Stitching photos for aligning engraving on a substrate larger than the bed area ?

[Ian Stewart-Koster]

I'll try and be succinct: is anyone able to offer advice on aligning one pass with another, when engraving something bigger than the laser bed?
Our 1300 x 700 bed laser is all I have at the moment. I've enquired all over SE Qld, and only ONE business has a machine that can engrave the size I need, and they don't like the idea. Plenty can cut big, but not engrave.

The image I need engraved on acrylic is 8x3 ft, 2400 x 930mm, to go in the middle of a 10 x 4 ft sheet, 3050 x 1220mm at 10mm thick.

So in theory I could fit it in 4 passes at full width, plus a bit of room for the head to turn around- 4 x 700mm bed depth = 2800mm and we need 2400 mm.

BUT if I did the first quarter, and clamped blocks to the bed, and then slid the acrylic along for the next quarter, I need to stop it in the right spot to pick up the next lot of engraving from. I've fiddled and a minor overlap is better than a minor empty missing line, but I'm seeking advice on this realignment process from anyone who's actually done it before!

As well as I can guess, I'd need to embed a hairline line in the image, every 700mm or 650mm, in an unimportant part of the margin, so that I can align that up with a test-firing to begin the next section - or am I wrong?

I also have to see if the machine can handle the memory needed to do this, of if each quarter has to be broken into thirds and etched like that...
___________
On our CNC router, I've handled a few jobs way bigger than the bed, but I have origins and marks all down pat and have no problem stepping along. But then vectors are a bit easier than a BMP...

Thanks in advance for any hints!

[Ian Stewart-Koster]

Looks like I'll have to chop it into much more than 4 - and much more than 7 pieces.
It can't handle the memory needed to deal with much more than 1 meg at the most, per section.
I tried sending 1/7 to the laser, and the first seventh was prepped to go OK, but the 2nd gave me an out of memory.
After dropping the scan interval to 0.35, it was happy, and would send it, but the image was too coarse for my liking, so I'll look at breaking it up more...

[Dave Sheldrake]

Doing stuff that size even in one hit becomes hit and miss due to expansion/contraction/warping Ian. Especially with acrylic

[Ian Stewart-Koster]

Thanks, Dave - that's where the mass of the 10mm stuff I'm hoping will be solid ebough absorb the heat and expansion at lower power might be negligible?
At any rate, the RAM in the controller, despite being connected by LAN, only likes a certain filesize, so I'll be chopping it up even more.

It's a shame it cannot jst fetch the lines of code as needed, from the PC, like a serial cable can (at least on our router it does fetch it as needed)

[Glen Monaghan]

While you can't send a whole bed's-worth or quarter of the image at once, you can still figure out how much you can do in one shot, such as a fifth of the bed, and send that much, then the next piece, and so on until you have sent and engraved a bed-full, all with no registration problems. Then you have to move the substrate and start over, ideally with no gap and at most a scan or two of overlap. Ideally, you could create a precise set of registration keys, such as precise stop blocks solidly attached or holes through the substrate in a margin area that can be trimmed away later, precisely positioned at intervals equal to the bed size. Then position mating stops or pins to the bed where they can engage the keys on the substrate. After completing each bed-full of engraving, reposition the substrate to engage the next set of keys. If you don't have anyplace you can safely add keys, you could try measuring along the edges of the substrate and marking or taping bed lengths that way, then mark or tape a corresponding reference on the bed itself, and do your best to line up the edge marks to the bed references by eyeball each time. I did this once with a two-part acrylic reverse engraving job and though I had ruined it because I had what looked like about 2 scan lines of overlap, very visible looking down at it on the bed. To my surprise, the overlap was completely invisible when I removed the acrylic and looked at the image through the front side. Can't say you'd be so lucky with overlap because it may depend on the design itself, but I've not experimented further.

[Andrew Holloway]

By SE QLD, I assume you mean Australia? Try Steve at Acrylic Online, he's in Brisbane and has an SP3000, plenty big enough.

