Why do torque sensing e-bikes have cutoff delays?

[hobbyvac]

There is probably a better name for it but its the annoying feature where you stop pedaling and the motor runs on for another 1-3 seconds. I have an early TSDZ2 that does not do this but all newer torque sensing mid drives I've tried will run on for a while after you stop pedaling. I was able to tune out some of it with a Bafang mag ultra but not all of it. I won't consider buying any new bike if I can't turn off the delay. I think its dangerous on trails especially in high power settings and you shouldn't have to adapt to it. Which motors are tunable or don't have this stupid feature? The newer Bafang Ultra's are no longer programmable.

I like a natural and transparent feel. New bikes all feel over boosted to me because they kick in at 15-20% power instead of ramping up smooth, and then over run a couple seconds when you stop pedaling. Who thinks that's a good idea?? I don't want the motor to modify my power inputs or timing, just do what I'm doing only add a little more.

 

[amberwolf]

It depends on what it uses to detect crank rotation.

If it is *only* detecting torque and tracking that with assist output, there will be virtually no delay in response. The catch with these is that it is possible for a single point of failure in the sensor path to create a voltage that causes the system to just always run and not be able to shut it off (except by ebrakes or power switch, etc--stopping pedalling doesn't stop the system--the same thing can happen with a throttle-only system, or a throttle-overrides-all system).

If it also has a cadence sensor it uses to detect crank rotation, whether or not it uses that to determine assist response amounts, then it will have a delay based on how many poles the sensor has (how many magnets in it's "ring", etc), and whatever it's programmed to wait for after it receives no more pulses from that sensor. The fewer the poles, the slower the response time, in addition to the wait time. They use this partly as a safety mechanism to prevent runaway operation in the event of torque signal path failure.


As an alternate method of control, you might want to check out the Cycle Analyst V3.x, as the newer versions allow using just the torque sensor rather than cadence, AFAICR, so no delay in response. It's an add on to any motor system with a throttle input and the ability to respond *only* to the throttle, as long as the torque sensor on the motor system outputs a 0-5V range signal that can be calibrated in the CA's setup. (the TSDZ2 sensor requires the TSDZ2 controller to drive and read it, so it wouldn't work with this, for instance).

Generally you want to use a "dumb" controller with the CA, so the CA can do all the processing and limiting of the controller inputs. A smart controller with the ability to do it's own processing and limiting may interfere with what the CA is trying to do, and make setting the system up more complicated, and system behavior less predictable.

I actually use the CA with just a cadence sensor for a pretty natural feel on my SB Cruiser trike. (it has a torque sensor but I am not using it--just the cadence sensor built into it). In my case because it's a cadence sensor with a middling pole count it has some delay, but it's short....and all i have to do to cut off "instantly" any motor power is pedal backwards a teeny tiny bit (like operating a coaster brake except there's no braking from it on my system).

 

[hobbyvac]

Thanks for the info, I figured it wasn't simple.

 

[E-HP]

I have an early TSDZ2 that does not do this but all newer torque sensing mid drives I've tried will run on for a while after you stop pedaling.

Which drives have you tried?

 

[AHicks]

The Bafang shift sensor, stock on many/most Ultra equipped bikes, will defeat any delay set up in the controller parameters. The downside of that switch is when you are at low speed with low cadence, and finding yourself in a position where you need a lot of power RIGHT NOW. Downshifting is going to trigger that shift sensor and could stall the bike (in the low speed/low cadence scenario). A MUCH better work around in that scenario is to use some of the gobs of torque you have on hand, and quickly increase available power by going up a couple of PAS levels. Won't hurt a thing, and in my experience will generally work out pretty well.

The rest of the time, let the shift sensor due it's thing. I've been riding bikes for quite a while, so that little delay while shifting, on my part, is something I don't even think about, so I don't notice that switch is even there for the most part. I will say that a LOT of guys do have some trouble with the shift timing. It's different, and may take a few miles to get used to. My take anyway, FWIW. -Al

 

[hobbyvac]

I've tried the Bafang Mag ultra from Luna with many tweaks and was able to shorten the delay some. A newer Bafang Ultra 620 that's not programmable with CAN bus, Turbo Levo which had a Brose motor I think. Bosch CX, some Yamaha and several others at a demo day. They all had long delays or decay after pedaling and a high startup point of say 15% instead of ramping up from zero. I never use thumb throttles so not commenting on that, just pedal response. I know brake switches and shift sensors can defeat the cutoff delay but they are just bandaids that aren't needed if you don't have long delays. You shouldn't add more tech to cover up for flaws that shouldn't be there. I know you don't need the cutoff delay because my cheap old TSDZ2 doesn't have it and it shifts perfectly with no need for brake or shift switches. Humans shouldn't adapt to technology flaws, it should adapt to us.

One thing is that my speed sensor magnet got ripped off so I haven't had a speedo since new. I think the TSDZ2 has both a torque senor strain gauge and a crank position sensor internally? I never hacked the controller or changed settings. What possible reason could there be to want your motor to run on after you stop pedaling??????

