutch lever to switch modes could be valuable but simple speed throttle is pretty twitchy and accelerates an unloaded wheel violently. At least on high amp cheap older controllers.

What about a speed+current throttle but program it extremely progressive so the first half of the travel is very forgiving but you can still have tons of amps last 50% throttle?

You could set it up such that the throttle curve gets more progressive and less sensitive the further the clutch is pulled. This would make the unloaded motor way easier to control but not sacrifice the high amp torque where you need it.

 

[speedmd]

Using the clutch lever to switch modes could be valuable but simple speed throttle is pretty twitchy and accelerates an unloaded wheel violently. At least on high amp cheap older controllers.

The throttle would default to speed only when clutch is pulled. You would tame the twitchy behavior (mostly felt by the clutch and flywheel) with a slide of 3 speed selector via profile settings. Once dialed up, it would be much easier to dial in RPM and flywheel energy on hand.

When clutch is fully released, it would be in current throttle and its intensity controlled by a separate slide or three speed switch. You could then set torque levels set by the switch to act much like a 125, 250 or 500cc if you wish. Thinking the two selectors would act independently to take max advantage.

 

[speedmd]

Using the clutch lever to switch modes could be valuable but simple speed throttle is pretty twitchy and accelerates an unloaded wheel violently. At least on high amp cheap older controllers.

The throttle would default to speed only when clutch is pulled. You would tame the twitchy behavior (mostly felt by the clutch and flywheel) with a slide or a 3 speed selector with desired ramp profile settings. Once dialed up, RPM -stored flywheel energy is easier to hold- dial in energy on hand with a speed throttle set up. Simple led tac could show this energy level for increased repeatability. Certainly, easier when you can hear the motor rev, but more silent electric drive may need something.

When clutch is fully released, it would be in current throttle and its intensity controlled by a separate slide or three speed switch. You could then set torque levels set by the switch to act much like a 125, 250 or 500cc if you wish. Thinking the two selectors would act independently to take max advantage.

 

[bikerpete]

Using the clutch lever to switch modes could be valuable but simple speed throttle is pretty twitchy and accelerates an unloaded wheel violently. At least on high amp cheap older controllers.

What about a speed+current throttle but program it extremely progressive so the first half of the travel is very forgiving but you can still have tons of amps last 50% throttle?

You could set it up such that the throttle curve gets more progressive and less sensitive the further the clutch is pulled. This would make the unloaded motor way easier to control but not sacrifice the high amp torque where you need it.

This sounds kind of a promising avenue.
I'm still not quite sure what would happen as the clutch engages and you keep constant throttle position? I think what you want is that the current is able to rapidly (read "almost instantly") increase to the limit set by the throttle position. How that works with the underlying progressive throttle ramp I don't know.
It's that thing about almost replicating some of the ICE torque/load characteristics.

I wonder the effect of having a speed+current throttle but the clutch actually modifies the blending of speed & current priority? The internal maths of the speed+torque, of which I have no understanding - maths is not my strong point!
So with clutch engaged it's very current dominant. As the clutch disengages it becomes increasingly speed dominant. Might make it easy to control RPM when unloaded, but give nice controllable drive on throttle alone?
Combined with the progressive throttle ramp it could be quite a nice feel?
I think it might be important that the throttle ramp stays consistent throughout, so you're not trying to juggle two dynamically changing controls which both relate to the amount of drive to the wheel.

That's very similar to:

The throttle would default to speed only when clutch is pulled. You would tame the twitchy behavior (mostly felt by the clutch and flywheel) with a slide or a 3 speed selector with desired ramp profile settings. Once dialed up, RPM -stored flywheel energy is easier to hold- dial in energy on hand with a speed throttle set up. Simple led tac could show this energy level for increased repeatability. Certainly, easier when you can hear the motor rev, but more silent electric drive may need something.

When clutch is fully released, it would be in current throttle and its intensity controlled by a separate slide or three speed switch. You could then set torque levels set by the switch to act much like a 125, 250 or 500cc if you wish. Thinking the two selectors would act independently to take max advantage.

But it would be a progressive changeover instead of trying to deal with a discrete switch action. Good control is reliant on nice analog feel, we're analog creatures.

