Torque vs cadence based PAS riding experience

[afzal]

I had this query in mind for some time and there was a brief discussion about this @ https://endless-sphere.com/forums/viewtopic.php?f=3&t=114541#p1693809, to avoid hijacking it, starting a new topic.

I would like to hear from people who have rode ebike with cadence as well torque based PAS on how they compare the riding experience. Also how a combined torque plus cadence assist fare in comparison.

Background: I am developing a pedelec controller that would provide a bionic experience to the rider without using torque or force sensor. If a satisfactory riding experience could be provided by the controller, I intend to sell it and thus have a way to fill my stomach

I have rode on-off & cadence based assist type hub motor ones, not yet torque based assist ones (there is only one locally available here, a middrive one & the costliest).

Almost all e-bicycles available around my location are on-off PAS type, problem is that rider feels like controller decides how it should move rather than the rider (varying PAS levels does help a little). Motor provides same level of assist irrespective of pedal speed or effort, slow moving is next to impossible and inability to change PAS levels quickly in traffic being another issue.

And while playing with VESC that has cadence based assist support, riding has been a pain (unless throttle is used) with a gearless e-cycle, lot of steep inclines here. So I had to add on-off type PAS assist support in VESC to alleviate the struggle, though I despised on-off type, for my usual route that was better than cadence assist.

 

[afzal]

Relevant posts pulled from the other topic,

Do you get a natural cycling feel when using cadence based PAS (as opposed to using torque controlled one) ?, During uphill & start, does the cadence based PAS assist provide a comfortable riding experience ?

Depends on what you feel is natural, since natural for a bike is no assist. There's a few ways to adjust the cadence based PAS with the CA to make it feel more natural. If you normally downshift when riding up a grade, then the CA has the ability to increase assistance when your cadence increases, so you don't have to make up all of the difference when load increases. You can adjust the ramp to what feels best for you, but it's not going to be like a torque PAS where you need to increase your own effort to get more assist, which may feel more natural, too. From a dead stop, the cadence based PAS doesn't provide any assist, where a torque based PAS does, but that's what a throttle is for.

Do you get a natural cycling feel when using cadence based PAS (as opposed to using torque controlled one) ?, During uphill & start, does the cadence based PAS assist provide a comfortable riding experience ?

For me, personally, on startups, because my trike plus me weigh over 500lbs, the answer is no, because my knees and right ankle aren't able to do normal pedalling under loads, so a torque sensor would work better for that (I'm slowly and intermittently working on a solution to do that in my Nano Tidbits thread).

But for most people that can pedal normally, it does work well, as long as they are shifting into the correct gearing for the pedalling for the conditions they're under at that moment.

For uphills, as long as I am already moving, then yes, it works perfectly fine, because I have enough motor power to compensate for my legs' lack of that.


To me, it is much more natural than a throttle, and far more controllable than the on/off full-power style of PAS control (which I've used on other bikes like my Fusin Test Bike that had a medium-large geared hub in the rear wheel), even when the assist level is selectable via buttons from the display. (having to change that level all the time while riding in traffic is distracting, unnatural, and sometimes actively dangerous for me).


A combination of torque and cadence works "best" for more situations, but is more complicated, and the Cycle Analyst doesn't use the torque sensor in the way necessary to make it most natural and easiest to use. That's what I want to change with my Nano addon if I ever get that done.

The torque and cadence sensors together with the CA work great once you get teh cranks moving, so if you're in a low enough gear at startup, then it does work like having bionic legs, basically like you describe, and you can setup the balance between human input and motor output just about any way you want.

My only problem with it is at startup from a complete stop, and only because of my knees/ankle plus the very heavy weight of my cargo trike and loads plus me.


But there are torque sensors like the Erider BB type that provide a simple throttle signal that can directly control any typical controller, even if it has no PAS input (if it does have one, then it's cadence output can connect to that if you like, but I don't know how the setup would work unless the controller is fine-tunable). Using the Erider BB with the CA, though, would let you tune it however you wanted, with the Erider as the throttle input and not the torque sensor input. (still using the Erider cadence output to the PAS cadence input). Then the CA throttle output to the controller throttle output.

I have never used the Erider BB, but it seems like a more universal solution than the Thun or TDCM that I have used so far (which don't output a standard throttle signal and have to go thru the torque sensor input on the CA).

By natural for ebikes, I meant a bicycle on steroids feel :wink: i.e. for example, when 1Nm torque is applied on the pedal, wheel should get 5Nm push, as torque exerted on the pedal is what the rider feels as the effort.

I would like to know from people who have rode ebike with cadence as well torque based PAS & how they compare the ride experience between these two. Probably I will start another thread instead of hijacking this one.

