New "Lingbei MM28.250 torque sensor mid drive"

Continue to believe in the nonsense that the manufacturers of the motor of lingbei write. But I want to remove noodles from your ears and show how people are being fooled. Above, it is written that the torque sensor can not stand on the axis, but on the clutch spindle. Well, I bothered to disassemble and this part, which consists of a pair of bearings, a large gear and a clutch response part, which is attached HARDLY by bolts. There is no torque sensor here. I wrote about this above, when I bought this motor and drove it. Also you can see 4 wires going to the pas sensor.
 

Attachments

  • IMG_20180818_203728.jpg
    IMG_20180818_203728.jpg
    83.5 KB · Views: 4,435
  • IMG_20180818_203714.jpg
    IMG_20180818_203714.jpg
    79.3 KB · Views: 4,435
  • IMG_20180818_203633.jpg
    IMG_20180818_203633.jpg
    96.9 KB · Views: 4,435
  • IMG_20180818_203545.jpg
    IMG_20180818_203545.jpg
    108.3 KB · Views: 4,435
Kisazul said:
Also you can see 4 wires going to the pas sensor.
I can say that PAS of TSDZ2 also has 4 wires: 5V; GND; PAS_1_SIGNAL; PAS_2_SIGNAL.

Since PAS_1_SIGNAL and PAS_2_SIGNAL are not in phase, we can use that to check for forward or backwards pedal rotation, other way only PAS_1_SIGNAL would be enough to calc the pedal cadence.

This is my code to calc pedal cadence on TSDZ2, from that wires/PAS_1_SIGNAL and PAS_2_SIGNAL and filter out when rotating backwards:
Code:
  /****************************************************************************/
   // calc PAS timming between each positive pulses, in PWM cycles ticks
   // calc PAS on and off timming of each pulse, in PWM cycles ticks
   ui16_pas_counter++;

   // detect PAS signal changes
   if ((PAS1__PORT->IDR & PAS1__PIN) == 0)
   {
     ui8_pas_state = 0;
   }
   else
   {
     ui8_pas_state = 1;
   }

   // PAS signal did change
   if (ui8_pas_state != ui8_pas_state_old)
   {
     ui8_pas_state_old = ui8_pas_state;

     // consider only when PAS signal transition from 0 to 1
     if (ui8_pas_state == 1)
     {
       // limit PAS cadence to be less than PAS_ABSOLUTE_MAX_CADENCE_PWM_CYCLE_TICKS
       // also PAS cadence should be zero if rotating backwards
       if ((ui16_pas_counter < ((uint16_t) PAS_ABSOLUTE_MAX_CADENCE_PWM_CYCLE_TICKS)) ||
           (ui8_pas_direction))
       {
         ui16_pas_pwm_cycles_ticks = (uint16_t) PAS_ABSOLUTE_MIN_CADENCE_PWM_CYCLE_TICKS;
       }
       else
       {
         ui16_pas_pwm_cycles_ticks = ui16_pas_counter;
       }

       ui16_pas_counter = 0;
     }
     else
     {
       // PAS cadence should be zero if rotating backwards
       if ((PAS2__PORT->IDR & PAS2__PIN) != 0)
       {
         ui8_pas_direction = 1;
         ui16_pas_pwm_cycles_ticks = (uint16_t) PAS_ABSOLUTE_MIN_CADENCE_PWM_CYCLE_TICKS;
       }
       else
       {
         ui8_pas_direction = 0;
       }
     }

   // limit min PAS cadence
   if (ui16_pas_counter > ((uint16_t) PAS_ABSOLUTE_MIN_CADENCE_PWM_CYCLE_TICKS))
   {
     ui16_pas_pwm_cycles_ticks = (uint16_t) PAS_ABSOLUTE_MIN_CADENCE_PWM_CYCLE_TICKS;
     ui16_pas_counter = 0;
     ui8_pas_direction = 0;
   }
   /****************************************************************************/

void read_pas_cadence (void)
{
  // cadence in RPM =  60 / (ui16_pas_timer2_ticks * PAS_NUMBER_MAGNETS * 0.000064)
  if (ui16_pas_pwm_cycles_ticks >= ((uint16_t) PAS_ABSOLUTE_MIN_CADENCE_PWM_CYCLE_TICKS)) { ui8_pas_cadence_rpm = 0; }
  else
  {
    ui8_pas_cadence_rpm = (uint8_t) (60 / (((float) ui16_pas_pwm_cycles_ticks) * ((float) PAS_NUMBER_MAGNETS) * 0.000064));

    if (ui8_pas_cadence_rpm > ((uint8_t) PAS_MAX_CADENCE_RPM))
    {
      ui8_pas_cadence_rpm = ((uint8_t) PAS_MAX_CADENCE_RPM);
    }
  }
}
 