[Ian Stewart-Koster]

Thanks, Andrew- I had a few places recomended when I asked in a facebook Aussie laser & signwriting group before Christmas.
Plenty had machines big enough to cut, but all I contacted said they would not engrave at that size - except Laws laser.
They're 'looking in to it', but I figured I'd test the water as well, with what I have here, until the new flat bed we've ordered, arrives. (probably still 4-5 weeks away)
I appreciate the contact!

Glen, the idea of predrilling holes to align in the laser frame is a good one - I could do that on the cnc router easily and accurately enough,
though I was thinking I'd just clamp blocks to the frame, as guides, to slide the acrylic sheet between, and laser an inconspicuous scratch mark at the end of each segment, that I can test align with a test spike of beam blast. I agree, a minor overlap of passes is not obvious when put upright and backlit!

I'll have to undo some sheeting from the shed wall to give the stuff room to come out the back of the laser, as I'd not allowed clearance for a 3 metre job like this when I first put the laser where it is!

[Keith Downing]

Another possibility is to add a layer of masking to the front of the acrylic. You can then "start" your second pass/section at a power low enough it shows up on the masking ONLY which gives you as many tries as you need to get it aligned correctly. Then raise the power back to the correct level, and start your second section after everything is aligned perfectly the way you want it.

I have used this method several times for items where I have to paint fill a section, then continue lasering the piece. But I see no reason it wouldn't work for your purposes, maybe you could even do this in conjunction with Glen's idea.

[Kev Williams]

I align similar to you Keith, except instead of tape I use pieces black laser brass and just fire a couple of dots a fair distance apart, then move the material. If you have a red pointer you can trust (like I do), slew the head over to the dots. The LED will hardy be seen on the black but you'll know it when it hits the gold dot! For more accuracy I use a 2x plastic loupe (from HF) and a bright flashlight aimed at the dots, the bright light tempers down the LED so it's much easier to see exactly where it is-- I can center the light to a gold dot to within .002"... Anyway, when I'm satisfied with the dot/dot alignment, I jot down the xy coordinates. Slew to the next dot, find it, note those coordinates. Now I create guidelines at those coordinates, and place the reference dots exactly on the guideline intersections. With luck, both reference dots will align well to the intersections. However-- this will bring into play any rotational or angular discrepancies with the material or within the machine. "Very close" alignment I've found to be fine usually, but if one point is off by .010" or so, I'll rotate the job or the material and re-align as needed till they match.

That all said, in normal practice when joining text & such, I don't usually create dots, I'll just pick points already engraved, such as the farthest top and bottom right corners of a letter or graphic, and red-dot those points.

Word of advice when doing multi section work, that I found out the hard way: As we all pretty much know, our laser beams don't necessarily fire straight down. Because of this, AVOID IF AT ALL POSSIBLE ROTATING THE SUBTRATE. The reason: issues with light reflection/refraction. Since the beam is likely firing on an angle, the sides of the kerfs will not be angled identically, one side will be steeper than the other. What happened to me was, on a 3' long Cermarked logo I had to 180, I joined the 2 sections perfectly, BUT, the entire left half was darker than the right half-- And when rotated 180°, guess what- the left half was still darker than the right half! It was all to do with the angle the kerfs reflected back ambient light. Plex will do this too, to what degree depends on your laser...

The big logo plate was eventually clearcoated, which helped a ton. Even so, I could still notice the difference... If you must rotate the work, test engrave!

[Ian Stewart-Koster]

Keith- THANKS! That's so obvious I'm surprised I'd not thought of it!
Simple...

Kev, thanks, too. I've checked the beam angle on this machne, and even when cutting through 25mm acrylic, I'm happy with a near-enough-to-90-degrees-on-all sides angle. (It took a lot of tweaking to get that alignment- and write down the steps so I don't forget for next time I want to know which way to turn the knobs!) The dot-on-brass makes sense- better than the line I was thinking of engraving. However I don't have a diode pointer on this one - only one on the little ULS. Rotating the bed & image - I did think about it but discounted it for reasons of micro-misalignment. I only want to have variables of alignment in the Y or up/down direction, not the X as well.

