Hall sensor issue, after switch

[Ilya]

Hey

I never had an electrical problem before, so this is completely new to me, but I want to learn.

I have a 36V, 250W, unknown ananda rear motor hub ebike , that recently started behaving as if it had a bad contact. It worked flawlessly for the most part of 550 km, apart from some resolved mechanical issues.

A couple of weeks ago power transmission would cease for short intervals with the motor making a rusty noise. Then it got more frequent up to the current point where there are only short intervals with motor power. Initially I thought road bumps (bad contact?) or upward slopes (temperature?) could affect the problem, but it could not always be reproduced. The Ananda D16 screen showed an error symbol with some printed out error code 25 (hall sensor).

I ended up buying the "generic motor tester" and a cable to connect it with. Turns out the blue hall sensor was always lit. When checking with a multimeter the voltage cycled between 0-2V. Other two 0-3.5V. Aha, sure enough, bad hall sensor I thought, so I opened the motor hub (easier said than done). I checked all the connections OK despite the silicone. Removing the hall was supposed to be the easy part according to the guides I watched, but the wires disconnected from the sensor as I tried to grab it with some pliers. Drilling it to dust was the worst part, as I could not identify the specific sensor afterwards. Connections OK after soldering an amazon bought S49E (3-6.5V operation) and applying new silicone.

A test with the "generic motor tester" showed a pattern to repeat consistenly with 6 unique entries, with the blue now turning on and off from time to time. Problem resolved I thought. But now I have a quite consistent Error 25 from start, only rarely the motor gets an impulse to power. Same result with the "generic motor tester" after assembling everything. Voltage checks shows the same old two 0-3.5V, and the new blue 0-4.5V. Is this bad?

- Do I have to change the other two to the same sensor? Change the new one to one of the old types? If so, any idea which one?

- When I look at it now the "generic motor tester"-pattern is somewhat sketchy:

0 : off
1 : on
Blue-Green-Yellow
000
001
101 ---? should be 011?
111
110
010 ---? should be 100?

The "generic motor tester"-manual states the normal pattern to be:
000
100
110
111
011
001

for 60 and 120 deg both.


It seems like my pattern can't be compatible with motor operation? Seems like my pattern holds three zeroes or ones in a row per color, but that they overlap out of phase. The replaced/new blue sensor needs to move down 2 places. Sensor out of phase (if that's a thing)? What can I do to resolve this?

 

[amberwolf]

I have a 36V, 250W, unknown ananda rear motor hub ebike , that recently started behaving as if it had a bad contact. It worked flawlessly for the most part of 550 km, apart from some resolved mechanical issues.

A couple of weeks ago power transmission would cease for short intervals with the motor making a rusty noise. Then it got more frequent up to the current point where there are only short intervals with motor power. Initially I thought road bumps (bad contact?) or upward slopes (temperature?) could affect the problem, but it could not always be reproduced. The Ananda D16 screen showed an error symbol with some printed out error code 25 (hall sensor).

I ended up buying the "generic motor tester" and a cable to connect it with. Turns out the blue hall sensor was always lit. When checking with a multimeter the voltage cycled between 0-2V. Other two 0-3.5V. Aha, sure enough, bad hall sensor I thought,

FWIW, it's not usualy the sensor, unless it gets overheated or water damaged. It's usually the connection between the sensor and the controller, most often at the axle wire exit from the motor, or at the connector between motor and controller, but sometimes it's actually the pullup resistor inside the controller that's poorly soldered and has come loose (very rare).

so I opened the motor hub (easier said than done). I checked all the connections OK despite the silicone. Removing the hall was supposed to be the easy part according to the guides I watched, but the wires disconnected from the sensor as I tried to grab it with some pliers. Drilling it to dust was the worst part, as I could not identify the specific sensor afterwards. Connections OK after soldering an amazon bought S49E (3-6.5V operation) and applying new silicone.



A test with the "generic motor tester" showed a pattern to repeat consistenly with 6 unique entries, with the blue now turning on and off from time to time. Problem resolved I thought. But now I have a quite consistent Error 25 from start, only rarely the motor gets an impulse to power. Same result with the "generic motor tester" after assembling everything. Voltage checks shows the same old two 0-3.5V, and the new blue 0-4.5V. Is this bad?