So you want to work harder when climbing a hill. You can do that with either cadence or torque based PAS, but you won't have the assist from a standing start with cadence. I don't do the math between my legs and assist, since I don't care about it, and I mainly care about how much or little effort I add to the equation.

I like my legs providing the same level of effort on flat ground up to about and 8% grade and try not to go below 15mph. I never shift from 46:13 until the grade greater than 8%, above that I downshift and cadence increases, and so does the assist, and downshift more for more assist. I have to pedal harder, or adjust the assist up (or more easily apply some throttle) when the grade is above 15% or so (or just twist the throttle and stop pedaling when I'm lazy).

I don't think there's anything "natural" once you add assist. You need to aim for what feels and works for you. I got the CA with the intention of moving to a torque based PAS if the cadence PAS wasn't good enough, but truthfully you can replicate a toque based PAS or cadence based PAS just by manipulating the throttle alone.

 

[AHicks]

Here's how I think of it, kinda like a keyboard (think user interface). A junk keyboard controlling the nicest computer imaginable is going to leave you with a bad taste in your mouth when you sit down for a session at that console. Scrap that junk keyboard and replace it with a nice one chosen for it's ease of use, and your experience will likely be light years better!

I've ridden with torque sensing, on a Bafang Ultra based bike with the Frey "Smooth" tune installed. THAT bike, in my experience is the ultimate in well done torque sensing. Prior to the installation of the Frey Smooth tune, running the OEM Bafang programming, that same bike felt like that junk keyboard above. NOT a very pleasant experience.

Now, for cadence based PAS, I've ridden several bikes that used a speed based firmware in the controller. In PAS1 for instance, you start off pedaling, and as long as you are pedaling (makes no difference how hard) you will accelerate to 10-12 mph. If you want to go faster, you switch to a higher PAS. All fine and dandy. But what if traffic won't allow you to run that 10 -12mph that PAS 1 offers? What if you want to go 6mph on a narrow trail for instance?

Enter a KT controller and display. This is the equivalent of the much nicer keyboard - AND - unlike the cheap cadence based controllers, able to be set up for many of the users 'druthers. On top of that, the real icing on the cake, is that the KT has an option to be power based rather than speed based. When enabled, using totally different logic, speed will no longer have any relevance. It will now be power based - and further - YOU have control over how much power, using the various available parameter settings.

NOW, that 6mph speed is easily maintained, and that same amount of power will not change regardless of your speed - unless you have a speed limit set (YOUR OPTION!). Need more assist/power? Go up a PAS level.

My point to all this, is that it's about the software that's been installed in the controller. There is software that's been dreamed up by those that have never ridden a bike before, and there is the opposite, software that's VERY well done. Clearly the result of a lot of thought and experience. The kicker? When talking cadence based PAS, there is no difference in the hardware being used. It's ALL about the firmware within the controller. The logic itself, the method used to control it (speed vs. power based for instance), and how adjustable it is - to allow riders options that can make a night and day difference in the riding experience....just like the keyboards above. If you want to make a difference, IMHO, it's about the software (firmware). The hardware works just fine as is. -Al

 

[dogman dan]

I really like torque based. When I feel some torque on my feet, I creep up the throttle. Modern torque based PAS might be just as good. But cadence based PAS sucks balls in my opinion. Your brain and a throttle is a really fast computer. Any kind of lag in the adjustment just causes discomfort. Having to increase your cadence is discomfort.

What I liked best about e bikes was leaving my cadence fairly high, and watts output easy, but not too easy for the entire ride. Up hill, on the flat, into the wind, always that same easy 75 watts, always that same fairly quick cadence, but never having it get too easy till on a downhill, when I'd stop pedaling for a bit.

But thats old, sick dogman. When young and fit, I never shifted out of high gear, never sat on the seat, and could break a chain.

 

[AHicks]

I really like torque based. When I feel some torque on my feet, I creep up the throttle. Modern torque based PAS might be just as good. But cadence based PAS sucks balls in my opinion. Your brain and a throttle is a really fast computer. Any kind of lag in the adjustment just causes discomfort. Having to increase your cadence is discomfort.

What I liked best about e bikes was leaving my cadence fairly high, and watts output easy, but not too easy for the entire ride. Up hill, on the flat, into the wind, always that same easy 75 watts, always that same fairly quick cadence, but never having it get too easy till on a downhill, when I'd stop pedaling for a bit.

But thats old, sick dogman. When young and fit, I never shifted out of high gear, never sat on the seat, and could break a chain.

This is the beauty of POWER based (vs. speed based) cadence sensing PAS. It WORKS! With the KT you can set up PAS1 to something that works for you, based on preferred cadence or whatever, and change it to something higher as/if required. I set mine up to start with anything from just under 100w, to maybe 150w.

