I am interested in one (probably the 240), but I don't think Ian plans to do any further procurement. Seeed Studio is apparently planning to carry them, but I have not seen them show up there yet, nor do I know what kind of pricing they will have. If they offer them a a decent price (and don't gouge like some aliexpress prices I have seen), and their shipping is reasonable, Seeed will be a good trusted source. But then again the machines are still not available there yet.
I just looked at my solderpaste gerber in gcprevue, and there is no board outline from the dimensions layer. I don't know what might be different about your itead cam export, but it must be different than mine. I just use the standard eagle CAM-processor (I get my boards at myro, but gerbers are gerbers). I also use an older eagle version 4.16 on XP, as years ago we paid a shit-load of money for the full pro version (as we needed big boards with 6+ layers), and with the autorouter (which we have only used once), and have really never needed to spend another thousand bucks to upgrade it (except for a few more-recent libraries we can't open). To export gerbers: File/CAM-processor File/Open/Job (select our 6/4/2-layer.cam file, which just defines solderpaste-top as a .SPT, RS274X...) Process-Job...
To shrink the solderpaste (which I think is important for the cameo cuts), open your board in eagle, click on DRC, click the MASKS tab -- it has settings for STOP (solderMASK), and CREAM (SolderPASTE) layers. They do the opposite: --- when you select a positive number for the STOP layer, it makes the mask wider, so your soldermask has some clearance around your pads to compensate for soldermask registration. --- when you select a positive number for the CREAM layer, it makes the mask narrower, so you can compensate for stencil registration, and also the spreading of the gooey, oozing, disgusting, solderpaste, by shrinking it a bit. I have STOP set to 5mil-min/100%/5-mil-max (so it is always 5mil wider). I have CREAM set to 2mil-min/100%/2-mil-max (so it is always 2mil narrower).
If you have tight holes blowing out, it might be helped by shrinking the cream layer. However, even if the paste spreads across all of the pins in one giant rectangle, it will likely reflow quite nicely with no shorts.
As jesuscf pointed out, not all transparency film cuts the same. He had good luck with Apollo write-on stuff (and bad luck with Canaon) -- I am using Apollo Inkjet stuff, but have tried 3M and Highland (also 3M) sheets, and they were all much the same for me (the Apollo probably a tad better). I can't comment on current prices as all three boxes of this stuff has been sitting in the closet since the late 90s, when we actually gave presentations with overhead projectors :)
You can look at your solderpaste gerber in gcprevue before you cut it, but note that editing the DIMENSIONS layer in your eagle board file will have NO effect on your cameo cut. Gerber2graphtec only looks at the outer BOUNDS of ACTUAL CUTS (see the generated pdf), and adds 1 inch extra on all sides to trim it out of the sheet (you can change this trim number in the py file).
Unless your board is large, I don't know why you care where the stencil is "centered" on your sheet of cheap throw-away mylar. As long as your board fits in the 8.5x11 inch sheet, you will be happy to have it cut for you, and you will happily throw the scrap away.
As for the incredibly-long time it takes, if you watch the knife, it will very often move to the upper-right corner for what Peter calls a "training" cut -- this orients the blade (which rotates freely in the holder), to get in the correct position for the next cut or two. This is, I think, the key to the program's cut accuracy. When I cut small rectangles with sil-studio, I could see the effect of the knife rotating as the cut transitioned 90-degrees, and it resulted in totally unusable stencils. Peter's program may be slow, but it works very, very well.
Previously, I cut a small stencil that took a half-hour. The 4.5x4.75-inch one of which I just posted pictures, took two hours. If you have a little board and it is going for four hours, there may be something wrong. It does look like the board should be cut twice though, so if you see it doing the same cuts twice I think that is normal. If it is cutting them more than once, well, i dunno...
To check you blade depth, you can use the blue cursor buttons to move the head to an empty area, and tell it to do a test cut. Then peel out the little section with an xacto to see if the cut looks clean. This is actually two laminated sheets of the printable vinyl, about 8-mils thick total. I used a blade depth of 3. [attachment=3]
Tell it to find the registration marks (manual worked for me, but auto got confused), then tell it to cut. It peels off nicely. Registration was slightly off -- next time I will make a second jpg with no cut lines, and bring it in after building the cut layer. That way there will be no small black lines visible if reg is off a bit (and reg of the text is not that critical anyway). [attachment=2]
And stick it on the chassis. [attachment=1]
Not bad at all. Cost me about a buck for one. Smaller ones are dirt cheap. Doing a plastic panel for a pactec box? Just drill out rough openings in the blank panel and stick a nice overlay on top. Custom label for the recess in the top of your enclosure? No problem. Use your imagination. [attachment=0]
After finally getting stencils working with my Cameo cutter, I wanted to try making chassis overlays. Over the years I have spent good money getting polycarbonate overlays (7 to 10-mil) with rear silkscreen for connector identification and such, a flood-fill color, backed with 3M 467 adhesive, and die-cut with all the openings and outer shape. It's a great way to allow multiple overlays to work for different products using a single chassis with a flexible common hole pattern, but pricey.
