Revolt motor sensors - bad hall position and wiring

larsb

1 MW
Joined
Dec 10, 2014
Messages
2,820
Location
Gothenburg, Sweden
I have twitching and cutoffs on my revolt RV120 motor / kelly KLS7230 controller combo. When loading motor enough i get motor stalling, it cuts off a half second to come back with a jerk.

Measuring battery amps show stalling starts at 35A of loading. If i go easy on the throttle until higher revs then it's possible to get up to max throttle and get good acceleration but not from slow speeds. Even when running "well" the motor still screams quite loudly.

Max phase amps are now set at 20% out of 300A on the Kelly so current is at max 60A. This is only to make my bike driveable, if i increase current then the stalling is almost unavoidable.
BMS is 120A so no issue there and battery can easily deliver 300A

Hall signals does not look good even without stall. See below scope curves (Hall sensor signals are 0-12V, not 0-5V as i'd expected.)

In the first readout it seems that the red signal is inverted for a number of periods before coming back on track.
the offset between the two sensors line up so that they are 180 degrees apart most of the time which is not correct, how can this be?
Image10.png

In the second readout the red signal has symmetries with double peaks of "high" signal and a too long low in between
The blue signal has something of the opposite with double lows and a too long high.
View attachment 1
- There is no pull-up resistor or capacitor in the hall circuit. I looked at the datasheet for the SS411A halls which has a 10k resistor on the diagram for the simplest circuit:

- The hall wires are routed at 20-30mm distance from the phase wires outside of motor but inside they are on top of the wire/coils with no screening.

My guess is that the hall signals are affected by stray fields from the coils. Does someone have experience with these kinds of issues:

Are these signals possible to improve?
Is it possible to tell the root cause from these readouts?
 
If the missing pulses have a period ... that is they skip at the same spot every revolution it may be the magnet ring is out of round and the halls are not close enough to trigger at some point in the revolution. Shimming the hall slots to bring them closer to the magnets may help. Also one or more weak magnets may cause this.
 
Yes, make sure there isn't something wrong with the magnets or the position of the hall sensors. I've never seen hall signals look like that.

Adding a 10k-20k pull up resistor on each hall signal line won't hurt.

Is there a way to see the location of the hall sensors?
 
These halls are directed towards the motor magnets. Placement of halls is below the stator teeth, see pic. The 120degree positions are offset by some millimeters between them so there's definitely some error in the placement.
hall position.JPG
Having seen the inside of the motor i wonder if sensors can function without getting interference from the coils?
Wires to hall sensors are right on top of the coils and they're not even twisted or equal lengths, no shielding.

On my first RV120 motor the halls are fastened by jammed cigarette filters, on the second motor (measured one) the halls are fastened by jammed bamboo bbq spits.

However professional these fixation methods might seem I suspect they have something to do with the bad hall performance since the halls might vibrate and/or come loose :roll:
I did something like 10 measurements and they all look different. Therefore i assume that the errror is caused by interference but it's only a guess since i am looking at my first scope readout.
I'd like to hear what someone who has some experience thinks about it..
 
The Hall looks like it is rather far from the magnet. All my motors have a grove\slot cut in the lamination for the Halls to sit in that insures they are close to the magnets. Your problem is more likely weak magnets or Halls to far from the magnet. I added small capacitors across the Hall inputs inside the controller once and it cleaned up all the fast spikes on the oscilloscope trace but did not make any difference to the motor performance at all.
 
The switching halls I've worked with are current sinks, so without a pullup resistor, they may not have a consistent signal output, as the signal line from there to your controller may have enough induced noise on it to give false signals. Doesnt' really explain the wierd signal you get on your scope, though.

But the sensors being so far from the magnets, they might be being triggered by teh flux in the stator teeth instead, or in addition to the magnets.

That might explain the 180 out of phase, if the signals you're looking at are showing the current flow for different phases?
 
I hooked up my motor to an RLS encoder and all the issues with bad stuttering, noise, vibrations, heating etc are gone.

Safe conclusion from this is that poor sensor function is root cause for my motor issues. Probably a combination of the hall placement deep in the slots with interference from coil fields and/or the hall wiring getting interference from coils.

As several users have reported similar issues with the RV120´s i think that the sensor design is flawed to the point of killing motor performance. I would advice everyone with a poorly performing Revolt motor to do a scope readout, if readouts are like mine then ask for a warranty replacement by Revolt. Don't do like me and fix their design!
 
larsb said:
I hooked up my motor to an RLS encoder and all the issues with bad stuttering, noise, vibrations, heating etc are gone.

Which encoder did you use?

Seems like it would be possible to make the hall sensors work but would take some grinding, gluing, etc. If the face of the hall sensor was even with the stator tooth, it should work fine.
 
This is the encoder:
https://www.rls.si/rmc22-commutation-and-incremental-encoder

I think i'll write a guide for installation and usage of this to give something back to the ES users. It was simple enough to install it, time will tell how good it is.
I agree that hall sensors are good enough when they work, and are a simpler/cheaper choice when done well.
 
i am having a similar problem with my Cyclone 3000W motor. only the hall is triggering three times for one commutation. something like this:
Code:
      _    ________    _
_____| |_|          |_| |_______

i think it has something to do with the rotor, and how it's build. will post a picture later.
the encoder you used could be a good alternative for me i think. how do you like it?
 
Good to know, Thanks. do you know if the I2C pins (for reprogramming) are accessible on the RCM22?

i am not sure what i will use. the RCM22 (RLS) or the AMT312S (CUI)

the RCM has more options, but the AMT is easy reprogrammable by just unplugging the encoder and connecting it to a UART terminal.
 
Back
Top