Isn't it Cuuuuute..?....
Several updates: the "Little Cube" is coming along nicely, but as I construct it, several questions pose themselves:
Given the crude construction techniques, how can I insure that the machine will have adequate precision to make parts as good as those used to make the first one i.e accuracy won't decline with each copy...
Given the crude construction techniques, how can I insure that the machine will have adequate precision to make parts as good as those used to make the first one i.e accuracy won't decline with each copy...
The most likely problem is parallelograming the frame during assembly - out of parallel linear bearings are easier to spot - and a significant error will cause binding in the travels.
This can be addressed during assembly with some measurement and a proper bolt tightening sequence, but a subtle parallelogram will be harder to detect and correct.
Applying this principle, can I make the machine MORE accurate at each iteration? If so, to what limit?
I think I have an answer to this as well:
Yes, using a technique from land surveying - if a test program is built with measurement points in a rectangular grid and the error for each point is logged, then a compensation table can be emplaced to counter irregularities in positioning accuracy due to out-of-square or non-parallel problems.
Two approaches then present themselves: do all compensation with the control software , or iteratively suggest adjustments - then retest until certain tolerances are met and compensate between those tolerances and the practical limit of the positioning system - this should yield the best accuracy a given system is capable of...
The main caveat to this is such a system would compensate for static errors, I can think of no practical way to compensate for dynamic error due to deflection, sag, or harmonics although proper telemetry would likely allow for all that - this may seem to be overkill - but if this form factor of machine is to make its own parts some of them may call for very precise positioning (I'm thinking SMT pick and place f'rinstace... ;-) ) so I believe it is a problem meriting careful consideration
And where all these maunderings lead me is:
What positioning system? This first cube will be a ball screw/stepper positioning system, but I think machine 2 will be a Linear stepper solution.
For the attractive merits of greater mechanical simplicity, much more precise positioning, and potentially closed loop operation inherent in a linear stepper system.
the questions posed above are satisfied by this solution and the upper threshold for accuracy becomes potentially very high if a mechanism to loop the accuracy upwards iteratively is implemented.
I think I have an answer to this as well:
Yes, using a technique from land surveying - if a test program is built with measurement points in a rectangular grid and the error for each point is logged, then a compensation table can be emplaced to counter irregularities in positioning accuracy due to out-of-square or non-parallel problems.
Two approaches then present themselves: do all compensation with the control software , or iteratively suggest adjustments - then retest until certain tolerances are met and compensate between those tolerances and the practical limit of the positioning system - this should yield the best accuracy a given system is capable of...
The main caveat to this is such a system would compensate for static errors, I can think of no practical way to compensate for dynamic error due to deflection, sag, or harmonics although proper telemetry would likely allow for all that - this may seem to be overkill - but if this form factor of machine is to make its own parts some of them may call for very precise positioning (I'm thinking SMT pick and place f'rinstace... ;-) ) so I believe it is a problem meriting careful consideration
And where all these maunderings lead me is:
What positioning system? This first cube will be a ball screw/stepper positioning system, but I think machine 2 will be a Linear stepper solution.
For the attractive merits of greater mechanical simplicity, much more precise positioning, and potentially closed loop operation inherent in a linear stepper system.
the questions posed above are satisfied by this solution and the upper threshold for accuracy becomes potentially very high if a mechanism to loop the accuracy upwards iteratively is implemented.
.
No comments:
Post a Comment