Guilloche!

[BMeyers]

Hessel

I'm sort of working along with a book on Guilloche and Rose Engines and trying to duplicate some of the simpler patterns, of course no guarantee that this is the right or best way since I'm really using the wizard in unintended ways.

I used the Elliptical Order box to set the number of lobes, and slid the coefficient slider all the way to the right to get the "gently undulating" curve.  I've been setting both the inner and outer envelopes the same for now, but real rose engines used a variety of shapes.  Search for Bill Oombs for a rose engine simulator, trying that out is on my list for the future.  I set phase% to zero, only because it makes it easier to read and match up the numbers on successive traces.

Fill density is set to 1, Fill type to line.

The maximum radius will need to be adjusted to be greater than where you want the trace.  Use the Outband and Inband Radius sliders (if you click on them you can use the keyboard arrow keys for fine adjustments) to get the size where you want it.  When the outer ring looks right (it's actually the middle one in the screenshot) then click  next.

I closed Vexx before the second screenshot so some values might be a little different.  The Maximum Radius will change, set it back to your first value or something that gets the outer envelope outside your work, in this example I made it a little larger.  You'll do this every time, different values for max radius might make the slidering easier as the pattern gets smaller.

Use the InBand Radius to get the second trace close to the first one, then use the Phase slider to align the high spots so they touch.  You need to set both the inner and outer phase the same (or maybe not, if you want a different effect).  Then zoom in and get the two bumps almost touching, and adjust the phase if necessary.  When it looks good then click next and repeat, alternating a phase of zero and the value that lines them up.  After a couple passes it goes fairly fast.  Repeat until you get the desired effect or get tired of repeating :)

Hope that helps!  After rewatching the Vexx videos I'm going to try the celtic knot cutout again, maybe with the snip tool and reworking the knot dxf to make sure all the shapes are closed.  We'll see.  My inspiration is attached (this one is from a swiss watch company via a link on cnccookbook.com) but I'll never get that good.  No special reason for the celtic knot other than I like them and had this particular dxf sitting around.  Why try something easy when there's something much harder sitting around :)

The diamond dragon (aka drag bit) is on it's way, a couple 120 degree bits from drillman1 on eBay are here (shipped and at the post office half an hour after ordering Saturday afternoon on a US holiday weekend, highly recommend him if you eBay in the states), and I used a christmas gift card I'd forgotten about to order some 1.5"/38mm 20g brass "stamping" disks. I expect my problems are more operator than material, but brass is the traditional practice material.  I don't think I'll be trying gold, silver, or platinum or ordering a $100K US MADE Rose Engine...

Kirk

wonderful design.

[ArtF]

Thats awesome..

[Mooselake]

Incredible, but absolutely not mine, it's from a swiss watch company and done on a real rose engine.  I wish I could do something like that.  It's the dial on a probably $10K US watch, no price provided since if you have to ask...

The diamond drag bit came today, and with no preparation except changing the collet and thinking half the bit extension was enough (worked out to 5mm "cut depth", really spring compression), at 200mm/min both plunge and feed rate did the attached photo; took about 15 minutes.  It's the same 25mm flourished test pattern I've now cut at 0.05 and 0.1mm doc and both 90 and 20 degree vbits.  I could hear little thumps where it crossed other lines, but couldn't feel (it's so slow, and non-spinning, I gently held a finger on the cutter)  them, and it doesn't appear to have lost any steps.  First time the flourishes and center detail came out.  Pretty good for $50 US, I think, plus it'll engrave on plastic, steel including stainless, and glass from other user's reports.  Tried to convince Mrs. Moose that since it was a diamond she could share it as her upcoming birthday present, which worked about as well as you may have expected.

I'm going to try the 96 lobe barleycorn for grins, then see what lobe count (no idea what the official term is) will give a nice effect. 

As always the photo doesn't do it justice, just a 100Weq LED desk lamp at an angle and a cell phone. The original changes as you change your viewing angle, and has a better jeweled effect than my vbit cuts.

