Obtaining Dimensional Accuracy of Prints With Max v2

[Pyropainter]

I am having issues with printed dimensions not measuring as expected. This is what I have done so far. I have created four single wall cylinders. The first 10mm in diameter then consecutively doubling the diameter to 80mm. I used Cura and the Jaris setting to create a print with a full bottom layer and 1 MM wall about 10mm tall. The extrusion rate was calibrated on the 10mm cylinder to achieve a wall thickness fo 1mm within .03mm. All cylinders were printed at this setting. Each consecutive print measured less than the previous by a greater amount. The greatest difference being roughly 2.5mm. The wall thickness also measures smaller as the overall print diameter increased. I have recently calibrated the bed flatness to within .02mm at the X,Y,Z tower and bed center.

I have done research regarding increasing the dimensional accuracy of prints. But have not yet put that research to use. From what I have read it involves changing the "#define Delta_Diagonal_Rod" length. In the firmware. When changing the rod length the following formula is used, New rod length = (Expected Measurement / Actual Measurement) x Old Rod Length. This printer has the diagonal rods with Traxis ball ends from Trick Laser. The current Repetier setting is 269mm. One reason for not pursuing this on my own is, I cannot see how this will improve wall thickness accuracy.....
Not being a wizard at math. Will I be able to correct the outer dimensions by the method described, even though the inaccuracies do not appear to be similar throughout the 4 cylinder tests? I am guessing that the inaccuracy changes as some trigonometric function or percent.
Will this change also improve the wall thickness accuracy I am experiencing? If not. Do you have any suggestions as to what will?

As always your help is appreciated. Thank you.

[Polygonhell]

You can change the diagonal rod length BUT it needs to reflect the actual rod length, given the arms are cast nylon, I would expect the given value to be accurate to within a few tenths of a mm.
If you change it to accommodate other errors, you run into issues because it introduces doming in the motion path, if you then try and tune this bowed motion out at the towers and center using delta radius, you'll end up introducing high or low spots between the towers, because the two domed motions interact.

[JFettig]

In order to get dimensional accuracy out of my printer I printed in PETG which has minimal shrink, and I had to adjust the diagonal rod length significantly shorter than actual. Each time you make a change you have to change the horizontal radius as well as tower rotations if you're working with that as well.

Its been printing great since I've dialed it in despite the warnings from polygonhell.

First off - only print a single wall part to dial in extrusion width/multiplier, etc.

Secondly - only measure the OUTER diameter dimensions of the parts and locations when calibrating scale. If you measure the inner dimensions, as polygonhell pointed out to me, the inner perimeters shrink considerably. So printing something in ABS or Nylon, your inside diameters of holes will always be small. If you print a low shrink material like PETG it'll be spot on.

The points between towers that he speaks of can be dialed out by using tower rotations. Its a real pain to dial that in and you need an indicator. I measure the entire bed and then make adjustments.

[Polygonhell]

One other thing t note when measuring, is you want to measure near the top of the part, ideally over 10mm from the base, if you run your bed excessively hot (Which many do to increase adhesion), you get an effect whereby a part will exhibit an "elephants foot" near the bed. The part near the base will have a slight hour glass shape, which can throw off measurements.

[Pyropainter]

Is it typical of a delta style machine when it comes to dimensional accuracies for the error to increase as the dimension of the printed object increases? I would not expect it to be true of a Cartesian style printer but, unsure of a delta. The error that I saw between the 10mm object and the 80mm object was at least 1.5mm.

Thank you for your help.

[Jimustanguitar]

Is it typical of a delta style machine when it comes to dimensional accuracies for the error to increase as the dimension of the printed object increases? I would not expect it to be true of a Cartesian style printer but, unsure of a delta. The error that I saw between the 10mm object and the 80mm object was at least 1.5mm.

The same GCode will lay down the same plastic on a delta or a cartesian machine, and I believe that you'd see the same inaccuracy equally in both. There are very very slight differences if you were to measure the motion, but they're decimal places away from what you're seeing and measuring on your parts. What's happening here is the plastic is contracting when it cools. This happens to all plastic things, google shrinkage in the casting and molding industry. Different shell thicknesses, infill percentages, printing speeds, and printed shapes will all change the amount and direction of shrinkage slightly. If you need a part to be absolutely perfect, you'll have to print it, measure it, adjust the model file, and lather rinse repeat until you're satisfied with the tolerance.

Most printers are more accurate than the material that they're using, if that makes any sense.


Changing the arm length in the firmware will only warp the flat plane that the machine moves in for each layer. Where did you find the info on this fix and the formula? I don't believe it to be true, personally.

