PDA

View Full Version : 3D print Quality improvement



Evan
01-09-2016, 10:16 PM
I have doing some testing to see what can be done to improve quality on my printer. I have not yet tried anything with nozzle size and I know that should help a lot. Looking at the horse model it occurred to me that it might just look a lot better if I made it a different way. It was being printed from bottom to top which of course creates a stack of thin layers perpendicular to the Z axis. Instead I cut the model in half from top to bottom and am now printing it as symmetrical halves. It is easy to glue the halves together. More importantly it totally changes the appearance because the layers are now at 90 degrees to to the X,Y plane. This produces a very different look and in my opinion it looks far better. I also cut the layer thickness to 0.12 mm (0.0048") and to 0.10 mm (0.004"). That takes a lot longer to print but time is cheap. The first model was thickness of 0.20 mm.

This is how I changed the model in CAD:

http://ixian.ca/pics12/horsequal1.jpg

These are the results. The first is the one I have already shown, on the left. The middle is slightly smaller and the layer thickness is 0.12 mm. The right is with layer thickness at 0.10 mm, the lowest the printer will do. It approximately doubles the print time over the left most but to me it is well worth it.

This is with "normal" lighting:

http://ixian.ca/pics12/horsequal2.jpg

This is with lighting intentionally placed to show edge shadows as much as possible:

http://ixian.ca/pics12/horsequal3.jpg

This has turned out better than I expected. For this particular model the shapes of the layer edges just seems to suit the model far more as well as being far less visible. I am quite happy with this.

boslab
01-09-2016, 10:26 PM
Could they be chemically smoothed out, like your Perspex or acrylic polishing method?
I'm guessing it's possible
Mark

Evan
01-09-2016, 10:39 PM
Yes, but I would need to use Carbon Tet and that absolutely requires very good ventilation. I can't do that safely in this apartment this time of year.

Zahnrad Kopf
01-10-2016, 12:07 AM
Evan, this is actually one of the not so small details that can make or break one on the professional printers, as well. Kudos for getting this far and figuring that out. I have to say - your last model there looks really good! < applause > Frankly, the bit that I could choose to comment on ( and I do not choose to ) is rather obviously not anything to do with your work, but rather is the model itself. In short, orientation plays no small part in part quality. Below are both an SLA and SLS model of something I am working on. The SLA model was printed on "rough" settings to save time ( other more important items were in the same build ) so I'm polishing it out before giving to the casting house.


http://i1340.photobucket.com/albums/o736/Zahnrad_Kopf/10EE/6AC3FE5A-A44F-4D83-B9A9-24EAFEACD7F3_zpshmc5abpf.jpg


http://i1340.photobucket.com/albums/o736/Zahnrad_Kopf/10EE/ADA9956D-5446-40DA-A3B3-677D68EDB948_zpsfdlkajyw.jpg


http://i1340.photobucket.com/albums/o736/Zahnrad_Kopf/10EE/3CAE66AF-6C26-4A2D-BC7F-E2262D25764B_zpsvhznsi28.jpg


http://i1340.photobucket.com/albums/o736/Zahnrad_Kopf/10EE/7BC19BFE-E3DB-4E6A-98E1-EFA8209BCBE5_zpsegxnoa81.jpg

Left Handed Spud Wrench
01-10-2016, 12:16 AM
Hah! Good show, Evan! This is something of a secret weapon that I use when printing structural parts. I orient the layers to provide maximum strength. It also does have a striking impact on the visual quality. This is something a great many printing folks aren't figuring out for themselves... this drives sales of compounds like XT-C and the chemical polishing methods.

A little effort at the model stage or even at the pre-slice stage can save hours of sanding and chemical polishing.

RichR
01-10-2016, 12:49 AM
Evan
Based on what I can see in your pictures, the one printed using the 0.10mm doesn't appear to need any smoothing at all.

wierdscience
01-10-2016, 12:57 AM
The one on the left looks like a bad video image,the center one much better with the right one better still.Nice job!

Evan
01-10-2016, 01:37 AM
Thanks guys. Still much more to learn. I am very curious how it will look at maximum resolution with a smaller nozzle. I also still have a lot of playing to do with various plastics I have in stock. I just picked up a small roll of fully transparent PLA to see what it looks like and I have the two kinds of nylon and the T-Glase. Plenty to keep me busy this winter. Right now though I have to print some more stars for the kids at church. They really enjoyed painting them for themselves and parents before Christmas so we are going to give them some more play time next Friday.

Edwin Dirnbeck
01-10-2016, 10:36 AM
Am I missing something here?
It looks like the one on the left was printed in the usual bottom up way.The other two look great. It just goes to show that you cant keep a good man down.Thanks. Edwin Dirnbeck

Evan
01-11-2016, 02:19 PM
Oops, I posted this in the other thread when it really belongs in here.

Well, it is even better now. The limitation on this model is now the CAD model, not the printer. That is what I like to see, especially when you consider that this is a small model. Make it twice as large and the detail area available goes up by a factor of four.


It is the centre one, which should be easily seen. The left one is the previous "best" model. I changed the design a bit so it would be obvious.


http://ixian.ca/pics12/horse5.jpg


This was done with a layer thickness of 0.09 mm and the model placed at a 45 degree angle to both the X and Y axis. By placing the model at that angle it reduces slight changes in the positioning that can be caused by small amounts of slack or differences in the mechanical drives. It is a form of mechanical anti-aliasing. Think of it in terms of drawing a line on a digital image where the line is just slightly off the X or Y axis. Then compare that to a line drawn well off both axes. Really, placing it more than a few degrees off axis alignment is all that is needed. It needn't be at 45 degrees, any significant amount away from attempted perfect axis alignment will do the trick. It's the same reason that all digital values have a maximum accuracy of plus/minus 1. That shows the worst along each axis.