It's amazing such a small change (speed based to power based) can make in the rider experience. If you ride a power based setup (Bafang mid drives have this option as well), you are spoiled!

 

[afzal]

In the hindsight, topic subject should have been "Torque vs cadence controlled PAS riding experience", mea culpa.

Now, for cadence based PAS, I've ridden several bikes that used a speed based firmware in the controller. In PAS1 for instance, you start off pedaling, and as long as you are pedaling (makes no difference how hard) you will accelerate to 10-12 mph. If you want to go faster, you switch to a higher PAS. All fine and dandy. But what if traffic won't allow you to run that 10 -12mph that PAS 1 offers? What if you want to go 6mph on a narrow trail for instance?

Based on the behavior above, I think it is not cadence control (though is cadence based PAS), rather the on-off type control mentioned by amberwolf & the Basic PAS, constant power/throttle types as in ebikes.ca

To remove the confusion I myself caused due to the terms used, in brief,

PAS detection can be 2 of types,
1. Basic PAS
2. Torque sensor

while, PAS control can be 3 types,
1. On-off
2. Cadence
3. Torque

(Left out Cadence plus Torque control scheme to keep the above list simple)

Both On-Off & Cadence type control scheme uses the same Basic PAS which detects pedal speed. In the case of On-Off type, once pedal rpm is above a certain value there would be a constant power/throttle, else no assist, while in Cadence control type, speed or torque (& perhaps power?) is proportional to the pedal rpm. e.g. in VESC cadence control detects the pedal rpm & controls the motor torque in proportion by varying the motor current.

hope I did not confuse more :wink:

Enter a KT controller and display. This is the equivalent of the much nicer keyboard - AND - unlike the cheap cadence based controllers, able to be set up for many of the users 'druthers. On top of that, the real icing on the cake, is that the KT has an option to be power based rather than speed based. When enabled, using totally different logic, speed will no longer have any relevance. It will now be power based - and further - YOU have control over how much power, using the various available parameter settings.

NOW, that 6mph speed is easily maintained, and that same amount of power will not change regardless of your speed - unless you have a speed limit set (YOUR OPTION!). Need more assist/power? Go up a PAS level.

I thought KT can have a torque sensor input as well, was torque sensor used in the above scenario ?, even otherwise, cadence control type can achieve it as mentioned in ebikes.ca link above

... If you want to make a difference, IMHO, it's about the software (firmware). The hardware works just fine as is. -Al

Thanks for the input, most of my career was in software development, hope I can make a difference

 

[afzal]

This is the beauty of POWER based (vs. speed based) cadence sensing PAS. It WORKS! With the KT..

I thought KT can have a torque sensor input as well, was torque sensor used in the above scenario ?

Okay, I take it that you were not using torque sensor

 

[ilu]

I would categorize the terms slightly differently:

PAS detection
1. Basic PAS (only measures pedal rotation on/off)
2. Cadence speed
3. Torque amount

PAS control can take into account one or more of these and in addition variables set through the display unit connected to the controller. KT controllers by default use only basic PAS detection mode, but control modes are different. In speed control mode you can choose a target maximum speed from the display, and motor power is at maximum until this speed is reached. In power control mode you choose the desired power level from the display, and the motor power is the same regardless of speed.

With Open Source firmware for KT controllers it is possible to utilize cadence measurement or torque sensor reading in the power control and the formulae for these can be finely tuned by the user. Naturally with more fine tuning more natural assist feeling can be achieved.

In my opinion a torque sensor is required for the best feeling, although straightforward torque sensor reading to motor power ratio is not optimal, but also cadence reading needs to be taken into account to modify this ratio.

 

[AHicks]

Any PAS/cadence sensor I've ever seen does NOT measure speed. They're used to detect crank movement, period. The crank is moving, or it's not. On-Off.

A variable done with the cadence sensor is crank sensitivity. How far does the crank need to turn prior to the controller turning on the power? The controller does this by counting the pulses generated as a magnet goes past the sensor. If it turns on with the first magnet passing the sensor, this makes for a pretty sensitive setup, great for low speed control often enjoyed by those riding trails for instance. This vs. those set up to limit power until the crank has turned much further - from 1/4 turn to 1 full turn. Downside of the one magnet/first pulse power up is accidental starts. For instance if you stopped with a magnet just coming up on the sensor, and you bump the crank accidentally while stopped, you can get a short burst of power that's completely unexpected. Some people don't care for that potential, and many will suggest the controller should be set so a minimum of 2 magnets must pass the sensor prior to power up.

 

[ilu]

Sensors themself generally don't output the cadence amount reading, it depends on the controller software which is detecting the sensor signal. For example check the KT Controller Open Source firmware I mentioned.