 

[amberwolf]

The Bafang shift sensor, stock on many/most Ultra equipped bikes, will defeat any delay set up in the controller parameters. The downside of that switch is when you are at low speed with low cadence, and finding yourself in a position where you need a lot of power RIGHT NOW. Downshifting is going to trigger that shift sensor and could stall the bike (in the low speed/low cadence scenario). A MUCH better work around in that scenario is to use some of the gobs of torque you have on hand, and quickly increase available power by going up a couple of PAS levels. Won't hurt a thing, and in my experience will generally work out pretty well.

The rest of the time, let the shift sensor due it's thing. I've been riding bikes for quite a while, so that little delay while shifting, on my part, is something I don't even think about, so I don't notice that switch is even there for the most part. I will say that a LOT of guys do have some trouble with the shift timing. It's different, and may take a few miles to get used to. My take anyway, FWIW. -Al

The problem I see with all those is that they don't address *ceasing power* when you stop pedalling, which is the issue the OP needs to address.

Well, with the possible exception of the shift sensor, assuming that by "defeat any delay" it means it will instantly shut off power to the motor...but you would need to still be pedalling to do gear shifting on most bikes, or risk a chain jam or derailment depending on the derailer system, and the OP needs to stop the motor when stopping pedalling. I suppose the shift sensor input on the controller could be connected to a button on the handlebars instead for manual deactivation, or a device (arduino, etc) could be created to pass the cadence sensor thru to create a signal to activate the shift sensor signal whenever pedalling is stopped.

 

[amberwolf]

Humans shouldn't adapt to technology flaws, it should adapt to us.

That is correct--it's the whole purpose of technology. But it is not the philosophy of many of the technology creators, in any field. I used to do beta testing for various music creation / composition products, and generally they didn't care what people actually did with the stuff, only that they (the company) got to make whatever they felt like they wanted to sell...whether that product could actually be used for it's intended purpose (easily, or even at all in some cases) by it's intended audience was irrelevant to them. There's some technology creators out there that do care and do try to fix flaws or prevent them from being built in in the first place, but not many AFAICT.

Sometimes it's a money thing...but more often it's just their way of doing things.

What possible reason could there be to want your motor to run on after you stop pedaling??????

None, but the common design of the systems with the cadence (PAS) sensor being used to detect whether or not the cranks are moving means there is an inherent delay of some amount in either starting or stopping. The amount of actual delay between controller detecting that and reacting may be longer because of poor software design.

There could be ways created to "instantly" detect crank motion start/stop but they require different hardware than is presently used on those systems that have the delays.

The TSDZ2 sensor system has a large-diameter coil of wire that's energized by the controller which then detects changes in the induced field in the coil, and such a system can be made sensitive enough to detect virtually any crank motion (and apparently is so in the TSDZ2), in addtion to it's cadence sensor. I don't know that this is why there's a difference in this system vs the others you've tried, but it's my best SWAG.

If a system has a torque sensor of other designs in addition to cadence, then the system could be made sensitive enough to detect if you are or are not moving the cranks, but under some conditions wouldnt' be reliable enough to depend on--it would either not stop even though the cranks were stopped, because torque was still applied just a teensy bit due to the way these function and are calibrated, or would stop even when still pedalling slowly because insufficient signal was detected, unless the right combination of cadence and torque detection was used.


It's not an impossible task...just not one important enough to the system designers to tackle (even though it is of critical importance to riders in traffic and other precision control situations). THe typical response I hear / read for this and other such things is "oh, you'll just get used to it", which is not true if you are in a situation where the delay in power shutdown causes a fatal collision.

(there are even systems where the ebrake does not instantly cut power, but has some small delay, or where throttle release does the same thing, and many systems where release of ebrake or engagement of throttle does not instantly deliver power again, any of which could also be fatal in the wrong (if rare) situation).

 

[E-HP]

None, but the common design of the systems with the cadence (PAS) sensor being used to detect whether or not the cranks are moving means there is an inherent delay of some amount in either starting or stopping. The amount of actual delay between controller detecting that and reacting may be longer because of poor software design.

I used it for a couple of years, when I had the controller set for slide regen. I actually set a fairly high delay via the Cycle Analyst, since as soon as assist would stop, braking would start, so I didn't want to start braking just because I stopped my legs to pull up a sock or something. I generally ride with my fingers on the levers anyway, so it never caused an issue.

And, I never use PAS offroad.

 

[amberwolf]

Ah, well, that's an unusual situation I haven't seen before, and is a case of human adapting to technology limitations.

So I guess I'll have to say "almost none".

However, if you could have, would you have had it cease motor output when crank motion stopped, if you could have done this without causing braking?


Speaking further to the forcing of human adapting to technology instead of technology adapting to human: Using ebrake levers to stop a motor from outputtting power isn't intuitive, unless you also want to slow or stop the bike. It can be learned but that is human adapting to technology limitation instead of fixing the limitation. (it could be said that a feature of the technology was being used to fix a limitation of the technology, but there are better more intuitive ways to do that).