As I said earlier - I really don't think a visual RPM indicator will be of any use at all - that's the last place you want to be looking as you're sizing up an obstacle while trying to regain balance and planning where you want to drop the front tyre. You fix your eyes on the target, then go there.
Also, trying to find the correct point to switch from one mode to another is going to be a constant challenge, the range of movement between clutch engaged and released is very small and depends on load (are you holding against a steep hill? A shallow slope? Balancing front-on and actively driving the front into the obstacle while stationary? Rolling down hill but building RPM to hop over a ditch?). Too many variables even before you think about clutch wear, oil viscosity change with temperature etc.

You'd have the same factors to deal with on a variable speed/torque, but at least things are all changing progressively, not in discrete jumps.

 

[bikerpete]

This raises a couple of questions immediately for me, and I'll probably come up with more as I process it.
1. What do you think an "ideal" map would look like? Is it an inclined plane? A ridge running up and back?
2. How would a controller achieve the varying automatic load response of an ICE? Is it via a complex "3D" speed/torque map instead of a 2D type formula?
Thanks for sharing that perspective, it'll give me hours of contemplation I think.

Sorry I don't have super definitive answers to these questions. I'm far from an expert. I just enjoy thinking out loud through these engineering problems with you guys.

I wish I knew what an ideal map would look like. If we are considering an ICE bike to be the gold standard of rideability then an ideal map would look like the map of the gas bike. This is why I mentioned the Dyno testing and datalogging of the gas motor. You need to know what your target curve looks like.

I think the torque might have to be limited and based on rpm to create the most "natural" feel. This is again assuming ICE feel is your goal. Its a bit of a shame to de-tune the nice torque of the electric motor but the goal is precision control not max power.

As for how the controller achieves a varying load response, that might kinda be the result of an RPM based torque curve. Seems to me if the delivered current were tied to the RPM then a clutch dump from 6000 rpm would be violent and a clutch dump from 2000rpm would be milder. This would feel like what we are used to. The ramp up RPM would be controllable at low rpm and the power wouldn't let you down at high rpm. If you needed to get to the power quicker you could slip the clutch a little bit to get up into the power. This sounds a lot like the experience of riding a gas bike to me.

The current+speed control logic is what will allow you to hold an RPM with the clutch in without the motor just continuing to spin up. Without that almost any throttle will cause the RPM to continue rising. Same with the wheel not running away in the air. I believe you will need to get the speed data from the motor output because its motor speed you are concerned with not wheel speed. I don't recall if the Nuc current+speed control is speed or rpm.

If it were me building this I would just put it together with a flywheel and clutch and a Nuc with current+speed control and see what happens. It might just be great. If not then you are probably going to have to build a device that looks at rpm data and buffers the throttle signal between your hand and the controller to "trick" it into doing what you want.

There's some good insights in there I think.

I really like the thinking around tying torque to RPM. With a progressive speed+torque throttle it should be really controllable and quite intuitive how exciting things are about to get.
The tricky part might be how to set this up so you still retain the ability to take advantage of the high torque/low rpm that is so unique to electric. So you can creep up a steep hill, then with a simple twist of throttle drive hard up an obstacle mid slope. That's the sort of thing that takes a fair bit of skill to slip the clutch on an ICE to have the power available when you need it. It would be ideal to avoid that need and just tap the big torque at low rpm inherent in electric.
It's a somewhat contradictory set of requirements.

 

[speedmd]

We could eliminate the speed throttle all together if speed ramps could be fed back to limit current control in some way. Much like traction control works on some systems. Once a specified amount of slip is detected, the speed is not increased significantly. Time (Ramps), intensity, and slip angles all adjustable. You would need torque and speed sensors and controller that understood how to deal with them.

 

[DanGT86]

Yeah, its definitely going to require some control based on rpm/speed. Otherwise the motor is just going to constantly run all over the place while you are slipping the clutch. If you command 50 motor amps output and then drag the clutch the motor will just continue speeding up until the controller achieves that 50amps. Who knows? Maybe that's a good thing?

This will for sure require finely adjusted PID control. It has to know to ignore the constant erratic transients like the wheel chattering for traction or the chain slop oscillating wildly as the suspension bounces.

I can see the general principles involved but the programming of such things is way out of my league.

Perhaps a simple current ramp with a set rate of amps/sec that results in output equal to an ICE motor is the solution. Keep it simple and just save some cash for the clutches you plan to burn through.

I wonder if Electric Motion would tell us what their throttle logic is like?

