I am looking for a partner to produce an add-on torque sensor for the BBS01

[Woosh]

Ditching the usual assist levels has transformed my bike but for the rest.... Can you talk about functionality and benefits?

adding a torque sensor to the BBS01's bottom bracket axle has only a small market potential because practically all the BBS01 owners are happy with what they have got. The BBS01 is dependable and relatively cheap. It would be hard to convince existing owners to fork out something like £40-£50 for a new axle and a new display and spend hours fitting it.
However, the same component can be potentially assembled into a new kit, and offered at competitive price because assemblers save money on a Bafang LCD.
The main benefit is of course the ability to do away with the throttle, albeit that the latter is retained in this project. This is possible because of the flexibility of a programmable user interface. You can modify the torque multiplying factor to suit you, your speed, the gradient and the motor.
It is possible to execute a hill start for example, something quite difficult without a throttle.
It will also be easy to add more sensors to the basic torque sensor like a temperature sensor or a heart rate sensor.

 

[guerney]

It would be hard to convince existing owners to fork out something like £40-£50 for a new axle and a new display and spend hours fitting it.
However, the same component can be potentially assembled into a new kit, and offered at competitive price because assemblers save money on a Bafang LCD.

As I predicted!

I bet any TS mod Woosh makes will be a Woosh kit customer exclusive.

...in this post!

 

[Woosh]

guerney's predictions are way too far ahead.
I may hit a lot of troubles making a good replacement axle.
I guess we'll know when the samples arrive.

 

[guerney]

guerney's predictions are way too far ahead.
I may hit a lot of troubles making a good replacement axle.
I guess we'll know when the samples arrive.

Will it be closed source code too? It's great @StuartsProjects used Github for his alarm LoRa range extender.

 

[Woosh]

Will it be closed source code too?

not really. The code for the HX711 is already on the net and it easy to modify.
The user interface will probably be written in micropython, there are plenty of more experienced programmers than me so I expect rapid progress once the hardware is available.It's up to the individuals to choose where and when they want to publish.

 

[guerney]

It's up to the individuals to choose where and when they want to publish.

True. But I choose to not contribute to any project so closed and exclusive.

 

[Woosh]

you see too much into this project, honestly. I will publish the schematic diagram so anyone wants to make a better device, can.

 

[StuartsProjects]

Will it be closed source code too? It's great @StuartsProjects used Github for his alarm LoRa range extender.

There are good reasons for that.

I was mainly interested in building one for my own use, its useful to have and gives you something to do.

You might think then, why not sell them ?

Although it would be cheap to build it is a radio transmitter and the regulations around UKCA & CE etc are designed to prevent small businesses competing with big business.

To get a device that has a radio transmitter through all the testing needed would be circa £25K. A cost which is of little consequence for big business, and simply ignored by them in the far East, but its a major investment for a small business in the UK.

So if its just not on to sell such a built board, you might as well give the design away. I do sell the PCBs, which have to be lead free of course.

 

[guerney]

you see too much into this project, honestly. I will publish the schematic diagram so anyone wants to make a better device, can.

Part of your target market is hobbyists. The availability of spare parts without having to buy entire kits, plus open source code, has made the TSDZ2 more popular than it would have been. Toseven's firmware needs work, you don't want to fall into the same trap. That being said, this is your project and of course you're free to do what you wish with your work.

 

[Woosh]

I don't try to emulate the success of the TSDZ2.
As we would do away with the current assist levels and throttle, take a look at this VESC project:
VESC – Open Source ESC | Benjamin's robotics (vedder.se)
If combined, it would make for a cheaper and more flexible kit.
If this project succeeds, I can produce the same torque sensor + controller for ordinary hub kits.

 

[cyclebuddy]

adding a torque sensor to the BBS01's bottom bracket axle has only a small market potential...

Will the BBS survive at all once the newer M325/M625 part of Bafangs offering has been further fleshed out? CanBus/battery restrictive updates of the BBS's they may be, but still cadence sensor and otherwise very, very similar.

But I notice references in the M325/M625 manual... "only with torque sensor fitted"... and this in the fault table which too suggests a TS option may be coming soon?

 

[Woosh]

Will the BBS survive at all once the newer M325/M625 part of Bafangs offering has been further fleshed out?

yes, I should think so. The BBS01B fulfils the need for a basic pedelecs, so there is no reason to remove it. Would they fit a torque sensor in the BBS01B? Quite possibly I reckon.

 

[Woosh]

Just a quick update for those who follow my little project.
The strain gauges are more sensitive than I thought, so that's good.
I use two high precision BF350, I superglued the first BF350 to the axle for measuring and taped the second for temperature compensation. The ADC is a 24-bit HX711 running on 5V.
The problem I come across is this:
the baseline tare at zero torque drifts quite considerably with time.
Over half an hour, it can cause as much as 15% error in the reading. I can of course remove the drift in code but prefer to know the source of the drift and fix it. Is it the reference voltage applied to the Wheatstone bridge? The HX711 is supposed to have stable E+

 

[saneagle]

Just a quick update for those who follow my little project.
The strain gauges are more sensitive than I thought, so that's good.
I use two high precision BF350, I superglued the first BF350 to the axle for measuring and taped the second for temperature compensation. The ADC is a 24-bit HX711 running on 5V.
The problem I come across is this:
the baseline tare at zero torque drifts quite considerably with time.
Over half an hour, it can cause as much as 15% error in the reading. I can of course remove the drift in code but prefer to know the source of the drift and fix it. Is it the reference voltage applied to the Wheatstone bridge? The HX711 is supposed to have stable E+

We used to use two at right angles when I did my engineering degree. Don't forget that you're measuring the connecting wire too, so keep it short an as thick as possible.

 

[Woosh]

We used to use two at right angles when I did my engineering degree. Don't forget that you're measuring the connecting wire too, so keep it short an as thick as possible.

I use only one FB350 for sensing so the orientation of the second FB350 is not important, it's just there so that effect of the temperature on the resistance of the strain gauges cancels each other out.

I think I found the source of the drift BTW. I'll fix it tonight.
It's funny, when you share a problem, you see it better, clearer yourself!

 

[lenny]