New Geared Hubmotor variable-regen e-braking system that still freewheels, by ChargeBike

Mclewis1 said:

Regen with a specialized freewheeling SX2 using a Superharness and enabled via a simple brake sensor trigger so I can keep my hydraulics? ... perfect, I'll be waiting.

Click to expand...

That's the hope, and in fact it's possible to have it work without even a simple brake sensor too, as long as you don't simultaneously brake and throttle at the same time. The system knows when you are using the throttle (and/or pedal assist) and in that scenario the disc rotor isn't spinning at all.

When you aren't powering the motor then the disc rotor should be spinning at the normal rate of a planet carrier (which is slightly slower than the hub shell). If the controller system sees that the disc rotor suddenly stops spinning when there isn't power flowing to the motor, it knows that the user has applied the brakes and can automatically go into the braking control mode.

It's pretty crazy. Alon from Charge has been playing with that mode on his own bike to good effect and I believe will have one of the demo builds at eurobike running without any brake sensor at all, but I think for a commercial product it is still prudent to have a digital add-on ebrake signal.

offGridDownUnder said:

If Grin is on it, I would expect there is some benefit, but I don't see it yet. The system still requires the electronics, a trickle from the battery to overcome cogging (usually mentioned as the reason for a freewheel in the motor) uses less power than is regained from the resulting regen, and it's mechanically as or more complicated.

Click to expand...

I'll always be a diehard champion of the direct drive motor with all its simplicity. But it seems a larger portion of the potential ebike riders prefer a geared motor because because of the smaller size, better torque density, and freewheeling. This new option has the advantages of a geared motor of (good torque, small size, perfect freewheeling) while ALSO allowing upwards of 95% of stopping energy to go into regenerative braking rather than brake pad wear.

But on top of this, there are three pretty unique traits that no other system has:

1) The regen braking force automatically tracks how hard the rider presses their existing brakes, no extra sensors or control mode needed. You just throw the hub on your bike and brake like you did before the motor was added, and regen just 'happens'.

2) When the batteries are either full or can't take a charge because say they are too cold, the system can automatically steer some or all of the braking energy back into heating the discs rather than to the batteries which can't absorb it. You can squeeze the brakes and get a given amount of braking torque under any scenario, but the control electronics can decide how much of that goes into regen vs how much goes to heat.

3) A large number of the failures we've had with locking the clutch on conventional geared motors is a result of the back and forth alternating torque. This makes it hard to secure the axle and and also gives reliability problems with the link between the planet carrier and shaft. With this new system, the torque on the motor axle is only ever in one direction, and all the regen torque instead gets coupled to the frame via the disc caliper. For something smaller like the SX2 motors just the axle flats alone should be sufficient with no torque arm required.

It IS true though that there is still some amount of brake pad wear due to the slow slip rate, while a traditional regen system can activate regen with zero contact at all. But this should still be like a 10x or even 20x increase in the brake pad longevity. That pad wear hasn't been empirically quantified but it stands to reason that if the disc rotor is spinning 10-20 times slower than normal when you brake, then you are getting 10-20 times less wear on the pads.

Anyways time will see what impact this option will have in the years ahead! My hope is that many of those who were formerly drawn towards mid-drive systems can be brought over to the hub motor camp with a technology like this once it's mature.

The livestream is starting in 10 minutes if you want to check in for the video release and follow-up conversation!

justin_le said:

But it seems a larger portion of the potential ebike riders prefer a geared motor because because of the smaller size, better torque density, and freewheeling.

Click to expand...

On reflection, a success like this may spur more efforts towards gear longevity, and repair-ability. My first experiments with my current "mid-drive" mount used a GMAC and now an SX2. The next phase of my own trike involves mounting an All-Axle as the "mid-drive", as I'm designing for a 20-year working lifespan - I think the gears are a weakness, although valuable. I'm able to bypass this by using sprockets to provide the gear reduction, but for many bi/tri-cycles internal gears are more practical.