 

[trazor]

Interesting conversation. To me is obvious that an open source and powerful controller will be needed to modify and upload custom firmware to try these ideas. Have somebody tried something like this in a motorcycle?

https://www.aliexpress.com/item/1005002911195871.html?spm=a2g0o.productlist.0.0.2a8a47d6UCR1sL
or
https://flipsky.net/collections/electronic-products/products/flipsky-75100-with-aluminum-case-based-on-vesc-for-electric-skateboard-scooter-ebike-speed-controller

I don't see other options other than VESC based.

 

[bikerpete]

Perhaps a simple current ramp with a set rate of amps/sec that results in output equal to an ICE motor is the solution. Keep it simple and just save some cash for the clutches you plan to burn through.

I wonder if Electric Motion would tell us what their throttle logic is like?

Doesn't that just take us back to the problem that when you suddenly demand a big, fast current increase it simply isn't available? Such as when you bring rpm up (no load so low current) then drop the clutch (big sudden current demand).
Or am I misunderstanding what you have in mind?

Oil bath clutches withstand an incredible amount of slipping in a trials situation. 400+ hours of almost constant clutch usage is not at all uncommon. By that stage most of the bike is getting pretty tired. Mine has nearly 300 hours on it and is working as well or better than ever. I'd guess the friction plates are perhaps 1/3 worn? And I slip the clutch much more than strictly needed as I've been spending a lot of time working on improving my clutch control.
The thing is there's usually not a great deal of load or high RPM while slipping the clutch. It's often close to idle, or virtually stationary with just enough drive to stay still against front brake while you crank the RPM, or a fraction of a second where there is high rpm and you're firing out the clutch. Hillclimbs, particularly with poor traction are possibly the hardest on the clutch as you use slip it to enable instant variation of drive along the way. Electric with it's high torque reduces the need for that, if the problem of runaway RPM increase when the wheel loses traction is solved.

As for EM's throttle logic, not sure it's that interesting really as it doesn't seem to be that great.
I have wondered if the EM Silixcon controllers are locked or if it's possible to connect to them and read/write. Presumably they have a route for sowftware upgrades? But perhaps not - they don't have a great reputation for post-sale support from what I've heard.

To me is obvious that an open source and powerful controller will be needed to modify and upload custom firmware to try these ideas. Have somebody tried something like this in a motorcycle?

I haven't looked into the VESC much, maybe it's got the capability? I don't know.

The bigger problem is probably finding someone with the skills, interest and time to do the programming. That's going to be a long process involving some interesting and possibly innovative solutions I suspect.

 

[DanGT86]

Doesn't that just take us back to the problem that when you suddenly demand a big, fast current increase it simply isn't available? Such as when you bring rpm up (no load so low current) then drop the clutch (big sudden current demand).
Or am I misunderstanding what you have in mind?

It might take us back to the original problems. I guess I'm just thinking that if you allowed a current/torque ramp equal to or more aggressive than a gas motor you could just tame it with the clutch and see how it feels. We might be overthinking it.

As for controllers, My gut feeling is that it will be hard to find a controller thats programable enough. I think the most reasonable solution is to make a throttle buffer device using an arduino or raspberry pi or some other microcontroller. Feed it the rpm, clutch, load etc. your throttle signal goes into it and it sends the new calculated throttle signal out to the controller. This way you can make all the changes you need and not have to find a full controller manufacturer to cater to your unique needs.

 

[owhite]

Interesting conversation. To me is obvious that an open source and powerful controller will be needed to modify and upload custom firmware to try these ideas. Have somebody tried something like this in a motorcycle?

https://www.aliexpress.com/item/1005002911195871.html?spm=a2g0o.productlist.0.0.2a8a47d6UCR1sL
or
https://flipsky.net/collections/electronic-products/products/flipsky-75100-with-aluminum-case-based-on-vesc-for-electric-skateboard-scooter-ebike-speed-controller

I don't see other options other than VESC based.

Trazor and others on this thread -- sorry I'm a bit of a late comer to the conversation, and I know that we're probably just kicking around hypothetical ideas -- but let me mention my experience with VESCs is that they are extremely flexible.

They export and import data via the serial. I have so far just used it to read data, but I have seen other posts of using the UART to control the board. I dont completely track what it is that you want to accomplish but using an external controller to read and write commands is certainly possible.

Also, let me confirm that it is possible to customize the firmware, i have read several posts on this as well. This of course require someone with strong electrical engineering training to take the plunge.

