search

danielnilsson9 / bbs-fw

Open source firmware for multiple electric bike motor controllers.
GNU General Public License v3.0
183 stars 37 forks source link

[BBS02] Speed sensor loses tracking #114

Open MrHumanRebel opened 1 year ago

MrHumanRebel commented 1 year ago

Recently rode the bike with the latest 1.3.1 version for a long run, I noticed around every 10-15 minutes the speed sensor data is not shown properly for a few seconds. For example cruising at constant 35 km/h and sor a few moments firstly a lower speed is shown like 15-20 km/h and then a spike up to the 50s, then it is back to normal. Might be a sensor pulse filtering issue, but I'm just guessing at this point, I have no more information about his anomaly.

The speed sensor and the magnet is located perfectly to each other, and working (and so blinking) perfectly fine. This is with a BBS02B.

Thanks in advance!

danielnilsson9 commented 1 year ago

No idea. I think there is one other guy on a BBS02 who had similar issue too along with a lot of other issues. Haven't seen a this on my BBSHD or TSDZ2. I can leave the issue open but if this was a common firmware issue on the BBS02 I would have expected quite a few reports.

I own no bike with BBS02 so I have no experience of how it runs on that controller, support was added blindly by me and then tested by others.

kb79000 commented 1 year ago

Hi, i did not notice this problem on my bbs02b

silenec commented 1 year ago

Yeah I did notice this on my last longer ride with road speed limited PAS levels. The motor randomly cuts power for no apparent reason and spins up again while the speed on the display goes up to 50s. Not very often though as you say so it is hard to diagnose.

ronmicro commented 1 year ago

Yeah I did notice this on my last longer ride with road speed limited PAS levels. The motor randomly cuts power for no apparent reason and spins up again while the speed on the display goes up to 50s. Not very often though as you say so it is hard to diagnose.

My bbs02b with v1.3.1 has this same problem of random power cuts, it happens about 2 or 3 times in a 10 mile ride

danielnilsson9 commented 1 year ago

Alright, definitely sounds like a firmware issue then. Was anyone of you running 1.2? Was the issue present there too or is it new in 1.3?

MrHumanRebel commented 1 year ago

Alright, definitely sounds like a firmware issue then. Was anyone of you running 1.2? Was the issue present there too or is it new in 1.3?

Personally, I started using this firmware from 1.3.1 so I have no information regarding this.

ronmicro commented 1 year ago

I started from 1.3.1 too but i just tried version 1.2.0 on a long ride and it still happened a couple of times

Larswagner1987 commented 1 year ago

Ich habe auch mit 1.3.1 angefangen, aber ich habe gerade Version 1.2.0 auf einer langen Fahrt ausprobiert und es ist immer noch ein paar Mal passiert

that's a shame, do you have any idea what could be wrong, when does it happen, is there a trigger or did you notice something?

Gonespa commented 1 year ago

Hi,

I Have 1.2.0 too and noticed the same. I though this was related to the fact that i have max current set to 33 amps in a bbs02 and it was hitting some kind of extreme condition but I see it might be due to other thing. In my case, I noticed that behaviour when locking the rotor and demanding max current for a few seconds. I managed to see an error coming in the log.

Larswagner1987 commented 1 year ago

I also suspect that it has something to do with the current, every time the motor is supposed to give current it happens, on the road it's just not a problem but as soon as there is an increase it starts as soon as he has to give more current then he interrupts 1-2 seconds and then the line comes back, I have a bbs02b motor 48v 750Watt

dav0000000 commented 1 year ago

Are you guys using pas or throttle when it happens and if you are in SPORT, is it still in SPORT after the problem?

Larswagner1987 commented 1 year ago

ich benutze nur pas und die normal , also kein Gas oder Sport

Gonespa commented 1 year ago

I have only noticed this when using the throttle. I am not 100% sure about the mode, but I would say thay the mode is not changing. If it happened I would probably not notice it since I have configured the "standard" mode as full power mode (and the one I use 99% of the time0 and "sport" mode as a tamed down one.