I notice there was mention of a superharness. Is that really necessary? I’m imagining only a controller (ex, BaseRunner) capable of regen is needed.

slaphappygamer said:

I notice there was mention of a superharness. Is that really necessary? I’m imagining only a controller (ex, BaseRunner) capable of regen is needed.

Click to expand...

From what I gather it needs extra logic beyond just being regen capable, for measuring disc brake slip etc. Sounds like eventually they may get this functionality into the runner series controllers but ASI doesn't have it at the moment. So the superharness is the solution they have for now. I imagine they also plan on getting it into the Cycle Analyst as well though I've read it's really hard to get new features for the CA because of the outdated chip they use.The other solution they mentioned in the QA is that they've written firmware for a couple open source controllers with the extra logic added. Vesc and Lishui I believe it was.

This seems really clever. It's similar to a two speed gearbox I created a while back that used a planetary gear and a sprag clutch so that with the motor in reverse the output would be a different speed but the same direction. This application is really neat. There's just two things I'm a bit unsure of:
1) If the controller fails, faults etc... do you lose braking?
2) is there a max brake force limited by the motor torque; I know you can probably get enough for road use but for serious mountain biking where you actively use the brake for control?

The control algorithm seems pretty simple for the controller... If you have a half decent FOC then you know the motor speed and you know the wheel speed so you know the disc speed... Then torque (q current) is just proportional to the disc speed with an offset to prevent the disc locking... Really neat.

> 1) If the controller fails, faults etc... do you lose braking?

They mentioned on the livestream that they have a relay where if the controller/battery dies, it just shorts the phase wires. This locks up the motor enough to provide enough braking power to comply with European regulations for having a reliable mechanical brake.

I missed the Q&A section. I see that there are 3 of the 6 bolts that secure the rotor to the motor shell.
Are those 3 bolts secured with locktite or similar to keep water out? Maybe that isn’t even an issue.

Would it be a problem when it comes time to replace the brake rotor? Since 3 bolts make contact with the clutch, would it be an issue to realign the bolts if the wheel moves. I’m assuming after removing the 3 bolts that secure the brake rotor to the clutch, that when the wheel moves, the position of the motor shell, in relation to the clutch, would change then those 3 bolt holes on the clutch would be realigning to the 3 bolt holes in the motor shell.

lnanek said:

> 1) If the controller fails, faults etc... do you lose braking?

They mentioned on the livestream that they have a relay where if the controller/battery dies, it just shorts the phase wires. This locks up the motor enough to provide enough braking power to comply with European regulations for having a reliable mechanical brake.

Click to expand...

If the motor is big enough. They mentioned the braking torque requirement (70 Nm, or maybe less ; I think there was a correction in the video, but I was left with the impression that there was no conclusion on the right number), and that some smaller motors can't meet it.

Epithemeus said:

From what I gather it needs extra logic beyond just being regen capable, for measuring disc brake slip etc. Sounds like eventually they may get this functionality into the runner series controllers but ASI doesn't have it at the moment. So the superharness is the solution they have for now. I imagine they also plan on getting it into the Cycle Analyst as well though I've read it's really hard to get new features for the CA because of the outdated chip they use.The other solution they mentioned in the QA is that they've written firmware for a couple open source controllers with the extra logic added. Vesc and Lishui I believe it was.

Click to expand...

Thanks Epithemeus, this reply is correct on all point! In the interest of moving forward with development within an existing ecosystem, the Superharness gives us a nice platform that we have full firmware control over. Ideally it could be a firmware option for the ASI controllers and we'd have it natively supported without the harness using a Phaserunner or Baserunner, but that requires a collaboration with ASI that can (we can tell you from experience) be very slow moving!

I imagine that the people at Charge may make their custom VESC firmware available for DIY'ers wanting to experiment with the technology too, though that would be up to them.

