hub motor hall sensor woes - need help

I have a NOS Crystalyte 5303 front hub motor. I thought it would be interesting connect it to a cheap 72V 1500W Sunwin controller. Bench test was smooth. On the street it was more than interesting; actually a little scary but the bike died climbing the first hill. Turns out one of the motor hall sensor died. And...the controller also quit! I chalked it up to a bad combination. I replace the bad hall sensor and bench tested the motor with a "Brushless Tester" and then with a Crystalyte 48v40a controller. All good, so I stored the fixed 5303.

Now I want to use the 5303 on a tandem. I get it out and test it with the "Tester". Phase wires, fine. All 3 hall senors, bad. When all 3 Halls don't work it's usually a broken power (red or black) wire. But no....4.85V measured at all 3 hall sensor's pin 1&2. A magnet near each hall doesn't trigger the tester's LED. A new hall connected to the tester triggers when a magnet is close, so the tester is good.

What fried all 3 hall sensors? I'll replace them, but how do I prevent a repeat.

I think the Crystalyte 48v40a controller works on motors with or without hall sensors. Could it have fried all 3 hall sensors and then continued to work in the sensorless mode? I'm lost.

Firedog said:

But no....4.85V measured at all 3 hall sensor's pin 1&2.

Click to expand...

AFAICR, all the hall sensors I have used / tested so far, for motor halls, have been the following wireing, assuming you have the bevelled-edge face up, and the wires pointing "down".

The left wire is power in, the middle is ground, and hte right wire is signal out.
(like the images in this search
https://www.google.com/search?newwindow=1&q=hall+sensor+pinout&tbm=isch&source=univ&client=firefox-b-1&sa=X&ved=0ahUKEwik_7-xmobiAhVNCTQIHZ0OA8sQsAQIFA )

So if yours are like this, then anything other than 0v on the middle wire means a broken ground (either inside or outside the sensor itself).

Also, usually you get a slightly lower voltage on the power input pin, than you do on the signal out. Getting identical voltages on both is unusual, even if those are the right pins to see it on.

What fried all 3 hall sensors? I'll replace them, but how do I prevent a repeat.

Click to expand...

Unless it got REALLY hot (probably so hot you'd smell it, and too hot to touch the motor casing), then the only other thing that could toast all of them at once is a voltage on the power input that exceeds their ratings. A lot of these can handle up to 24v, so not much besides a phase-wire short to the hall power wire could do that--and that would also damage the controller's 5v supply, probably frying the MCU too (and probably the throttle, as it's run off the same 5v supply). Since those still work, then this is an unlikely failure mode.

But... if the first failure you had was caused by such a short, then that could've killed that controller, and acutally killed all the sensors (or damaged them so they failed shortly after).

If so, then you would need to examine your entire phase / hall cable for breaks in the insulation on each wire. If your 5303 is like my 5304, it uses teflon insulation, which when aged, especially when heated, cracks easily, and pressure at bends can force the cracks open and wires stick out and touch, etc.

Thanks for responding amberwolf. I think you got me on to the right path.

I have the wires connected correctly and all beveled edges pointing up. All 5 wires are continuous from the connector to the individual sensor's leg. None of the wires is shorted to any other. None of the wires is shorted to the axle, except the red!!!! With the tester on, the whole hub and axle is 4.86v compared to the black wire.

When I replace that hall sensor months ago, I must have it's + leg not insulated from the armature.

I will be a day or so before I can trace it down, but do my symptoms square with this discovery? I'm not sure

I connected a good hall sensor to the tester. If no magnet is near, the tester's LED is off and the voltage on the output (#3) is very near the voltage of Vcc (4.86v). If a magnet is near, #3 drops to 0 volts and the tester LED goes on. I shorted the red (#1) to the wheel expecting the sensor to stop working, but it doesn't.

I'll report the fix if I find it. Thanks again.

I fixed the one red wire that was shorted to the armature. I re-tested the continuity of all 5 wires from the connector to the sensor; checked that no wire was shorted to any other and that none were still shorted to the armature (ground). All good. Unfortunately, none of the halls will change state when near a magnet. I guess the grounding of their 5 volt supply (pin 1) damaged all 3.

It will be a week or more to get and install new halls.

Unless you have the battery voltage negative wire grounded to the bike frame, there should be no ground connection to the axle or stator to short that red wire to anything else.

However, if one of the motor windings, or phase wires (or battery positive) shorted to the stator, axle, or bike frame, even momentarily, at the same time as the 5v wire, it could definitely fry the halls *and* the controller proviidng the 5v, and anything else connected to the 5v (like a throttle).

So all three Hall sensors were fried. I replaced them being careful no lead was grounded or shorted to another. The Crystalyte 5303 motor now works perfectly and wow!. I mounted it to a tandem with a 35A max controller using using 54V it has more power than should should be legal. (It isn't legal in most places) I mounted much lighter Crystalyte 408 with the same controller and even that set up is has more power and top speed than necessary. We RV with the ebikes instead of towing a car and weight is important. Not sure I'll find a good use for the 5303.

Still trying to figure out what killed the 3 halls sensors and the controller simultaneously. The red line (5v) short to the hub caused no problem on the test bench, because the controller was electrically isolated from the motor. On the real bike, the controller may have be grounded to the frame. I suspect the controller's 5v power is blown, although I didn't see visual damage. I haven't checked with a voltmeter yet. If that's it, any suggestions to fix it? Still trying to learn.

Amberwolf, thanks again for you help.

If it's only the 5v, and the 12v is still ok, then there's often a "78L05" three pin device that converts the 12v to 5v. It's a TO92 case, if you google that you'll see images to show you what it looks like. It may have a different part number on it, but it'll probably be some variation of that 78x05.

The 12v is provided from the "keyswitch / ignition / doorlock" wire that puts battery votlage to some big resistors, which connect to the input of either a 7812 or an LM317 in a TO220 package (also three pin, looks like the FETs do, but is not near the FETs).

There's a number of threads that show the "LVPS" schematic of generic controllers; most of them are similar enough to use for testing purposes.