- Do I have to change the other two to the same sensor? Change the new one to one of the old types? If so, any idea which one?

- When I look at it now the "generic motor tester"-pattern is somewhat sketchy:

0 : off
1 : on
Blue-Green-Yellow
000
001
101 ---? should be 011?
111
110
010 ---? should be 100?

None of them is stuck on or off, but you could have a wrong polarity sensor or wrong type. Most of the sensors in these are the S411A type, which is a bipolar latching type, which toggles the output every time the polarity it senses changes (but otherwise holds the output the same as it last was).

They also don't have any actual "output" signal, they only ground the output whenever they're "on". Any voltage comes from a pullup resistor inside the controller itself (which is why they can't be validly tested without either being connected to the powered-on controller, or a properly-designed tester that provides such a pullup, or a manual pullup resistor to the hall 5v power line, unless the motor hall board has these pullups built in--but there are reasons not to put them there, and rather to put them in the controller itself, so it's uncommon to find them in the motor)

An S49E is a linear sensor, so it doesn't switch on and off, it outputs a variable voltage depending on the strength and polarity of the signal. With no field they're at half supply voltage, and full field they're at either nearly full supply or nearly ground based on polarity. They're usually used in throttles for this reason. It also doesn't have an open-collector output, so the pullup from the controller interferes with it's attempt to operate, making it even worse.


Since you can't read the existing sensor's numbers, your best bet is to get the S411A type. If you go ahead and get at least three, then if you have any problems with the other two, you can replace them all.

The "generic motor tester"-manual states the normal pattern to be:
000
100
110
111
011
001
for 60 and 120 deg both.

The patterns for 60 vs 120 are different. 120 will never see all on or all off, at least one will always be on, which is a reason to use it over 60 (a simple connection or power issue can cause all on or all off, causing incorrect position sensing with 60--either of those conditions in a 120 system are invalid and the controller "knows" it so wont' try to operate the motor and instead indicate a failure, if it's even a bit more advanced than primitive. ).

Most motors that use 60 flip the center one upside down, so that it's small face and it's numbers face away from the magnets instead of toward them, but the outer two do face the magnets. They may also space the sensors farther apart, but not always.

 

[Ilya]

Thank you very much for your insights.

I will switch the S49E for a SS41 (does it have to be S411A specifically?) .

Since the pattern changed after the sensor change I guess the error was concentrated in the area of that hall sensor unless I moved cables around during the disassembly, but I will check connections out of the motor just in case.

 

[amberwolf]

As long as it's a bipolar (latching if necessary) open-collector hall that can run on less than 5v, but can tolerate up to 20-30v, it should work.

There are versions of the 4xx series that have variations on those properties, such as ones that have their own pullup resistors, or that have higher supply voltage needs, etc., that wont' work correctly in this application.


Rarely the halls are not bipolar but are unipolar (only respond to one polarity), or they are or are not latching...if that's the case then you'll have to replace all three with the new kind to get identical results from each.

 

[TommyCat]

Whist you wait for your new hall sensors. Curious cat wants to know if swapping your BLUE and GREEN wires to your test box corrects the sequence?


Regards,
T.C.

 

[Ilya]

That's an interesting observation, maybe it would work. But given sensor placement, would it hold for all conditions and be robust enough? In this case it would be a hassle to do the rewiring motor-side, but it could probably be done easily as a temp solution at the point of connection to the controller wire.



I guess I'm a bit trigger happy when it comes to sensors, ordering them and not doing proper research first. Perhaps it's because snow is approaching and it's my only winter ready bike for the daily commute. I initially ordered the SS41, which is arriving tomorrow, but I have my doubts about it so I ordered SS411A as well. In the meantime a local store has SS460S, but at least I didn't buy it (yet). In the end, I can sacrifice a couple of days if it means picking a more durable component.

