On 10/26/15 10:00 AM, John Kasunich wrote:
> On Mon, Oct 26, 2015, at 11:43 AM, Sebastian Kuzminsky wrote:
>> On 10/26/15 8:50 AM, andy pugh wrote:
>>> Some systems have many indexes and a way to distinguish between them.
>>> Sometimes the indices are different distances apart (some linear scales) or
>>> there may be a secodary feedback system (robots with encoder + index and a
>>> potentiometer).
>>>
>>> I don't think that LinuxCNC can work with these systems at the moment as
>>> the home_offset is a static value read from the INI.
...
>>
>> I don't understand what problem you're solving, or what your solution
>> entails.  Can you give more details?
>
> Imagine a lathe Z axis.  The leadscrew has an encoder with an index pulse.
>
> It can be inconvenient to home to a fixed location.  If you put the home
> switch at the headstock end you might run into the chuck if a larger-than-
> normal chuck is mounted.  (Or the workpiece, if one is loaded).  If you put
> the switch at the other end you might run into the tailstock depending on
> where it is located.
>
> Now imagine that you have some kind of coarse absolute position
> measurement.  Perhaps linear potentiometer connected to the carriage.
> It can't measure position to 0.001", but it can do +/- 0.050".  The screw
> pitch is 0.200".  So the pot can tell you where you are to within one turn
> of the screw.
>
> In theory, it should be possible to home by moving the carriage approximately
> one turn of the screw in either direction, to find an index pulse.  The pot 
> tells
> you what turn you are on, the index tells you exactly where you are in that 
> turn.
>
> But there needs to be a mechanism for LinuxCNC to use the info from the pot.

Thanks John, that example makes it clear.

I now see the utility of having HOME_OFFSET settable from HAL, and I 
support Andy's suggestion of making it into a pin, to be driven by the 
pot or some other suitable sensor.


-- 
Sebastian Kuzminsky

------------------------------------------------------------------------------
_______________________________________________
Emc-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-developers

Reply via email to