Shrinkage

[JFettig]

How is everyone dealing with shrinkage?

I made a few test parts and measured them, they all measured .015-.025" small, and for the size parts it equated to about .7%, so out of curiosity when I was making a larger part(top section spacer with temp controller cutout for my rostock), I decided to scale it up by .7%, turns out it was too large. More and more I'm finding it to be closer to an even .015-.025" shrinkage. I even made 4 single wall cylinders testing some other settings, but I had S3D build up 2mm before switching to another cylinder and its jagged, like it shrunk, then laid the next 2mm on top of that, then shrunk(made it kind of conic) so each 2mm section has a barb look to it.
[img]http://i.imgur.com/VRnUjPt.jpg[/img]
I'm not worried about the jagged part - I can fix that, its shrinkage and dimensional accuracy I'm after!

How do you combat this to make the most accurate prints? I'd love to have a solid way to compensate for this to make accurate prints!


Thanks,
Jon

[bot]

I'm guessing this is ABS.

This is completely normal behaviour for thermoplastic extrusion. As plastic cools or heats, it contracts or expands. If you are trying to make one solid piece, it has to heat or cool at the same rate or there will be differences in the cooling which leads to warping, cracking, etc etc.

How to accomplish this with a 3D printer? Heated chamber, hgher print temps, faster print speeds... welcome to the machine.

[Eaglezsoar]

I'm guessing this is ABS.

This is completely normal behaviour for thermoplastic extrusion. As plastic cools or heats, it contracts or expands. If you are trying to make one solid piece, it has to heat or cool at the same rate or there will be differences in the cooling which leads to warping, cracking, etc etc.

How to accomplish this with a 3D printer? Heated chamber, hgher print temps, faster print speeds... welcome to the machine.

Welcome to the wondrous world of 3D Printing. It is difficult under the best of circumstances to make an adjustment for shinkage that is perfect.
The adjustments would change depending on the type of fillament such as PLA or ABS and also between colors of the different types.
The room temperature makes a difference, etc.
Bot gave you the best way to ensure the shrinkage remains stable.

[JFettig]

Enclosure is in progress.

I'm well aware of how things shrink and grow with temp changes, if I couldn't handle that, I'm not sure how I would have gotten that engineering degree
ABS and PLA behave the same, my test is a .75" cylinder that always comes out at .725-.730", 20mm cube comes out .015-.025 small, even larger parts do the same, always by .015-.025", its not like the machine configuration is incorrect or anything.

Like I said above, the jagged shinkage is obvious, but a definitely good example, There has to be a better way to get accurately sized parts.

[JFettig]

I made a quick PLA test print of circles .55mm*4 thick(4 extrusion widths) and .5" tall, .5" diameter, 1", 2", 3", 4" 5", and 6" outside diameter.

[img]http://i.imgur.com/PO06aA6.jpg[/img]

Generally when plastic shrinks, it shrinks a percentage, when anything shrinks that is generally how it works. Shrink on this is very consistent. Wall thickness is spot on(I calibrated with single wall .75" cylinder). Diameters of each ring is consistently .015 to .020" small. Its kind of frustrating trying to design working prototypes when I can't just print them, or even apply shrink to make them work properly.

Are there any things in calibration that I can work with to get an accurate part?

[Jimustanguitar]

I'm no expert, but my theory is that the shrinkage is a predictable percentage - we're just thinking about it wrong... If you were to mold a 100% solid part, it would shrink by 'exactly' that factor. But when you're making a 10% solid part that's extruded .55mm at a time... I'd guess that the variation of the overall part is mostly only the shrinkage of the outside shell.

If you make everything with 2 perimeters, I bet you'd see about the same shrinkage in the overall part regardless of the size. Perhaps shrinkage can be planned for more like a set tolerance than a coefficient or factor...

[JFettig]

You have it pretty close however I'm finding that its not shrinkage I'm battling at all. Yes, you get shrinkage on the .55mm, but you're actually extruding that wider while molten if you've calibrated the extruder in the conventional way.

A friend and I did some math(actually sketches in solidworks) and found that if I increase steps by 1% it'll get x-y perfect, but the problem is that Z is already correct, so next I looked at the arm length, I decreased that by 1%(based on calcs), ran a part and it increased in size by .010", so I'm still .015" small on this particular part, I'm ran it again at 2% decrease in arm length and the part is only .005" small, I'll drop the arm by 2.5% to get me to final dimension then I'll run another ring test or similar.