Spent a few hours being net beaned and figured out how to run Bill Oombs' open source rose engine simulator on my W10 laptop.  It's written in Java, compile your own for non-fruit computers, and a great visualization (makes me want to look closer at NetBeans) of how the cut will end up.  I played with barleycorn until hauled off to dinner; it allows setting phase and position on every cut.  While mostly intended for ornamental turning he does provide a couple PDFs describing how to do a spreadsheet to get the proper manually entered values for a couple popular guilloche patterns.  It will provide g code, but for a unique machine Mr. Oombs built that uses (iirc) X, Z, and C and way different cutters (they call them cutting frames, I think what he's calling a drill is a conventional spindle).  If I can resurrect my 2000-2002 Java memories I might take a look at his g code generation and see if I can make it produce X-Y-Z code.

There seems to be several techniques of creating guilloche patterns.  There's the spirograph emulators, Art's very nice unique style, rose engines with rosettes of unlimited shapes, and geometric chucks (which can also be mounted on a rose engine) which seem to be closely related to spirograph patterns.  T. S. Bazley's Index to the Geometric Chuck, published in 1875 where he used one to generate pen and ink drawings of a very large number of patterns and numbered them.  Alan Battersby in the UK did a geometric chuck simulator that would reproduce Bazley patterns but his site has gone dark and archive.org doesn't seem to have captured the downloads.  I have it up north, but not currently accessible.

I did say it was a rabbit hole...

Kirk

[Mooselake]

Today's effort, with the diamond drag at 300mm/minute and the usual 25mm square.  Read that bigger routers do these at 4000mm/minute, might try one at the machine's 1000mm/minute max speed, and see what happens.  Plan to keep the mouse over the non-physical (and really a black square) red mushroom button just in case.

This is the aforementioned celtic knot dxf with a line circular array on top, started with a line from center straight up, then did a 100 line circular array; could have done more.  Much panning, zooming, and snipping later all the lines outside the knot were removed and this resulted.  These lines would snip where the guilloche patterns would not.

I did get into a bit of an issue with the 3D rotate (the other CAD program uses middle click to pan, Vexx uses a right click), and couldn't figure out how to get back to upright and flat, so got as close as I could and just lived with it.

Got the $7US tool setter installed, reminding me how much I dislike crimping dupont connector pins, didn't help that the insulation was bigger than the pins were designed for. In retrospect I should have stripped it off and had both crimps go to bare copper.  Took about 10 pins to get two that worked, more than normal, maybe need to do more than 2 a year to get better.  Now need to play around with grbl commands to use it properly since bCNC's tool set tab is set up for one in a fixed position and drove the (conductive, diamond won't work) bit into the top of the cheap setter and then tried to move it around.

Despite being Amazon Prime the brass disks might be here Monday.  The "ground shipped to offer such a low price" tool setter came yesterday.  Ordered Tuesday; the hurricane might have slowed things down.  Non-prime pool algaecide ordered the day after came before the prime shipments.  Also have a batch of 50 1/32 to 1/8" mostly ball end mills, "slightly used" due Monday, and the 1/8"/3mm 7328 cast acrylic is already here for the lithophane party.

Thinking of trying a rotated array of gearotic gears next, just for fun.

Kirk

PS: searching for how to photograph engravings gives many hits on how to make them but nothing on how to actually take a decent picture.  This one sparkles and the effect changes with the viewing angle

[Mooselake]

The 3x3 gear array crashed Carbide Create multiple times, maybe it was too many small vectors.  So for fun here's today's project.  I tried an online gcode optimizer on the second one, and while it reduced the rapid horizontal move total length it didn't change the engrave time noticeably.  What it really needs is eliminating a number of unnecessary up and down movements by optimizing the cut paths.  For example the dial circles could have been two cuts, not a series of ups and downs distracted by the tic marks.  Oh, well, still took under 5 minutes although I've done some non-gearotic engraving that took well over an hour, a raster to vector converted pencil sketch that I worked on almost 10 years ago.  Too many small artifacts that didn't get edited out.