I need to read up on motors and hall sensors in general, but do you see any performance considerations to account for between latching and non-latching? SS41 seems to be bipolar non-latching so I would probably have to change them all, which I want to avoid given the difficulties I had before. And it seems like in terms of durability SS411A might be the better choice? given the 3.8-30V vs narrower 4.5-20V range. I guess a voltage closer to 5V should be expected in practice, but a higher "tolerance" just in case? SS41 has a min operating temp of -55C vs -40C, that's beyond biking temps anyway (+-150 Gauss). SS460S is latching, 3-24V - 40C. (+-30 Gauss)

The data sheets show some differences between switching behavior and temps. SS411A, SS460S seem to show stable characteristics regardless of temp. I figure component temps can get high, while 85C seems to be a cut off point for SS41 where it gets less sensitive to the field.

I don't yet understand the relevance of water damage (short circuit?) and subsequent permanent damage to the sensor (potentially getting 32V from the main line?). Is it advisable to glue the sensor for durability (water resistance, less damage from vibration)?

 

[amberwolf]

That's an interesting observation, maybe it would work. But given sensor placement, would it hold for all conditions and be robust enough? In this case it would be a hassle to do the rewiring motor-side, but it could probably be done easily as a temp solution at the point of connection to the controller wire.

If the wire combination when it worked was the same as now, changing the combination would just make it not work in some way, when you have the right sensor in place.

What it will do now depends on what the sensor you have in place is doing relative to what the old one did. If it's output is opposite of what it was before, then changing the wire combination might make it behave correctly, but my guess is it is not as simple as that, because of the way the linear sensor works vs the bipolar types. I never did that test, so I don't know exactly what the output difference is (plus we don't know which specific sensor type it had before, or how it behaved).

So, the test shouldn't hurt anything, but it may not change it for the better.

I guess I'm a bit trigger happy when it comes to sensors, ordering them and not doing proper research first. Perhaps it's because snow is approaching and it's my only winter ready bike for the daily commute. I initially ordered the SS41, which is arriving tomorrow, but I have my doubts about it so I ordered SS411A as well. In the meantime a local store has SS460S, but at least I didn't buy it (yet). In the end, I can sacrifice a couple of days if it means picking a more durable component.

As you noted, with some analysis:

The 411A goes down further in supply voltage than the 41, so it will operate correctly even when the 5v to the hall sensors drops as low as 3.8v, where the 41 is only guaranteed down to 4.5v. Many controllers use a diode in the hall supply line which drops the 5v down to 4.5v or less to start with, not coutning any voltage drops for whatever reason between there and the motor.

The higher voltage capability of 30v vs 20 means it can also handle spikes in the supply induced by phase currents in the wiring better.

The 41 can sink twice the current the 411 can, but that's not usually an issue in this usage. (sometimes it makes for better noise immunity in ground (on) state, when there is a lot of phase current noise induced into the hall signal lines).

Neither one is latching, but that's not required for most systems. Latching just means that it holds the last state until it's triggered again. If the magnets on a motor are discontinuous, with spaces between them, latching is important to tell the controller to keep that state until the next magnet comes along instead of accidentally changing to a new phase state that may not smoothly keep the motor spinning. Yours is built with continuous magnets around the rotor, so latching isnt' required.

It's likely that even if you used a latching sensor and your others aren't, it would operate correctly, as long as they are all bipolar or all unipolar, etc.

The 460S can handle even lower voltages, but not as high a max. It's latching, but it takes much less magnetic field to trigger it, so it may be triggered more often by the constantly-switching phase fields than the others. The 41 would be most immune to this, but the 411A is a little more immune than the 460S.

If you ever have a motor that doesn't switch the halls properly, using one that's more sensitive may help, but it will also be more sensitive to the fields of the phases, so they work better on separate "encoder" rings like the one this motor has:

Really strange KT controller issue, help please!

Hi! I have the following issue: Sometimes the motor assistance completely cuts-out and does only reingage when I stop pedalling for a second or two and/or release the throttle and then start again. The display stays on, it's only the motor that cuts out. It seems to can happen in any assist...

endless-sphere.com

but most of those use linear sensors on a sin/cos sensor arrangement with the fields on the ring magnet gradually changing from one polarity to the other rather than suddenly like on the rotor magnets.


