End stop switch inconsistent: a fix

[dtgriscom]

I've discovered a problem with the end stops on my Rostock MAX v2. On my Z tower, when the end stop screw contacts the end stop switch's lever, they touch at an angled part of the lever:

Factory end stop switch lever, installed, with end stop screw precariously on curve of lever

Any wobbling of the screw from side to side significantly changes the height where the switch actuates. This means that turning the screw a given amount will change the end stop position in unpredictable ways. Sometimes it will even move in the opposite direction you expect.

This probably happens on this tower because of the exact geometry between the switch and the Cheapskate; it could have happened to any (or all) of my towers. To check if this is happening to you, fire up Repetier-Host, connect to your printer, and switch to the Print Panel. Home All, and then adjust the Z axis up slowly until the switches activate and motion stops. Then look at how the screw contacts the switch lever. If it's contacting on that initial curve on the end of the lever, then wobbling of your end stop screw can cause trouble.

To fix this, I decided to flatten the ends of the switch levers. The switch levers are relatively easy to remove; they have a forked end with points that fit into divots in the body of the switch. Just gently squeeze the fork with needle nose pliers and they'll rotate out:

Factory end stop switch lever, top view, with attachment fork on left

A side view shows the curve at the end which is the problem:

Factory end stop switch lever, side view, with contact curve on right

Use pliers to flatten this end, being careful not to distort the rest of the lever:

Fixed end stop switch lever, side view

When you reinstall the lever, you'll find that you have a much broader contact area between the screw and the lever, so sideways wobbling of the screw won't matter any more:

Fixed end stop switch lever, installed

FixedLeverInstalled.jpg (31.14 KiB) Viewed 5314 times

Note that it's possible (although I think unlikely) you'll break one of your end switches this way, so don't do the surgery unless you're OK with having a few days off of printing until a new one arrives.

Hope this helps someone,
Dan

[Eaglezsoar]

Good idea! The other thing you can do is epoxy a cap to the top of the screw which makes the contact area larger and if it has some ridges on the side
you could use it to turn the screw.

[Tonkabot]

I Noticed that my Max V2 endstop switches are already the flat actuator version.

[brent]

Good idea! The other thing you can do is epoxy a cap to the top of the screw which makes the contact area larger and if it has some ridges on the side
you could use it to turn the screw.

I like this idea Eagle.

[Polygonhell]

FWIW my fix for the same issue on the original Max was to turn the switch around, so the screw hits the minimum amount of lever.

[lordbinky]

FWIW my fix for the same issue on the original Max was to turn the switch around, so the screw hits the minimum amount of lever.

I was going to say the same thing

Now I just adjust the endstops in software, since I swapped out the end stop switches for 3144 hall effect sensors. No board or anything, just the ~$1 sensors. Had to run that third wire for the endstop then hot glued & zip tied it to the melamine, and glued magnets to the cheapskate in line with the sensor. I like that the cheapskate doesn't make physical contact and even have the possibility of banging stuff around and the repeatability of the setup makes me confident any issues I have aren't endstop related.

[dtgriscom]

FWIW my fix for the same issue on the original Max was to turn the switch around, so the screw hits the minimum amount of lever.

At least for me, this would put the end stop screw hitting right on the hinge of the switch, which doesn't seem as sensitive to position.

Glad to hear SMC is going to straight lever actuators.

[Polygonhell]

You want as little lever as possible for the most repeatable result, but it does need to beat least over the switch itself.