Larswagner1987 commented 1 year ago

I have no gas in use, I tried again to generate the error and it only comes when the engine draws power, then the power goes away and comes back then it's fine and at some point it starts again, it's weird, it can but just look at my display and see how the watt reading goes to 0 in time and then is full again. must be a small thing, but where only in the log is there nothing in it for errors.

danielnilsson9 commented 1 year ago

@Gonespa Probably not related to this issue unless others also have their current set that high. It's been reported before that it's not possible to set current above 25A for BBS02 because some other hardware protection is triggered outside the control of this firmware, that is probably what you are seeing. Set max current to 25A and it should work normally.

Larswagner1987 commented 1 year ago

@danielnilsson9 I have mine set to 23a and it can be 25a, could it be because the engine had the misfires as a result???

Gonespa commented 1 year ago

@Gonespa Probably not related to this issue unless others also have their current set that high. It's been reported before that it's not possible to set current above 25A for BBS02 because some other hardware protection is triggered outside the control of this firmware, that is probably what you are seeing. Set max current to 25A and it should work normally.

Agreed, I only mention it since the described behaviour looks exactly as I saw. I mean, behaves just like when the motor control micro returns an error: power disconnects completely for 1 or 2 seconds and the resets. Looks to me it is returning some kind of error. I was thinking, is it possible to save in permanent storage "the first error" returned, so that it can be retrieved like a car MCU does with the app?.

Anyways, other than that extreme condition, I also noticed this a few times when using the bike in much less stressing conditions.

Larswagner1987 commented 1 year ago

@danielnilsson9 The only thing I have done now is connect the laptop and start and then the error comes, but it only shows Motor target current changed to 0% for 1 second and then it goes back to normal, and once it says Failed to read voltage from motor controller, everything was fine so far.

Larswagner1987 commented 1 year ago

@danielnilsson9

I've tried everything now and it's a pity that this error is there, as soon as the laptop is connected it doesn't come up anymore as soon as it's not connected it comes up which is strange, the cables are new. I tried different settings but it didn't work, I don't know what the problem is anymore, sometimes it's there immediately, sometimes it's only there after 4 km, then it doesn't work at all, it just breaks off and then the performance comes back after 1-2 Seconds, I've got the original bbs02 version back on it now, if you find the bug and then have a new version I'll try it.

danielnilsson9 commented 1 year ago

I see, yes you are probably better off with the original firmware. These issues are not likely to be fixes as I don't have a bbs02 and are not doing any work on that motor. So unless someone else decides to do some investigation here it will probably stay this way.

Larswagner1987 commented 1 year ago

@danielnilsson9

ok shame, like I said, it's great except for the error, I can only send you a bbs02 controller or the original hex file, can you do something with it then?

silenec commented 1 year ago

I got back from another ride and I can say this bug appears way more often and / or is more noticeable when going uphill. I can pretty much predict some situations where it happens. It almost always happens when approaching a hill.

I think we should test some bare minimum firmware - no temperature monitoring, no throttle, no keep current, no push walk, no ramping anything that is not necessary, maybe even no speed sensor if possible etc. and see when it starts working properly.

Larswagner1987 commented 1 year ago

@silenec then try it and let me know, I definitely need the speed sensor and the shift sensor too, I don't need the rest like the gas, I hope that the error trigger is found, the original version is really not nice to use there this version is perfect. thank you already 👍

danielnilsson9 commented 1 year ago

@Larswagner1987 Thanks but I have a BBS02 controller and firmware but diagnosing things that only occurs while riding on high power is hard with a controller on my desk.

@silenec That makes sense, I was never able to reproduce that issue, I have ridden > 100km since then using PAS and throttle on high power on my BBSHD but it's rock solid, I have not seen a single glitch.

Weird thing is that there is not really any BBS02 specific code here in this firmware apart from pin assignments and motor MCU configuration. Search "BBS02" in all files and you will find very little. The motor MCU config was directly mirrored from the original BBS02 firmware.