So I have this Cameo that can cut up to 24x12, and an inkjet that can print up to 11x17, and I have found inkjet-printable adhesive-backed plastic sheets in up to 11x17 -- can I make my own overlays?
The Cameo's included Silhouette-Studio program let me get the job done. First off, Silhouette-Studio has lots of oddities, but it also provides some important "drawing" tools (for drawing your cut lines, that is) -- not just the basics like line, polygon, rectangle/square, and ellipse/circle, but the ability to copy-and-paste-in-front (exactly in front, not offset a bit), and then move the object precisely. I use this a lot for arrays of connector holes. But besides just letting you define a cut pattern, they have a feature called "Print-n-Cut." Now this was designed so crafters could print a pretty picture of a unicorn or something, print it from sil-studio to a printer (maybe on a thin cardstock paper), ADDING REGISTRATION MARKS, then define a cut layer (which for many projects is just an outside edge), then load the printed item on the mat, and load it into the Cameo. But you are not limited to just an outside edge. In fact, I found a lot of railroad model makers printing and cutting intricate window frames and stuff.
Here's the important part: the Cameo has an optical sensor near the knife, and it can find the registration marks on the thing you just printed/loaded, and get precisely-oriented for its cut pass. You then tell it to cut, and bada-boom-bada-bing, it makes nice.
The Print-n-Cut adds a step to the process of creating your cut file -- first, you open a jpeg (or other bitmap) file. Then you add cutlines as needed. You add registration marks, print it out, and so on as previously described. So for my bitmap file, I wanted the text for my connector definitions, and thin lines where my cuts were to go. My artwork came from Adobe Illustrator, but I could not export a jpg with decent resolution. I could, however, open the .ai file in Photoshop, and save-as a jpg. I used 600 dpi (lower-res was not as clean, and higher res made sil-studio hang). I had all of my text plus hairlines identifying where to cut. This worked great for an 8.5x11 sheet but I also wanted to do 11x17 -- this also made sil-studio hang on file open (probably a bug, since their workspace is 24x12), but I could fake it out by dividing the image in two in photoshop, opening them both in sil-studio, copy one, paste into the other, align visually, group, and then I had my 11x17 image ready to go. Clumsy, but whatever gets the job done.
I opened my image, added reg marks (you need to avoid printing in keep-out areas), and started adding the cut lines for the hole pattern and the perimeter of the overlay. I put as many overlays on a single sheet as I could fit. After the hole pattern for one is done, select the whole group, copy, paste-in-front, move, repeat -- it was pretty easy. There were a few annoying things along the way but I am happy with the results.
The inkjet material I used I found at texascraft.com -- WVF1117KH white waterproof adhesive-backed vinyl 11x17 sheets, $1.44 ea in 100s (but they also sell 10 packs or single sheets). It is also available in 8.5x11. They also have laser vinyl, but my laser does not do 11x17, nor does it do color, and I want to try color overlays next. This stuff is about 4-mils thick including adhesive, which is a bit flimsy, so after printing a sheet, I stuck it onto a second sheet for added thickness, before cutting. Enough talking:
In Sil-Studio, the black is the jpg text, and the red is the cutlines I added. [attachment=3]
Some settings. [attachment=2]
Printing it on Big-Brother. [attachment=1]
Loading it the Cameo (on the optional 24x12 cutting mat) [attachment=0]
Some of the paste smeared -- no big whoop -- it'll reflow nicely even if the pins all look shorted right now. You are more concerned with just having enough of it more or less. Some of those SOIC pads look a little long? I made a hybrid part so I can use normal 150-mil parts, or the wider 209-mil SOP/EIJ parts. Likewise, my MAX202 is able to use the 150-mil narrow or 300-mil wide parts. Why have procurement problems in the future? [attachment=3]
Tweezers for the small bits (an old fart like me will put on two pairs of reading glasses). [attachment=2]
Little suction thingy for chips. [attachment=1]
Time to fry this puppy up on my Tilt-n-Drain Griddle from Walmart. Turned out great! [attachment=0]
Make a "nest" with old boards or whatever, align stencil to your board, and tape it down so it hinges. [attachment=5]
Ready to go -- drop a board into the nest. [attachment=4]
Stencil drops exactly in place. [attachment=3]
