Mid-drive vs hub motor: Experiences

We've covered more ground on the ebikes the past two months than we did on the pedal bikes for the past two years. 😁 Best invention ever!
I don't think ebikes would have come along without the latest battery technology. The old nickel-cadmium batts just would never have worked like these new packs.
 
Actually hub motors can have really nice torque sensing bottom brackets. See eBikes.ca
 
Using a throttle doesn't contribute to the rider's health condition. Why not to buy an electric motorbike or drive a car :) I'm an advocate of pedalling.


I'm not sure if the speed really matters. I would rather agree with @AHicks and @Ken M that the hub motor is fed with constant power depending on the PAS level and the speed is the outcome of different resistances met on a ride. I can't still understand why the hub motor in my e-bike obviously "weakens" for a while when I'm stalled with a squall. I can hear and feel it even if the motor is almost noiseless.

Why does it seem that so many believe that if an ebike has a throttle the rider doesn't / won't pedal? A throttle is just an infinite assist control device so you can pedal as activey as you desire and just use the throttle to either control your top speed or maintain the same effort level.

Let me mention this. PAS is cool but they use an array of sensors to predict what assist level is needed / desired. I just don't drink the koolaid that there is a better way to control assist level than a throttle - simple, time tested, more reliable, etc. Just facts not industry hype.
 
Let me mention this. PAS is cool but they use an array of sensors to predict what assist level is needed / desired. I just don't drink the koolaid that there is a better way to control assist level than a throttle - simple, time tested, more reliable, etc. Just facts not industry hype.

This is just incorrect, plain and simple. There's only one PAS sensor on any bike I'm familiar with, and it's only purpose is to tell the controller the pedals are turning. Anything that happens from there is done by the controller and it's programming. Which was my point earlier.

An example of controller programming might be the top speed associated with a particular PAS level. That's not always the case. On some controllers that's an option that can be turned on or off. Very similar to setting your wheel size or max speed. Those are both examples of controller programming as well.

Some controllers have some very well thought out logic behind the controller programming, while others are much more basic. -Al
 
I must add my original post (the hub motor bike part) is related to my bike. The manufacturer assures me there is a cadence sensor and the controller reacts to increased cadence with more power to the motor. It is a 2019 e-bike. It may explain why the motor on the bike stalled by a wind gust intermittently loses its power.
YMMV.

Additional note: If the hub motor were getting the constant amount of juice at given PAS level then the energy consumption would be constant. It is not, however, what I observe. I took a ride 25 km upwind followed by a downwind return way in PAS 2 of 3. The energy consumption was significantly higher on the upwind route. Why?
 
Last edited:
Is this a difference between Hub and mid-drive or a difference between cadence and torque sensors? I know that hubs generally don't have torque sensors, but my Juiced CCS is a geared hub motor and it has a torque sensor. It's one of the reasons I bought the bike.

I know everyone has their preferences and I won't try to say my preference is the best for anyone but myself. But for me, I greatly prefer a torque sensor. It provides me with the riding experience I want.

I ride a Juiced ocean current with a500W geared hub, this is second bike I've had with a hub drive and torque sensor, the first one was a Bionx direct drive hub, I like torque sensors because they respond quickly to how you pedal, just coasting along the motor just gives light assist, suddenly stomp on the pedal or start to climb a hill the assist ramps up to suit.

Hill climbing, for a steep hill I use Sport mode and first gear, I still have to put some effort in, throttle is basically useless for this unless I am walking the bike up a hill.

My other bike has a bbshd mid drive, climbing the same hill in first gear in level 4 out of 9, while pulling a loaded shopping trolley is pretty effortless, I weight a bit under 90kg or 200lb, this on a 50lb bike and up to 40lb of shopping, not bad.

Along the flat the Juiced bike will cruise at whatever speed I settle on, fast or slow, the bbshd wants to settle in to a set speed range depending on the assist level, I normally use 3 and the bike will cruise along to about 27kph, push harder it will get to 32kph, I'll dial it down to 2 if there are lots of people around.

I had four stents fitted about four years ago, and I have to do the exercise, so currently the last twelve months I've done 700km in the car and 5000km on the bike or near 15000km on both bikes in a bit over two years, and another 9000km on the Bionx in three years before that.
 
I own hub motors of various power levels and one bafang mid drive, all with cadence sensors. I've seen a big variance in how these bikes pedal.

The cadence sensor itself has lot of possibilities. It puts out a string of pulses. From this the controller can determine that the pedals spin, that they are spinning in the right direction, and how fast they are spinning.