 

[AHicks]

Sensors themself generally don't output the cadence amount reading, it depends on the controller software which is detecting the sensor signal. For example check the KT Controller Open Source firmware I mentioned.

Let's try and keep this relevant.

1. It'll be a cold day before I go through all that open source info looking for a point YOU are trying to make. I would even question it's relevance here.

2. If you have the time and money, ANYTHING is possible. I'm talking about a STOCK KT controller, and over the counter cadence sensors.

 

[amberwolf]

Any PAS/cadence sensor I've ever seen does NOT measure speed. They're used to detect crank movement, period. The crank is moving, or it's not. On-Off.

There are systems that do use your cadence (crank speed) to control the amount assist. They dont' need special cadence sensors...just a controller that actually measures the speed of the cranks using the signal already coming from any of the typical over the coutner PAS cadence sensors.

You can do it with the Cycle Analyst v3, for instance, on any controller that uses a throttle input and doesnt' require it's own PAS input. I do this on my SB Cruiser trike.

I recall that there is a mode in the open-source firmware for KT controllers (and possibly others) that does this; I need to look in the thread for that to find the reference.

There have been others mentioned now and then over the years that do it (usually prebuilt OEM bikes), but I can't remember any of their names.


Most cadence-only PAS sensor controllers don't do this, and o1nly detect whether or not the cranks are moving, as you say, and either assist or not (at whatever level the system is set to).

BTW, just because you yourself arent' willing to do something that will let you use a system as cadence-controlled assist amount, doesnt' mean it isn't relevant, doesn't exist, or that others aren't using them or can't use them.


EDIT: A bit of googling around and I found a site
https://leedsbikes.com/leed-e-bike-kits/electric-bike-throttle-control-pedal-assist/
that specifies down the page a ways that their PAS bikes do indeed use the speed of pedalling to control the speed of the bike.

What does pedal assist mean? Well, there are two different styles to consider: cadence sensor or torque sensor. The most common is the cadence sensor. This is the type we offer on our 500 Series Kits. Cadence sensor systems usually come with a ring of magnets mounted on the pedal crank and a sensor fixed to the bottom bracket.

As you pedal, the sensor detects your cadence and increases the speed output from the motor accordingly. Our PAS comes with above industry standards with twelve magnets. This means that our kit features a rapid-response system turning the motor on immediately as the user begins the motion of pedaling. A PAS cadence system is installed on the pedal crank

I poked around a little more and didnt' find another specific example of a commercially available prebuilt system that does cadence-control of assist in a variable way, but I haven't found a good set of search terms that find much relevant results either. (lots of PAS and cadence discussions, but not much that even mentions this mode at all, though I know it exists out there). Even grin's site doesn't come up, though I know the CAv3 does it.


I found a search term that does bring up some info on some systems, but it is used in two completely different ways: Torque Simulation. In regards to PAS, it is used in at least some cases to mean the cadence-controlled variable assist. But it also seems to be used for some sinewave controllers to mean something to do with the less noisy operation they have over squarewave, and/or being a "fake torque (current) control" of the motor instead of the common voltage control most controllers use (evne most sinewave).

I'll keep poking around, and post any further links to such systems I can find, for anyone that wishes to use one (or just read about them).

This might be the KT OSFW info but it is not clear on how it operates for the user:
https://opensourceebikefirmware.bitbucket.io/windows_instructions/

Ride modes

Throttle
Torquesensor with assistance linar to human power
Torque-Simulation ( fixed current steps with cadence dependent ramp)

This thread
https://endless-sphere.com/forums/viewtopic.php?t=62374
discusses the S06S's torque simulation, which appears to be cadence-controlled variable assist.

I think this one does it too:
https://www.pswpower.com/products/36v-48v-1000w-1500w-brushless-dc-torque-simulation-square-ebike-electric-bicycle-hub-motor-controller-with-right-output-33
but the info on that page is insufficient to tell.

There are some pages that say the torque simulation only means that PAS controls the power instead of the speed, but still uses the separate levels to control how much PAS gives how much power (meaning not variable based on cadence), so the term torque simulation seems to be used for two completely different things regarding PAS.

Very confusing.

Easier if they just all called it cadence-controlled variable assist or something even more clear.

 

[AHicks]

AmberWolf-
KT uses speed based control or "imitation torque" control to describe power based in the manuals. Your choice. "Imitation torque control" is Chinglish for power based speed control. See the P3 parameter in any of the manuals. The OEM controllers are NOT able to handle a signal from a torque sensor - unless you want to mess around with the "open source" stuff.