This solder paste expired a year ago, but after taking it out of the solder fridge (don't put this in with food), letting it warm up for a couple of days, and giving it a good stir, it worked just fine. A bit thick, and I would not even guess what could be used as a solvent to thin it -- it got the job done. This is a $40 250g tub from stencilsunlimited.com, which has an air-exclusion inner cup -- get the leaded no-clean stuff (water-soluble flux needs to be cleaned, and I am too lazy to run it through the dishwasher and get out the air compressor). Lead-free needs a higher reflow temp, and unless you have a good oven (and are using a high-temp laminate like HR370 as well), you will have more problems. Using lead-free on cheap proto FR4 and cranking up the heat may leave you with a board that is partly de-laminated. Just use lead, even though it is getting harder to find. The tiny 30-ish gram leaded Mechanics paste that is popular is out of stock at sparkfun -- it was cheaper at dx.com anyway, and dx has a replacement or two. There are lots of other places. [attachment=2]
A little paint scaper works for me as a solderpaste squeegee. [attachment=1]
I can't help you with the error you are getting -- I just installed the packages with whatever defaults they requested, in whatever directories they desired, and made the minimal tweaks that jesuscf indicated. I try to understand and learn as little as possible about windows; I have hated windows since 3.1. But I can't do everything on macs, so I have win machines around too. One of my win7 machines does not update win-explorer windows -- I can delete a file, but it is still there until I do a view/refresh -- I can change a filename, but need to do a view/refresh. So fricking annoying. Google shows me that many folks have this issue, going back over two years. Hmm, the filesystem is like computer-design-101 guys. Also, why would microsoft ever, ever, ever, ever, think that when I unzip a file, that I would want all files set to today's date -- gotta "unblock" first, whatever that means. Stupid stuff. But I digress.
First, I got gerber2graphtec cutting a little test file (emphasis on little for your first one, like maybe one chip and one resistor). Once I got it running, I just cleaned it up with a top-level directory in which to copy my gerber, and the little batch file that kept it simple. I put the py stuff in a subdir, and tweaked it to keep the intermediate files. Probably only want to keep the pdf.
>>> if you look at the pdf, you see that it only extends to the edges of the furthest cuts -- YOUR BOARD OUTLINE IS NOT INVOLVED. Peter Monta, who wrote gerber2graphtec (google it), told me that the default edge around the cuts is one inch, but it could be changed in his py file (it's pretty easy to see it). I was going to increase it a bit, but then you start to run out of space on your mylar sheet when the board gets bigger.
>>> speaking of the mylar sheet, put a full one in, not a little scrap -- it's cheap stuff. As you found, it does not cut from the upper-left corner as you might think. In fact, your paste file will be rotated 180-degrees, and the 0,0 origin will be in the upper-right. This fact, plus the 1-inch border, probably explains why your stencil did not cut where you expected it.
>>> as for cutting depth, using this office-supply-store overhead transparency film (which I measured to be about 3.5 mil thick), I thought I was initially cutting with a blade depth of 1, but I might have used 2 on my last one. I need to run a couple of tests to be sure. If the depth is too shallow, it may not cut all the way through (though note that gerber2graphtec seems to cut the entire board twice, for good luck). If the blade depth is too deep, you cut into your mat more, and they say that a too-deep blade can get the tip snapped off. You cannot change speed or anything else without digging into peter's code. I end up with a few holes that need the little scrap poked out with an exacto and scraped off, but it's a minor thing.
I will post some pics of my last stencil here shortly.
If you followed the instructions that jesuscf provided previously, you should be able to get it running. I have it on a win7 latptop and it works well. A couple of things:
When you open up a command window, you will probably not be in the directory with your gerber2graphtec executable. So you need to cd to the proper directory first (it say not recognized since it won't be in your path, but will run fine from its directory).
The > character shows up in a couple of places, first as the prompt just before the cursor, as in C:windows-sucks>
Also, it is used to "redirect" output from stdout (the screen) to a file or to be the input to another program, as in DIR > pud.txt
The > characters in the example have nothing to do with directory slashes.
