rider119 said:
From a mechanical standpoint, how does a sin/cos encoder function? Is it similar to a hall?
Depends on the specific design, but basically the UVW uses the actual (digital) motor hall sensors to sense the actual motor magnets passing them, and these switch on or off in a six-step pattern based on where the magnets are in relation to each sensor. This gives a fairly granular position reading whose resolution is determined by the number of magnets vs circumference of rotor vs number of halls / steps. It's "good enough" to get the position for smooth startups on good controllers (crappy controllers may grind at full throttle from a stop in some rotor positions...but typically that's only the generic ebike type controllers, not anything with FOC or that could be programmed with any seriousness).
The SIN/COS has it's own ring magnet that is magnetized in a smoothly-transitioning N/S manner continously around it's cirumference, and has two analog ("linear") hall sensors that continuously change output voltage based on that magnetization as it passes them to give a pretty exact position reading of the rotor. As long as the controller can correctly and fully utilize this information, it can tell pretty exactly where the rotor is (vs the granular position data from the UVW system).
But the UVW system actually tells the controller (in theory, anyway) exactly where a specific rotor magnet set is in relation to the phase coils / stator teeth...if the controller is well-designed, it should be just as good (better, without calibration) as a SIN/COS encoder.
In reality...it depends on the motor design, sensor placement, the sensors themselves / the encoder itself, the wiring, the controller design in hardware and software, any settings that exist, etc.
If interested, somewhere here
https://endless-sphere.com/forums/viewtopic.php?f=30&t=32838
are pics of the SIN/COS encoder in a powerchair BLDC motor, along with my method of installing UVW hall sensors so I could use it with a regular ebike controller.
I do have an IXXAT cable and DVT software and have had it up and running. I picked up a Gen4 Size2 recently for cheap that I'm hoping to learn on before investing in a large Size4/6. I've read many a threads about the horror stories and I may end up adding to it, but time will tell!
As long as you know what you're getting into.
I have an old Sevcon from a Zero, but I don't have the motor it went with, and have no way to change it's settings to work with any of the motors I do have (even if I had any idea what I was doing with it :lol: ). I don't remember what model it is, but I can check, if it's something you might be interested in.
FWIW, a size2 is nearly 300A, IIRC. What kind of stuff are you trying to do that this isn't enough for? :lol:
And yes, I understand that the controller is the driving factor in how the power is delivered, but as you stated, the UVW gives off a square waveform vs the parabolic waveform of a sin/cos wave.
Just remember...the *position sensor* waveform is not the *motor power* waveform. Theoretically the SC type is finer resolution for more accurate position data than the UVW...but that depends on how the controller actually works.
What you should do is find a controller and motor system that is designed to do the specific thing you want it to do, under the conditions you want it to work that way, and get those.
As long as what you are using is a true FOC controller, with torque (current) throttle control, and at least some tuneability in it's settings for phase current vs throttle input, etc., it can very likely be setup to do whatever you want. A good FOC controller will be able to read feedback from the phase signals via it's current sensors as well as position data from the position sensors to probably get about as fine a control over everything regardless of position sensor type.
How true it is with each specific controller out there, I have no idea. :/ ATM I only have direct experience with one FOC controller on my trike, and that was Incememed's SFOC5, which is still in beta testing stages, and as such still left a fair bit to be desired at the time I had to stop testing it well over a year ago (I hear it's better now...but no experience). I've got a pair of Lebowski-brained Honda IMA controllers in build progress, but as yet not finished to test on the trike--I expect they'll do just fine based on others' results with them, though...and they're a damn sight easier to setup than a Sevcon even if I have to build them myself. :lol:
(and the source code is now publicly available so if I had any clue how to code I could modify that to do anything I wanted that they don't yet do).
