In this case everything is referenced to B-. Not true 3 phase.
EDIT:
EV Inverter Output Reference to Battery Negative
In an EV traction inverter, the output reference point for each motor phase is the battery’s negative terminal (common ground), and the inverter’s switching network connects each phase to either the positive or negative rail of the battery DC link.
How the reference works
Most EV traction inverters use a three‑phase voltage source inverter (VSI) with three half‑bridges. Each half‑bridge has an upper and lower semiconductor switch. The lower switch of each half‑bridge is connected to the battery negative terminal (often called the “common” or “ground” rail) evkx.net+1. This means that the inverter’s output for each phase is referenced to battery negative.
I wonder if it isn’t “Y” connected with center ground. Only way I can think of it.
If it gets to be important, I can connect a motor on the bench to measure some things. It all takes so much time and I’m hoping for a quick fix.
Also, listen to the sound. I just realized that’s not the wheels hitting brake parts. That’s the motor… what could be making that sound? It does not make that sound when I move the wheels without power…
Did I have you test A + B from controller with encoder unplugged? I haven’t been up to connecting a controller here. Most Sevcon inputs are about 1/2 B+ voltage.
As phase reversal had no change, it is almost certainly an encoder issue. I don’t understand it not causing a fault. Have you checked for blinking led before turning off key, which clears faults.
How about faults showing wrench on dash?
Looking like controller problem as you suspected earlier. Not showing any faults has me confused.
I believe that any shorted or open fets in output bridge would certainly throw a fault code. Phase wires disconnected, I expect all outputs to be about 1/2 B+ referenced B-. And open circuit between them using low voltage meter.
I noticed. Motor fighting itself trying to run without proper encoder signal. IMO I wish I had a car to disconnect encoder for comparison.
I’m not up to connecting test motor today. I have some health issues going on.
FFS!!! Modified part… made it fit. Scared the crap out of me… @Inwo@AssyRequired@JarJarJava . Thank you all for the assistance on this… Might want a faster controller at some point tho…
Faster controller, try a SpeedBBX … once you’ve had one, you’ll never go back.
And maybe a couple tubes of JB Weld.
Truth be told, an unlocked controller is the way to go IMO. You can get everything adaped for your needs.Torque, top speed, acceleration, regen braking level, and so on.
This may be a long read, but it will be here to supplement my memory tomorrow!
I was going to make some tests but my bench battery was dead. I would like to get data to Byron for next time. That is, next time if you are running. I have my doubts that this is a real fix.
Tbh, I don’t understand how the “B” output is generated 90 degree from “A”.
There is only one cog wheel. It can trigger on leading edge or falling edge. 180 deg difference. It can’t use time, as time varies with speed. A second hall pickup can still only read one edge or the other. I’m sure that I’m missing something simple.
In any case, after a battery charge, I can see if AB inputs are at 1/2 B+, like the other inputs.
Then I can do your voltage tests of AB with motor turning. Expected result is 1/2 of the 1/2 B+. Inputs being pulled low 50% duty cycle. I can also see if I can spin motor slow enough to “see” each cog. That will be a go-no go test of encoder. In my best guess.
What twisting the encoder does, I have no earthly idea.
The only way to change timing is rotation. I’m quite sure of this.
Twisting can align the hall sensors to the cog wheel. But what can cause this to be off? I’ve used the Amazon golf cart sensors myself. Possibly some cart motors have a different diameter wheel.
I’ll add more if I get my bench motor running.
Cart now works in both directions without any issues. No faults on dash or controller. I cut the new speed sensor to be able to spin. The bolt hardware got in the way of the motor metal, so it was not able to spin. I tried it on the new sensor, but I have a hunch it would also work on the old sensor as well (if I cut it to size).
This is how it goes for me sometimes. Things just don’t make sense, but we make it happen! I’m open to do tests for you - for research purposes.
Rather than cutting off the tab, an alt solution might be to swap pins 2 and 3 in the plug. inspect the plug on the encoder and see if the pins will pull out easy enough. (special tool may be needed)
The next puzzle - I have no idea how it got this way. Perhaps the controller was swapped by the PO and the programming was wrong? (guessing there may be an encoder direction setting in there somewhere).
I’m stumped as well.
It’s not a phasing problem, as twisting encoder won’t change timing.
Maybe I linked to a non-compatible replacement. A search suggests that some golfcarts use a 5v encoder, and possbly a different diameter wheel. Low confidence in that, but it does seem that some of the chinese clones are wonky. Is it exactly the same length?
Sevcon is very particular about the waveform from encoder. I needed to research on how the 90° signal is produced.
The encoder has two hall devices spaced exacly 1/4 tooth apart. The oem Gem encoder is a honeywell part. I used to buy them from Digikey until I found the cheap Amazon ones for golfcarts. Even though they worked for me, may not be an exact part.
I would like to see an oem encoder tried on this motor to put a fork in it.
Or I can test the oem one to see if it failed.
This is how it goes for me sometimes. Things just don’t make sense, but we make it happen! I’m open to do tests for you - for research purposes.