Kisazul said:
Continue to believe in the nonsense that the manufacturers of the motor of lingbei write. But I want to remove noodles from your ears and show how people are being fooled. Above, it is written that the torque sensor can not stand on the axis, but on the clutch spindle. Well, I bothered to disassemble and this part, which consists of a pair of bearings, a large gear and a clutch response part, which is attached HARDLY by bolts. There is no torque sensor here. I wrote about this above, when I bought this motor and drove it. Also you can see 4 wires going to the pas sensor.

I see you are passionate about this....passionate, confident, and angry? Strong words!

You got this motor early on, correct? - When it was harder to tell about the torque sensing...

Below is the BBS02 PAS:
View attachment 2

notice the location of it... pretty easy to access I suppose...

at least easier then embedding it into the shaft/axle/spindle housing...

such as in the MM28 as shown in your picture below:


now if they copied the BBS design, why place the PAS sensor in such a place?

Lets take a look at the TSDZ2 torque sensor below:
tsdz2torquesensor.png

it seems to look alot more like the MM28 sensor, and is mounted in practically the same location...

I also would be interested to see where the wires go (the additional part of the sensor, perhaps) in the MM28 from where we see them...

Also note, you can definitely tell it is a torque sensor on the newer version... I was trying to politely tell you this, but you are really determined this is not a torque sensor...

While you have it open, maybe I can send the new programming for you to update your motor? I will get it when I can, I am laid up with an injury now, but would be happy to provide you with it when I am able.

fantasy2, that goes for you too, I didn't break into the new motor and copy the code, but I will when I am able - and until then I will just get them to send it to me, that shouldn't be a problem now that I think of it, hopefully I won't have to open a motor right now... I have too much going on.

Also, maybe we can compare the old and new code, to get an idea of the torque sensing code, and see if someone can modify/improve that!! eh? now I am really dreaming, hahah
 
eyebyesickle said:
Lets take a look at the TSDZ2 torque sensor below.

it seems to look alot more like the MM28 sensor, and is mounted in practically the same location...

Well, sure, if by "look alot more like" you mean "a small circuit board with electronic thingys on it" and if by "mounted in practically the same location" you mean "in the area around the bottom bracket".

However, the TSDZ2 sensor is mounted in a milled recess in the crank spindle and presumably measures the deflection of the spindle due to pedaling force. The MM28 sensor is mounted to the housing and is not exposed to and hence cannot sense pedaling forces.
 
-dg said:
eyebyesickle said:
Lets take a look at the TSDZ2 torque sensor below.

it seems to look alot more like the MM28 sensor, and is mounted in practically the same location...

Well, sure, if by "look alot more like" you mean "a small circuit board with electronic thingys on it" and if by "mounted in practically the same location" you mean "in the area around the bottom bracket".

However, the TSDZ2 sensor is mounted in a milled recess in the crank spindle and presumably measures the deflection of the spindle due to pedaling force. The MM28 sensor is mounted to the housing and is not exposed to and hence cannot sense pedaling forces.

Wow... sounds good but the difference between the BBS and the LingBei sensor location, and feel - is night and day - and the style and location is much more similar to the TSDZ2. Any amount of sarcasm isn't going to change that... are you going just by pictures or do you have both motors? Have you tested the new version of the MM28? I think I'm almost done 'discussing' this...
 
It would be interesting to remove the circuit board, remove it's silicone coating and make clear pictures. This should give us the answer. Measuring it on an oscilloscope would be nice as well. There are many methods to measure torque for eBikes. Tongsheng uses only one of them.

Interesting paper about several methods:
http://lup.lub.lu.se/student-papers/record/5044786/file/5044800.pdf
https://repository.tudelft.nl/islandora/object/uuid:74ee44a2-08b9-480b-9f29-663ab61a45ba/datastream/OBJ/download

As you can read, some methods use a non-contact PCB just like shown on the picture.
So as long as we don't fully understand how the claimed torque measure mechanism works, we shouldn't jump to conclusions too quickly.