Gruesome said:

If the motor is big enough. They mentioned the braking torque requirement (70 Nm, or maybe less ; I think there was a correction in the video, but I was left with the impression that there was no conclusion on the right number), and that some smaller motors can't meet it.

Click to expand...

The EU rules for EN15194 specify that the rear brake needs to be able to apply a 220 Newton stopping force, so the actual torque that translates to depends on the wheel size:

220N on a 29" wheel is 81 Nm
220N on a 26" wheel is 72 Nm
220N on a 20" wheel is 56 Nm

So even a small/medium sized geared motor would generally have no problem meeting EN15194 in a 20" wheel size, but if you want this compliant in a 29" wheel then it would require one of the larger motors in the eZee/GMAC/G62 power realm.

There isn't anything we can find in the EN15194 wording that prevents the braking force from coming primarily from regen. The way it is written it's like the authors weren't even considering it as a possibility, which is kinda surprising in a document that is supposed to be a continent wide ebike standard!

justin_le said:

The EU rules for EN15194 specify that the rear brake needs to be able to apply a 220 Newton stopping force, so the actual torque that translates to depends on the wheel size:

220N on a 29" wheel is 81 Nm
220N on a 26" wheel is 72 Nm
220N on a 20" wheel is 56 Nm

So even a small/medium sized geared motor would generally have no problem meeting EN15194 in a 20" wheel size, but if you want this compliant in a 29" wheel then it would require one of the larger motors in the eZee/GMAC/G62 power realm.

There isn't anything we can find in the EN15194 wording that prevents the braking force from coming primarily from regen. The way it is written it's like the authors weren't even considering it as a possibility, which is kinda surprising in a document that is supposed to be a continent wide ebike standard!

Click to expand...

Might I add that in an emergency brake, when both front and rear brakes are slammed, the physics and weight distribution will cause the rear wheel to skid at relatively low braking torque, the almost infinite strength of a mechanical brake won't give any advantage over Freegen, Freegen's ABS on the other hand...

Well, 220N is about the braking force you can get with 25kg loading on a rubber wheel on asphalt (friction coefficient of about 0.9; unless you ride a Tesla of course, then you get to use way higher friction coefficients, to enable those advertised sub-2.7s 0-60 mph times...).

Alon-ChargeBike said:

Might I add that in an emergency brake, when both front and rear brakes are slammed, the physics and weight distribution will cause the rear wheel to skid at relatively low braking torque, the almost infinite strength of a mechanical brake won't give any advantage over Freegen, Freegen's ABS on the other hand...

Click to expand...

Since the "slippage" of the brake rotor in a FreeGen setup is being constantly monitored (and acted upon) with respect to the actual motion of the fixed part of the hub (and therefore the wheel/tire) shouldn't it be relatively easy to add a form of ABS?

This sounds quite obvious but I had to ask.

It is exciting to see a novel idea show up and get a thorough technical explanation of all the compounding benefits when applied to an existing hub motor design. Personally rooting for a vesc option since I have that hardware already.

Mclewis1 said:

Since the "slippage" of the brake rotor in a FreeGen setup is being constantly monitored (and acted upon) with respect to the actual motion of the fixed part of the hub (and therefore the wheel/tire) shouldn't it be relatively easy to add a form of ABS?

This sounds quite obvious but I had to ask.

Click to expand...

I think they replied yes to this question in a video. They said it could be added via software update. Probably in a future version?

Is there any chance in the future of getting a GMAC with thinner laminations and steel gears so that it can handle 5+ kW peak for short periods? It would be nice to have a pedalable vehicle that still has no cogging torque losses when the battery is dead, but can also do regen and reliably/regularly make enough power to accelerate like a car without failure.

Also, is there any possibility for a front wheel version of a GMAC in the future as well for trikes?

In any event, this freegen system is promising. It may be in my Milan SL's future once available.

justin_le said:

My hope is that many of those who were formerly drawn towards mid-drive systems can be brought over to the hub motor camp with a technology like this once it's mature.

