Accurate low cost method of home/zero Axis setting

[picclock]

Hi
This is idea I read about some time ago and am now implementing on a mill. The limit microswitches are reliable but not so good accuracy wise. This method works by latching the microswitch high when it is opened, but pulling the output low by means of an optosensed disc attached to the ballscrew. So the sequence of home events is the ballscrew rotates until the home microswitch opens. The controller then reverses the screw to sense and set the home position when it closes. However a small opto circuit makes the closing event only occur when a slotted disc aligns with the opto interrupter. Using a 36mm disc  and a 5mm pitch ballscrew I have calculated that the accuracy to be .0025mm or a tenth of a thou. The slot in the disc has to be radial but can be of a reasonable size, say 10 degrees of arc as the position is sensed by the edge.

I have designed a circuit to do this and am currently waiting for parts from overseas to build and test. The pcb is made to work with my DDCSV.

Best Regards

picclock

[awemawson]

Heidenhain encoders do a similar thing as implemented on Bridgeport Interacts and indeed also on my Beaver Partsmaster.

They drive to the limit switch then look for a 'once per rev' index on the encoder.

[mc]

What Picclock describes is essentially a poor mans index homing system.

I considered implementing it on my original CNC lathe for the X axis, but the opto slot sensor I used never showed any noticeable variance on a 0.01mm division DTI, so I never bothered, and I never had any tolerance issues.

Does it really make that much difference that you can home to a tenth accuracy, when your stepper is only likely to be accurate to within a half step depending on load/stiction?

[DICKEYBIRD]

...by means of an optosensed disc attached to the ballscrew.

Yay, somebody else "gets it!"   I've been using a similar scheme on my retro'd ORAC for years and it works very well.  The only difference is that I fitted double ended stepper motors & mounted the disc/opto sensor on 'tother end. 

[RotarySMP]

Have you also mapped the ball screw lead error? Having a home position repeatable to a tenth will not add much to the complete system accuracy if the home position is a foot or more from the commonly used area of the table.

If you start needing that sort micron level accuracy, you would be better off closing the position loop with a glass scale encoder. The Heidenhain LS-403's on my MAHO were delivered with test reports for each scale, mapping the error every 10mm (it never exceeds a µm). They also have home indices at micron accurracy.  You could program this error compensation table into LinuxCNC.

On the Maho, they even went as far as an Invar linkage to ensure the Y axis scale is referenced from the spindle end, and heat from the gearbox is compensated for.

https://forum.linuxcnc.org/12-milling/33035-retrofitting-a-1986-maho-mh400e?start=320#106096
I am not at a skill level to need such accuracy, but it is interesting so see the length a machine builder went to.
Mark

[DICKEYBIRD]

I haven't mapped my screws yet although it looks pretty easy with the Acorn & Centroid software.  The only reason I put the extra effort into an accurate homing system is to ease the tool offsets process.  If I set up the 26 tools referencing them from an accurate & repeatable home position, life is much easier.  The total offset from the home position to each tool's X & Y location being (almost) exactly the same distance each time means more to me than minor variations along the way.  After all, most tools are pretty similar and will be measured at approx. the same spot on the screw.

[picclock]

The absolute accuracy is not the reason here, its the repeatability of getting back to the exact same position  - within a tenth or so :-).   
I am going to fit the disc on the end of the 'plum' type coupler and the pcb will go underneath.

Best regards
picclock

[hanermo]

Repeatability is extremely desirable.
Cheap optos deliver about 2 microns in repeatability and accuracy.

[picclock]

@hanermo
2 microns - do you do realise that that is 0.08 thousands of an inch? Where did you get that information ? Which device are you referring to ?
spec sheets on most devices indicate transition at around 0.1mm, which is still quite good.

Best regards

picclock

[DICKEYBIRD]

Wow, I'll have a bagful of those cheap optos please!

[mc]

As much as I like to disagree with Hanermo, his figures are probably not far off, given my experience of slot sensors.

Looking at the datasheet for the current version of the slot sensor I used (Panasonic PM-45 series), it lists hysteresis as 0.05mm or better, and repeatability as 0.01mm or better.
Given I never noticed any variance on a DTI gauge with 0.01mm intervals, I'd put repeatability at better than 0.005mm. However due to the stepping nature of stepper motors, the stepping probably contributes to the repeatability.

[awemawson]

Would those 2 microns possibly be the travel of the ball screw when located by the sensor. So linear travel of screw related to angular travel of opto sensor 

[cnc-it]

Heidenhain encoders do a similar thing as implemented on Bridgeport Interacts and indeed also on my Beaver Partsmaster.

They drive to the limit switch then look for a 'once per rev' index on the encoder.

   

Yes I found this out the hard way on my Partsmaster. After removing the Z axis encoder for repair I pranged the tool change carousel as the Z axis tool change position had changed by around 3mm! Luckily I got away without any damage     

[BillTodd]

Typically, the linear sensor acts as a trigger for locating a rotation sensor  , either a motor encoder pulse or a separate lead screw sensor.