I need to see this shaft in real life but things don't add up if it's just a PAS sensor.
The slot they made to accommodate the PCB is very deep. Twice the size of what you need for a simple PAS PCB with 2x hall sensor.
Also, the shaft seems to contain two rings next to each other. One with a clear magnet and close to the PCB we see, but the other one is deeper inside, if that makes sense.



-dg said:
However, the TSDZ2 sensor is mounted in a milled recess in the crank spindle and presumably measures the deflection of the spindle due to pedaling force. The MM28 sensor is mounted to the housing and is not exposed to and hence cannot sense pedaling forces.
It's clear you don't understand it, please read the papers.
 
Some clarification pictures below.

If what I'm thinking is actually true(hard to see some details on the pictures), they use two rings of magnets and measure a phase shift between the two HALL sensors that respond to magnets in the rings. When they are in sync(hence the alignment) it means there is no torque applied. If this sync shifts, a torque is applied.

They could use the same signal for PAS as well, as these rings will make pulse trains.

Now, to confirm this, you need to connect these pulses to an oscilloscope and look for signal shifts. The longer the shift, the higher the torque.
It's a well known method and cool and clean method actually and is mentioned in the papers I shared.

lingbei torque sensor.jpg
Untitled.jpg
 
If it's not a torque sensor (which I'm operating like it is since I've felt it) then why would they place the sensor there (was my point in the other post that dude burned me up on haha.) My reasoning that if it was just a PAS why not copy bafang on that too...

I obviously am not sure how it works, just that it does, and also that there is no way it's just torque simulated etc...

It's the programming people, and on the first version it really is hardly noticeable... It's more noticeable now but still not great as I stated way back.

LingBei does also offer standalone BB torque sensors and I had my guy in China verify the MM28 was a torque sensor unit for real before we bought some...

If the torque sensor function sucks... I won't fight you, that's opinion. But to say it's not a torque sensor is going against the fact that it IS a torque sensor, is what I am trying to tell you gentleman.
 
stancecoke said:
Question to stancecoke ??
Hey Stancecoke what do you think, i know that you did lots of "surgeries" on many different kind of torque sensors.

Seeing the inside of the Linbei now, i do not understand how torque should be measured at this spot.
I mean i could think of that a hall-sensor measures the bend of the axle, but why at this spot?.... cause bend can be measure much more reliable in a vertical line, where the most force is applied by stepping on the pedaling.

If it is a torque sensor, and i can imagine it really is.... PAS and torque measured at the same spots, its probably a really primitive way to measure bend of the axle. so it might just give a vague signal that the axle ist bent or not.
But i can hardly believe that this method can deliver a realiable torque signal about exact force on the pedal axle.

Looking at the photos its sure that torque is not measured by chain force, but by axle bendfor me now.

Still asking
stancecoke said:
stancecoke
what do you think with your gathered experience ?
 
eyebyesickle said:
If the torque sensor function sucks... I won't fight you, that's opinion. But to say it's not a torque sensor is going against the fact that it IS a torque sensor, is what I am trying to tell you gentleman.

So it sound you have one of these motors ?

What happens if you put the bike on a bike-stand, with rear wheel in the air
--> bring the pedal into vertical position
-->and just push or step on the lower pedal without turning it clock or counterclockwise

will the motor start spinning ?

If it does start the motor..... it measures any kinf of axle bend.

If not.... its probably a very unrealiable method,...has bad configuration,...is not fully developed yet,.......or anything else
 
notger said:
Seeing the inside of the Linbei now, i do not understand how torque should be measured at this spot.
I mean i could think of that a hall-sensor measures the bend of the axle, but why at this spot?.... cause bend can be measure much more reliable in a vertical line, where the most force is applied by stepping on the pedaling.

If it is a torque sensor, and i can imagine it really is.... PAS and torque measured at the same spots, its probably a really primitive way to measure bend of the axle. so it might just give a vague signal that the axle ist bent or not.
But i can hardly believe that this method can deliver a realiable torque signal about exact force on the pedal axle.
It is a pretty basic but advanced way to measure torque. It's used in high tech industry for turbines as well, although they have more ways to do it nowadays. Using strain gauses is the primitive way to measure.. But yeah, the technique has been around since the '50's.

Measuring torque with axle bent is a very good way to measure it...

There are detailed descriptions in the two papers I shared.

What I am suspecting here, is that the torque sensor is not sensitive enough and they have a hard time translating it's signal. They said they will release a new controller, so there is a good chance they do this so they can amplify the signal for a better full range measurement.