Stopping pedalling should just stop producing forward motive power the instant pedalling stops, and just coast, for the typical bicycle rider's built-in training, the same way that unassisted pedalling does.

Another example of technology limitation that's "easily" fixed is variable regen in the CA (and some motor controllers) controlled by throttle input after using a brake lever to engage it, thus using a slower non-intuitive two-step process for something that should be quick, one step and ingrained-reactive (like using a brake lever already is for a bicycle rider).

Instead, it could allow a second analog input from a brake lever to directly control the regen level. Rather than adapt to the technology, I "fixed" this problem for myself on the SB Cruiser with a "second throttle" controlled by the brake lever and a relay to switch between the two engaged by the ebrake switch on the lever, so the process is the same for me as if it were just a normal brake lever, so I have better reaction time than someone that has to do the two step process (even if they've gotten used to it...and in a complete panic stop there is a chance they will not remember the throttle step if they first trained on normal bicycles, but there's no chance of that with the corrected technology ).

 

[E-HP]

Ah, well, that's an unusual situation I haven't seen before, and is a case of human adapting to technology limitations.

So I guess I'll have to say "almost none".

However, if you could have, would you have had it cease motor output when crank motion stopped, if you could have done this without causing braking?

I was using my coasting button for a while, that applied 50W so the bike would perpetually coast. I got rid of it when the pot was malfunctioning, and found the CA "solution". I'm back to cadence PAS with ramping and instant off now, since I'm not using regen.

 

[BalorNG]

That's due to power averaging. I've independantly rediscovered it when I was programming my own series hybrid pedal assist.
If you simply instantly amplifiy rider's power strokes and ease to nearly zero in dead spots you create double peak chain tension (or actual triple-quadruple!) and lose quite a bit of motor efficiencincy because it does not 'like' such a regime at all - much better is to take peak torque and 'smear' it around the 360 deg of crank rotation at least. Human power is delivered by a 'sine wave'!
In theory, it should be possible to programm instant responce into the system, but you do that on your own risks - much greater drivetrain wear and chaince of bending chainrings/breaking chains, greater chaince of wheel slip, less range.

 

[afzal]

Torque exerted by a rider is pulsating within one pedal rotation. There would be 2 peaks (around the instant where each leg naturally exerts max effort, i.e. when pushing down) in each rotation.


If the controller instructs the motor to provide torque in proportion to the raw torque sensor signal, ride would be jerky. To avoid this, i.e. to smoothen out the torque sensor feedback within a rotation, a low pass filter and/or ramp function would be used. And this by its nature, causes the delay.


More powerful the electric drive train is, more apparent the effect would be.

 

[BicycleJoe]

I'm on my third year with a Luna X-1, Bafang M600. It is a mountain-bike and I ride roads and some very intense difficult trails. My bike has about two seconds of run-on power after I stop pedaling. At first I got surprised by this unexpected energy sending me. I quickly came to expect the added run-on. I already have one or two fingers on each brake lever, always. Knowing it will continue to propel for a bit after I stop pedaling is key.

I only bothered to chime in on this weeks old thread because I love the extra couple seconds of power. There are at least a dozen challenging sections of my home single-track trail where I rely on the run-on to get me past the obstacles. On narrow uphill trail sections with big rocks and stumps lining the trail bed you have to stop pedaling at times to avoid whacking a trail feature. On an incline, on a heavy ebike, your momentum is critical. Without the extra couple seconds of run on these bikes would be hobbled on the tough stuff, reletive to a pedal bike.

The run-on is there for reason. I understand that this is not the OP's use case. I wish I knew how to solve this issue for them.

 

[amberwolf]

FWIW, you don't need the "run on" to get the power needed, you would just keep applying throttle or pedals for that same amount of time, and get the same result (without the hazard that the run-on creates in certain situations).

 

[afzal]

I'm on my third year with a Luna X-1, Bafang M600... I wish I knew how to solve this issue for them.

It runs VESC, right ?, then I can point you on the changes to be made on yours, but whether it is possible to flash it, you have to check with your seller.

It seems VESC M600 code does both low pass filter & ramping.

1. to remove the low pass filter, change the low pass filter third argument from 0.1 to 1.0
2. remove the code inside if condition

 

[Chalo]

I like a natural and transparent feel.

Me too. That's why I use throttle only and a wholly independent pedal drive. Pedal assist is a cool gimmick, but it's a needlessly hard/expensive problem to make it feel like amplified muscle power.

 

[hobbyvac]

Exercise is my main objective so throttles are out of the question. I have motorcycles for that. My TSDZ2 mid drive is pretty close to feeling like bionic legs and I love it. Most others just have the annoying cutoff delay which ruins the experience for me. Seems like a simple adjustment and I'm surprised so few complain about it.

 

[Chalo]

Exercise is my main objective so throttles are out of the question.

There's a proven kind of bike for that. Ask at any bike shop.