[Ian Stewart-Koster]

P.S. Kev, why was the stainless or cermarked plaque clearcoated? Surely that'd be the place it'd fail - or was it for antigraffitti purposes?

[Kev Williams]

it was clearcoated to eliminate the 2-tone look it had. It was also an indoor sign so weathering wasn't an issue

to test for reflection distortions and such, draw a 1/2" x 2" rectangle and center it vertically on the left half of a 1 x 4" piece of plex, with a 1mm left margin. Color the square 70% black, then engrave it as a photo. Rotate the plex 180 then engrave it again. With luck the 2 halves will look identical other than the 2mm overlap in the middle, which will give you an indication of how precise your seams will need to join up...

[Ian Stewart-Koster]

Thanks, Kev.

Update on the big job... so far we're about an hour and a half from the halfway mark. Touch wood all continues to go well, and I'll get to bed before sunrise!
I broke it up onto 25 skinny BMPs, and then stitched or realigned them in the Goldenlaser software, down to the last pixel, making up 4 separate beds' worth of small files.

Using Photoshop & setting guides, then using the snap to guides option, and maing the canvas 2cm wider, and drawing a 3 pixel wide line in the margin, then cropping, with the snap to guides, means I had half the line in each file, and aligning them was pretty OK once you got the hang of it.

Also I could use the test pulse to check the home position of the head with the last line on the previous quadrant's worth of etches, then tweak the whole lump of acrylic by a nudge or half a hair as required.

I G-clamped guideblocks on the bed, left front, left rear, right front, and onto the table and other obstacles that were in the way, so the sheet could not get slewed sideways while in progress, but could only move parallel to itself in one direction.

The home position is back left of the bed... this is one time the front left would have been a great help, and starting the job from the back of the laser, instead of all in front and going through to the back...

because checking the pulse home location with the last line on the last piece before moving the substrate is really hard when you can't get your body or eyes enywhere near the head. I had to use a mirror on a piece of rod, to look at it up close- and a torch - to check sideways alignment.

The idea mentioned above of using masking tape for a test of the new section is great - but with this laser- again - you can't get near the head at the back of the bed when there's 8 feet or more of substrate poking out the front!

One comment - you need to pause it every 15 minutes to dust the acrylic fluff off the air nozzle-I use a microfibre cloth - and also have a huge fan set up blasting in the front to blow away and cool the residue before it can settle. I also undid the shed walls to give us room to put the sheet as we moved along.

Photos later... I'm really looking forward to reaching the halfway mark!

[Ian Stewart-Koster]

Partial photos:
(13 hours engraving...)

This is what the start of an 8 foot x 3 ft engraving piece looks like, on a 10 x 4 ft sheet...

20180205_164440_Medium.jpg

Moving along...

20180205_164421_Medium.jpg

Close-up of the interesting end- the train is about 2'6" long.


20180206_171234_Medium.jpg

Aligning the next quarter - the pulsed dot lined up pretty well in the lasered alignment line. (or vice versa)

20180206_025217_001_Medium.jpg
And another angle:
20180206_025210_001_Medium.jpg

The finished piece, sitting on an 8x4 piece of MDF, (overhanging each end by a foot) waiting to have the ACM front & back glued on:
20180206_171241_Medium.jpg

I think it was worth the time it took... but jeepers, 13 hours for 2.4metres of length in Y axis- near enough to 3 mm per minute, at a scan gap of 0.167, 144 ppi. 300 mm/s speed, 1500 acceleration, Energy/power 30% in the software.

If I tried to go faster, thre detail was left behind and the width of the machine started to be a hindrance with the stopping and turning around.

The front side of ACM about to be glued on:

20180206_171257_Medium.jpg

Photos of it installed and framed will have to wait till tomorrow- I have some sleep to catch up on!

[Scott Shepherd]

Well done Ian! That's a serious undertaking and looks like you nailed it.