KT motor controllers -- Flexible OpenSource firmware for BMSBattery S/Kunteng KT motor controllers (0.25kW up to 5kW)

[honya96]

I'm in the process of modifying my S06S for higher voltages. Can anyone tell me the purpose of C1/C5/C14? (https://github.com/KingQueenWong/bmsbat ... roller.PDF upper right corner) I wonder if their voltage rating needs to be raised if the max input voltage is, say, doubled.

I guess they dont, Vbst should be from some regulator so it remains the same..

Yes, the FETEs are controlled with around 15V from the LM317, however one leg of C1/C5/C14 (the negative one in fact) sits directly on the motor phase, so it must be involved in the load circuitry somehow.

But i dont recommend trying this. Buy 72v 12fet and cut case to ~50mm

Surely you jest? The 12FET PCB extends the entire length of the case: https://opensourceebikefirmware.bitbuck ... s/15-1.png

Lenght, but not width, that can be cut.

[cencen]

It would be easy to create another non technical dashboard screen.

But then its main purpose is to know and change what's going on inside that black or silver box.

For me it will replace the original display, there are cheap smartphones for under 50€/$ out there so i will just mount one of them and have more flexibility and insight than any commercially available display could ever provide.

For almost the same price that is.

Just have to find one that looks really crappy, so it doesn't get stolen

Hi Xnyle,

In the beginning, while I was thinking of using this open-source firmware I was attracted by the idea of controlling the ebike from a mobile phone. I made the replica Lishui BT adapter and tested it on the controller which showed some limited information on the mobile phone display. I strongly believe this new BT app is a great starting point to a complete BT display solution for this project and thanks very much for all your efforts.

As a beta tester, I tested BluOsec app with the Torque-Simulation mode.

I understand that this version of the app is technical and intended for the developers, debuggers.
I would like to know more about some values and terms like: Correction value, Setpoint, PAS values
Offroad Enabled

App works just fine and shows all the values except the SPEED! No speed value!
I got a correction value of -20, -24 is that normal?

I think changing values on the run has not been implemented yet. I’ve tried to change some values but found out that only max speed value can be changed.

I’m in favor of a normal use dashboard display (on mobile phone), perhaps if possible, showing a combination of technical and non-technical information. As you have mentioned a cheap smartphone will really be an attractive replacement for any other display.

Cencen

[Xnyle]

Not sure if you're using the latest versions.

You can change everything that's grey in the Java tool and as far as i know everything works.

If something is not working, please make a screenshot and tell us what you did, what you expected, what you got instead, what you think might be the reason you got something different.

For instance "Speed" refers to two different values on the Fragments, Erps speed and external sensor speed. If you don't have a sensor the latter one will always be zero.

Also please make sure you always use the latest version/commit and always use the apk of that very commit. Stuff is still changing in communication protocol, so old versions might not work.

If you'd like a quick response better post in the German forum, no one will blame you for posting English there

[Xnyle]

I would like to know more about some values and terms like: Correction value, Setpoint, PAS values
Offroad Enabled

Yes documentation is key, but my motivation to write a good documentation on top is... well

Correction value is an adjustment to the motor specific angle (the angle between what the hallsensors say and where the motor really is).
It should be zero. You should adjust the angle until it is. Value is normally between 200 and 250.

Setpoint is duty cycle, ask stancecoke why it is/was called this way

Offroad enabled is illegal in Germany If enabled, we will all go to jail, so don't ask about it.

PAS values, which ones?

[stancecoke]

Today, I broke the shaft of my BMS-Battery bottom bracket torquesensor, too. Sadly I can't find casainho's pictures with his broken part, to compare if they broke in the same area...

I've welded the shaft and I hope this will work until I receive my new Sempu 3rd gen.

regards
stancecoke

[geofft]

I've welded the shaft and I hope this will work until I receive my new Sempu 3rd gen.

Glad to see I'm not the only one with a mig welder amongst his toys..

Decent looking weld but it will get a stern test in that location...

[geofft]

Just thought I'd show you my latest attempt at taming the simmering series resistor and LM317:-

Bought a couple of these:-

https://www.ebay.co.uk/itm/Through-Hole ... 2749.l2649

...unfortunately, as you can see, they're not cheap. Splayed the legs out to fit the resistor pcb holes:-

....also rotated the LM317 so it could be clamped against the case for improved heatsinking. Initially intended to 'cross its legs' but then realised that re-wiring the outer pcb tracks was an easy and neater solution. Had to add an insulator as the LM317 tab is not isolated.

