PDA

View Full Version : Precision Prototyping



Joe Pelonio
12-16-2008, 5:48 PM
I have a customer that sent me a nice DXF file for some part prototypes to be laser cut from 1/16" ABS. I did a few sample first and he found them to be off by 60/100ths which is not acceptable for his application. I don't have a digital caliper, and don't want to drive all the way to Harbor Freight or elsewhere to buy one today, so I printed it from the DXF actual size and laid a laser cut part over it. Turns out that 60/1000s is not much to the eye. I'll do some experimenting by contouring slightly until it fits on the paper exactly but thought maybe someone has a better idea? Is the beam width too variable for this kind of precision? In 4 years I never had a job with such a tight tolerance before.

James Stokes
12-16-2008, 6:04 PM
with a laser and acrylic I think you are going to have a hard time getting any closer. If the tolerances have to be that close I think I would let it go by.

Margaret Turco
12-16-2008, 6:04 PM
Hi Joe,
That's what I want to be doing! But ABS causes problems because some of it burns off when you laser. I would off-set .045 outside the entire part to start with. Even if you had calipers, such thick material will measure differently from the top to the bottom of the cut. I used to have to make parts +/- .006", but from Nomex, not ABS.

Can you talk him into using another material like polypro?

Bob Tate
12-16-2008, 6:07 PM
I am assuming you are imaging from Corel. If you are using another app, this may still apply. I normally image from Adobe Illustrator. Even a hairline stroke, which normally prints to the center of the line, changes your dimension. My brother owns a machine shop. He has asked me to make parts on occasion. He has always remarked about the accuracy of the parts I give him. I have found it necessary to use the Illustrator feature that allows the stroke to be set to the outside. It is kind of like a tool path those CNC guys have to think about. It worked for me.
Bob

Joe Pelonio
12-16-2008, 10:21 PM
Yes, Bob, I'm importing his file into Corel.

If the roads are not too bad Thursday he'll bring me a digital caliper to use
and try to get it better before running the 200.

David Darnell
12-16-2008, 11:16 PM
Ummm, do you mean .060 inch?

Andrey Anfimov
12-17-2008, 4:52 AM
Joe,
you may use Autocad. There is one useful function - OFFSET (Menu Modify-Offset) . You can draw new shape around old one and you can specify offset very easy and very accurate.

Martin Boekers
12-17-2008, 11:15 AM
question? will the size vary by how hot the laser is (power settings), the speed, frequency? Just how acuurate is the focusing method as well as the conical shape of the beam. I imaging a smaller focal length lens may help.

It seems to me that there are a few variables that might be dificult to control to get machine grade tolerences.

I haven't done much that critical, just sizing to press fit coins into acrylic, which is, if it's tight it's good.

Marty

john welsh
12-17-2008, 11:46 AM
from what im reading it doesn't matter how many times you print the pattern out the laser cut piece will never meet the tolerances you require?

printers always print slightly skewed , under or over sized . this is also the reason some basic laser camera system find it difficult to locate and cut pre printed images etc .

if you print direct from the dxf file into coral to the laser there should be no variations and the laser beam is vastly more accurate than any other form of cut.
for example routing.

i hope the info helps in someway

Joe Pelonio
12-17-2008, 11:51 AM
It would seem that once I get the artwork adjusted they should all cut the same, or at least close enough. I agree that the paper version may not be that accurate either, hence the caliper. Also, he's dealing in metric and I in inches and I hate to change over just for one job. The digital caliper does either with the push of a button. I'm probably going to use Corel to do a contour of .03 and see what happens.

Doug Griffith
12-17-2008, 11:51 AM
I've had a lot of success vectoring parts well within .010" tolerances. In the machining world, that is a very large number. It is the angle of the kerf that always causes me problems.

The .060" you mention is just under 1/32" which is .0625". Your laser should easily get tighter tolerances than that. I offset the line .005" to compensate for the diameter of the beam which is around .010". I then run a part, measure it, adjust the pre-offset file, add the offset, and vector again.

Another thing to consider is that you may need to scale the file disproportionately to compensate for your laser's X and X axis cutting differently.