Does the speed on you display show incorrect value each time this happens?

silenec commented 1 year ago

Sometimes I see speed glitches on my display when it happens but sometimes I do not yet the motor still stops. It is weird. Maybe something in the NEC MCU does not like how the new FW behaves? Maybe there are some limits like maximum change of current / % of requested power per second or something and the limits might be different for BBSHD? And when something goes over the limit, NEC MCU "resets" and starts again? I have no idea honestly.

danielnilsson9 commented 1 year ago

Ye maybe, I was thinking along these lines too as there is actually an error code reported from NEC MCU on BBS02 described here https://github.com/danielnilsson9/bbs-fw/wiki/NEC-Motor-Control-Protocol which I guessed was an overcurrent error but could be anything really. It's not exported to any error code sent to the display so you will never se an error when it occurs. Would be interesting to know if that error is set when this issue occurs.

NEC MCU shutting down doesn't explain the speed sensor reading issue though, since speed sensor reading is done by this firmware..

silenec commented 1 year ago

Now that I think about it... I put new chain on an older cassette once and the chain used to skip. The motor somehow detected that and every time the chain skipped, the motor did exactly this - stopped and spinned up again. Same thing when my rear tire slipped on ice in winter - motor stopped and started again. Not sure how that worked exactly. That was a few years ago on stock FW. Might it be the "chain skip protection" (if there is one) getting triggered by glitchy speed sensor values? Is the speed sensor reading reported to the NEC MCU or not? I guess not so this would not make sense then... hmm

dav0000000 commented 1 year ago

I had speed sensor rise with a watchdog reset a while back. Only way I knew was SPORT mode always reset to STANDARD as the system restarted.

Larswagner1987 commented 1 year ago

Hello

have you tested everything? a friend of mine thinks it could be something with the magnets, because when you drive and the power is gone, he took a closer look and noticed that it could be in the pas leagues, but there could also be incorrect information, the second could be there it's all mirrored what's wrong with the current values, since the BBs motor does a mini reset and then it's full again, or could it be that the memory/ram is too full and that's why it's happening?

ronmicro commented 1 year ago

I tried disabling the speed sensor in the config tool today but made no difference, still a lot of drops over a long ride

silenec commented 1 year ago

We could also try if increasing the Stop Delay (ms) changes anything in case the default value is too low or if lowering the Current Ramp (A/s) to some low value has any effect.

danielnilsson9 commented 1 year ago

Maybe its RAM related anyways, its the only thing I can think of that is different between BBS02 and BBSHD, BBSHD has 4kb of ram, BBS02 has 2kb and there is only like 4 bytes of free ram left when compiling for BBS02. As per my current understanding that should be ok as everything is allocated upfront but maybe I have misunderstood something or there is a compiler bug causing these issues.

I will fix issue #86 and see if that helps, it's worth a shot. I'm also considering trying to compile with the Keil compiler and see if that makes any difference.

MrHumanRebel commented 1 year ago

Thanks @danielnilsson9 in advance for your hard work and patience debugging these stuff!

Larswagner1987 commented 1 year ago

@danielnilsson9 Thanks also from me and my wife, you are doing a great job, you are the best. Thank you thank you thank you.

silenec commented 1 year ago

Meanwhile I went back to stock FW and the skipping is gone, no speed jumps on the lcd. So just to confirm, definitely not a hardware / sensor issue.

MrHumanRebel commented 1 year ago

@silenec Do you have the stock FW for the 48V BBS02B? If so can you share it with me somehow? Just for backup purposes, I plan to further drive this FW by @danielnilsson9.

silenec commented 1 year ago

I used the biktrix one named CRR1094824E010015.3_170815.hex https://biktrix.stonly.com/kb/guide/en/52v-programming-for-bbs02-motor-K2gUq2nWAX/Steps/1132765 It is a "52V" FW but it works fine with my 12S battery (minimum LVC is 38V, same as with the stock-stock FW before I flashed anything).

