Help diagnosing controller/motor problem - not starting

[gertjan]

Hi e-bikers

I am looking for some suggestions on trouble shooting my e-bike problem. Last week on the way home the motor started sputtering. Didn't start smoothly and when it ran, it was noisy at low rpms. At this point it does not start up anymore at all. However, I can _sometimes_ get it going by starting it forward by hand (at some significant speed). It will run ok, although it seems a little less smooth than normal. It doesn't reliable run like this though. Turning off power and restarting improves the odds of it running like this.

My guess is always Hall sensor connectors. I have two sets of connectors - my controller (situated in bike front panier) to the wiring attached to frame, then a set near the rear hub (an Ezee geared motor, which I think I got from Grin Tech). After suffering lots of problems with the single row amphenol connectors (corrosion), i replaced the rear set with a 5 pin mini-XLR two years back. The phase wiring is Anderson power pole. Near the controller I still have the flat 5 pin connectors for Hall and bullet connectors for power. I am convinced the connectors are good this time.

Next guess: Hall sensors. Got out my multi-meter and probed the back of the connector of the controller cable i.e. as close to the controller as I can get. I am still not sure why, but initially I got a strange reading 5 V every where. That was Friday morning before work. Confusing. On the way home, I picked up a new set of Hall sensors at Grin just in case and also brought a scope home.

I made the mistake disconnecting the Hall sensors at the controller and trying to measure the Hall sensors by supplying 5V from a USB power supply and measuring the 3 signals (wrt to gnd) by spinning the wheel backwards (yes, it's geared after all). That is a mistake because after checking the specs for some Honeywell hall sensors I realized they probably have open collector outputs. This means you have to connect them to your controller to see the 5 V pulse. Internally (i think) the controller has a weak pull up to 5 V. The Hall sensors either ground the signal (when output transistor is on) or let it sit a 5V (transistor off). The signals then probably go into a comparator. Feel free to correct me.

So with better leads attached to the back of the controller connector, I see nice 5 V pulse trains on all 3 hall wires. I have two probes for the scope and can see them out of phase, each 50% duty cycle. The fact that they are open collector and that I am seeing the signals ALSO means they are indeed pulled to 5 V in the controller and there cannot be a problem with the wiring from the PCB to the connector I am probing. Right ?

Oh - my throttle seems to produce 0.2 --> 4.3 V as I press it, so thats seems ok. Plus a throttle issue doesn't really explain sputtering and start up problems.

So now I am stumped. What is next ? I am suspicious of the controller but i am not sure what to test next. Could it be a motor issue despite healthy looking Hall signal ? Anything else I should check on the motor side ?

Any suggestions are greatly appreciated. I need this thing for my daily commute.

Gertjan

 

[amberwolf]

I am convinced the connectors are good this time.

It could still be the wires themselves. I've seen wires broken inside the insulation jacket, and I've even seen wires spliced inside them from the factory, where they jsut twisted the ends together and let it feed thru the cable-making process. This is fine, except that if there was moisture or finger oils or whatever embedded in teh twisted connection, it'll corrode at least enough to make a poor connection, if not an outright open.

The other thing that can happen is a wire broken right at the back of teh contact, both with crimp and solder contacts. Unlikely that this would happen on your new connectors as well as the old ones, but it does happen.

Next guess: Hall sensors. Got out my multi-meter and probed the back of the connector of the controller cable i.e. as close to the controller as I can get. I am still not sure why, but initially I got a strange reading 5 V every where.

That usually means either :

--all the hall sensors are not working (no power would be the main one for all of them failing to work; it's not likely they'd all die at the same time electronically, unless severely overheated), and thus not grounding the signal lines when activated by magnets,

*or*

-- all the signal lines are broken (open) from the controller to the halls (unlikely)

*or*

--the ground is broken.

Internally (i think) the controller has a weak pull up to 5 V. The Hall sensors either ground the signal (when output transistor is on) or let it sit a 5V (transistor off).

Yes. You can use an external pullup resistor to V+ to the hall signal under test to get a valid result, without connecting to the controller. Most halls can take up to 20-25V for power, so you can use a 9v battery as a power source.

