A3144 hall endstops

I'm going to do some little PCBs to hold these guys in the T slots for my Ordbot Hadron, but I'm not exactly sure which orientation is best for accurate, repeatable, end stopping.

Datasheet here:

[pdf1.alldatasheet.com]



Should the magnet face approach the face of the hall sensor dead on? Or is it better to slide the magnet face over the face of the hall sensor?

I notice more hysteresis when going face to face rather than sliding over the face from the side. I would have just gone face to face, but in a video I noticed someone setting them up to slide over the face, and thought maybe there was a good reason for that. It would seem to me it would be more tempermental dealing with another range of motion, being the distance between the sensor and magnet when slid together and also the sliding distance to the magnet, but I thought I'd see what the consensus was before I etch some boards.

Why not use a photo sensor? I have been using them for xyz axis and are simple to mount with a flag. The hall sensor seem more finicky to get setup correctly.

@PrintThings:
Can you post the video that you are referring too.

This is an example of a linear Hall effect sensor:

Hall effect sensor tutorial - 3D printer end stops
[www.youtube.com]

I'm not interested in opto end stops because they seem to be the most complex and finicky. Using an A3144 is dead simple and they cost $0.20 each.

The videos I was refering to are here, it appears that both the X and Y endstops are mounted so the magnet will slide over the face of the hall sensor.

[www.youtube.com]

I'm just going to go face to face though. I made a quick little setup on my lathe with some clamps and a dial indicator to hold the sensor in place and measure as I progressed the tailstock chuck that has a metal rod with a magnet on the end of it. It's within 0.001" each time, the 0.0005 variation or so also only seems to happen from the different speed at which the magnet approaches. I noticed as I was progressing the magnet very slowely, I could stop it right where it was about to engage and then it would go about 5 seconds later without moving it any more. I think that with a homing command, having the velocity constant for the homing, it will probably stop in the same spot even more accurately, so I'm happy with that.

Between the previous paragraph and this one I decided to test my theory. I set up my gropro camera on 240 frames per second and fed the magnet in a constant rate about 20 times, then watched it frame by frame. As far as I can tell the trigger point is exactly the same every time.

Such a A3144 sensor is not afraid of optical interference and is easier to install