With all the mods and substitutions that are possible for a given table design, how do you test whether or not a change is made for the better? How do you compare designs and their changes that others may make to see if it might benefit yours?

Is it possible to come up with a procedure, perhaps some free g-code, or other objective tests that can measure those characteristics that are important for a "good" router table? If that was available here, each of us could try it out and post the results that might give us a more empirical result about the changes we make.

As a newbie, I can only shoot from the hip here, but the test suite might include a standardized measurement of rapids, repeatability, object "trueness" (like a circle), measured resolution (next to calculated), etc. It might be nice to come up with an overall score for comparison. It may or may not involve actually cutting something (meaning the test would cost in materials, and therefore be somewhat discouraging to run) and standardized measurement procedures would have to be followed.

I think this would be useful for a number of reasons. Scores between builds made without substitution would be a measurement of the builder's accuracy of components, and a pointer to where improvement would be worthwhile. Score differences between axes would give a builder an idea on where improvements would make the most difference. Differences between components would give a more objective judgment about the effectiveness of newer steppers, drives, and belts, screws, or racks. It might even be fun to compare a score against factory models.

Forgive me if there is already such a thing. If so, I am happy to try it and post the results as soon as my build is finished. If not, are there any thoughts on how a test like this might be done? What tests have you found useful? Anyone interested in a "CNCZone benchmark?"

I think the difficulty with your proposal is that even if multiple people used the exact same design, the results will be different enough to render the score unusable.

You could take the exact same parts and get two different results out of them because one was made by one person and the other one made by someone else. Even if the same person made them they could be slightly different enough to skew the results.

It would be nice to say "here is the best machine design" but I guarantee you that you would get 98 out of 100 people to find a better way to design it. It may only be better because their method is something they are used to and have perfected, and that's why it works better.

This is exactly why you probably won't see two machines here that are the same. Some are similar, but not the same.

I think it would be helpful to somehow rate various components and assemblies, but I don't know if there is a way to do it so that the results are consistent and meaningful.

Gary

Thanks for your post, LaserImage. However, I do disagree with the idea that there is nothing we can do to measure improvement.

"I think the difficulty with your proposal is that even if multiple people used the exact same design, the results will be different enough to render the score unusable."

Why? Reproducing parts from the exact same g-code should produce the same part, with slight differences between machines. It is measurement of the resulting differences between the parts that is of interest.

"You could take the exact same parts and get two different results out of them because one was made by one person and the other one made by someone else. Even if the same person made them they could be slightly different enough to skew the results."

The "persons" being tested here are machines, not persons, that are supposed to have some measure of repeatability. I am simply trying to suggest a standardized way to measure that repeatability.

"It would be nice to say "here is the best machine design" but I guarantee you that you would get 98 out of 100 people to find a better way to design it."

I'm not trying to say one machine is the best. If we don't have a way to measure the differences between one design or mod to the next, how do you know if "the better way" is indeed better? If going from an ACME leadscrew to ballscrews only improves accuracy (in the broadest sense of the word) a small amount, it would certainly help make the decision on where to make the next mod. If you look at many of the threads about this choice of stepper/driver/bearing or that mod, the authors do that all the time, by naming some result informally. I am just suggesting that there might be a way to measure it that we could all use.

Pretty much any machine can be made to cut very accurately just by making very careful adjustments. I think that the main factors that come into play are rigidity, and manufacturing skill (and accuracy). And if the machine has enough adjustments, the second one can be adjusted to work fine. Which comes back to rigidity. This includes not only the machine frame, but all the moving components as well. All screws and nuts must be backlash free and mounted so no unwanted movement can occur. Most mods that would make the machine more accurate would be the result of making the machine frame stiffer, or removing backlash from the drive system.

You mention going from acme to ballscrews. Both come in several different grades with different accuracies. You can easily buy acme screws that are more accurate than certain ballscrews. The one common factor is that as accuracy increases, cost goes up even faster.

Is it possible to come up with a procedure, perhaps some free g-code, or other objective tests that can measure those characteristics that are important for a "good" router table? If that was available here, each of us could try it out and post the results that might give us a more empirical result about the changes we make.


Steer clear of subjective terms and stick with quantifiable terms. For example, for speed:

Rate it according to rapid, and cutting various materials.
Rate it in terms of accuracy by manufacturing a part and measuring the results, repeating enough times to get a quantifiable result.