Cheers,
Doug

Albert Nix
12-17-2008, 11:56 AM
Joe I run into the same problem cuting out templates for moulding blades.
First "so far" my customer has never emailed a file to me that was a perfect match to the part. It is always off somewhere. Now they deliver a drawing or a piece of the molding. Then I scan it to laser master. Even then to get with in 003 or 004 I have to allow about 006 to 008 on the horizonal plain and 005 on the vertical. After doing that I cut one from a piece of scrap 1/8 plex then measure it up make minor adjustments if I have to then cut the good part. The matieral and laser time isnt much but the time can be ruff. Most off mine have not had to be super close yet. The cabnet folks are the ones that want them perfect.

Doug Griffith
12-17-2008, 12:02 PM
the laser beam is vastly more accurate than any other form of cut. for example routing.

I have to disagree with this. CNC routers can be very accurate and hold machine tolerances. The shear nature of using a laser to melt away material with a conical beam is not as "predictable" as physically removing material using a dead straight cutter. It's all about a quality machine with good fixturing. "any other form of cut" is also too broad. Wire EDM can be accurate to .00005".

Cheers,
Doug

Dave Johnson29
12-17-2008, 12:49 PM
The .060" you mention is just under 1/32" which is .0625".


Hi Doug,

I think you mean just under 1/16" as 1/32" is 0.03125"

Doug Griffith
12-17-2008, 12:56 PM
Hi Doug,

I think you mean just under 1/16" as 1/32" is 0.03125"


Whoops!!! Thanks for the correction. I thought it felt wrong. 1/16" is a HUGE number.

Cheers,
Doug

Pete Hunter
12-17-2008, 1:07 PM
We only cut polypro. With our laser. I have found that is consistently .030 small due to the laser beam width and the amount of material that has melted in the process. I still always run a sample and check it to see if there are any adjustments needed for the size.

Rodne Gold
12-17-2008, 1:26 PM
Doug , you 100% right re the CnC thing , but the bit diameter is the limiting factor , you cannot get inside angles of cut outs with less than 1/2 the bit diameter radius (fillet)
The offset command in Corel is called contour , but I find ACADS much better , an offset for tool width is not always possible , try offsetting to the inside of a heart shape with a tight crack tween it's top butt cheeks....

Doug Griffith
12-17-2008, 1:56 PM
try offsetting to the inside of a heart shape with a tight crack tween it's top butt cheeks....

Funny and true!

Besides not having to build complex fixtures to hold material down, the laser's ability to get deep in those butt cheeks and make almost sharp corners is one of it's greatest abilities. These 2 abilities are a great selling point for semi-precision jobs. I've seen companies spend many, many hours building fixtures etc... just to have to scrap the whole thing because of a design change to the part. I always say "you wouldn't have this problem if you used a laser in the first place". For acrylic jobs, there also is also no need to clean up or flame polish.

One thing I find myself doing, if the job can't be vector cut, is cutting fixture components for CNC routers. Sometimes it is a lot faster and the sharp corners come in handy.

Cheers,
Doug

Brian Robison
12-17-2008, 4:29 PM
Joe,
I think with a 2" lens I get a kerf of about .009".
Try to just increase with the percentage tool to compensate for the kerf until you hit the magic size.
I think I've only had to increase to about 102% or so.

Joe Pelonio
12-17-2008, 5:52 PM
Thanks, Brian, but the placement of shapes within the piece are also important, and in this case, the outer perimeter (rectangle) is too small by .06, while one inner cut, a smaller rectangle is too large by .006, and the 3 small circles are fine as is. I figured contour is the only way to fix it while maintaining the positioning.

Albert Nix
12-18-2008, 6:35 AM
I need to try the contour in corel one time. Using the percent route you will more offset on a 2" box than a 1" box so you would be constantly changing the value.

Scott Shepherd
12-18-2008, 8:49 AM
I must be missing something, I make precision parts all the time. I have a customer who I do nothing for but that and the quantities are decent sizes for the same thing. I've run table after table of things and measured them and they were all within a couple of thousandths.

It's all been acrylics or engraving plastics. I have done some ABS in 3/16" and it worked well.

Joe Hayes
12-18-2008, 9:08 AM
We only cut polypro. With our laser. I have found that is consistently .030 small due to the laser beam width and the amount of material that has melted in the process. I still always run a sample and check it to see if there are any adjustments needed for the size.

