A Cunning Plan For Accurate Homing With Mach 3?

[magicniner]

If we assume a machine with all limit switches on one input and a separate input for each home switch and wire it up with any old decent Normally Closed micro-switch at end of travel we get home positions but with accuracy and repeatability which is often dismissed by owners of pro systems and rarely proves good enough to pick up halfway through a finishing operation.

But what if we add a cam somewhere on each stepper shaft or ball-screw which operates a second switch wired in parallel to the end of travel switch for that axis?
The cam switch will operate once per turn but the home circuit will only open with both switches open, tune the cam switch to open/close a fraction of a turn after the linearly actuated switch is open and the accuracy and repeatability should be significantly better than is achievable with just linear switching.

Sound feasible?

- Nick

[dodgygeeza]

you mean like using the index pulse on an encoder that is fitted to the motor, and making the homing more accurate? this is a standard feature, but will need an encoder.

[magicniner]

you mean like using the index pulse on an encoder that is fitted to the motor, and making the homing more accurate? this is a standard feature, but will need an encoder.

Yes, well spotted, can you spot that what's described above specifically doesn't require an encoder and is within reach of everyone, without adding an encoder?
So I will not need an encoder :D

[routercnc]

Should work. Would increase the homing accuracy by the ratio of the perimeter of the cam 'sweep' to the ballscrew pitch length. But wouldn't like to have the microswitches continually opening and closing during normal jogging and machining. Swap to contact-less prox sensors (still cheap) reading off the 'cam' instead.

Could put it on the back of the stepper?

[magicniner]

But wouldn't like to have the microswitches continually opening and closing during normal jogging and machining. Swap to contact-less prox sensors (still cheap) reading off the 'cam' instead.

Could put it on the back of the stepper?

Good ideas that anyone implementing this could make use of, the main thing is that it's an easy add-on for existing systems which can be implemented at little or no cost and with easily understood and accessible technology.

[m_c]

It's been done.

Have a read of this thread - http://www.machsupport.com/forum/index.php/topic,18323.0.html

Just had a quick scan, and look at the exchanges between me and Dickeybird starting around page 3. He implemented it on his lathe.

[magicniner]

It's been done.
.

I hadn't seen it used before but given that I based this on the fact that closed loop servo systems do a similar thing it's not exactly a surprise.
What is a surprise is that it's not in a sticky on most home CNC sites and that there are so many homing switch queries where it doesn't get a mention.
But hey, the next time I have a thought that might help instead of posting it with a clear title I'll keep it to myself ;-)

[m_c]

The thing is, it's not really needed provided you pick good quality homing switches, and mount and actuate them sensibly.

If you have a read of the thread, those of us who fitted good quality optos, had excellent repeatability. Certainly in the couple years I had that lathe, homing accuracy was never a problem. I could re-home in the middle of a job, and accuracy was more than good enough, but it do show up how much flex there was. Even on my bigger lathe that uses the servo index pulse for homing, recutting something after homing will still remove more material, due to flex, not inaccurate homing.

[Boyan Silyavski]

I have done some experiments on homing and best real life repeatability was achieved when the switches were properly mounted so that they were perpendicular to metal that activates them.

From reading in forums it seems expensive vs cheap switch is not really important when properly done. Surprisingly cheap mechanical micro switches seem to give best results.:stupid:
At least initially.

I don't see something better than an absolute encoder though, speaking for a diy machine. Or a hard stop :joker: