Managing material shrinkage

[yellowshark 153]

Does anyone know what actions one can take to reduce shrinkage?

PLA vs ABS vs Nylons – any significant differences

Bed temp. higher or lower?

Extruder temp. higher or lower?

 

Fans higher or lower?

Layer depth higher or lower?

 

Print speed higher or lower?

 

Infill % higher or lower?

 

I am guessing that keeping the chamber temp. as close to ambient temp. as possible will help?

 

[kris 29]

Difficult to do anything against shrinkage since it is a material property that more or less behaves linear with temperature.

One thing you can do however, is to make sure that whatever you print cools uniformly. Hence heated build platform & heated build volume.

For that I use a greenhouse to make sure the part I print stays uniformly at a higher temperature until it is completed.

 

[gr5 2,167]

Shrinkage should be simple but in practice it's complicated. Please specify which of the 5 or so common issues you care about:

1) part is wrong size dimensions

2) corners lift or part won't stay on bed

3) overhangs lift and get hit by nozzle

4) bridging strings rip/tear/break

5) topfill rips/tears/breaks

6) other?

I looked up the density versus temperature curve for ABS and PLA and they aren't perfectly linear but close enough. I was surprised. I expected a different result below the glass temp than above it. Also they both had almost IDENTICAL graphs! Yet ABS seems so much worse! Well - it isn't. It just has a different glass temperature.

What? you ask?

2) corners lifting, or part coming off glass issue:

This is the most important issue because if your part comes off the bed it's always a complete failure. Some people can tolerate the other issues.

PLA has a very low glass temp - you print around 220C, glass temp around 60C, room temp around 20C. So it is basically a liquid above 80C or at least very soft (like clay). The upper printed layers cool and pull hard on the lower layers lifting the corners. But this effect doesn't start until the upper layers are below 60C or so. Above 60C the material simply thins out a bit and since it is a liquid it just flows around and doesn't pull very hard at all on the lower solid layers. So the cooling/shringing/lifting-corners effect only happens from 60C to 20C. That's only a delta of 40C.

ABS glass temp is closer to 100C. So this happens from 100C to 20C or 80C! Much more of a problem for issue #2 above! Very roughly twice is bad. One fix is a heated chamber (if chamber is at 60C now you have 100C to 40C or only 50C delta - about the same as PLA). But people have learned to instead use hairspray on glass with heated bed which STICKS LIKE HELL and brim and this keeps the corners from lifting by just having it stick so well. So it doesn't adress shrinkage - it addresses getting it to stick so hard you can pick up that little UM robot and swing the entire printer around the room without it coming off.

THIS IS JUST ONE ISSUE PARTLY EXPLAINED. There are many other issues. Also I've printed nylons but I'm not certain what their characteristics are. They definitely have shrinking issues also. They have higher glass temps but seem to shrink less than ABS. Not sure though.

Here's another issue: PLA I would expect to be worse than ABS when it comes to other issues: when printing a vertical circular hole, the PLA cools so fast (faster than ABS because ABS is usually printed with no fan and because the layer below is warmer maybe) that it's like putting down a tight rubber band that pulls that circle inward and makes the hole too small. ABS turns to solid so fast that I suspect it doesn't have this problem.

When printing PLA you want 100% fan (actually you want 500% - yes 5 fans print much better). With ABS you mostly want 0% fan with the exception of certain overhangs and bridging. The old slicers tuned to ABS turn the fan on and off constantly - sometimes 5 times on one layer. Too much fan with ABS and you get bad bonding between layers.

 

[yellowshark 153]

George, a great post, thanks for that. Sorry for delay in responding but I was away for 3 days at the weekend and following your post I wanted to do some testing. To comment on your post.

 

1. 
   My issue is solely no. 1 dimensional accuracy. I have the other issues you mentioned pretty well dialled.
    
   

 

Of course dimensional accuracy is more complicated that it sounds. Is the Vernier gauge accurate? - mine has never been calibrated. Also physically taking the measurements requires some expertise. Then of course if you are creating a part to fit within a 3^rd party assembly, just how accurate DOES it need to be?

 

1. 
   Anyway your comments on temps. got me thinking, as did Kris’s comment on the consistent nature of the environment. Two of my core settings are non-changeable; a print speed of 60 mm/s and a layer thickness of .200. Print bed temp. I used my normal 60 degrees.
    
   

 

For the other, extruder temp., I chose 225 for my first test (before starting this thread)- unusual for me because I never go over 220 and normally use 205-210. But, probably for the 1^st time, I was using Colorfabb black filament. I found a test block I had done for this some months ago and it actually was done at 60 and .200 and the temp was 225 (i.e. the temp that gave me the best surface finish).

So for my 2^nd test I reduced the extruder temp to 210 and I changed my post print cooling process. Normally , once finished, I immediately remove the glass bed and leave on a glass table to cool. For my new process I left the print bed alone until the part temp. was reading 40 degrees; no scientific reason for 40, I probably just got bored at that time. During the cooling I had also removed the top from the printer. Ambient temp. was about 23 degrees.

The result was a noticeable improvement in accuracy.

 

1. 
   I had made two changes so I redid the test but this time using my original cooling process, whip the print bed straight out. The accuracy fell.
    
   

 

So I deduce that lowering the extruder temp. and leaving the part in the printer environment for a period of time are both beneficial to dimensional accuracy. Maybe I should do another test waiting for the part to get to ambient temp. before removal.

 

1. 
   I did 11 tests in all. From these I found that
    
   

 

 

• 
  I could not determine any difference in accuracy between 50% fan and 100% fan. 0% fan seemed to give me worse results although bizarrely gave me the most accurate Z axis measurement of the day.
   
  
• 
  I also ran some tests at 205 degrees. There was little difference in the results between 205 and 210. At no time did either temp. deliver best readings for the three axes, but 210 delivered more best times for two of the three axes.
   
  
• 
  For the first half or so of the tests I had walls set to 3.6 (all tests had 100% infill). I then started to play with reducing the walls. I am inclined to say that using 1 wall gave the most accuracy – not enough tests really to be sure
   
  
• 
  For the last few tests I swapped to Colorfabb Dutch Orange filament and I am inclined to say that was marginally more accurate than the Black – again not enough tests really to be sure.
   
  
• 
  All of the Z axis measurements were oversized; target was 3.3mm. All of the X and Y measurements were undersized. Maybe I have a problem with my Z stage?
   
   
  

 

 

1. 
   My most accurate readings were received with
    
   

 

 

• 
  Fan 50%
   
  
• 
  Extruder temp. 210
   
  
• 
  No. of walls 1
   
  
• 
  Filament Dutch orange (shame as really I need to use Black)
   
  
• 
  Cooling process New
   
  

 

Measurement errors were

X 80 microns

Y 100 microns

Z 120 microns

Assuming 100% accuracy in the Vernier gauge.