This is what is happening with the two hall sensor outputs:
torque-measurements.jpg
 
notger said:
eyebyesickle said:
If the torque sensor function sucks... I won't fight you, that's opinion. But to say it's not a torque sensor is going against the fact that it IS a torque sensor, is what I am trying to tell you gentleman.

So it sound you have one of these motors ?

What happens if you put the bike on a bike-stand, with rear wheel in the air
--> bring the pedal into vertical position
-->and just push or step on the lower pedal without turning it clock or counterclockwise

will the motor start spinning ?

If it does start the motor..... it measures any kinf of axle bend.

If not.... its probably a very unrealiable method,...has bad configuration,...is not fully developed yet,.......or anything else

Damn, I think that the Yamaha PW motor I also have isn't torque sensor based when it has to follow your description!
It doesn't start to spin when I apply torque on the pedals but don't rotate them, so it has to be pure PAS, right?

No of course not. These controllers use BOTH PAS and torque. The motor has to be sure that you're actually moving before it will use the torque sensor input. Why? Otherwise it will constantly push you while you're waiting at the traffic lights and have one foot on the pedal.

If your motor starts moving by just pushing on the pedal, you have a bad and primitive configuration that is not fully developed.


And please everyone, please read into the subject before making any claims without understanding all the details. This isn't rocket science. I'm not able to break open my motor again today but maybe Kisazul can help out. Connect an oscilloscope to the two output wires, you should see two sine waves shifting.


I really don't get it why people are responding so offensive and paranoid for the last few pages. Yes we know it doesn't work optimal right now, but it's a new motor. The manufacturer acknowledged is and they're actively fixing and improving the motor(which is pretty unique for a chinese motor). I was hoping for a more constructive thread over here when I started this topic instead of bashing this motor for no reason.
 
notger said:
Hey Stancecoke what do you think, i know that you did lots of "surgeries" on many different kind of torque sensors.

It can work the way fantasy2 explained. That's an usual way to measure torque contactless from a turning shaft. Normally you have a big distance between the two sensor areas to get as much twist as possible of the shaft. If the sleeve is made from steel, the phase shift must be very small in this case. You will have to use an oszilloscope to see, if there is an torque dependend phase shift between both rectangular signals. If this would work properly, I think other manufactors would do it, as this is a very simple and cheap solution.

regards
stancecoke

1438758635-95-konzeptdarstellung-iph.jpg.626x0.jpg
 
The Tongsheng torque sensor (photo a few posts back) actually measures torsional strain ( "twist" ) in a tube surrounding the pedal spindle.
The pedal spindle is solid and has an external spline near the left-hand end. This spline feeds the torque into the tube, and the tube drives the main gear through the sprag clutch at its right-hand end.
The sensor is mounted in a slot milled in the tube, and is measuring a sideways displacement of the target relative to the body of the sensor. This displacement occurs because the torque causes the tube to "flex torsionally".
(On a side topic: the controller both powers and communicates with the torque sensor through two coils of wire: one fixed and one rotating with the pedal spindle, with only two wires from the controller to the fixed coil.. quite clever! This is needed because the torque sensor is rotating with the pedal spindle so cannot be "hard-wired").

If the Lingbei torque sensor is measuring torsional displacement between two targets, and the targets are close together, is it possible that one of the targets is mounted on a tube surrounding the pedal spindle and connected near the other end.. to give a greater length of spindle between the attachment points of the two targets? Hard to judge without having one in my hands.
Also, if the two targets are both on the pedal spindle, and both on the same side of the ratchet that transfers the "pedal power" from the spindle to the gearing, does that mean it is a single-sided torque sensor?
Or are the targets both on a driven tube like the TSDZ2, therefore a double sided torque sensor??

So many questions.. it will be interesting to see what the final answer is!
 
I'll check. I've been preoccupied with an injury.

I do want to mention something concerning some of the previous discussion here. I try to be polite and respectful when stating something I know... Halfway as a matter of routine as to not be cocky or abrasive, halfway as to not antagonize officially educated engineers etc. I don't know if this is coming off as uncertainty, or people are just extra skeptical... but just an FYI, I won't be putting much energy into proving/debating/arguing about this motor. So please, let's relax a little bit if we are getting this thread fired up again. Sound reasonable?
 
As I understand it the new firmware is complete, debugged and tested and has been for 3 weeks. I was expecting a sample 2 weeks ago but had to delay it because I was overseas. I should have it in the next 10 days. Here's the specification:

52V "500W"
9 Mosfets (original has 6) so I guess this should be the higher current version. It is identified by 2.0 after the serial number.
Lingbei say: Improved torque program & better riding experience optimized on start & climbing.
External programming port precludes the need to open motor to re-program.