The temperature ranges are not too important for most common usages, but if you are pushing a motor hard it can quickly reach 50-70C, as happens on my SB Cruiser heavy cargo trike, or on my old DayGlo Avenger with the little Fusin motor I was regularly overloading.

I don't yet understand the relevance of water damage (short circuit?) and subsequent permanent damage to the sensor (potentially getting 32V from the main line?).

Water damage usually means corrosion, and that can eat away the metal leads of the sensors, as they are at the outer edge of the stator where water will get on them as the rotor spins even if they're mounted at the "top" of the stator..and if they are at the bottom of the stator water sitting in the motor pooled around them may eat them away faster (especially if you have road salt issues and a lot of water accumulates in the motor over time.

I dont' see any sign of that inside your motor.


If there's enough water and conductive contaminants inside the motor, it could also conduct phase voltages into the sensors, which can damage them (and the controller itself), but this is usually secondary to the corrosion damage, and I don't recall a specific instance of it diagnosed as such here on ES.

Is it advisable to glue the sensor for durability (water resistance, less damage from vibration)?

They're usually glued in place to keep them from vibrating out of the slot, but if they're soldered to a board with such short leads as on yours, that will keep it from happening on it's own.

 

[TommyCat]

That's an interesting observation, maybe it would work. But given sensor placement, would it hold for all conditions and be robust enough? In this case it would be a hassle to do the rewiring motor-side, but it could probably be done easily as a temp solution to at the point of connection to the controller wire.

This is just a thought experiment… as your hall sensor(s) is on the outs. Any wire swapping would be done at the controller connection. Had this in mind if you were just testing with the test box just to see. Don’t make any extra effort on this… Might have been of more interest if you absolutely had to get it going NOW.

I need to read up on motors and hall sensors in general, but do you see any performance considerations to account for between latching and non-latching?

As always try to go with the OEM model that the manufacturer installed. If the part number absolutely cannot be determined, you could do your own testing with a magnet to test between bi-polar, (needs both north and south poles to change states) and unipolar needing just one pole that depends on gauss strength to change state.
The higher operating voltage seems a non issue to me as you point out. A little extra lower voltage operation would be nice.

Note: If you do all three, and having a 60 degree system. Make sure you match the positions of the replacements to how the originals were. I.E. the direction that the faces point (beveled with part stamp) as far as tword or away from magnet. Unfortunately I’ve seen that in this configuration the middle sensor can be different to the outer two…

Is it advisable to glue the sensor for durability (water resistance, less damage from vibration)?

Absolutely, often a high temperature epoxy is used to help keep them in place. With silicone on the wiring for extra water resistance if desired.

 

[Ilya]

While waiting for the 411A I tested 460S with the generic tester only to discover that I managed to short circuit the red and black inputs during its installation. I resolved that, but the blue is always on (with some changes in light intensity, basically Blue 0 below is "0.5"). Toast (or some other issue)?

BGY
100
101
001
011
010
110

Just in case I also checked the hall connections with the external motor-controller wire which seem to be OK.

Tried testing with a (rather weak) magnet, but no flipping. (I guess I need to test with an even stronger magnet to overcome the field that's there?).

Hopefully testing the 411A tomorrow.

A bit worrisome that it's always on, same behavior ("0.5"-1) as before I installed the linear sensor.

Finally found the motor ID EP1020822-1011100005-01 (Ananda M145)

 

[TommyCat]

Finally found the motor ID EP1020822-1011100005-01 (Ananda M145)

Some motor manufactures offer hall sensor replacement boards, with all three hall sensors connected... just sayin.

A bit worrisome that it's always on, same behavior ("0.5"-1) as before I installed the linear sensor.

Growing suspicious of your tester...
If you swap two of the hall sensor signal wires to tester. Does the issue stay with the identified bad hall sensor, or follow the test lead?

I might not be understanding exactly how your testing...
See this testing thread to see if we are on the same page on the how and whys. Or use the multi-meter testing procedures mentioned.

Testing BLDC motor's Phase Wiring - Hall Sensors and Wiring. - Electricbike.com Ebike Forum

Tried testing with a (rather weak) magnet, but no flipping. (I guess I need to test with an even stronger magnet to overcome the field that's there?).