Click to expand...

I think it would help a whole lot if freegen hubs could be offered in thru axle variants. This because, for example, 99.9% of all aerodynamic narrow q factor bikes (like gravel and modern road) use 142mm x 12mm rather than 135mm. Furthermore, the most desirable drivetrains like 12-13 speed are only on 142mm x 12mm.

Aerodynamic bikes with narrow q factor and 12-13 speed should be much better at regen than the usual upright mtn bike with wide q factor and 8 to 11 speed.

P.S. Some power consumption figures for various types of bikes to add to the discussion (from highest to lowest power consumption using default tire configuration provided by calculator):














Too bad road bikes (which have the lowest power consumption when using the drops) are not commonly found in 135mm (re: the road bike world on a mass scale went from 130mm (e.g. Ultegra 6800 11 speed --->https://bike.shimano.com/en-EU/product/component/ultegra-6800.html ) straight to 142mm x 12mm thru axle (e.g. Ultegra 8000 11 speed-->https://bike.shimano.com/en-EU/product/component/ultegra-r8000.html).

ebike4healthandfitness said:

I think it would help a whole lot if freegen hubs could be offered in thru axle variants.

Click to expand...

For sure. The good news is that at Eurobike this year there were at least a half dozen companies showing off thru-axle geared motors (up from just a couple last year), though most were still in testing/development stage and weren't yet in full production.

Of those we're hoping that at least one or two will have an internal structure that lends itself to the freegen modification. The things that rule it out would be using steeply angled helical gears with thrust bearings (since regen puts the axial thrust in the opposite direction), double stage reductions, or if the gearing assembly is on the drive side and not the disc side of the motor.

That said I don't think there will be any shortage of people interested in this even with the limitation of slotted dropout frames! So the first focus will be just on that.

The Toecutter said:

Is there any chance in the future of getting a GMAC with thinner laminations and steel gears so that it can handle 5+ kW peak for short periods? It would be nice to have a pedalable vehicle that still has no cogging torque losses when the battery is dead, but can also do regen and reliably/regularly make enough power to accelerate like a car without failure.

Click to expand...

The MAC style motors have quite high torque with relatively narrow gears and would be the prime candidate for having at least one metal gear if offered in a freegen variant, so that will be on the roadmap for testing out for sure.

Thinner laminations don't impact the torque or acceleration or your ability to dump in 5kW for short times where you are dealing almost entirely with copper losses as the prime heat source. That would mostly just help with the efficiency at low torque levels and high RPM's.

Through axle and one sided gear motors would be a nice option for lower speed trikes and mobility devices

Here are my findings after experimenting with freegen on a MAC geared motor (front). Regen is not strong enough on hills or slow speeds to SAFELY substitute a front disc brake. So I wrote a LISP script on VESC to activate electronic handbrake on low RPM where it sends positive current thru 2 phases to generate holding torque. Handbrake will still slip at 100A under 100 RPM so you need to send a handbrake signal continuously on the VESC (10hz is fine). This has the effect of ABS slip/hold...however the e-handbrake will overheat the motor rather quickly rendering it useless as a motor AND brake. It will also make a horrific groaning sound as it struggles to brake the motor.

The only safe and practical solution was to mount a small hydraulic brake (from a 1/5 rc model car) plumbed into the main brake line using the bleed port from the main caliper. The 80mm aluminum rotor is connected to the clutch planet carrier thru a hub. I removed the spider from a 203mm shimano ice-tech rotor and mounted that directly to the motor hub. No compromise braking and regen. Both brakes are activated by the same lever and can be modulated like normal.

Great information and happy to see you still here!

Here is the LISP code for the VESC. It will activate the e-handbrake using 100% of your max brake current when brake signal maxxed out (you should have a variable voltage brake sensor rather than on/off). May bake your motor. Use at your own risk.