What is polypro? Where do you get it? I thought I had heard about every product but that is a new one on me.

Brian Robison
12-18-2008, 9:18 AM
Joe H., it's polypropylene sheets.
Scott, are you using a 2" lens?

Conrad Fiore
12-18-2008, 10:51 AM
Joe,
I am a mechanical engraver and also do light machining. We have CNC engraving machines that we use to profile from DXF files created in our cad program. Might I suggest that if ALL your samples are off exactly the same amount, you should have your customer revise his DXF drawing to allow for what you are off as measured by your customer. That should correct your offset. Do another set of samples and have your customer check again and if needed make any minor corrections to the file again. If you get consistant measurements in ALL your parts, this will correct your problem.

Kenneth Hertzog
12-18-2008, 12:23 PM
Joe
I use a cad program quite a bit. I also use the cad program to cut parts on the laser. My laser will vector cut from a cad program. when you copy and paste within the cad program you can enlarge the copy the .060 needed to make your piece bigger.

ken

Joe Pelonio
12-18-2008, 12:37 PM
Thanks Conrad and Kenneth. I have e-mailed and asked him to do that, make the adjustments and then tomorrow, if the roads permit, he'll bring the
caliper and I'll have new samples to measure and adjust if needed while he's here.

Scott Shepherd
12-18-2008, 1:14 PM
Scott, are you using a 2" lens?

Yes, 2.0 lens.

Rodne Gold
12-18-2008, 2:09 PM
Enlargement will not work for inside holes if the precision departure is due to kerf width , it will enlarge inside holes and make them even more imprecise. If there are no inside cutouts and you only have an outside cut , enlargement will work , however enlargement and offsetting or countouring are not the same , even if it is only an outside cut. A contour or offset is a line or curve passing thru a set of points uniformly distant from a original , and will not have the same shape as an enlarged original in most cases (unless the item is a very simple shape) , the offset is the better way to go.
Use 2 layers , offset on a non original and keep the original as a master , offsets can change with materials and circumstances.
You might find in corel that the item you want to contour is not "joined" and you can use smart fill to generate a copy of it that IS joined and will offset. Corel is disgusting at contouring and offsetting , you often get spikes all over the place and very wierd results...thats why i like AutoCads better...
On our CnC machines we do not mess with the drawing , the drivers we use have tool diameter offsets as a function of the driver and have routines where you can use a much smaller or v shaped bit for inside corner sharpening.
I must say , considering a bad spot size would be 10 /1000's , a 60 thou departure does not sound like a kerf width problem alone. 60 thou is 1.5 mm and thats a LOT of error.......
As any rate , if you are doing a lot of this work on thin stuff , a 1.5" lens will work much better than stock , both with kerf(cut) width and power density leading to less heat affected zones , combine that with the right air assist pressure and direction and you will get very much more precise cuts. Check your focus of the lens and machine you using , often Autofocus isnt spot on and neither are the measuring tools...

Richard Rumancik
12-18-2008, 5:09 PM
Joe

I can cut parts to +/- .005 and better with some effort and +/- .010 without being too fussy. I do some wood craft kits of .060" basswood and hold within a few thousandths. But you have to be systematic.

You will definitely need to compensate for kerf width. One way is to cut some lines into the ABS at required speed/power and measure the kerf width with a feeler gage. You can get a feeler gage for $5.00. If you measure a .006" wide kerf, offset the line in the Corel file by .003" all around.

I would be reluctant to ask my supplier to fudge the drawing for me. Assuming they supplied an accurate drawing, I consider it my problem to make any adjustments necessary to make the part.

The second thing you should do is calibrate the laser for travel in the x and y directions. I have explained how to do this in other threads. Basically, take a sheet of poster paper and scribe some lines say 22" apart in x axis and 16" apart in the y axis with the laser. (Choose numbers to suit your machine.) Then measure the distance between the lines. I use a precision 24" steel rule and an optical loupe. A tape measure won't work for this - you will be wasting your time. The rule cost me $40.

If you are very lucky, your laser will be dead on. My laser is out in the x axis and bang-on in the y axis.