To further explain this, a percentage changed on the vertical axis yields a consistent change in x-y motion, however it changes the Z by a percentage - it ended up adding 4mm to travel - I'm lucky I didn't start a print. I measured a 13" tall part I made and it was spot on so I skipped this and went to the arm length, checked the theory and it almost holds true, at the largest diameters, I might be a hair large - I'll test this with the 6" circle.

[BenTheRighteous]

Can you just scale the model in Repetier Host to be 1.01 in the x and y dimension only?

[geneb]

JFettig, you changed the DELTA_DIAGONAL_ROD length, right?

g.

[JFettig]

Can you just scale the model in Repetier Host to be 1.01 in the x and y dimension only?

Yes but that will scale the position of features rather than just their size.

JFettig, you changed the DELTA_DIAGONAL_ROD length, right?

g.

Yes, that is correct, Matter Control just says "Diagonal rod length [mm]"

Its pretty much dead nuts with my diagonal rod length at 268.6mm on my .75mm diameter cylinder. I will run all the way up to 6" to see how it does next.

[JFettig]

I take this post back

Its basically scaling by a constant, not by percentage

[geneb]

What you've done is basically step #4 here:
http://minow.blogspot.com/

Do you think your cylinder tests work better than the 100x2x2 test object mentioned above?

g.

[JFettig]

I haven't tried the 100x2x2 test, is it just a 2x2mm profile 100mm outside square? I can certainly try it.

What I like about the cylinder is you can measure its accuracy in more than just X and Y dimensions, measure all around it by spinning it. What I did with the 7 cylinders above is marked them with colored sharpies to know what orientation is pointed at which tower.

[geneb]

Actually, it's a 100mm long object, 2mm thick by 2mm wide....which might be what you wrote, but my parser crashed a while back so I'm punting.

g.

[JFettig]

yeah, not at all what I wrote would you make 2 or more in multiple directions on the print bed?

[Holy1]

How is everyone dealing with shrinkage?

Thanks,
Jon

I find not swimming in cold water helps!

[JFettig]

How is everyone dealing with shrinkage?

Thanks,
Jon

I find not swimming in cold water helps!

I expected this to be the first reply

[JFettig]

Just a heads up - I made a larger part last night with some bigger off center holes after calibrating the rod length and it came out nearly perfect! I'm .005" large at most.

X and Y directions aren't matching 1:1 but its pretty darn close. Part is 7.5" by 4.01" with 2 2.5" holes 3.5" apart. I'm finally really happy with the machine

I haven't found a real good calibration guide, I think I might write something up. I have a much easier way to set the towers and horizontal radius too(I posted it somewhere already).

[geneb]

The minow blog is currently the gold standard for delta calibration, but if you can improve or clarify it, do so!

g.

[JFettig]

So I'm struggling. I messed with firmware and eeprom stuff trying to get more out of my calibration, all seemed well until I measured a part and it seems like it reverted back to stock delta arm length settings... however in the eeprom it still shows 268.6mm. So I spent a ton of time trying to calibrate and no matter what I put in that field, my parts come out exactly the same every time, .015-.020" small again.

What is going on? Is there something that is burned in not changing? Is a value over-riding the other values? I tried changing the configuration.h file setting and that doesn't change anything, in Matter Control I change the configuration - eeprom setting for it, no dice. Still printing exactly the same size. How do I wipe the RAMBO board clean and start again? In Arduino software, I run the EEPROM clear program to wipe it, shut the machine off, wait a few seconds, turn it back on then load the Repetier firmware, load up matter control, everything appears to be reset. plug in my old values for z height, horizontal radius, etc. print a part, its small, adjust diagonal rod length, print, no change...

Ideas?

[BenTheRighteous]

The EEPROM_clear program only wipes the first 512 b of memory, but the RAMBo has 4096 b.

I have no idea if that's part of your problem, but it was something I found out when I was having USB troubles and was trying to wipe everything and start fresh.

Here's what I used to clear my EEPROM: http://forum.seemecnc.com/viewtopic.php ... 079#p56321

[JFettig]

Thanks Ben! It seems that worked - to get the eeprom settings to work.

However I'm having major issues getting the print to size, I did what I did before, I had to only adjust the delta arm length by .4mm before, now I'm a full 3mm and I'm still not to size.. makes me wonder if something else is wrong.

Jon