However, most controllers, maybe 99% of controllers, only know that the pedals are spinning forward. What they do with that can vary. I think the better controller assign a speed limit for each assist level, tapering off the motor current as the bike approaches that limit. If ridden at the speed cusp, both the rider/motor will help power the bike and the pedal effort feels natural.

I have owned some poorly designed systems where the above speed limits were set too high. Then the rider can't find a bike speed that doesn't feel overly assisted. In those cases, I replaced the controller/display. Not something most people want to do.
 
i see the same thing as ahicks is seeing - you pick a level of PAS and the motor gives that level of assist regardless of cadence but tapers off as you hit the target speed of that particular PAS setting. for me if i want to do any sort of pedalling i undershoot my target speed with PAS and then add as much pedalling as i care to, and i ride the same way on either my older surface 604 colt (the one with the older bb torque/cadence sensor) or my ebikeling traditional magnet wheel / hall sensor kit. the torque part of the colt's sensor seems to mildly moderate the level of assist but it is a pretty subtle effect so if you are not feeling very 'motivated' you can get away with providing almost no torque and it will still give you a lot of assist.
Good description. I used to get minimal excersize on my hub drive. It encouraged me to be lazy.
 
This is just incorrect, plain and simple. There's only one PAS sensor on any bike I'm familiar with, and it's only purpose is to tell the controller the pedals are turning. Anything that happens from there is done by the controller and it's programming. Which was my point earlier.

An example of controller programming might be the top speed associated with a particular PAS level. That's not always the case. On some controllers that's an option that can be turned on or off. Very similar to setting your wheel size or max speed. Those are both examples of controller programming as well.

Some controllers have some very well thought out logic behind the controller programming, while others are much more basic. -Al

Really??? .... Only one sensor? I don't really like telling anyone directly that they are wrong but I can assure you that the Bosch, Brose, and Yamaha mid-drive motors use cadence, torque, and speed sensors. PIM actually had a gear sensor on the DD hub motor drive system that primarily used cadence. Most good PAS system also have a thermal sensor that will limit assist if the motor is getting too hot. The world is not flat, we do seem to be impacting the climate....and there is more than one sensor used on most ebikes to control the assist provided.
 
I own hub motors of various power levels and one bafang mid drive, all with cadence sensors. I've seen a big variance in how these bikes pedal.

The cadence sensor itself has lot of possibilities. It puts out a string of pulses. From this the controller can determine that the pedals spin, that they are spinning in the right direction, and how fast they are spinning.

However, most controllers, maybe 99% of controllers, only know that the pedals are spinning forward. What they do with that can vary. I think the better controller assign a speed limit for each assist level, tapering off the motor current as the bike approaches that limit. If ridden at the speed cusp, both the rider/motor will help power the bike and the pedal effort feels natural.

I have owned some poorly designed systems where the above speed limits were set too high. Then the rider can't find a bike speed that doesn't feel overly assisted. In those cases, I replaced the controller/display. Not something most people want to do.

You are correct but the industry doesn't want to tell people that their cadence systems are really just on/off switches for the assist levels programmed - you just get the wattage set by the assist level up to the speed set by programming. In my opinion that is so inferior to a throttle that provides essentially infinite assist level control.
 
Really??? .... Only one sensor? I don't really like telling anyone directly that they are wrong but I can assure you that the Bosch, Brose, and Yamaha mid-drive motors use cadence, torque, and speed sensors. PIM actually had a gear sensor on the DD hub motor drive system that primarily used cadence. Most good PAS system also have a thermal sensor that will limit assist if the motor is getting too hot. The world is not flat, we do seem to be impacting the climate....and there is more than one sensor used on most ebikes to control the assist provided.

Ken,
All due respect, you said "PAS is cool but they use an array of sensors ".

A PAS sensor is a PAS sensor. It tells the bike if the crank is moving, period (OK crank direction and speed as well, but it's beyond me what that info might be used for as far as useful controller input). I feel pretty confident in saying that a thermal sensor, if the bike is equipped with one, is not associated with the PAS sensor or PAS circuit in any manner. It's a totally independent input the controller monitors. A bike with a gear sensor or speedometer on it would be another example of other circuits not associated with PAS.

A PAS based SYSTEM, that includes thermal sensing, a gear indicator, speedometer, and many other features is entirely possible, BUT, a PAS system designed without them will function just fine....

While Bosch, Brose, Yamaha, and other mid drives may be equipped with these features, that does not make them a necessary function on ALL PAS based systems - even good ones.