The end result looks pretty neat:-

...all seems to run nice and cool so far but haven't given it a long test yet.

[stancecoke]

Nice work! Xnyle and me, we both have replaced the resistor and the LM317 by a DC/DC converter and both have trouble with them... I burned 4! converters while experimenting with switching off the PWM while no power is wanted and switching it on again when power is needed while the bike is moving... I buyed a pack of ten, so I was able to fix the controller fast

The new function (now stable ) can be found in a new branch. It needs a motor specific constant defined in the config.h
Of course this function is only interesting for direct drive motors.

regards
stancecoke

[Xnyle]

Why those converters blew up is still unclear. Maybe it doesn't even have anything to do with switching off PWM but rather with voltage spikes "backfiring" when testing without load.

I still wouldn't call it "stable" though until there isn't a good explanation what's actually happening.

[geofft]

Nice work! Xnyle and me, we both have replaced the resistor and the LM317 by a DC/DC converter and both have trouble with them... I burned 4! converters while experimenting with switching off the PWM while no power is wanted and switching it on again when power is needed while the bike is moving... I buyed a pack of ten, so I was able to fix the controller fast

I did try a 12/5v buck converter a while back, but lost confidence in it after a couple of random mosfet failures and reverted to the stock setup. I think in our application these supplies need to be super reliable, so maybe sticking with the LM317 is the best (least bad..?) option...

[stancecoke]

so maybe sticking with the LM317 is the best (least bad..?) option...

Hm, perhaps you are right, but I never had problems with the buck converter until I started experimenting with disabling/enabling the PWM. And it seems logical that there are current/voltage peaks when the PWM does not start syncron to BEMF...

So I have to take a longer testride to prove the concept of the new branch...

regards
stancecoke

[casainho]

Today, I broke the shaft of my BMS-Battery bottom bracket torquesensor, too. Sadly I can't find casainho's pictures with his broken part, to compare if they broke in the same area...

I can confirm that looking at your pictures, mine broke on the same place!!

[stancecoke]

I took a testride with the torque-override branch now. It works as expected, you get assistance from standstill and the pure torque signal overrides the torquesensor-mode (factor*torque*cadence) if it's higher.
meaning of the loop state:
5: PAS timeout
6: torquesensor-mode
10: torque override

regards
stancecoke

[kkm]

Latest firmware compiling with errors, if "speed sensor = external" marked in configurator:

ACAsetPoint.c:73: syntax error: token -> 'if' ; column 3
ACAsetPoint.c:75: warning 85: in function aca_setpoint unreferenced function argument : 'ui16_time_ticks_between_speed_interrupt'
ACAsetPoint.c:75: warning 85: in function aca_setpoint unreferenced function argument : 'ui16_time_ticks_between_pas_interrupt'
ACAsetPoint.c:75: warning 85: in function aca_setpoint unreferenced function argument : 'setpoint_old'
sdcpp.exe: fatal error: when writing output to : Broken pipe

BTW, about "external" (internal on geared motors with white wire in motor cable) sensor - some Bafang drive (i have SWX02 and BPM) has 6 magnets, and generate six pulses per wheel revolution. Original firmware show speed correct with "P2=6" KT-LCD3 setting, this firmware show incorrect "speed*6" :(

[randysway]

First, thank you all for your work on the firmware and documentation surrounding the KT controllers and their analogues.

I've flashed it - but have been unable to test .... because I hadn't taken any good (helpful) photos of the internal wiring/pads before rewiring it. Heh.

It's the 18 Fet controller, KTE-18S5-2RX, which I think (?) has the same pad assignments as the 12/15 Fet models.

At any rate, I'm primarily interested in the Brake + Throttle pads. Although, a list of the stock connections would be phenomenal for myself and the community. I'm 98% sure that this information is nowhere else online.

[stancecoke]

Latest firmware compiling with errors, if "speed sensor = external" marked in configurator:

there was just a semicolon missing in line 69 . I've fixed it at github. But the refactored code is not proven yet. Perhaps you can test it and give us feedback.

BTW, about "external" (internal on geared motors with white wire in motor cable) sensor - some Bafang drive (i have SWX02 and BPM) has 6 magnets, and generate six pulses per wheel revolution. Original firmware show speed correct with "P2=6" KT-LCD3 setting, this firmware show incorrect "speed*6"

feel free to add the parameter PULSES_PER_WHEEL_REVOLUTION to the code
Until we have not implemented this feature you can use a workaround by setting the wheel circumference to one sixth of its real value.