Here is what I did: I have a folder C:gerber-files-to-cut and in there I have another folder, gerber2graphtec, which has the py stuff. I copy the file I wish to cut (xyz.SPT -- my eagle cam is set to use .spt for solderpaste-top) into this folder. I created a simple batch file which I just double-click and then type in xyz.spt and hit return -- then I go have lunch (my last stencil took 2 hours to cut, but it was perfect). I also set eagle to shrink the paste layer by 2-mils and it all worked great. Fryed up a new proto pcb on my cheap walmart skillet with paste that expired a year ago -- worked great. Here is my batch file:
@echo off rem This is for printer named "Cameo" which must be named and shared in control panel. rem This is for computer named "i4004" echo. echo This script will send gerber file to Cameo for cutting solder stencil. echo Generates intermediate files .pdf, .pic, and .graphtec.txt echo. echo Set your Cameo blade to 1 and turn cutter on. echo Load mylar sheet (landscape) at upper-left corner of mat. echo. set /p filename= Enter filename of gerber (eg: test.spt): gerber2graphtecgerber2graphtec.py %filename% > %filename%.graphtec.txt copy /B %filename%.graphtec.txt \i4004Cameo
I am seriously considering getting this machine (probably the 240; maybe the 220), and have been keenly following this discussion. It has been interesting to see the problems/solutions that you brave early-adopters have run into, and I thank you for all the great insight you have given to us guys lurking in the wings.
I realize that not all of the problems/issues have been overcome, but would you folks who have and use the machine recommend it (for r&d just-a-few-off or low-vol-dozens-a-day use) ?
And a couple of other questions, if I may:
- can a 24mm reel be used (with offsets to accommodate it)? There are quite a few chips on 24mm reels.
- how are the wells at the front used for loading chips? Can you program different parts in different wells (eg: can I have a TQFP-64 in wells 1-5, and an SSOP-28 in wells 6-x?) Do you keep adding chips to the wells as the machine consumes them? Do you just push them into one corner and hope for the best alignment?
- has anyone run into issues with part height? Yeah, most res/cap/ic stuff is a similar height, but what about tall electrolytic caps (of varying heights)? Anyone try smt headers?
Nice cuts. We need to compile a list of good vs. bad stock to use.
What was your blade setting? I just used the min of 1 since gerber2graphtec cuts the whole thing twice, and I heard that if your blade is deeper than it needs to be, you run the risk of snapping of a bit of the tip (ruining the blade -- not sure if that is true, as the cutting mat is soft, but I'll play nice).
Were the two stocks the same thickness? If so then one polymer must just be tougher than the other.
I have three brands of transparency stock, but I won't be back in the office until Monday to get you actual product numbers. It may not be too relevant since they are probably circa 1995 or so, when I actually used an overhead projector for presentations! One of them is a 3M number-number, another is an Apollo somethin-somethin, and I forget the other one. Using a micrometer, they all measured about 4-mil thick. One was probably for a laser, and one for an inkjet... I tried cuts on all three and they seemed the same.
They don't prominently tell you what the specific material is for transparencies, but polyester (mylar) seems to be typical. I am not a plastics expert by any means -- in the past, for die-cut overlays, I have spec'd polycarbonate 0.010 material with 3M 467 adhesive, and for switch overlays (which would lead to cracks after a while in polycarbonate), I have spec'd polyester of probably 0.007 thickness. That is the extent of my professional plastics knowledge. Other than that my only experience in plastics is with PVC for plumbing and irrigation lines, and ABS for the poopy stuff :)
I have not tried Canon-brand transparency stock. You might just try another brand from the office supply store.
There is also some slightly-thicker (maybe 6-mil?) clear plastic stuff that is used for covers for reports 'n stuff. Also in the office supply store beside the binding machines and combs etc. I have not tried that yet but want to try one (as I have some of that, also left from years ago) just to try a thicker application of solder paste.
Let us know what you find out in materials. I will post the brands/models I have on hand when I get in next week.
Here are some pics from cutting a stencil with gerber2graphtec on the cameo. Compare this to the sh%$ that sil-studio did in my OP.
The only cameo setting is the physical blade depth, which is set to 1 (minimum). The media is polyester (mylar) overhead transparency sheet from any office store. I measured it to be about 4-mils thick. The pics are of the cut sheet resting on a textured black powder-coated chassis.
A pic of resistors: 0402 (upper-left), 0805 (lower-left), 0603 (upper-right), 1206 (lower-right): (the 0805 has closer-than-norm pads as we use it for solder-jumper also) [attachment=3]
50-mil pitch SOIC and SOT23: (26-mil-wide hole with 24-mil space) [attachment=2]
0.65mm (26-mil) TSSOP (bottom) and 25-mil QSOP (top): (~15-mil-wide hole with ~11-mil space) [attachment=1]
0.5mm (20-mil) hdmi conn and SOT23: (~12-mil-wide hole with ~8-mil space) [attachment=0]
Note that the holes are wider than they should be, but this is actually a stricter test, with thinner between-hole segments. I did not shrink the cream layer before generating gerber from eagle. Some folks have said to bring it in by about 2-mil (I presume that is per-side) which will probably be about right, and I will try that on my next stencil.
But just look at the thin bits of mylar between the 0.5mm pads! They must be only a few mils wide -- very accurate cutting indeed. I am quite pleased.