How to handle LARGE project file? (74"x10")

General MeshCAM Questions

How to handle LARGE project file? (74"x10")

Postby cnczane » Tue Jul 18, 2017 4:46 pm

I want to carve oaken sideboards for my pickup truck. I haven't found anything suitable at Walmart...

I've run into this before when I tried my first large project (30x30 topo map of Wisconsin), but got around it by reducing the resolution to an only slightly unacceptable level to scrape by. However, to get the current 74"x10" model to the required "<3000 pixel" limit would require a resolution I can't tolerate.

I am going to have to break the project into two sub-projects, but I very much hope to do that by editing the final G-code of the major project rather than submitting two separate projects to MeshCAM to treat as independent of each other, which they would not be.

Any insights/advice for large-format projects would be appreciated.

Thanks!
--
C.Z.
If you have not received a reply from me in over a year, I am not ignoing you: more likely I am fallen asleep under a tree. Again. Please poke me if you think it worth your trouble.
cnczane
 
Posts: 277
Joined: Sun Sep 21, 2008 4:15 am

Re: How to handle LARGE project file? (74"x10")

Postby larynx » Wed Jul 19, 2017 3:53 am

The only "larger than my table" projects I have done involved me making "keys" or "dovetail" cuts to align and put them together again - but yes, involved separate gcode files for each piece. I would think it is easier than manually splitting the gcode and moving the board. My opinion - good or bad - is that it would still be easier to have two files for one board even if moving it - use dxf includes that overlap - positioning must be fantastic of course.

I would also be interested if anyone has another way of doing it easier - without having a slide table axis on your setup to re-position the piece for you, of course.
I am not aligned with nor do I have any relationship with MeshCAM or its staff other than being a user [that sounds a little like I have a dependency problem]
larynx
 
Posts: 64
Joined: Thu Oct 29, 2015 2:56 am
Location: East Central Florida

Re: How to handle LARGE project file? (74"x10")

Postby cnczane » Wed Jul 19, 2017 5:43 pm

@larynx

Thanks, just what I was hoping for. Not so much an exact how-to, but more how-to-think about it.

As I read your comments, I started imagining clamping some kind of "indexer" to the table, maybe using a single dowel to locate into predrilled holes in the piece, but with my luck, immediately also imagined "revealing" those holes with the cutter. Now I'm wondering if it might be possible to position the piece accurately by visually aligning "knife" cuts across the piece and a clamp. Maybe putting a flat "feature" in the center of one edge to remain unmachined and used for aligning left-right zeroes?

EDIT: rather, leaving it UN-machined in the first half [with an exact DXF exclude], but machining it off during the 2nd, after it's been used for alignment.
If you have not received a reply from me in over a year, I am not ignoing you: more likely I am fallen asleep under a tree. Again. Please poke me if you think it worth your trouble.
cnczane
 
Posts: 277
Joined: Sun Sep 21, 2008 4:15 am

Re: How to handle LARGE project file? (74"x10")

Postby cnczane » Fri Jul 21, 2017 10:57 pm

I decided to try it by breaking my big image file into two slightly overlapping halves and machine them by adjusting zeroes.

But this part is maddening: NO MATTER how I save off the png file, when MC5 opens it, it's always slightly LONGER.

For example, I saved a 10x37.125x80dpi image file and opened it in MC5. In the dialog window, at 80dpi, the imported model is calculated to be 10x37.125, exactly.

However, when I check Geometry->Properties, MC5 reports that the file is 10x37.159 meaning my file is going to be stretched.

But I don't want it stretched, not even by 0.034.

My disappointment is not little because otherwise this approach showed a lot of promise.
If you have not received a reply from me in over a year, I am not ignoing you: more likely I am fallen asleep under a tree. Again. Please poke me if you think it worth your trouble.
cnczane
 
Posts: 277
Joined: Sun Sep 21, 2008 4:15 am

Re: How to handle LARGE project file? (74"x10")

Postby cnczane » Sat Jul 22, 2017 1:51 pm

I'm still disappointed: I think MeshCAM ought to be able to handle whatever resolution I throw at it, as long as the hardware is there to handle it.