Speaking of "open source" while some might be interested in the potential, my bet is that less than 1 in 500 will be willing to get his hands dirty messing with it. So fine, there is a lot of potential there, but speaking in practical terms, I believe darn few would ever attempt it. That thought is why I question it's relevancy when it comes to practical application.

Been messing with KT conversions for a while now, so I've sort of been there done that with much of it.

I am NOT familiar with the Grin options, basically because I'm boycotting them until the update their display. Would LOVE to mess with one of their Gmac systems some day, IF they ever decide to bring the display up to current standards. If you say they can measure speed using the cadence sensor, I find that a bit of a stretch, but I'll take your word for it. My bet is though, they're using/getting a speed signal related to the motor rpm, not cadence speed.

Regarding this:
"As you pedal, the (cadence) sensor detects your cadence and increases the speed output from the motor accordingly."

I believe this fellow doesn't really understand his system. I've never seen or heard of a system that can/will do that. He says he's using a 12 magnet ring. What would happen for instance, if somebody installed a PAS sensor with more or less magnets on the ring?

Further, to reinforce this gentleman's lack of understanding, this part of his quote proves he's struggling with how a PAS system works regarding sensitivity.

"This means that our kit features a rapid-response system turning the motor on immediately as the user begins the motion of pedaling."

It DOES NOT turn the system on immediately. It starts counting PAS sensor signals immediately and powers the system up when the programmed number of pulses has been achieved. Clearly I could be wrong here, but I doubt that seriously. IF what he said were true, he would VERY likely be deling with frequent false start issues, triggered accidentally when the crank was bumped.

-Al

 

[amberwolf]

Thanks for the clarification of the torque simulation definition; it is still not clear that all systems that use this term use it the same way, though. I have never used any system that uses this term so have no direct experience with what they might mean by it.

You're probably right about the OSFW; the same is also true of other DIY solutions like using a CA to add functionality to controllers that don't have certain things (like the cadence-controlled variable assist we're talking about).

I am NOT familiar with the Grin options, basically because I'm boycotting them until the update their display. Would LOVE to mess with one of their Gmac systems some day, IF they ever decide to bring the display up to current standards.

They dont' have a display, in the sense that the controllers you use (KT, etc) have a display that communicates with the controller and is required to make the controller work as designed with all features.

Their controllers (at least, the Grin-specific ones, Grinfineon, Phaserunner, Baserunner, etc) don't use a display of any kind. Grinfineons don't have any settings to change, and the PR and BR use a setup program to change their settings, from a computer or device, non-realtime. (they may have finished the phone app but I can't remember if that allows realtime settings changes or data display; if so, then those can use your phone as a display in a way similar to the KT/etc LCDs).


If your'e referring to the CA (cycle analyst), it's not a display; it's a completely separate and independent computer, not part of any controller system, and doesnt' do the things a "controller display" does.

You might personally call it semantics, but it is a very significant difference I want to be sure is clear to other readers; among other things, there have been numerous noobs that wanted to use the CA instead of their controller's display but couldnt' because they need that display to access controller settings as they ride (they had no other way to do it).

FWIW, if you are boycotting everything else they sell just because you don't want to use the CA due to it's screen design, well, that's up to you. I understand it even if I don't choose to do so.

If you say they can measure speed using the cadence sensor, I find that a bit of a stretch, but I'll take your word for it.

It's most definitely true. It's very easy to setup in the CA; you can see the various modes right on the CAv3 info page where they show the setup menus.
https://ebikes.ca/product-info/grin-products/cycle-analyst-3.html
If you're really interested in the details, they also have a manual, and videos.

My bet is though, they're using/getting a speed signal related to the motor rpm, not cadence speed.

Not for PAS, they are not. That would not allow you to control the system. That would just get a speed signal for the motor RPM (and perhaps vehicle or bike speed), and that does happen on a completely separate input for a completely separate purpose.

I built my entire trike myself, and used the CAv3 to read a cadence sensor that controls the assist amount directly via pedal RPM. I discuss it now and then in various places in my SB Cruiser thread, mostly the last 2-3 years (I forget when I started using it this way) but that's a lot of reading for you unless you're really interested.

I ride it using the pedals to control the speed (I dont' recall if I used the power, current, or speed control method, but I am effectively controlling speed with the pedals).

I have a throttle for certain startup situations and in the case of leg/foot injury preventing me from pedalling, but I hardly ever have to use it.

Regarding this:
"As you pedal, the (cadence) sensor detects your cadence and increases the speed output from the motor accordingly."

I believe this fellow doesn't really understand his system. I've never seen or heard of a system that can/will do that. He says he's using a 12 magnet ring. What would happen for instance, if somebody installed a PAS sensor with more or less magnets on the ring?

Then the speed it detects would be higher or lower, and the amount of assist it provides would be more or less, proportionally.