Finally let me mention Trazor showed a very good example of a commercial VESC, and I launched a thread about a DIY VESC over here. I definitely recommend using the commercial one over a DIY for reasons I can explain if needed.

 

[bikerpete]

It might take us back to the original problems. I guess I'm just thinking that if you allowed a current/torque ramp equal to or more aggressive than a gas motor you could just tame it with the clutch and see how it feels. We might be overthinking it.

As for controllers, My gut feeling is that it will be hard to find a controller thats programable enough. I think the most reasonable solution is to make a throttle buffer device using an arduino or raspberry pi or some other microcontroller. Feed it the rpm, clutch, load etc. your throttle signal goes into it and it sends the new calculated throttle signal out to the controller. This way you can make all the changes you need and not have to find a full controller manufacturer to cater to your unique needs.

I'm probably not really understanding something here.
If you had a current/torque ramp the same as that of an ICE under load wouldn't that make it's response pretty radical under no load again?

The throttle buffer idea is certainly intriguing. It'll take me a while to process that idea and consider the minimum inputs you might be able to get away with and still get an effective system.

Trazor and others on this thread -- sorry I'm a bit of a late comer to the conversation, and I know that we're probably just kicking around hypothetical ideas -- but let me mention my experience with VESCs is that they are extremely flexible.

They export and import data via the serial. I have so far just used it to read data, but I have seen other posts of using the UART to control the board. I dont completely track what it is that you want to accomplish but using an external controller to read and write commands is certainly possible.

Also, let me confirm that it is possible to customize the firmware, i have read several posts on this as well. This of course require someone with strong electrical engineering training to take the plunge.

Finally let me mention Trazor showed a very good example of a commercial VESC, and I launched a thread about a DIY VESC over here. I definitely recommend using the commercial one over a DIY for reasons I can explain if needed.

That's great, thanks.
It doesn't look like the Flipsky really has enough power capability, although it might be enough in something like the small dob or a full DIY lightweight trials.
Now to find that unique person with the skills, knowledge and interest to take on the programming project.

 

[trazor]

That's great, thanks.
It doesn't look like the Flipsky really has enough power capability, although it might be enough in something like the small dob or a full DIY lightweight trials.
Now to find that unique person with the skills, knowledge and interest to take on the programming project.

I don't think it will be extremely difficult. What is proposed here is to add some analog inputs, tweak throttle curves, PID values, etc. It isn't super low level stuff.

I've a question for @owhite:

What do you consider is the best commercial VESC based (open source firmware) that can handle power around 5KW?
I can't find any that is designed for 20S (nominal 72v, max 84). Seems the market is very focused on skateboards.

 

[owhite]

What do you consider is the best commercial VESC based (open source firmware) that can handle power around 5KW?
I can't find any that is designed for 20S (nominal 72v, max 84). Seems the market is very focused on skateboards.

So I recognize that youre asking about 20s, but let me work my way down in case other people stumble on this thread.

For 24S, 200A VESCs:

I thought maybe this would be reasonable: GO-FOC HI200/HV200.

These HI200s are appearing all over the place on aliexpress. Unfortunately I ordered one from the above link it never arrived. So strike one in the customer service category. Also, Florida citizen Barncat has his not happy experience with one. As stated in my post I wondered if this was basically a configuration problem -- which would have been nice to test if one arrived.

The only other thing I know of is the TRAMPA VESC 100V 250A. I have not tried it, but given the price I am gonna guess it'll at least a) arrive at your house and b) have good support.

And of course just because I am one of those guys I will mention the 10kW DIY version, simply to state you a) better love electronics b) like supply chain issues and c) enjoy the smell of exploding capacitors because youre gonna get all three if you go this route. It does work though, I pull 10kW out of it all the time.

For 20s, 100A VESCs:

The flipsky 75100 looks promising. Flipsky seems to be a company that has been around for a while. I'm buying one for my next build with a QS165, but I dont think that will be on the road until late summer.

Then after doing a web search I found the UBOX 75V 100A. I had no previous knowledge of them, but the board seems reasonable.

Personally I think the price of the 20s VESCs make them very attractive for our hobby. My view is controllers are a consumable item -- they're flakey and it's better to get a cheap one and expect it's gonna explode at some point -- rather than getting an expensive one and hating yourself when it blows.