There's a small gotcha in the "2" in the magnified picture, one side is an S that descends below the bottom line.  These were un-snippable in Vexx.  Not significant since it's the old editor.

This is clock dial from the Auggie wizard with a simple guilloche pattern overlaid on top, diamond dragged at 800mm/minute and with the spring compressed 5mm (half way), trying 2.5 was too shallow a cut.  That 5mm, plus the 1mm safe height, consumes most of the cut time as it makes 200 trips up and down.

Kirk

[Hessel Oosten]

And another SUPERIOR guilloch?  video:

https://www.youtube.com/watch?v=b-6B776Hw3I

See eventually from ~ 8 minutes.

Hessel

[Mooselake]

Thanks, haven't seen that one yet.  I'll watch it later today, have some appointments and maybe bring a new dog home from the animal shelter to keep the moosedog company

70 1.5"/38mm brass blanks arrived today, 50 that were lost in shipping for almost two weeks and 20 that I ordered yesterday to replace some of them (funny how that makes the missing ones appear).  Need to make a fixture and see what engraved brass looks like

Kirk

[ArtF]

>>couldn't figure out how to get back to upright and flat

  Double click on Vexx's screen will return to flat plane x,y.

Results are looking great...

Art

[Mooselake]

Hessel, thanks for the video, we watched it yesterday evening.  Now I'm motivated to try straight line guilloche too.  This presenter was less down on CNC than my other sources

I've been trying to avoid doing any actual programming (something about retired and wanting to do new stuff), but am becoming interested in the Processing environment, mostly because it looks like it handles most of the visualization details along with simplifying vector output.  I was mostly a real time and OS programmer, missed all that gui stuff.  The challenge with simulating a straight line machine is the pattern bars; sawtooth and sine should be straightforward but DIY pattern bars would be trickier especially trying to avoid lots of short vectors.  Phasing doesn't look to hard, figuring out how to represent horizontal movement, particularly trying to do some of it automatically looks about middle difficulty.  Of course, that's from a non-graphical programmer and could be totally wrong.

I'm starting to dream about how to add geometric chuck patterns to Art's style of envelope.  Not sure if that qualifies as nightmares or not.  The math looks pretty straightforward, which means I have no clue about what it would take  :)

Kirk

[Mooselake]

This may not look like much, but the svg file was generated with a 30 line program written in processing that creates a triangle straight line machine pattern bar and then steps it across the workpiece.  The scaling is way off, one might say non-existent, no way to change parameters except recompiling, no phasing, masking, spacing changes, pattern bar changes, etc.    Lots more to go but actually achieved a CNCable file!  First code I've written in quite a while, first attempt at processing too.  Not the easiest way, but it allows for other styles, or even customized, patterns.  Who knows, might evolve into a straight line guilloche machine simulator. 

25mm sq, 90 degree diamond drag on anodized aluminum, 800 mm/minute.  Still haven't got the lighting thing, very sparkly in real life.

For grins here's the program, probably needs some comments :) :

Code: Select all

// create a PShape representing a triangle patternbar
//  with pitch spaced teeth, eventually this will be generalized 
//  to create additional types

import processing.svg.*;

PShape patternbar;
float pitch=20, pattern_spacing=10;
String output_file = "testZigZag.svg";

void setup() {
  size(400, 400);
  background( 0); // #b5a642);
  stroke(255);
  noLoop();
  beginRecord(SVG, output_file);
  patternbar = createShape();
  patternbar.beginShape();
  patternbar.noFill();
  patternbar.stroke(255);
  patternbar.strokeWeight(1);

&nbsp; for (float y=0; y<width; y++) {
&nbsp; &nbsp; patternbar.vertex(abs((y % pitch) - pitch/2), y);
&nbsp; }
&nbsp; patternbar.endShape();
}

void draw() {
&nbsp; for (float&nbsp; i=0; i < width; i += pattern_spacing) {
&nbsp; &nbsp; shape(patternbar,i,0);
&nbsp; }
&nbsp; &nbsp; 
&nbsp; endRecord();
}

Kirk

[Hessel Oosten]

evolve into a straight line guilloche machine simulator

Now I'm motivated to try straight line guilloche too

Who knows ? May be... Art is reading this thread ?. ;)

It really looks great -and- promising !