Additionally, the less magnets in it the slower it's response time to changes in pedal speed, like when you first start pedalling from a stop.

But if it has any user-changeable settings, then it is likely that the system has a setup menu like the KT and others, that lets you change the number of poles (or magnets) in the PAS sensor specifically for this reason. If the number doesn't match then it will still work, it just won't be the same proportion of assist.

Further, to reinforce this gentleman's lack of understanding, this part of his quote proves he's struggling with how a PAS system works regarding sensitivity.

"This means that our kit features a rapid-response system turning the motor on immediately as the user begins the motion of pedaling."

It DOES NOT turn the system on immediately. It starts counting PAS sensor signals immediately and powers the system up when the programmed number of pulses has been achieved. Clearly I could be wrong here, but I doubt that seriously.

You could be right.

Personally, I know that different people certainly have different definitions of "immediately". It will definitely take some amount of time to begin responding, but if the controller he is using has settings for how quickly it begins to respond, and those settings go as low as a single pulse to trigger startup, it could be virtually immediate response as you wouldnt' have to rotate the cranks very far at all with 12 magnets. Just 1/12 of a rotation, the angle between two numbers on an analog 12-hour clock.

IF what he said were true, he would VERY likely be deling with frequent false start issues, triggered accidentally when the crank was bumped.

It might very well have such false starts. The CA lets you tune things, so they are less common, I had it happen a lot when first setting mine up for as instant a response as possible.

I eventually tuned mine for less of that, and to get better response I just move the cranks backwards a bit first then forward. Because of my specific chainline, a backpedal doesn't work like on a regular bike, it causes sag in the top of the chainline that gives me the slack to quickly move the cranks forward again and get started from a stop without putting pressure on my bad knees and ankle, under most situations (the throttle is there when this won't work).

 

[donn]

The GRIN Pedal Assist page goes over a lot of this at some length, and maybe a little easier to follow.

And while playing with VESC that has cadence based assist support, riding has been a pain (unless throttle is used) with a gearless e-cycle, lot of steep inclines here. So I had to add on-off type PAS assist support in VESC to alleviate the struggle, though I despised on-off type, for my usual route that was better than cadence assist.

If I follow you here, GRIN's comment on this from that page:

In the case of a bike with a single speed gear, you might want the very opposite behavior, where the power is highest at low pedal RPMs when you are just starting off from a standstill or slowing down as a result of encountering a hill, and then decrease when you are pedaling fast and up to speed.

I don't see in there how they propose to make that work, though.

Unless you can incorporate other data from the motor and wheel system. I've been assuming that the cadence algorithm is used to as the equivalent of a throttle input that adjusts the motor voltage. Motor RPM follows from that voltage, but the load is also a factor, and thus wheel speed. If your cadence indicates that you should be going 5mph, your ultra sophisticated cadence control could up the voltage until you're actually going 5mph, which voltage will vary depending on load. That would take you up the hill, where a cadence control that doesn't sense motor RPM will bog down, because its 5mph voltage is really 5mph only if there's no load. Maybe.

 

[dogman dan]

I still insist that if you don't have a torque sensor, use the one on the bottom of your feet. It's pretty sensitive, and connected to the best computer on earth, your brain. It may take a lot of practice, say 5000 miles, but if you are paying attention, you can get the perfect ride. Once you aren't thinking so hard about it, it's like landing the plane while talking to the tower on the radio. Humans are quite capable.

Pick a speed. Pick a gear that matches that speed, at the cadence you prefer. Then when you feel more pressure on your feet increase the throttle, when you feel less, decrease it. Ride with cadence always same, speed always same, with of course adjustments for conditions, such as WOT across that busy 4 lane intersection.

 

[afzal]

The GRIN Pedal Assist page goes over a lot of this at some length, and maybe a little easier to follow.

And while playing with VESC that has cadence based assist support, riding has been a pain (unless throttle is used) with a gearless e-cycle, lot of steep inclines here. So I had to add on-off type PAS assist support in VESC to alleviate the struggle, though I despised on-off type, for my usual route that was better than cadence assist.

If I follow you here, GRIN's comment on this from that page:

In the case of a bike with a single speed gear, you might want the very opposite behavior, where the power is highest at low pedal RPMs when you are just starting off from a standstill or slowing down as a result of encountering a hill, and then decrease when you are pedaling fast and up to speed.

I don't see in there how they propose to make that work, though.

I did give a thought on that, as that control scheme would prevent a superman feel on flats, did not pursue that path

 

[donn]

I did give a thought on that, as that control scheme would prevent a superman feel on flats, did not pursue that path

The motor RPM linked system I proposed would, though. Like any cadence system, though, it would have the annoying limitation that it would offer no help to initially achieve the cadence, so it would be better to get going before you hit the hill. Or use a throttle or the boost button that I believe some of them have.