I would say you need to have the motor's magnetic ring away from the sensors...

 

[Ilya]

Good suggestions. I think I found the entire motor as a relatively cheap replacement part, which is a comfort, but fixing things is better.

Will do the tester test you mentioned, by switching cables, just to be sure. The initial hall setup showed some discrepancies in the multimeter readings of blue (no more than 2V, others were 3.5V). The linear setup managed to cycle the blue lamp to real off but not in phase with the other readings. New voltage readings would be interesting.

Going to check your guide before additional testing

This is what I use + unused additional similar cable , but with a male connector. (could make it easier to check voltages with the controller in).

Basically following colors for phase to phase and hall sensor color to hall sensor color.

Wish I could remove the magnetic ring easily, but seems like there's a lot of glue involved.

 

[TommyCat]

I would say you need to have the motor's magnetic ring away from the sensors...

Not quite sure what you mean by "lot of glue involved". But to clarify, my suggestion is just about separating the hall sensor's location from the magnets for accurate testing. As in rotor separation from the stator... not magnets removal.
If the sensors weren't so stuck in the stator, you could just move the halls and PC board out and away from the magnets as another example.

I think I found the entire motor as a relatively cheap replacement part, which is a comfort, but fixing things is better.

Looking forward to your latest testing results.

 

[Ilya]

Ok, so with the generic tester plugged in I get 5V black to red.

(I didn't test with a 10k -ohm resistor.)

- Yellow, Green to Black spans either 0 or 3.3V (Yellow, Green to Red 1.7 and 5 V, which I guess is expected)
- Blue to Black spans either 0 or 1.5V.

I removed the 460S for the 411A (it gets trickier and trickier to fasten the sensors) with similar results.
Black to red 5V as expected.

- Yellow, Green to Black spans 0V and 3.3V. (Yellow, Green to Red 1.7 and 5V)
- Blue to Black spans either 0 or 1.2V (Blue to Red 0 and 3.8 V)

The blue is either half lit or lit. The new pattern appears to be exactly the same as with the 460S:

BGY
100
101
001
011
010
110

I moved around blue and green at the tester, with the result that green is either lit or half lit instead of blue.

I'm starting to think that the original hall sensor was ok after all since the results are more or less the same between 3 sensors (except the linear one). From my notes the voltage spans differ a slight bit between the senors (but then my sampling method didn't do science proper justice either).


I should also show a pic from a different angle. Is that corrosion? I removed some of the silicone to reveal the cables, pushed them around a bit but seems like there's no difference on the signals. Also tested moving around cables at the motor exit (where the metal spring covers the cable, but could not detect any changes). Seems robust to perturbations.

Did not manage to remove the stator from the rotor yet. The glue part is between the beige plastic and the metal. But from your pic I see that the pieces are not to be separated there. I tried to push, but seems to require a lot of force. Do I just grab the axis and pull away from the rotor? Remove gears first if they are present on the back side? (can't verify the back side right now).

Could the voltage drop Blue-Black as compared to the rest imply that there is an additional connection with some resistance between blue and black somewhere? (Leak?)

 

[TommyCat]

Ok, so with the generic tester plugged in I get 5V black to red.

Is all this testing with the motor disconnected from the controller?

 

[Ilya]

Correct

 

[TommyCat]

I moved around blue and green at the tester, with the result that green is either lit or half lit instead of blue.

From this description it sounds like the issue is indeed in the BLUE hall sensor circuit.

To me the description of LED bright to dim, verses 0 to 1.2vdc seems odd. Must be how the tester is set up.

I would recommend that you take your meter's test leads directly to the hall sensor's wiring legs to directly test the voltages there.

Here is a diagram to easily identify the inputs and output...

Hopefully this will identify the short, or issue that you suspected earlier. (Looks like other components on the P.C. board...)

Note: Your last two hall sensor sequencing information supports that you're working with a 120-degree board.
So, no worries about the middle sensor as far as that is concerned.

Updated diagram from the BLDC testing thread...

Look familiar? 1-2-1-2-1-2.