***edit*** This is all assuming that you've properly set the machine up to begin with and have everything properly calibrated as outlined in the assembly manual or the blog that JFettig mentions below. Otherwise, as Phil says, you're fighting wonky with wonky.

[JFettig]

Minnow blog is one place. Changing arm length, horizontal radius, and tower rotation can get you a more accurate machine. Mine is very accurate and consistent now

[Pyropainter]

Hello

I have found the calibration equation for delta arms in several places. This one is from Thingiverse "http://www.thingiverse.com/thing:745523". A calibration object is included. Go to the Instructions tab, you should find it at the bottom of the page. Some of the information found on the instructions page are relevant to Marlin, but I think Repetier is included. If you change the arm length you will also need to recalibrate the delta radius.

Should the dimensional error be the same regardless of the size of the print? I do realize that plastic shrinkage is given as percentage. I am more interested in the printer error.

JFettig.
You mentioned" Minnow blog". Is this found within the See Me CNC forum, or elsewhere. I tried to do a little searching but, no luck. Do you have a web address?

Thank you.

[JFettig]

http://minow.blogspot.com/

If your prints are too small, the rod length needs to shrink.

Mine ended up at 266.35mm to achieve accurate parts.. seems quite wrong to have to move them that far when they really do measure 269mm
128.8 horizontal radius
210.18 A
330 B
89.79 C

You have to use an indicator to figure out where to move the tower rotations, simply picking mine won't do the trick, it just gives you an idea what is reasonable for changes.

[stonewater]

would one of you guys be willing to give a little tutorial on tower rotation and how to measure an adjust it? perhaps it could become a stickey

Tom C

[JFettig]

Its been a while since I did it but I posted up here: http://forum.seemecnc.com/viewtopic.php ... 375#p58696

[teoman]

Pilot's software should be calculating all of these for you.

Just need to port it over to a rambo. Or write a gcode that dumps it to the sd card.

I have been wanting to do this but work just gets in the way.

[stonewater]

wanted to be clear on a question regarding dimensional accuracy.... I was trying to print a part that has a set of holes in it for magnets. the magnets are .875 wide and .385 thick. in order to get the internal holes correct I had to design the holes to be .910 and .400 respectively. I came by this set of numbers by printing a set of rectangular parts while altering the internal dimensions to get it correct. with the modified dimension that size the part comes off dimensionally correct. is this the diagonal rod length error that JFettig talked about and can it be corrected using the method listed on minowblog?

part of the reason I am asking is because when I print a 20 by 20 by 20 MM calibration cube it comes out x -19.92MM y - 19.92MM and z - 20.02MM that seems pretty good, but I can see the deviation getting larger as the part gets bigger...

Thanks

Tom C

[geneb]

http://hydraraptor.blogspot.com/2011/02/polyholes.html

g.

[JFettig]

What materials are you printing? Internal perimeters(holes) tend to shrink a lot more than external perimeters because the external perimeters have no where to go while internal perimeters have nothing holding them to size as the plastic cools so it likes to shrink more.

When I print a low shrink material such as PETG or Taulman 910, I can print directly to size. I find that it seems like the ABS I use is always .015" small regardless what size so I compensate it and I can do it pretty repeatably.

[stonewater]

I print in abs.... the stock hotend works great for abs 15 thousands is about half o what I am seeing, I am seeing about 30 thousands difference.

Tom C

[Polygonhell]

Internal holes pretty much always shrink, for several reasons, see Genes link for one discussion of the effect. Unfortunately by exactly how much depends on the plastic, the temperature it's printed at, the speed it's printed, the extrusion width/height and the material, so it's hard for the slicing software to compensate. I vaguely remember that at one point Slic3r used to try and accommodate for this effect and there were a lot of complaints because it became impossible to predict how something would come out, so it was removed.

I usually size holes up for fasteners, and I use the shrinkage to my advantage is I want to thread a part for say a 3mm fastener I'll actually print it at 3MM and the fasteners will pretty much tap the hole as they are screwed in.
If I need a press fit, I'll print test parts, to find the right size, or print to size and drill/ream to finish. Be careful however if your drilling to finish, it can generate enough heat to distort the part.

[stonewater]

http://hydraraptor.blogspot.com/2011/02/polyholes.html

g.

well that muddied the waters a bit. what I am left with is designing parts to compensate.... and things I download from thingiverse may be too small for me to use

Tom C

[Polygonhell]

I would just assume anything you download from Thingiverse is at best a work in progress. Many of the models have horrible issues.

[bvandiepenbos]

The newest Slic3r 1.2.9 has a new feature for x y compensation that is supposedly able to help hole size. Also on the website is a explanation and tips for getting correct hole sizes.
I have not printed with it yet.