Can mechanical homing switches make the grade?

Recent discussions of homing switches and various modifications in the homing switch technology used made me wonder just how accurate the lowly mechanical homing switch is. So I ran a test to see if I needed to consider an alternate technology.

Spoiler alert: I don’t see any reason to change.

My machine:
X-Carve 500mm x 500mm stock Inventables kit
Standard electronics (Arduino/gShield mounted on 24 volt power supply)
Mechanical homing switches (shielded wire modification)

My assumptions:

The Y axis is the least accurate of the three axes.

With the homing switch on the left side of the gantry, the right side of the gantry is less accurate than the left side as regards to homing. Therefore by placing the dial indicator toward the right side of the gantry I measure a worst case condition.

That your machine is setup and tuned as well as mine.

The test.

Part A - home the machine (take a baseline reading on the dial indicator)

Without moving the X axis or the Z axis move the Y axis away from home by (25,50,75,100,125,150,175,200,225,250 mm) and then home the machine again. Take a reading on the dial indicator each cycle.

Subtract the baseline reading from the dial indicator from each measurement, add the those results together, divide by ten which gives the average error term.

Part B - repeat the test above, but this time for each measurement move the X axis 100mm off the homing switch. Calculations the same as above.

After running this test I came up with a deviation of 0 +/- 0.00001 for part A and 0 +/- 0.00003 for part B.

Since the fourth decimal place was an estimation of the needle position between tick marks on the dial indicator, I consider that decimal place to contain errors.

Using this test I would be comfortable saying that, with mechanical switches, the X-Carve will home to within 0.001 inches of the same position each time you home your machine.

Your mileage may vary.

Here’s a picture of the test setup:


I got pretty much the same results.

1 Like

Yep, this is pretty close to what I got on my 1000x1000 machine.

Personally, given that I’m usually working in wood, and my acceptance that my machine is not a rock solid $200k machine (I wish) and considering my often lazy "use a random piece of paper’ and “line it up by eye” methods of setting zero; any errors introduced by the limit switches are not something I’m concerned about.

Agree with this. It would be better to put any extra effort into making a probe to help set start position rather than trying to get the home switches micron accurate.