 

[Ilya]

Will do that. The pattern might be correct now then? I don’t think I decided to register dim lights as 0.5 with the original sensor in my notes , but set them to 1. Maybe it was the same pattern. At least somehow the motor would start , but not be consistently on. Sometimes if it didnt work and I started it hours later after putting it in a slightly warmer than outside garage it would work. Then after running over some bumps it would start behaving like if it was almost ”out of gas”. If the pattern is right one would think it would only be a matter of adjusting software (if such rights were bestowed on the user). Or to amplify the signal. Eg if the blue voltage is higher than 1V supply 5V from red.

 

[TommyCat]

The pattern might be correct now then?

I believe so… yes. Pattern good, voltages no. Damaged wiring thru axle? Looking forward to voltage checks @ sensor directly.

 

[Ilya]

So, when testing today, dim-blue flickered from start for some reason. I seized to opportunity to check parts of the cable and seem to have found an error (if not -the- error). By bending the cable out of the motor at a non-intuitive location close to the controller connector in different ways I could toggle flickering from dim to off. Not very obvious, I could not detect this during my first tests, but then I gave up and went on to suspect the halls a bit too quick perhaps.

They all give 3.3V / 0V now, no dim blue. If I bend the cable at that location the dim behavior is back. This could also be consistent with observations of the sketchy motor behavior. Maybe stable for a couple of bumps, but also stable during error. At closer inspection there's a sharp cut mark close to that location, but interestingly not directly toward blue as far as I can see. In general the cable has seen better days.

The wiring through the axle seems robust still though, I tried this location first as it would be intuitive.

Next few tasks will be to secure the soldering of the new hall to make sure it holds up to vibration.
Reapply silicone.
Deal with the cable.

I think it's long enough to survive a cut, but the complicated part will be to pinpoint the exact location to not cut too much. Then in some way it has to be put together in a robust way. What would be the best way to deal with this?

Edit: Now I can't reproduce the dim behavior at all. Very odd. Chose to not resolder what must qualify for the worst soldering of the year award, tried to remove the connectors with a little force, but did seem quite stable, too risky to improve what works. At least something to consider in future errors. Motor seems to work in walk mode now.

Edit 2: Worked an entire ride now. Took some precautions with bumps and didn't use full motor power. Want the silicone to fasten a bit more. I expect the error to reoccur at some point, but willing to give it a chance and to verify if it's the cable.

 

[amberwolf]

Possibly useful info:

Cable/wire faults cause more problems than anything else in electrical systems. With "ebike" cables I've seen some wierd stuff, aside from the damage that can happen to them in various usages. The most common is just conductor broken inside the insulation; it usually happens where a cable bends or exits something, but it can happen anywhere.

A cut into the cable jacket can slice thru the wires inside, too--not always completely, and eventually can cause a break from vibration or bending, or moisture can get in thru the slice and corrode conductors till they break.


The least common (but I've still seen it more than once) is a splice in the cable--the manufacturer of the cable itself uses spools of individual wires fed into whatever machine casts the cable housing around the wire bundle, and the spools may not all run out at the same time. Sometimes they'll just twist the wire ends together on a splice from a new spool to an old one and keep the system making cable, so there can be an unsoldered splice hiding inside the cable--if it was wet or humid when this was done, that moisture is trapped inside and corrodes the exposed conductor. Sometimes they just stick the new end into feed and don't even bother splicing it, so there can be a gap between the old wire and the new one, so that wire doesn't even work in that length of cable when it gets used to make something.

 

[Ilya]

I tested the bike for a bit longer today. Despite being careful I ran over a small bump after which the problem reoccured. Stable operation until then. I could somewhat reliably correct it by bending or compressing the cable. But it would hold for a short while only. I think it has to be a cable issue. Sensitive to moisture temperature and what not as it seems to work sometimes afternoon after being idle in a garage.

I could theoretically strip the outer plastic to inspect the wires, and reapply a cover with some heat after repairing it. But I guess theres a best practice to this . Any ideas?

 

[TommyCat]

They all give 3.3V / 0V now, no dim blue. If I bend the cable at that location the dim behavior is back.