So with better leads attached to the back of the controller connector, I see nice 5 V pulse trains on all 3 hall wires. I have two probes for the scope and can see them out of phase, each 50% duty cycle. The fact that they are open collector and that I am seeing the signals ALSO means they are indeed pulled to 5 V in the controller and there cannot be a problem with the wiring from the PCB to the connector I am probing. Right ?

Yes.


So phase connections and wires are the next thing that could cause what you see. It's unlikley, but even a connection problem from the phase wires to the windings inside could cause it.

If you are using Anderson PP45s, then if you have thick enough phase wires, or insulation on them, that keeps them from floating inside the housings, they may have higher resistance than normal, causing heating and distortion of the housings that leads to further problems, eventually leading to connection failure.

It can also happen if the wires are tied down to the frame / etc close to the housings in such a way that the same thing happens, twisting the contacts or otherwise pressing them away or against the springs harder than the springs can push, so the springs cant' force alignment.

Bigger andersons, like the PP75s and SB50s and up, don't have this problem nearly as mcuh as the 45s, because of spring size vs strength ratio I guess.



There are a number of "bullet connectors" of different types. Some are pretty good, like the RC 5mm and larger types. Some are really crappy, like the automotive ones, and loosen over time or with plug/unplug cycles. Then the same resistance/heating/worsening problem happens as with the other (or any) connector.

 

[larsb]

I’ve had similar issues: one was eroded andersons, one was a phase power cable almost broken, one was a dying (but not dead) hall sensor, all of them causing intermittent connection or signal.

-check all cables and connectors
-borrow a controller from grin for 5 minutes to tick that cause off the list?

 

[dogman dan]

In your situation, a motor /controller/throttle tester can be a big time saver.


Unless your motor was smokin hot when the failure happened, or the next ride after a hot motor ride, your halls are hugely unlikely to fail. Sure, they can just die, but mostly people cook them.


So unless that is what you did, its definitely in your plugs and wiring. Doesn't sound like typical controller dying. But checking that is easy with one of those twenty buck testers.

 

[gertjan]

Amberwolf, thanks for your extensive reply. I am thinking however that seeing a clean pulse train on the hall probes is a definitive test that the entire signal chain into the pull up on the controller is fine. But I will take apart the phase signal cables today and perhaps just replace every bullet/anderson pp.

Larsb: the suggestion to borrow a controller from those nice people at Grin is a good one, except my controller has different connectors, types inherited from my first e-kit 10 years go (all since replaced but sticking with the connectors was the path of -no pun intended - least resistance). I am tempted just to buy one of their controllers - I am very curious if a sine wave controller is quieter on a geared motor - but I wish their phaserunner wasn't nearly 10x the cost of our competitors to the east (mind you, I have their satiator charger so I am getting there...).

Dan: I didn't know cheap testers existed. I see here: https://em3ev.com/shop/ebike-tester/ . Any testers anyone would recommend ? It would be well worth it.

One bit of info I omitted yesterday: I also put the scope on the phase cables (wrt to battery ground). I was expecting to see a nice +ve, then -ve, pulse train followed by a 'quiet' period. I was also expecting the pulse train to have a duty cycle that increased with throttle. What I see on the scope is hard to describe - I will try to get an image attached today - The signal looks very messy, has very long rising tails, there is a pulse train but the duty cycle doesn't vary. Not what I was expecting to see. I have been looking for images of real like phase signals to compare to but haven't found anything online.

Thanks again for the feedback.

G

 

[gertjan]

Well, I am both relieved and a little embarrassed. I replaced the (visually ok) phase wire bullet connectors near the controller on the phase signals... and the motor started back up. Since it wasn't starting, I presumed it had to be Hall sensor related. In retrospect, I guess you can get a similar symptom because of the high start up current. If I manually started the wheel, it presumably requires significantly less current and the crappy/corroded connectors were sufficient. Lessons learned: (1) it is often the connectors (2) it is almost certainly connectors. Sigh. In the meantime I have one of those little testers in the mail. An new toy never hurts.

For those curious to know what my phase signals look like on a scope, I have (hopefully) attached them below, at two different time scales.
Thanks again for the tips.

G