I determine the error as a percent and create a "compensated" CorelDraw file with a new name. This file is for production only - do not do drawing revisions.

In my case I usually stretch the file 100.22% in the x-axis ad leave the y axis at 100% (non-proportional stretch). This is not a nice method, and requires good "bookkeeping", but the only thing that works for me. There is a way of tweaking in the driver to calibrate, but that method is useless for precision work. I wish I could enter the correction factor into the firmware as it seems to be fairly stable . . .

The larger the rectangle you can draw (and measure) the better your correction factor will be. Using paper drawings won't work very well. A pen plotter or laser printer will have errors too.

If you want to try a quick-and-dirty method, here it is. Find your kerf width, draw a rectangle 7.500" x 7.500". Correct your drawing for kerf by offsetting 1/2kerf, so it might then be 7.503 x 7.503". Cut a part and measure it. (Make sure you mark x and y axes.) Find the % error using an 8" digital caliper. Stretch the CorelDraw file as required. It might not be the same in each axis.

If all of this seems like too much, then pass on the job. Once you have done it a few times it is not that intimidating.

One other thing - when you get an undimensioned drawing or dxf file from a supplier, ask them to draw a bounding box around the shape (and supply dimensions) so you can check that the file has imported correctly.

Just so we all speak the same language:

In the machining world, precise & precision refers to repeatability. You can have a machine that is precise and makes bad parts (repeatable and always wrong.)

accurate & accuracy means obtaining the intended dimension.

If your machine is precise, then you need to figure out how to make it accurate so that the parts are to spec.

I realize that in everyday language, these terms are not always used correctly.

Joe Pelonio
01-23-2009, 12:54 PM
We eventually managed to get the file correct for fitting, and this job has turned into production work, as they ordered a thousand+. I have a couple of the completed items so I could see what they turn into (yellow things).

All of the ABS has a warp to it, probably came off a roll. After a lot of thought and experimentation, I came up with the idea of a hole in the center, and wiring it to the vector grid, which has worked out nicely. There is some
scorching on top so I apply transfer tape before cutting so the sandblasted finish stays nice.

Dave Johnson29
01-23-2009, 1:16 PM
All of the ABS has a warp to it, probably came off a roll.


Joe, I think you will find it is extruded sheet with rolled texture on top.

Joe Pelonio
01-23-2009, 1:22 PM
Joe, I think you will find it is extruded sheet with rolled texture on top.
I don't know how it's made, but every 12"x12" piece has the same domed warp, about 3/8" higher in the center. I didn't think it was actually sandblasted, it's just what they call the finish.

Doug Griffith
01-23-2009, 3:28 PM
Hi Joe,
since the parts are rectangular with rounded corners, you could probably shave off a bunch of time and get better yield if you kissed the parts together. I would first cut all the radiused corners as small squares with the sides curved inwards (plus the holes). Then run a second pass with just straight lines in a back and forth grid pattern. It makes sense to optimize the cut paths for large quantities. Usually I run it as you have it because it takes just a few minutes to set up. Then, while the lasers running, I optimize the file for future runs.

Cheers,
Doug

Doug Griffith
01-23-2009, 3:40 PM
Joe, I think you will find it is extruded sheet with rolled texture on top.

From the photo, the texture looks like what is called "haircell". It is the most common texture applied to ABS as it is extruded. The warping seems odd to me. I've seen literally millions of pounds of extruded ABS and it is pretty darn flat. It could be the sheets were left in the sun or possibly put in a drying oven at some point. I doubt it though.

Joe Pelonio
01-23-2009, 4:38 PM
Hi Joe,
since the parts are rectangular with rounded corners, you could probably shave off a bunch of time and get better yield if you kissed the parts together. I would first cut all the radiused corners as small squares with the sides curved inwards (plus the holes). Then run a second pass with just straight lines in a back and forth grid pattern. It makes sense to optimize the cut paths for large quantities. Usually I run it as you have it because it takes just a few minutes to set up. Then, while the lasers running, I optimize the file for future runs.

Cheers,
Doug
I kept them far apart because they tend to flare up some even with air assist if there's a cut close by to let air in.

Optimizing would save time but then it's only about 20 minutes/sheet now and gives me time to remove the transfer tape and pack them up, and work on other things before I have to change material.