 

[donn]

I still insist that if you don't have a torque sensor, use the one on the bottom of your feet. It's pretty sensitive, and connected to the best computer on earth, your brain. It may take a lot of practice, say 5000 miles, but if you are paying attention, you can get the perfect ride. Once you aren't thinking so hard about it, it's like landing the plane while talking to the tower on the radio. Humans are quite capable.

Pick a speed. Pick a gear that matches that speed, at the cadence you prefer. Then when you feel more pressure on your feet increase the throttle, when you feel less, decrease it. Ride with cadence always same, speed always same, with of course adjustments for conditions, such as WOT across that busy 4 lane intersection.

RIght - the more I learn about PAS systems, the happier I am to not have one. I may not be as particular about it - when I can, I just let cruise control take over and settle for similar results.

But the OP's objective here isn't so much to enjoy cycling, as to discover a variation on PAS that others think they might enjoy. There's probably room for that.

 

[afzal]

I found a search term that does bring up some info on some systems, but it is used in two completely different ways: Torque Simulation. In regards to PAS, it is used in at least some cases to mean the cadence-controlled variable assist...

This might be the KT OSFW info but it is not clear on how it operates for the user:
https://opensourceebikefirmware.bitbucket.io/windows_instructions/

Ride modes

Throttle
Torquesensor with assistance linar to human power
Torque-Simulation ( fixed current steps with cadence dependent ramp)

... S06S's torque simulation, which appears to be cadence-controlled variable assist.

There are some pages that say the torque simulation only means that PAS controls the power instead of the speed, but still uses the separate levels to control how much PAS gives how much power (meaning not variable based on cadence), so the term torque simulation seems to be used for two completely different things regarding PAS.

AmberWolf-
KT uses speed based control or "imitation torque" control to describe power based in the manuals. Your choice. "Imitation torque control" is Chinglish for power based speed control. See the P3 parameter in any of the manuals.

Torque simulation mode in the KT open source firmware explanation: https://endless-sphere.com/forums/viewtopic.php?p=1321496#p1321496. But don't know whether the original KT firmware power based speed control behaves the same way, it can be confirmed by some one who has used both original firmware & open source firmware on KT.

 

[AHicks]

"But the OP's objective here isn't so much to enjoy cycling, as to discover a variation on PAS that others think they might enjoy. There's probably room for that."

Now that's an understatement!

Regarding controlling your power/speed with the cadence sensor. I don't see how that will work either and for the same reason mentioned just above. There's no way to factor changes in load. With power increasing as cadence is increased, what happens when a hill is encountered? I don't know about others, but I can say that MY cadence is going to slow considerably (no doubt). If we are using cadence to control power, when cadence slows due to an increase in load, power is decreased. When faced with an increased load/hill, decreased power is the LAST thing I would want to see in this situation!

As far as torque sensing, I have an enormous amount of respect for that. As an owner of a Bafang Ultra powered bike with UART based programming, I have a very healthy respect for everything going on in that controller as it's rolling down the street. Because the OEM programming was SO terrible, I was forced to become familiar with it. I made a few improvements, but it turned out they were just the tip of the iceberg. The Frey "Smooth" tune changed nearly every parameter (4 pages worth) from the settings it came with, and that tune flat works. There's little doubt here that programming is going to be tough to beat.

HOWEVER, regarding the MUCH simpler cadence only based systems, I couldn't agree more with the idea they can be EASILY improved on. Point being, let's not debate the advantages of torque sensing vs. more conventional PAS systems. I think it safe to assume the torque plan is the better one, BUT the more conventional PAS has plenty of room for improvements. That is the topic I would love to discuss further..... -Al

 

[E-HP]

There's no way to factor changes in load. With power increasing as cadence is increased, what happens when a hill is encountered? I don't know about others, but I can say that MY cadence is going to slow considerably (no doubt). If we are using cadence to control power, when cadence slows due to an increase in load, power is decreased. When faced with an increased load/hill, decreased power is the LAST thing I would want to see in this situation!

I think maybe between a speed and a current input, one may be able find a clever way to somehow address the increased load and speed. On the cadence, mine actually increased when I get to a hill, since I downshift. Wouldn't work and a single speed, but if you change gears, which increases cadence, the CA ramping can take care of a decent sized hill before any throttle is necessary.

 

[afzal]

I could not yet understand how KT power assist mode feels for the rider. Al, if you can explain on how the ride feels, it would be helpful. I think it would be difficult to write down that experience, but thought will still ask

Like while starting from standstill, would you feel motor providing push proportional to your pedalling speed & once you reach a certain pedalling speed whether it would provide a constant push . Or whether motor provides same level of push as soon as you as start pedal irrespective of the pedalling speed. During uphill, does the motor push gets reduced etc.