Point being I think, is that you were referring to a PAS based SYSTEM, where I took it that you were talking about some sort of complicated PAS sensor, which I know of as a pretty darn simple device.

And last, a top speed to be determined by PAS level is possible, and used frquently, but I'm here to tell you they don't all function that way. My bike for instance, has no speed associated with the PAS level at all. You can be pedaling your butt of in PAS 1 doing 20 mph - getting the SAME assist level that you would be getting in PAS 1 doing 5mph.

You might consider this prior to making broad statements regarding ALL e-bikes. -Al
 
Last edited:
A related question on mid vs hub...

The most common hub motors, like Bafang, freewheel when PAS is off, or the battery is dead.

I know there's other factors (bike weight etc) to take into account, but do all mid-drives add the same amount of resistance when PAS is off (or the battery is dead) or is there secret sauce that Specialized has compared to Bafang (for example) or does Bosch have less resistance than others (etc)?

I'm currently in the very early stages of considering adding a second e-bike, and was curious to hear from those who've ridden both on a regular basis.
 
Of hub vs mid drive, the biggest difference in feel between both will be the cadence sensor on the hub drive vs the torque sensor on the mid. Id greatly like to hear of someone who has ridden both, using a cadence sensor.
 
There are so many misconceptions regarding PAS effectiveness. Some system do actually provide the ability to get exercise or clown pedaling. This regarding torque sensing bottom brackets.


Torque Sensors
Example Bottom Bracket Torque Sensors


Torque sensing bottom brackets measure the force that the rider applies on the pedals in addition to their pedaling cadence. This enables pedal assist modes where the motor power varies in proportion with the rider's input power, something present in most of the higher-end ebike systems (eg Bosch, Shimano STEPs, BionX..) At the moment, nearly all aftermarket torque sensors are designed for threaded bottom bracket shells and use JIS square taper cranksets. If you have a press fit bottom bracket shell these won't work. Similarly if you have a splined hollow spindle crankset then you'll need to replace your crankset with one meant for a square taper interface.

Illustration of Spindle Torque Sensing on Thun/NCTE
NCTE (Thun):
The original stand alone ebike torque sensor was made for THUN by NCTE. This sensor measures the actual torque across the spindle. When you pedal with the left crank it registers a full torque signal, but when you are pushing with the right pedal the force goes straight to the chainring, it doesn't couple through the spindle. This has some unique advantages for certain mid-drive and tandem bike applications, but for most riders it means a delay in power assistance coming in if you start pedaling with the right leg. The delay is further compounded by the fact that these sensors only have the equivalent of 8 cadence pulses per revolution.

On the plus side, this style of sensor is the very good at showing an accurate report of the human power in watts over any gear range and even at extreme pedal force, as it is a true linear torque measurement right up to 200 Nm. It is also easy to install, being mechanically no different that a standard sealed cartridge BB. There is also an ISIS splined option from NCTE, although we have had mechanical problems with them.


ERider:
Illustration of Spider Torque Measurement on ERider Sensors
These sensors have a separate mount for the chainring spider that is independant of the right crank, allowing for accurate sensing of both left and right pedal contributions. They measure human torque and watts with good accuracy and the high pulse count 18 pole cadence sensor allows for a rapid PAS power engagement at the start of pedaling. They also feature a cable exit on the side bearing cap so there is no need to drill a hole through the frame for running the cable passage. These sensors are available for both 68mm and 73mm bottom bracket shells and are available with both 4 bolt 104 BCD and 5 bolt 130 BCD chainring standards.

With these ERider torque sensors you cannot reuse the original cranks from your bike as the right crank has no chainring spider. We include a zero 'Q' factor Miranda cranks with each package that works very well and helps to offset the additional spindle length of this sensor design . You need to move the chainrings from your original crankset to the appropriate 4 or 5 bolt spider mount that came with the kit.


Illustration of Spindle Flex on Sempu Sensors
Sempu: The Sempu sensor also responds to both left and right pedal torques, but unlike the TDCM or NCTE sensor it makes no attempt to isolate forces that put power to the wheel from all the other forces on the crank. Instead, it measures the general flex of the spindle and assumes that for the most part this will vary in proportion to how hard you are pedaling. In practice, this works surprisingly well. It also has 24 pole cadence sensor, 3 times more than the Thun / NCTE and twice the TDCM. These details make it by far the most responsive sensor for engaging immediately when you start pedaling from a standstill.

