[lfbrown]

I recently purchased the Roland JWX-10 to use in making jewerly item wax models. Despite many conversations with both the manufacturer and the software supplier (ProtoWizard)I cannot produce a two sided model where the two halves are not offset by two or three tenths of a mm. The offset amount increases the proportionally to the distance from the rotary motor .

I can produce a perfect rotary alignment model using the file supplied by the manufacturer for alignment purposes.

I can also place a wax sheet in the flip fixture, mill a hole completely through, flip it and hit the hole perfectly from the other side. Yet when I put in the wax and run my tool path file the two halves mis-match.

There is an offset window on the Roland driver which has a variety of offset calibrations. This is a quick way to re-locate back to any given set of coordinates. There are also a number of windows, however, which I have no clue as to their meaning or purpose. And, mysteriously, neither Roland nor ProtoWizard seem to be willing to divulge their purpose or how to calibrate them. For instance there is "Y wide" and "Y narrow"; "Z wide" and "Z narrow", among others. I can line up a manual mill pretty quickly but I'm new to this world.

If anyone thinks they can help me on the alignment problem or give some info on the offsets window I would appreciate it. If requested I can upload a screenshot of the offset window.

Thanks,

L F Brown

[c-c-cncboy]

Hey LF:
A problem of CNC machinery that's built out of sheetmetal, rather than "the old way" is alignment of a rotary axis. Imagine an arrow mounted by the feathers on the rotary, dead center. If the rotary is not pointing quite true along the XZ-plane, yet the arrow was initially aligned to the correct Y down the pointy end, there will be a "narrow Y" close to the rotary and a "wide Y" far down away from the rotary, due to the rotation. Similarly in the other plane for Z values.

You need to point the rotary perfectly. Because that's not easy to do, Roland provide a firmware based correction system. Punch in the numbers describing the awful results, let the firmware correct for them, is the theory. The theory works fine for XYZ work such as 2-sided flips and falls over on genuine rotary work.

The jewelry forums are full of heart-breaking and tear-stained JWX-10 posts. I don't believe you will ever achieve success by relying on the firmware system to correct such alignment errors. Your time could be better spent perfecting the actual alignment.

Terrence

[lfbrown]

Terrance,

Appreciated, Terrance, many thanks.

Current values for the Y wide/narrow are -180 in each box (factory preset). My contact at ProtoWizard says these numbers are way too high while the techs at Roland say they're just fine. The ProtoWizard guy didn't believe them and had me reset to 0 but that really blew the tolerances. Re-entering the original figures brought the offset down significantly.

Since I'm still getting a noticable offset at the wide end, however, it would seem I should make a change in the wide box. But + or -? And how much?

Your point is well taken re physically adjusting the Y alignment but that is problematic on the Roland as the rotary table fits closely over two pins meant to secure Y alignment in the XY plane. I assume these pins to be permanently placed so no adjustment . I can, however, adjust it in the ZY plane by shimming.

The problem is further compounded by the ProtoWizard fixturing system. An adapter is mounted to the rotary plate at the motor which then allows for various fixtures (flip unit, rotary ring arbor, etc) to be inserted in the 2 jaw collet affair with a bolt then anchoring the fixture. Despite close tolerances the ring arbor, which is way too long in my opinion, has a serious runout problem at the end where the part is mounted.

I have come to realize that I'm going to have to change my thinking from "within tolerances" to "close".

If you have any thoughts or strategies on wide/narrow numerical inputs or the runout problem let me know.

Thanks again for the help.

Les

[c-c-cncboy]

Hey LF,
These are generic comments, based on 33 years of precision CNC machinery building and operation. They apply to many brands and types:

Analysis of flipper hardware and many waxes shows that half the issues are flipper-related (overhang, flexibility, looseness in assembly, manufactured out-of-alignment, poorly set up by user ... ) but in looking at a lot of waxes from flipper users, a huge but unrecognized issue is radial runout of their XXX-10 or XXX-40 rotary head. This is an unfixable issue. Those are what they are. The flipper is being rotated, but not in a circle.

When you are doing the 0-degree A-side and 180-degree B-side cuts, rotary radial runout gives you Z and Y errors. When you flip the part-machined wax over 90-degrees for the rotary pass, those Z and Y errors become "new X" errors and "new Y" errors. And during the final rotary pass, runout adds "new Z" errors and additional "new Y" errors. And of course they combine (inline or diagonally) to make the user wonder what the heck is going on.

Compound that with issues of radial runout in the rotary (even before the flipper is considered) and life is quite complex.

Worse, Z-height issues (even if the XZ-plane alignment is good) and Y-setback issues (even if the XY-plane alignment is good) must be thrown in the melting pot. Wrong Y-setback means the 0 and 180 passes will never align and all the rotary toolpath calculations will be out of whack, loading the tiny toolbits with a depth of cut they did not expect.

So I can't give you any magic numbers to work with, but I can help with these (and the prior email) listings of the variables. Perfect them (when possible) one by one. You did the intelligent thing by noting, then making a big change in a couple of settings to see if it became better or worse. Follow through, every location, every angular direction, working to perfect them. Make sure you never move the machine from its workbench location unless you have placed little cups to re-center the feet later, again in the same place, giving the same twist to the chassis.

Oh, and I nearly forgot to mention rotary backlash. Please add that to your list and take a look at http://forum.artcam.com/viewtopic.php?t=2175&start=15

A couple of users with a year or two of CNC experience under their belts think they have solved all the problems. IMHO they only perceive half of the problems and have set low standards for acceptable alignment - too easily satisfied on less-than-critical workpieces. Some will tell you "You just need to 'tram' your machine". Others will tell you "I have turned these alignments from an art into a science".

So it's good news and bad news:
Head position alignment issues: possible to solve most (except where fixed dowels defeat you).
Head direction orientation issues: possible to solve some by shimming.
Rotary backlash issues: Sorry, try to confine yourself to unidirectional rotary motion, never step back, never try for a second finishing pass. The chosen software and its output style will take command, resistance is futile.
Radial runout issues: Sorry. So, so sorry.

Send me a private message so I don't have to watch my words so much. Mommy taught me never to say anything nasty in public.

Terrence

[3dwaxmill]

Les

You should make a post on the Protowizard forum
There are plenty of jewelers who have gone through the same problems you are experiencing and now are having great success with the JWX-10 & MDX-40.

These jewelers are there & willing to help you .


Here is the link http://www.protowizard.com/phpBB/index.php

Good luck and persevere it will come.

[lfbrown]

Hi Jeff,

I already have a thread going there under Milling Machines>Roland>Roland Initial Calibration.

I've been getting some advice but can't seem to get results by applying it.

My firm belief is that what these machines need more than anything is to come with a service manual. If it doesn't work right out of the box, and it doesn't work when you go through the three little pdf's on aligning the X,Y, and A axes, then you're pretty much left to your own devices.

It sadly has been my discovery that calling Roland (who have not been so helpfu) or ProtoWizard (who have genuinely tried to be helpful) they basically just refer you back to the pdf's.

Cheers,

Les