Another problem arises with splitting the file into two parts: scaling. With a terrain file such as what I have, it's impossible to split the file into two rectangles that have equal elevation ranges (so they'll scale identically in MeshCAM and match up when carved). I addressed this by adding fake max and min elevations in the corners of each half-model, to guarantee each model would be scaled identically. But that was before I found MC5's nasty resizing behavior.

I finally loaded the entire file into MeshCAM in desperation, reducing the dpi resolution little-by-little until it was accepted: the final result was 10x74x40dpi.

With the resolution warning occurring at ">3000 pixels", I'll guess that the maximum resolution one can specify is determined by: 10 * 3000 / (width * height), because that's what seems to fit: 10*3000/(10*74) ~= 40. (Again, I wish that MC would do this automatically, based on resources.)

As it is 1/40" is about twice as much as the stepover I'd like to use for cutting the piece out with an 1/8" roundnose bit, but I was surprised to see that the display did not appear to actually be bad. (I'm beginning to wonder if MC's "resolution" means something different than what I expect from, say, the sampling theorem. The theorem says that the sampling rate determines the highest frequency reconstructable from the samples. In terms of models, the size of the cutter determines how "fine" the resulting model can be; it's no good to develop a model with a high level of detail--aka spatial frequency--and carve it with a Fat Cutter: the cutter needs to be finer than the finest detail to be reproduced. In fact, it needs to be half the size at the largest, but finer is better. By settling on a 1/8" finishing cutter, I have already decided how much detail I will accept in the final carving, but I don't want MC setting artificial limits before that stage.)

I want 80dpi to correspond with my 1/8" bit and 1/80" stepover, but 40dpi was the best I could get. (I just went back and verified that indeed there is a resolution setting below which the detail of the model is simply unusable: in my case 20dpi.)

Regardless, I created two DXF files for separate machining regions, one 10x38" long (the "north"/long/+Y), and the other the remaining 10x36 + a 1/8" overlap ("south"/short/-Y). Following what I suggested I'd do earlier, I picked an origin for both regions "1 inch south of the north's southern-most edge". If I machine the "north" part first, the origin will be preserved, but will be cut away after repositioning and cutting the "south."

I haven't cut this. I'm still thinking about how to index the cuts, but am pretty sure I want to use the precision and accuracy of the machine to drill the locator holes after the initial positioning. Based on the sizes I've suggested, it seems to me that I will need to move the piece 36-1/8" "north" between cuts. I think to do this by drilling two 1/8" holes, 36-1/8" apart along the Y-axis, straddling the edge between the piece and the fixed support. This will leave me with two half-holes/half-cylinders in one edge of the finished piece which given the final use, won't be significant. If I cared, I could buy a bigger board and put them in waste. It occurs to me that I can reduce this to a single half-cylinder if I drill two half-holes in the support first, load the piece, and redrill only the south hole.

Oh, er,

A-men.
--
C'zane
If you have not received a reply from me in over a year, I am not ignoing you: more likely I am fallen asleep under a tree. Again. Please poke me if you think it worth your trouble.
cnczane
 
Posts: 277
Joined: Sun Sep 21, 2008 4:15 am

Re: How to handle LARGE project file? (74"x10")

Postby larynx » Mon Jul 24, 2017 4:37 am

Well, I'm "talky" tonight - it must be the beer.

I never used the image stuff, yet. I had planned on trying it for a couple things, but resizing by MeshCAM sounds like a deal breaker.

Because of [the beer] I sort of follow your post. Yeah, you make me wonder if I should have index holes in my spoil-board [it is actually wood strips glued to the aluminum slot top that get spoil-board cut - saved my A*@$ a few times and can re-machine if necessary, but I digress.

So do you think a couple of "locator pin holes" in the table is the way to go, or perhaps - I have in the past placed a few"stops" in my slots using M6x10mm and slot nuts that get tightened against a board after I have it aligned, so after I remove the board I can push another board up against them to duplicate another without re-zeroing the table - anyway....or just stops at one side and the end - so you can rotate and push against or am I off topic - damn another beer.

Must be a simpler method. Clamp a precut "L" board at the table end to reference after alignment or two blocks instead of my t-slot stuff, in case the cutter hits then it will just be wood. Sounds like you board is already sized in width/length and no leftover?