 

[amberwolf]

If I follow you here, GRIN's comment on this from that page:

In the case of a bike with a single speed gear, you might want the very opposite behavior, where the power is highest at low pedal RPMs when you are just starting off from a standstill or slowing down as a result of encountering a hill, and then decrease when you are pedaling fast and up to speed.

I don't see in there how they propose to make that work, though.

Some of the parameters can be made negative, so that they have greater influence at lowest pedal RPM, and less at higher pedal RPM. With single speed, assuming you're always providing pedal input and trying to maintain the same speed, then the lower RPMs usually mean you're having to pedal harder to maintain that same speed.

If you're not trying to maintain the same speed, it isn't as simple, and you may need to use some of the other settings in the various CA menus to help compensate for this--it still wont' work quite like a torque-sensing PAS would in those loaded situations, but it can still be pretty good, and very usable for most situations for most people.

Unless you can incorporate other data from the motor and wheel system. I've been assuming that the cadence algorithm is used to as the equivalent of a throttle input that adjusts the motor voltage.

Whether it adjust motor voltage or motor current depends on your controller (common controllers adjust motor voltage, FOC and other torque-controlling systems adjust motor current), but it also depends on the throttle mode you choose in the CA; since it monitors battery current it can do a fair approximation (in many situations) of controlling the "motor current" (really battery current) or power, rather than voltage (speed), whcih gives a differnet feel to the control you have over the bike.

However, you are essentially right, in that the CA takes whichever PAS mode you're using and creates a throttle output to the controller (because that's the only input it has to a controller to tell it to do anything), based on however you've set the CA PAS up plus all the other inputs to the CA for any limiting, etc., that you have set up in the CA.

Motor RPM follows from that voltage, but the load is also a factor, and thus wheel speed. If your cadence indicates that you should be going 5mph, your ultra sophisticated cadence control could up the voltage until you're actually going 5mph, which voltage will vary depending on load. That would take you up the hill, where a cadence control that doesn't sense motor RPM will bog down, because its 5mph voltage is really 5mph only if there's no load. Maybe.

That's partly where some of the other throttle modes come in, and the various other settings in both the PAS menus and the other limiting menus. There is a lot of tuning available, though it can take a while to get it just the way you want it, and some of it requires knowing a few of the tricks (like the parameters that can take negative values to cause significantly different behavior). I barely scratch the surface of it's ability with my trike's usage and my knowledge of it--Teklektik (wherever he went) could do a whole lot more with it, and Justin_LE could probably work out any particular setup in his sleep. :lol:

 

[AHicks]

I could not yet understand how KT power assist mode feels for the rider. Al, if you can explain on how the ride feels, it would be helpful. I think it would be difficult to write down that experience, but thought will still ask

Like while starting from standstill, would you feel motor providing push proportional to your pedalling speed & once you reach a certain pedalling speed whether it would provide a constant push . Or whether motor provides same level of push as soon as you as start pedal irrespective of the pedalling speed. During uphill, does the motor push gets reduced etc.

OK from a stop, this old man uses a touch of throttle to get the bike moving while I collect my balance. Not much, and the throttle is generally released within just a few feet (maybe 6'?). I may or may not use the throttle again for the remainder of the ride. From there, how quickly the bike accelerates will be based on the PAS level. Clearly I'm pedaling by this point, so contributing something in addition to what is being supplied by the motor. The amount of power supplied to the motor in PAS1 is adjustable using the parameters. I value low speed control, so my PAS1 will be set to provide something in the area of 75-150 watts. This will result in PAS2 having something in the range of roughly twice that amount , 3 will roughly double that, 4 takes quite a jump, and of course PAS5 will have everything available - up to the point where you have capped the max amps/wattage. You are NOT able to customize the power in each PAS level (just PAS1). Power levels from there are handled with a built in algorithm you have no control over but will be based on how you've set PAS 1. Additionally, max speed can be capped if you like (using available parameters), and the amount of power available to the motor won't let your speed exceed that point.

Because you are using power based PAS, the amount of power (from a practical standpoint) remains the same within each PAS level. The 75-150w you set for PAS1 will remain the same regardless of speed. If there is cadence available that power will be there from the first crank of the pedals to as fast as you want to go (25mph+!). This is why you keep the power in PAS1 relatively low. It's just a trickle of power, but generally all you need at speeds under 10mph - down to where it's hard to keep your balance. Same story in PAS2. You'll have about double what you have your PAS1 set at, and that amount of power will remain about the same regardless of speed, as long as the pedals are turning.


-Al