The Sempu installs almost as easily as a regular cartridge bottom bracket (see video here) and is available in non-standard bottom bracket sizes, with adapters for 73mm BB shells and also models for 84mm and 100mm wide bottom brackets used in Fatbikes. On the downside we have seen a higher than normal issue rate (approx 20%) with the Sempu sensors throughout most of 2019 with the torque signal periodically dropping out, and are holding off ordering more of these until Sempu can assure that their QC problems are fixed.


TDCM:
Illustration of Chain Pull Force on TDCM Sensors
The TDCM sensor doesn't measure the torque on the spindle, instead it uses a clever arrangement with a slotted bearing support to measure how much the right side spindle flexes backwards due to chain tension. This allows the sensor to respond equally well to left and right pedal forces as they both put tension on the chain, while ignoring the downwards force which would only be present from the right pedal. It includes a higher 12 pulse cadence sensor helps it respond quickly to small pedal rotations.

Because the operation of this sensor requires an alignment in the direction of the chain tension, it involves a more detailed installation process taking care to read and follow the instructions. The output signal also saturates at the equivalent of about 50 Nm of pedal torque, which limits it's accuracy as a human power meter. But for the sake of providing proportional PAS assistance, this sensor works very well.
Availability Update: As of 2018 the TDCM has gotten out of the torque sensor business and these bottom bracket sensors are no longer available.


Spindle Length
All the torque sensing bottom brackets are available in a range of spindle length options. Even though the size of bottom bracket shells on bike frames is largely standardized to 68mm or 73mm, the exact location where the chainrings will line up can vary significantly between cranksets. Bicycle bottom brackets are produced in many different spindle lengths to achieve a desired chainline.
 
Some bikes use a torque sensor that goes between the frame and the rear axle, rather than being in the bottom bracket, Ive only even seen them on a hub drive bike.
 
To add to this discussion

my throttle only bikes have been Some of/or possibly the most battery efficient bikes i have owned

to give a good example my first bike was a cheap cadence bike and very on/off, wide open way too much power in any pas level
i pulled the magnet disc away from the pedals and rode that bike throttle only
later put a cycle analyst and torque sensor on it and all the sudden i have half the battery range

it worked well and fine but i used WAY more battery running the torque sensor than throttle

so this ridiculous thought that throttles make you not pedal is nuts
anyone knows if you dont pedal at all you will run out the battery quickly

although i own torque and cadence bikes totally agree with ken that throttle only is a simple system and works well- and i would own a throttle only system over some of the very crappy programmed on/off cadence setups i have test ridden- i refuse to purchase crappy programmed cadence bikes

in addition to the people that need throttle bikes for health reasons
 
  • Like
Reactions: JRA
A related question on mid vs hub...

The most common hub motors, like Bafang, freewheel when PAS is off, or the battery is dead.

I know there's other factors (bike weight etc) to take into account, but do all mid-drives add the same amount of resistance when PAS is off (or the battery is dead) or is there secret sauce that Specialized has compared to Bafang (for example) or does Bosch have less resistance than others (etc)?

I'm currently in the very early stages of considering adding a second e-bike, and was curious to hear from those who've ridden both on a regular basis.
Mid motors come in either single-clutch or double-clutch flavor. As far as I can understand, the single clutch disconnects the crank in the way that allows coasting. If the system is double-clutch, the second clutch disconnects the crank from the motor completely, so you can ride without any pedalling assistance.

Now, Brose and Specialized motors as well as some Yamaha motors are double-clutch type. Shimano STEPS is single clutch. Not sure about Bosch Gen 4 but all older Bosch motors are single clutch. How can you tell? Look at the chain-ring. If the chainring is small, e.g. 15-18t, the motor would be single clutch type. If the chainring is 38-48t, that would be the double-clutch kind. The point is, in case the chainring is small, the crank is connected to the motor via internal gears. In case power is lost, you have to pedal through the internal motor gears which creates significant resistance. In case of double clutch motors, you just drive the rear wheel by pedalling, as if it were an acoustic bike.

Now, I rode both my bikes without power. The geared hub drive e-bike just behaves as a very heavy acoustic bike, the freewheeling is working properly. More interesting is the behaviour of the double-clutch mid-drive motor bike. I lost the power on my Vado 5 twice: Once it was a stuck display; the second time I simply run out of the battery. In both cases I could ride only pedalling. My speed upwind was some 12 km/h and with no wind some 18 km/h -- something I could achieve on an acoustic bike -- and I rode for 15 km without power. I can bet that would be much harder on a Bosch or Shimano system.

Direct Drive hub motors are probably very resistant un-powered but I have no experience with these.

The "secret sauce" for Brose/Specialized and some Yamaha motors is the double clutch.
 
Back