I did make an L-board that I used a few times, cut it on the table with exactly 2" left to the inside corner. The best part - thought I was smart, we know what that gets us - but it worked - is the 1/2" hole machined from the "L" inside corner exactly from the X & Y axis points. That allowed me to put a 1/2" bit with end bearing in the router, bring the router down close to the table, raise the "L" board - the hole - onto the bearing, drop the router closer to the table capturing the bearings and some bit in the hole. hold the "L" down by hand, zero X & Y, raise the router, insert into the hole in the "L" board a slot nut and bolt with washer and tighten so the "L" can't move. Then run the router 2" over [exactly aligning to the inner "L" edge] drop the bit to table and run it against the edge until the board was "aligned" in the other axis against the bearing. then clamp the "L" board in place and tighten everything. Now I had a perfect corner ZERO [rezero by offset 2" from original zero] I could push blocks into the "L" corner and go, go, go over and over.

It may sound complicated, but it only took me about two or three minutes to do this. Then change the bit and cut - can always offset zero again by bit diameter cause I know the zero - exactly. Maybe this helps? Maybe it is just confusing?
I am not aligned with nor do I have any relationship with MeshCAM or its staff other than being a user [that sounds a little like I have a dependency problem]
larynx
 
Posts: 64
Joined: Thu Oct 29, 2015 2:56 am
Location: East Central Florida

Re: How to handle LARGE project file? (74"x10")

Postby larynx » Tue Jul 25, 2017 1:26 pm

So you are making me think, and that can be a bad thing........

While reading a Randy post:

. Re: X,Y Position Question

. Postby Randy » Tue Dec 13, 2016 1:27 am
. Mike, you could either literally define the stock as your whole stock, and place the geometry where you want the cutting to go.

. The easier way that I've used, is to just let MeshCAM "shrink-wrap" the stock to the geometry so X0 Y0 is at the lower (front) left edge of the geometry itself. Then when I set up the machine, I'll zero the X and Y after I've placed the spindle
. where I want the geometry to be cut. That way, I can throw a random piece of stock on the machine and not worry about X and Y margins, centering, etc. I don't know how big your workpieces are, but for my little ones, I'll print out a
. full-size outline of the piece to set on the stock and use to position the spindle at X0 Y0.

. Randy

So this made me think, and remember my previous about the "L" board with the hole for the bit with bearing - wondering, could a hole be drilled in the board at the one end where it will be moved, and that hole [doesn't have to be full depth] then be used with a bit to align the new start position [X0Y0] for the following cut? It would be "spot on" so to speak to continue the cut. Obviousy we are talking a move involved, would still need the "side" push against to keep it straight during the move, but...

You like ideas, there's one to do something with, perhaps. Interested in what you finally end up doing. I guess you could print something like Randy stated, but the hole idea [notice not "whole" but "hole" as I do not wish to confuse one type of world with another :ugeek: ], I am trying to provide you with a more definitive point from the spindle's perspective after a move.

Anyone else?
I am not aligned with nor do I have any relationship with MeshCAM or its staff other than being a user [that sounds a little like I have a dependency problem]
larynx
 
Posts: 64
Joined: Thu Oct 29, 2015 2:56 am
Location: East Central Florida

Re: How to handle LARGE project file? (74"x10")

Postby cnczane » Thu Jul 27, 2017 4:06 am

@larynx
I've actually got an "L-shaped" corner piece! I decided that By-Gum!, if I clamped a board in there and the machine CUT X- and Y-axes, then THAT's where the machine thought they SHOULD be!! Of course, almost immediately I had to unclamp it so I could put in a bigger piece, and then faced the problem you mention of how to align an "edge" along X- or Y- with only the point source of the cutter.

To Randy's post, it's similar to what I do, which I've reported elsewhere in these pages. I never bother with an absolute, immovable, registration point for my models. Instead, I _always_ use the top-center of stock as xyz=0,0,0. This has one negative in that the point is often carved away during the rough pass and I've lost the Z=0 reference for the finish bit. Most of the time, I note the "last-Z" of the rougher, load the finish cutter and drop it to touch the same spot and redefine Z=last-Z for the finish cutter there too. Good to go. If that spot's inconvenient, I just stroll back through my G-code, looking for another handy spot where the Z-value of the rougher is known and use that.