This is really great to hear! Well done on finding the way to make the problem re-occur.

Now I can't reproduce the dim behavior at all. Very odd.

Intermittent problems are the worst.

I think it's long enough to survive a cut, but the complicated part will be to pinpoint the exact location to not cut too much. Then in some way it has to be put together in a robust way. What would be the best way to deal with this?

Measure the absolute maximum length you can cut out, mark it, then take small steps backwards from the end hoping the damage position shows before the marked part. Perhaps keep a meter on the bad wire testing resistance to help you determine the damaged part's position. Could it be in the connector's pin connections?

I spliced my motor's wiring cable to allow for a HIGO connector near the hub motor to allow for easy wheel/motor removal for maintenance or repair.

It's doesn't have the same type/size wires (my controller is internal to the motor hub...) but perhaps looking at it may give you some ideas about splicing or replacement....

1st Build & E-Bike rider-Magic Pie V5 (rear) -Bottle Batt. 52v*11.5ah-Huffy Parkside - Electricbike.com Ebike Forum

 

[Ilya]

I'm leaning toward that kind of solution. Youtube vids have the opinion that the entire cable should be removed, but that would be a real mess and not only because of the obscuring silicone. But, after some more testing I'm starting to suspect that the error could be where the cable meets the connector. If I compress or move around the cable in that area, I seem to get motor function for a bit. It could make sense since I separated it when I replaced the back tyre to wintery ones. Vague memory of it needing excessive force. Maybe the connector separated a bit from the cable. One scenario would be to cut it off, remove cable depris and insert the rest, but I have no clue about how the cable is attached to it. Seems pretty spacey, but so is an ethernet connector which "requires" a special tool. If it's too much work I could just remove that part entirely and attach a ready made with a connector.

 

[amberwolf]

Youtube vids have the opinion that the entire cable should be removed,

That's the easiest way to ensure waterproofing and that the whole cable is working; removing the existing one and replacing with a new one...or finding the damaged spot (if it's not at/in the connector) and pulling the remainder of the cable between that spot and the connnector back thru the axle, then splicing the wires inside the motor, removing the whole length between the damaged section and the original motor connnections inside.

But, after some more testing I'm starting to suspect that the error could be where the cable meets the connector. If I compress or move around the cable in that area, I seem to get motor function for a bit. It could make sense since I separated it when I replaced the back tyre to wintery ones. Vague memory of it needing excessive force. Maybe the connector separated a bit from the cable.

It's possible. If it's a waterproof connector, the wires are either crimped or more likely soldered into the contacts of the connector, then placed into a mold, and plastic injected into the mold to create the "overmolded" connector that makes it waterproof.

There's no good way to repair those; I've usually broken other wires (or cut them) while trying to cut away connector plastic enough to get to the contacts/wires to fix them, with many different kinds of overmolded connectors in various equipment.

What usually happens if the wire is pulled on instead of the connector housing grip area is the wires break right at the back of the contact. They may not exhibit any problems, especially if they're low-current, until vibration or movement or more pulling has created an actual gap in the conductors. At that point moving them around can make them reconnect, but it's not usually permanently fixed by this.

 

[Ilya]

Finally some time to continue this project. Managed to use another bike in the meantime. No snow in sight currently.

I cut off the small cable plastic a bit too far back. Hard to adjust the lengths with so many cables. Strategy was to twine it as much as possible and then to solder. Still, after adding some insulation there's some unprotected wire outside. If I were to cut it again I would probably consider removing the outer plastic and cutting the smaller wires at different points. In any case silicone at those locations will have to do.

Another round of shrink plastic and silicone:

Put some additional shrink plastic over the plastic cable all the way to the connector as well. Then some tape:

Had it out for a test run, seems to work. But really needs a longer ride to know for sure. This will of course be a weak point, and a matter of time before it breaks down, but hopefully not this year. Submerging the bike is probably never a good idea, but I can imagine that some testing needs to be done in rain.

The bump could likely also have been avoided by cutting down the outer plastic sheath even more, insert the shrink cable on one side, solder the connection straight, then slide the shrink wire over. But, the bump is ok, and perhaps the orthogonal twining helps wire strength.