Reality bites was Re: Step count / inch problem

Oops,

Part of that last reply left off...

Ballscrew "errors". Poor choice of words on my part. What you have
said, and the Thomson engineer confirmed is correct. I mis-spoke.

This is what I "meant" to convey:

Since the lead error is "constant"(even in the chreapies), in that it
either is too much or too little as a result of the rolling process
used to make the screws, it can be cancelled out easily. This means
it won't "stack up", as we have specified the "correct"(real) number
of steps per inch/unit. What we cannot cancel out by setting the
steps per inch/unit is the cyclical errors...

In other words, we can "un-slope" the line by adjusting our steps per
unit to match the rolled screw error(which eliminates the "stacking
error"), but the waviness(cyclical variations) will still be there...

Hope this helps.

Ballendo

P.S. The graph you looked at answers your question about the
relationship of cyclic to cumulative errors... If you define the
slope part of the graph as .004 (or whatever), you can interpolate to
get the value of the cyclical "bumps" by their relation to this
distance on the graph axes.

So what you need are more of these graphs; and the values used along
the axes when they were created.

--- In CAD_CAM_EDM_DRO@y..., "Carlos Guillermo" <carlos@v...> wrote:
> Hi Ballendo -
>
> The Thomson rolled ballscrew literature shows a graph of the type
> of variation you are likely to see with their ballscrews. They
> clearly show that the errors DO stack up for longer lengths (they
> refer to the "Cumulative Representative Lead"), and they also show
> the superimposed error component that is cyclical, like what you
> describe ("Lead Fluctuation"). The graph is of a sloped line,
> with waviness added. Just imagine taking a ballscrew with
> cyclical errors, and changing its temperature. The overall length
> will change, and every point along the way will shift
> proportionally (with even temperature changes), but the cyclical
> errors will still be there. The cyclical errors are in fact small
> compared to the cumulative errors. It's just part of the
> manufacturing process. I spoke to a Thomson engineer who said if
> you could use software to factor out this cumulative error, you'd
> compensate for the majority of the error. When they give
> accuracies of .004"/ft, they're mostly talking about the
> cumulative errors. I tried to get some numbers on the cyclical
> errors with no luck. I believe they will map out their more
> pricey ballscrews for a fee, however. Anybody have this kind of
> data? It would be interesting to compare the magnitude of the
> cyclical variations against the cumulative errors, and to see the
> error characteristics of an actual ballscrew (slope, waviness,
> arcing, etc)
>
> HTH,
>
> Carlos Guillermo
> VERVE Engineering & Design
>
> Ballendo wrote:
> <snip>
> Ballscrew errors, even with the cheapies, do not usually stack
> up; but instead are cyclical variations within shorter distances.
>
> <snip>