I mark the center of my stock with nothing fancier than a metal tape and two eyeballs, one for the left or bottom measurement, and one for the right or top, giving me, more often than not, TWO marks for where the center is. I typically still have a narrow, finish cutter in from the previous job so I center it between the marks and define G55 X=0,Y=0 there. (I always make sure that G54 is identical to G53 (with G10 L2 P1 X0Y0Z0) so that I can see G53's coordinates at any time by setting the mode to G54.

BTW, I also lie a little bit to MeshCAM so that it gives me a model that will be about 0.1" larger than my stock in both directions. That way my center or my rotation can be slightly off and I still get an acceptable result. "Acceptable" in terms of a piece of art, not a piece for a rocket motor to Mars.

Most often I cut end-to-end, if I bought a 13" piece of hardwood 17.5" long, I hate any of it going to the burnpile, so I try to cram art on there until it's busting over. In the case of my sideboards, I don't have any scrap to drill a hole in, BUT I have edges that are going to be "down" and thus not attention grabbers. I think I can drill the hole you imagine as a half-hole in the edge of the workpiece to match either of two half-holes at the positions necessary to do the full cut. So we agree that might work. AND, that end-to-end approach means that my clamping pieces do take a lot of collateral abuse from the cutters.

Now, back to your earlier post.

I use something similar to your "stop" screws for my T-slot table. Mine came with a bunch of T-slot extrusions screwed down next to each other, but then has about a 1/2" thick rubber mat glued to the top, between the slots. I bought 4 (they have a name... but what?) steel cam clamps with a low-profile, sideways action, parallel to the table, not perpendicular to it. Because they're steel, I want to stay away from them, so I always have clamping boards between the workpiece and those things. I loosen the clamps, push the board up next to them and then attempt to align the board to the, typically, X-axis by running the cutter back and forth to touch the cutter off at each far edge, until the error reduces to something acceptable. (And remember who's writing here.

Now. We're to Paragraph 6, beginning: "I did make an L-board..." which confused me no little bit. I really can't picture it, and I tried. And then I went and got my own beer... And then I think it must be the wrong kind... :^)

Seriously, I'll say that I'm concerned about two things in aligning an straight-edge guide for use in quickly squaring workpieces, which I've mentioned: firstly, I want to be sure the cutter isn't bending (so I use a 1/2" bit) and secondly that the adjustment move I just made one end changed the adjustment I already made on the other. (Thus my walking it in--which I admit was b.s.--I never do that--most of the time, with my stuff, "nearly aligned" is pretty good enough.

Thank you for your thoughts! (I may not always understand you, but I'm a fan.)

Cheers!
--
Zane of C
If you have not received a reply from me in over a year, I am not ignoing you: more likely I am fallen asleep under a tree. Again. Please poke me if you think it worth your trouble.
cnczane
 
Posts: 277
Joined: Sun Sep 21, 2008 4:15 am

Re: How to handle LARGE project file? (74"x10")

Postby larynx » Thu Jul 27, 2017 9:12 pm

:twisted: L-Board:

Pics - in order

1) PIC 1 - board as cut on machine
2) PIC 2 - hole as cut on machine that a cap bolt with washer can be placed with T-nut under
3) PIC 3 - bring router close to board [like my simulation - even used a collet for points 8-)
4) NOT DEPICTED - raise board onto bit bearing
5) PIC 4 - drop head closer to table to capture bearing - the hole is deep so the bolt head is in it and attached to t-nut [deeper than it looks in pic - use 3/4 board or 1/2 like this one and be careful]
6) NOT DEPICTED - hold board down
7) NOT DEPICTED - raise head
8) NOT DEPICTED - tighten t-nutted bolt - note board still able to be "rotated" cause the bolt/t-nut tightens in the slot withOUT clamping the board
...... - probably should have used a glued guided insert in hole for the bolt too [in case wood deforms over time]
9) NOT DEPICTED - move head to 2" over in pic [up & left in pic 1 - for me X &Y both] and zero axis [machine G54] - X0Y0 - does not have to be dropped to table or Z Zeroed
10) NOT DEPICTED - run head UP in pic 1 or my X-axis until near end of board and drop to table so bearing even or so near board edge
11) PIC 5 - rotate and slam edge of board against bearing [I mean rotate board against bearing]
12) NOT DEPICTED - clamp board edges with traditional clamps

Now I can place a block of to cut material into the L and clamp it secondly, knowing in MeshCAM if I set my axis for generating g-code to edge of board "0-0" that I can remove and replace and run program without change of zero. The larger than 1/2" hole in the "XY crease" of the board is there cause I can't cut an inside corner with a round bit and it protects any corners of wood placed in the L. Also good indication of straightness between X-Y - board should fit tight. I had one that I could see thin paper size air along a partial edge - told me the drum sanding was not perfect.