MrHumanRebel commented 1 year ago

Compiled a beta FW from the "uart_mem_opt" branch yesterday before it has been merged, and rode 100kms today without any issue, except this one... This issue still remains, cruising at 30km/h then random spikes to the 50s, the motor stops for a few seconds, then is resumes for 10-15 minutes, then it repeats.

danielnilsson9 commented 1 year ago

Thanks for testing! I figured it would not help while rereading all the comments in this thread. Must be something else. I'm gonna forward the error code from the NEC MCU to the display so we can know at least if that specific error is triggered and reported.

Edit: Display will now show error 09 (phase line error) if the suspected NEC error code is reported. Keep an eye on the display when power is interrupted.

Larswagner1987 commented 1 year ago

Hello, have you already found the error? so far everything is going well except for error 09, I would be happy to hear that everything works without problems, the software is simply perfect and is what you want, I can only say thank you to Daniel for his great work on it.

danielnilsson9 commented 1 year ago

Hello, have you already found the error? so far everything is going well except for error 09, I would be happy to hear that everything works without problems, the software is simply perfect and is what you want, I can only say thank you to Daniel for his great work on it.

No I don't think it is resolved, see comment above by MrHumanRebel. The issue is still remains, fixing ram usage did not have any effect.

Larswagner1987 commented 1 year ago

Hallo, hast du den Fehler schon gefunden? Bisher läuft bis auf Fehler 09 alles gut, es würde mich freuen zu hören, dass alles ohne Probleme funktioniert, die Software ist einfach perfekt und entspricht Ihren Wünschen, ich kann mich nur bei Daniel für die tolle Arbeit bedanken.

Nein, ich glaube nicht, dass es gelöst ist, siehe Kommentar oben von MrHumanRebel. Das Problem besteht immer noch, die Behebung der RAM-Nutzung hatte keine Auswirkung.

Okay👍

MrHumanRebel commented 1 year ago

Rode another 100+kms with my bike as I ride everyday to work and back. Funny story, some bug got into my eye and I looked down and saw the km/h jumping from 35 to 28.5 and back to 35. (the motor was going flawlessly as 28.5<35 so no dropout) So that means the speed sensor spikes not only happen to the + direction to the 50s (then the motor stops obviosly 50>35), but also to the - direction. I noticed this firstly today.

ronmicro commented 1 year ago

This might just be a coincidence but i changed my display from a DM03 to an 850C and didn't get any drops at all on my last 2 rides

Larswagner1987 commented 1 year ago

how do you mean that no pot get??? I do not understand that .

ronmicro commented 1 year ago

Sorry what did you mean by no pot get? it didn't translate well. Maybe some displays randomly send a value that makes it skip the switch case statement in process_bafang_display_write_pas()?

ronmicro commented 1 year ago

Changing that function a bit seems to have fixed it for my DM03 display, went 16 miles with no power cuts yesterday

silenec commented 1 year ago

ronmicro sounds good, is it still working for you? Can you share the modification?

ronmicro commented 1 year ago

Yes i've been on a few rides with no power cuts now. This is in the extcom.c file line 721

static int16_t process_bafang_display_write_pas()
{
    if (msg_len < 4)
    {
        return KEEP;
    }

    switch (msgbuf[2])
    {
    case 0x00:
        app_set_assist_level(ASSIST_0);
        break;
    case 0x01:
        app_set_assist_level(ASSIST_1);
        break;
    case 0x0b:
        app_set_assist_level(ASSIST_2);
        break;
    case 0x0c:
        app_set_assist_level(ASSIST_3);
        break;
    case 0x0d:
        app_set_assist_level(ASSIST_4);
        break;
    case 0x02:
        app_set_assist_level(ASSIST_5);
        break;
    case 0x15:
        app_set_assist_level(ASSIST_6);
        break;
    case 0x16:
        app_set_assist_level(ASSIST_7);
        break;
    case 0x17:
        app_set_assist_level(ASSIST_8);
        break;
    case 0x03:
        app_set_assist_level(ASSIST_9);
        break;
    case 0x06:
        app_set_assist_level(ASSIST_PUSH);
        break;
    }

    return 4;
}
Larswagner1987 commented 1 year ago

@ronmicro

Can you tell us exactly what you did so that there are no more failures? I would be happy if it finally works without the failures, thanks in advance.