It's the 18 Fet controller, KTE-18S5-2RX, which I think (?) has the same pad assignments as the 12/15 Fet models.

At any rate, I'm primarily interested in the Brake + Throttle pads. Although, a list of the stock connections would be phenomenal for myself and the community.

Welcome to the project! Can you post a photo of your controller and a source where to buy it?

BTW, I've optimized the torque override function in the related branch. The override is now ramped down with speed, so you get assistance from the direct torque signal mainly at acceleration from standstill. That avoids a pulsating assistance when you are already in motion.



regards
stancecoke

[randysway]

These are the photos I took when I first got the controller. It's external branding is "RisunMotors", and appears to be identical to one posted (by you stancecoke?) on the bitbucket site.

Branding changes AGAIN on the amazon listing - a company called "HalloMotor"
https://www.amazon.com/gp/product/B01MG ... UTF8&psc=1

[honya96]

These are the photos I took when I first got the controller. It's external branding is "RisunMotors", and appears to be identical to one posted (by you stancecoke?) on the bitbucket site.

Branding changes AGAIN on the amazon listing - a company called "HalloMotor"

luckily these 18fets are all the same, I got 3, each from different seller. Are you running the custom FW?

BTW I recommend to add the missing big capacitor near battery cables. I have added even more, It helps not to blow your fets by voltage spikes at high phase amp. I was able to run 100A from battery shortly.

and an interesting find from yesterday, for those who want to change fets, there are smd resistors for each fet, 6fet has 101 value 12fet has 510, 18fet has 310. you have to change them to suit your gate charge.
I have changed 6fet to csd19536kcs fets and put 390 resitors (went from 9.8mOhm to 2.3mOhm fets) If Iam correct, It should work and run 40A real easily. gonna test soon

[casainho]

BTW I recommend to add the missing big capacitor near battery cables. I have added even more, It helps not to blow your fets by voltage spikes at high phase amp. I was able to run 100A from battery shortly.

and an interesting find from yesterday, for those who want to change fets, there are smd resistors for each fet, 6fet has 101 value 12fet has 510, 18fet has 310. you have to change them to suit your gate charge.
I have changed 6fet to csd19536kcs fets and put 390 resitors (went from 9.8mOhm to 2.3mOhm fets) If Iam correct, It should work and run 40A real easily. gonna test soon

You guys should be writing wiki pages, structuring this kind of information!!

I bet this motor controllers will be on the market for next years!!

I hope to start soon a new project: develop firmware for the new color LCDs, the ones the starts at 60 euros but gives a great flexibility!! Big display should let us implement kind of that graphs in real time that Stancecoke is showing of current, or motor temperature, etc.

I think I am going or with Kunteng new color LCD or the Bafang color LCD. Both of this are popular, easy to buy and kind of cheap. Banfang, we already know are easy to develop for and has huge memory size of 512kbytes of flash memory!! (KT motor controllers and LCD3 has only 32kbytes if flash memory).
Also we know that Bafang will add Bluetooth to this LCDs on next versions.

All this LCDs communicate using UART just like KT, TSDZ2, Bafang, etc, so this LCDs will be compatible with an big range of motor controllers. So please invest in a good wiki documentation for KT motor controllers because they will for sure be in the market for next years.

On TSDZ2 firmware, we are alreay a few developer/users editing the wiki pages -- sharing the efforts between us make the project going further!!

[kkm]

The question about "max current" values in the latest (C#ROME-B ) firmware versions.

What method of calculation and document is true now?

"new" configh.h (~sep. 2018 in Master branch)

#define BATTERY_CURRENT_MAX_VALUE 150L
#define PHASE_CURRENT_MAX_VALUE 300L

https://opensourceebikefirmware.bitbuck ... ndex4.html

"Battery Current max: maximum battery current in deziAmps. A value of 150 means 15A.
Phase Current max: maximum phase current. Calculation like Battery Current max. The phase current is derived from the formula phase current = battery current / duty cycle internally."

"old" config/h (~aug. 2018)

#define BATTERY_CURRENT_MAX_VALUE 662L
#define PHASE_CURRENT_MAX_VALUE 532L

https://github.com/stancecoke/BMSBatter ... s_firmware

Reamde.md:

"Battery Current max: maximum battery current. Calculation: Value = desired current in ampere multiplied by the value from field Battery Current cal a minus value from field Battery Current cal b. Example with the default values for limiting to 15A: 15A *10 - (-312) = 462

Phase Current max: maximum phase current. Calculation like Battery Current max. The phase current is derived from the formula phase current = battery current / duty cycle internally."