This had a special place for me fast duplicating a few items with pre-set [planed and drum sanded] dimensions without resetting my axis positioning for each board. Yes, there are other ways and I adopter edge finders and zero plates, but for some applications this is down, dirty and fast within less than 0.001" which meant no after finishing - close step overs and depths set so no sanding. When I needed fast duplication, but can not afford to set table up complicated or leave jigs in place permanently, this worked for me. The short t-nutted bolts work great in the table for edge alignment for my x-axis, cause they only stick up maybe 5mm - but they get removed a lot - there is actually on in the table in the fourth picture from the left, just in the shadow of the L board - look closely. As you can see - I cut channels later in my "spoilboard" I can place a 1/8" square keyrod into so I now can put the rod in and put things against the rod to align X. That worked too. Different approaches do seem to work - depending on perhaps nothing except how I feel at the time or how much beer was consumed. Best thing is none of them did not work.

Again, this is good for "tolerance allows", and if tolerance does not allow then edge finding comes into play. I have converted a few templates to edge finding, some work get edge found on it's own and then clamped, works great too. Whatever works, works.

Always willing to try something new. Zane - please proof read this one for me................ 8-)
Attachments
f01.jpg
f01.jpg (111.07 KiB) Viewed 220 times
I am not aligned with nor do I have any relationship with MeshCAM or its staff other than being a user [that sounds a little like I have a dependency problem]
larynx
 
Posts: 64
Joined: Thu Oct 29, 2015 2:56 am
Location: East Central Florida

Re: How to handle LARGE project file? (74"x10")

Postby cnczane » Thu Jul 27, 2017 11:19 pm

@larynx

I. get. it. I actually get it!

What I get may not be what you mean, but... ;)

I understand that you have a Perfect Right-Angle "ell" with corner relief (good idea, among a number of them) and a countersunk hole that can take both a) a tight-fitting anchor/pivot bolt AND b) at least part of an overhead circular bearing on the end of the cutter head.

I understand that you put the bolt in the countersink and slide its nut loosely into the t-slot to the rough location to where you want X,Y=0,0 to be.

I understand that you move your cutter head (and the overhead bearing) to the exact location you want for X,Y=0,0, and then simultaneously raise the countersunk hole to meet the bearing being lowered, until the bearing eliminates both X- and Y- movement of the ell, but still freely allows rotation.

I understand why you said to be careful because there should be no clearance for the bearing in the hole, and there's a danger of cramming either the bolt or the bearing or the hole.

I understand that when you have effected this rotation-only status for the ell, you raise the cutter and move it out of the way, giving you access to the bolt head, which you tighten, locking the X and Y position of the ell, but leaving its rotation free, because your bolt and pocket are chosen with this in mind.

I understand that the only thing left is to fix the rotation and you move the cutterhead/bearing to the perfect location where, when the ell touches the bearing, wham-o!, perfect squareness results.

And I must say, THAT's a tool I'd buy... Of course, if I bought one, I might find out that my X-axis really is not square to my Y-axis... But as you point out, if you cut your own ell, the axes, as they are, are recorded in the ell, and all future alignments will get the axes at least aligned to the original.

Very, well done! I'm glad I asked for, and you provided, more detail. Taking it a hair further, you could, I suppose, drill a 1/2" hole at one fixed location in your table, epoxy a pin in there to drop your ell over any time you wanted, and then do the rotation thing to have a square.
If you have not received a reply from me in over a year, I am not ignoing you: more likely I am fallen asleep under a tree. Again. Please poke me if you think it worth your trouble.
cnczane
 
Posts: 277
Joined: Sun Sep 21, 2008 4:15 am

Next

Return to General

cron