Hessel

[ArtF]

It does look good. Seems its more a function of technique than program
from the most part.

Art

[Mooselake]

A straight line machine traces a machined pattern bar, which can be pretty much any shape but from what I've been able to tell, based on perusing online sales,  are mostly triangular waves, sine waves, and some special designs.  To be traditional they need to be traceable by a "rubber" that rubs along them, so no undercutting or aggressive angles.  Of course a CNC program isn't bound to that, but might as well try to stay close.  Right now I'm working with 100mm pattern bars but eventually that should be extended to 150 or 200mm, just change a couple variables...

It's progressing. Added sine waves, some sliders for pitch (mm per repeat), spacing, and amplitude (amplitude currently only for sine waves, triangle is fixed at 45 and depends on pitch) and a button to write the SVG file, wasted a couple days trying to put the controls in a separate window and gave up over variable scope issues, seems I couldn't pass a boolean back to the main window out of a button and just added some whitespace on the right of the window for them.  Processing's an overlay/preprocessor for java, and while I spent several years writing java EJBs in the early aughts my library's up north and memory's a little fuzzy on esoteric details.

To Do includes being able to switch patterns without recompiling, not hard but leads into another issue which is specifying pattern designs, spacing, and phase (aka up/down pattern bar displacement) on each line.  Besides the canned pattern bars need to add a capability to read a properly formatted SVG file as a pattern bar (it's only a couple lines of code), designed elsewhere.  Need to add workpiece scaling rather than a fixed 100mm square, zooming (need to see exact line placement in some designs and 3.8 pixels per mm (96 ppi) is too tiny for my eyes).  Also need to figure out how to do masking, as in remove all of the design that's outside a selected shape, will probably limit that to rectangles and ellipses including circles.  Still a long way from basket weaves which need variable spacing and phasing.  Other than that one zigzag pattern the CNC machine has remained idle.

Art, are booleans implemented in Vexx?  Specifically I'm trying to do intersections, remove everything outside a closed shape, and couldn't figure out how to make that work.  My patterns are multiple single lines, not a closed shapes,  but the outside shape would be closed.  All I really want to do right now is be able to use 38mm brass disks without aircutting the parts outside.  That would be tough with the spring loaded diamond drag.

My doc wants a biopsy of my knee, the surgeon says he'll have to open it up again rather than a needle or endoscopic (he didn't think much of my suggestion of a small hole saw...) and take pieces from various places, and is scheduled for 10/2.  Of course it's a rare (in prosthetic joints) and difficult to cure infection so essentially no guidance in the literature of how to get rid of it (they only found one sort of applicable case in the database), the biopsy is to see if the treatment has worked.  If it has, new knee joint about a month after the biopsy, if not then ??  Might be away from the router and programming desk for a while after that depending on how the knee bends and chair sitting goes.  Moose might have 4 knees, but the redundancy just didn't work out on this one.

Kirk

[Hessel Oosten]

Here today some "straight-line engine turning...", with the mill.. ;).

The "basket weave"".

Small diamond engraver, aluminum (polished previously), diam. 30, depth ~ 0.05 mm, feed 75mm/min.

Nice explanation of the pattern-making by watchmaker Roger Smith:

https://www.youtube.com/watch?v=LBiiDpexmA8

Hessel

[Mooselake]

Nice!

You're ahead of me!  I'm still working on a simulator and spent the afternoon trying to debug a basket weave pattern generator,  darn thing is having some display issues.  Had other distractions lately or maybe I'd have gotten further.  How did you create that pattern?

I've been using both Peter (?) Shapiro and Roger Smith's videos as inspiration and design advice, thanks in part to your links.  There doesn't seem to be a lot of online info on straight line machines, and I'm guessing at some of the terminology and looking at eBay and other sales offerings to get an idea of what was available as pattern bar designs

Kirk