In my case, the values 170/300 (in OSEC GUI ) give a full load power of 1250-1300W on a KT-LCD3 (36V battery). In the case of stock firmware, power was not more than 600-650W. I have 17A KT-SVPR controller from "pswpower".

Flexible OpenSource firmware works excellently on all my geared motors - Bafang SWX02, MXUS XF15R and XF08. Very quiet, there are no unpleasant vibrations and some resonances on the ebike frame (as it was on the stock firmware).
But ... Thumb throttle control... Very accurate and predictable work on the original firmware (and other Chinese controllers), the acceleration is predictable, the response of the thumb throttle is accurate, and it is easy to maintain the selected speed.

Now... I practically cannot use the throttle knob - poorly predicted work, it is difficult to control the chosen speed, "viscous" acceleration.
I only use the throttle knob, and do not use PAS.

This this due to the different types of throttle control? "Speed control mode" in stock, and "electric current" control in flexible firmware?

Small "Safety" issues:
1. Power on controller with thumb throttle at non-zero position - motor start. Dangerous... All other controllers blocked motor in this case
2. Motor start, if voltage on thumb throttle hall sensor > 4.2V (short circuit in throttle cable (betveen +5 and hall signal), or broken hall, and.... ) All other controllers blocked motor startup in this case...

[honya96]

the ones the starts at 60 euros but gives a great flexibility!! Big display should let us implement kind of that graphs in real time that Stancecoke is showing of current, or motor temperature, etc.

Great! Here it is for 52$
https://s.click.aliexpress.com/e/c8L1svHA

There is also a vertical version if you scroll down.

If you plan to change the layout completly it will be great to have batterry % not bars.

Will it help you if the microprocessor wasnt read protected? You can try contacting kunteng's sales man directly at szktdz.com maybe they will understand that they will sell more if you moddify it.

But I guess you want to start from scratch

[Xnyle]

"viscous" acceleration.

Might be because definition of max current changed: It's now value/cal_a (without cal_b) so if cal_a is 10, 20 Amps would be 200, if cal_a is 5 as on 12 and probably 18 FET controllers, 20 Amps would be 100.

Regarding your "requests", why don't you create a pull request if you have an improvement in mind?

[casainho]

Great! Here it is for 52$
https://s.click.aliexpress.com/e/c8L1svHA

There is also a vertical version if you scroll down.

I am waiting to receive the horizontal version of KT and the vertical version of Bafang (Bafang also has horizontal version). For what I could understand, seems the vertical are more popular on Bafang users.

I am afraid that KT color LCD has low flash memory... because TSDZ2 color LCD uses STM8 and has only 32kbytes, the user that is developing firmware to it says it has troubles because of the very limited amount of memory. I hope KT LCDs aren't cheap because of low memory.... on the other side, the Bagang LCD with 512kbytes of flash memory is much more than what I expected.

I hope soon to have both LCDs in hand and discover about this questions.

[honya96]

"viscous" acceleration.

Might be because definition of max current changed: It's now value/cal_a (without cal_b) so if cal_a is 10, 20 Amps would be 200, if cal_a is 5 as on 12 and probably 18 FET controllers, 20 Amps would be 100.

Regarding your "requests", why don't you create a pull request if you have an improvement in mind?

Don't really understand what he means but sounds a bit like a slow PI controll..? I had problems with this too. Slow reaction to adding and releasing throttle (few months back)

[randysway]

luckily these 18fets are all the same, I got 3, each from different seller. Are you running the custom FW?

Ha, yes and no. I built + flashed it without issue... but I haven't been able to run it yet. I figured I'd rewire all the external leads while I was adding the STLink + UART(bluetooth) jumpers. However, I forgot to keep track of the connections as I removed them from the board.

Reassembly was a bit of a guessing game... haven't gotten the controller working since. I'm assuming this is due to miss wiring either the LCD leads or Motor Hall leads.

BTW I recommend to add the missing big capacitor near battery cables. I have added even more, It helps not to blow your fets by voltage spikes at high phase amp. I was able to run 100A from battery shortly.

Good Call. I've been a little disappointed in the 38A upper limit I've been able to pull with it. Especially compared to the Variable Voltage 18 Fet Sunwin controller (2800W off a 13s9p!!).

BTW, If you have the pad assignments for Throttle, Hall V+ and Brake Low, written down somewhere, mind sharing them? I'm getting better at tracing leads back to the CPU pins, but I don't think practice is going remedy the headache.