All nice words but we'll see, anyway. I'm just a punter trying this as an alternative to my BBSHD and TSDZ2. I have 7000km on my BBSHD bike and 1000km on the TSDZ2 so I know what torque sensing feels like. I'm sure I'd be delighted with the open source firmware on the TSDZ2 if I could understand it but my base motor is so poorly made, it'd be lipstick on a pig. At this stage I see Lingbei as worth a try. I'll let you know.
Marty
 
I have been wondering when these improvements were going to be done, what do they say about it being able to handle 20/21/22 amps now instead of the previous version which was , what , only 17/18 amps ?


wheelyphased said:
As I understand it the new firmware is complete, debugged and tested and has been for 3 weeks. I was expecting a sample 2 weeks ago but had to delay it because I was overseas. I should have it in the next 10 days. Here's the specification:

52V "500W"
9 Mosfets (original has 6) so I guess this should be the higher current version. It is identified by 2.0 after the serial number.
Lingbei say: Improved torque program & better riding experience optimized on start & climbing.
External programming port precludes the need to open motor to re-program.

All nice words but we'll see, anyway. I'm just a punter trying this as an alternative to my BBSHD and TSDZ2. I have 7000km on my BBSHD bike and 1000km on the TSDZ2 so I know what torque sensing feels like. I'm sure I'd be delighted with the open source firmware on the TSDZ2 if I could understand it but my base motor is so poorly made, it'd be lipstick on a pig. At this stage I see Lingbei as worth a try. I'll let you know.
Marty
 
Are you going to be selling the newer version 2.0 with 52 volt 500 watt specs, and 9 mosfets and external programing cable ?

Also
Now that there are Tariff's on Chinese Bikes and cheep Chinese E-Bikes and parts, I have noticed from a link from someone that it looks like BMS Battery , in response to the Tariffs have discounted a , or more than one of their hub motors by 20% , which offsets most of the Tariff added cost of that item.

Is Lingbei going to do the same ? Discount the product some to offset much or some of the final cost ?

eyebyesickle said:
I'll check. I've been preoccupied with an injury.

I do want to mention something concerning some of the previous discussion here. I try to be polite and respectful when stating something I know... Halfway as a matter of routine as to not be cocky or abrasive, halfway as to not antagonize officially educated engineers etc. I don't know if this is coming off as uncertainty, or people are just extra skeptical... but just an FYI, I won't be putting much energy into proving/debating/arguing about this motor. So please, let's relax a little bit if we are getting this thread fired up again. Sound reasonable?
 
ScooterMan101 said:
I have been wondering when these improvements were going to be done, what do they say about it being able to handle 20/21/22 amps now instead of the previous version which was , what , only 17/18 amps ?

Yes, I think so. This is the higher current rating modification but I don't know what the new limit is. Unfortunately, that aspect of it I won't be reporting on because I ride only for fitness and it's very rarely I pull 8A. I'm a 5-6Wh/km rider.
 
I wonder if the original motors can be upgraded as well or if the hardware is significantly different that we have to buy a new controller.. some of us were beta testers in the end..
 
Here is a diagram of how the Lingbei Torque Sensing works. It was already posted actually, it looks

The new programming is much better. I'll have some here shortly! Yaay.

Better torque sensing and 22-23 amps makes a lighter weight BBS02 model in my book. I dont run them over 23 amps anyway, so its perfect for me...

Things are looking up for this motor. I also am trying to get my hands on the communication protocol for a couple reasons - 1 to implement my favorite display!

More soon
 

Attachments

  • lingbeitorquesensing.jpg
    lingbeitorquesensing.jpg
    46.4 KB · Views: 3,530
Did they tell you if the firmware can be used for the first released models with 6 MOSFET's?
 
fantasy2: I didn't ask, but I assume it may be pushing it for 6 mosfets... so I wouldn't try it - HOWEVER, I can ask for the new programming with toned down power for you to flash if you want

scooterman101: I don't think they would be lowering price like BMSbattery. Also, just a hunch, but I would guess BMS battery was going to put those hubs on sale anyway, and its not really for counteracting the tariff, its just marketing. Because I also think BMS battery circumvents the tariff through creative claiming when shipping to US customers...
 
how are motor noise, cadence, and weight in comparison to a TSDZ2 in the real world?
 
Back
Top