First test print with a Smoothieboard

[plasma]

I was finally able to convert from Rambo to a Smoothiboard. However after calibration my first test print leaves much to be desired. This 20mm cube I printed with a .2 layer height, Material PET+. The cube measures 20.03 x 20.07 x 20, indicating the the dimension is not an issue. As you can see from the pictures I cant get the edges of the cube to be without the bulges. Also some surface imperfections drive me crazy. Any advice would be welcomed

[Jimustanguitar]

I think you're over-extruding. Better processor or not, calibration still matters.

[plasma]

I think you're over-extruding. Better processor or not, calibration still matters.

As I said the calibration was done, I did few prints adjusting the extrution, anything under this setting gave me gaps in the print.

[Jimustanguitar]

It just looks "full". Especially in the second picture.

If you drop it to 95%, what's your print look like?

[Polygonhell]

What's your Jerk setting and how fast are you printing?

[plasma]

The above cube was printed at 30mm/s speed. As far as Jerk Movement its based on the config-overide file.

Code: Select all

DO NOT EDIT THIS FILE
;Steps per unit:
M92 X80.00000 Y80.00000 Z80.00000
;Acceleration mm/sec^2:
M204 S1000.00000 Z0.00000
;X- Junction Deviation, Z- Z junction deviation, S - Minimum Planner speed mm/sec:
M205 X0.01000 Z-1.00000 S0.00000
;Max feedrates in mm/sec, XYZ cartesian, ABC actuator:
M203 X500.00000 Y500.00000 Z500.00000 A500.00000 B500.00000 C500.00000
;Optional arm solution specific settings:
M665 L269.0000 R134.5883
;Motor currents:
M907 X1.20000 Y1.20000 Z1.20000 E1.20000
;Home offset (mm):
M206 X0.00 Y0.00 Z0.00
;Trim (mm):
M666 X-0.145 Y-0.237 Z0.000
;Max Z
M665 Z361.830
;E Steps per mm:
M92 E92.4000
;E Filament diameter:
M200 D0.0000
;E retract length, feedrate, zlift length, feedrate:
M207 S3.0000 F2700.0000 Z0.0000 Q6000.0000
;E retract recover length, feedrate:
M208 S0.0000 F480.0000
;E acceleration mm/sec^2:
M204 E500.0000
;PID settings:
M301 S0 P11.0100 I1.0800 D28.1500 X255.0000 Y255
;PID settings:
M301 S1 P87.8600 I3.0100 D641.8200 X255.0000 Y255
;ABC=Shimming data; D=Shimming; E=Depth map; Z=Master enable
M667 A0.0000 B0.0000 C0.0000 D1 E1 Z1

[Polygonhell]

You might want to try increasing your junction deviation, unfortunately it's not clear to me without sittling down with a pencil and working through the grbl code what that setting actually means in terms of absolute value.
With 90 degree corners the planner has to reduce speed to whatever the minimum initial speed is, if it's too low, the head will end up depositing extra plastic at the corners.

It also looks like you don't have enough cooling and I would tend to agree it looks a bit like over extrusion.

[Polygonhell]

Actually I just read the article on grbls path planner (it's what the smoothie one is based on). It's similar to the Mach constant velocity algorithm, IMO it's not a good choice for a 3D printer, for a Mill it makes a lot of sense.
Your specifying the maximum deviation from the given path, your going to run into issues with rounded corners if that number is too high and if it's too low you'll run into an issue where you nearly stop in corners, which will likely lead to over depositing of plastic because of the hysteresis in extrusion.

[plasma]

You might want to try increasing your junction deviation, unfortunately it's not clear to me without sittling down with a pencil and working through the grbl code what that setting actually means in terms of absolute value.
With 90 degree corners the planner has to reduce speed to whatever the minimum initial speed is, if it's too low, the head will end up depositing extra plastic at the corners.

It also looks like you don't have enough cooling and I would tend to agree it looks a bit like over extrusion.

It looks like you hit the nail on the head. I have increased the junction deviation from .01 to .05. The print improved dramatically, I'd say almost perfect. I wish I could calculate the proper setting without guessing, however the math is to over the head for me. For now I will keep on tweaking the Junction deviation until its perfect.

Thanks Polygonhell

PS I also increased the acceleration from default 1000 to 3000

[forrie]

Ahhhh...I've run into exactly the same problem....but give that a go!

[plasma]

Quick update. The final setting that gave me the best 20mm cube print was to increase junction-deviation to 0.35. The edges came out sharp and on the same plane as the cube surface. It would be nice to have a discussion where we can set up Smoothieware to be perfectly dialed in for Rostock Max 2.

[mvansomeren]

Quick update. The final setting that gave me the best 20mm cube print was to increase junction-deviation to 35. The edges came out sharp and on the same plane as the cube surface. It would be nice to have a discussion where we can set up Smoothieware to be perfectly dialed in for Rostock Max 2.

It would be great if you did a write up on what it takes to upgrade to a smoothieboard. I purchased one a while back but it's been sitting in the box. Having a write up on the details all in one place would be a great help to anyone considering switching over.

[KAS]

I would support that as-well. I'm sitting on two smoothieboards with GLCD's that I intended initially to upgrade the v2 and one for a kossel style build. After changing the v2 to 400 step motors and seeing the visual improvements, I'm not seeing a huge benefit to making the switch.

[ccavanaugh]

Quick update. The final setting that gave me the best 20mm cube print was to increase junction-deviation to 35. The edges came out sharp and on the same plane as the cube surface. It would be nice to have a discussion where we can set up Smoothieware to be perfectly dialed in for Rostock Max 2.

Was that 0.35 or 35.0? I've made the switch to a Smoothie as well with the last bits for the GLCD here this weekend.

[plasma]

Quick update. The final setting that gave me the best 20mm cube print was to increase junction-deviation to 35. The edges came out sharp and on the same plane as the cube surface. It would be nice to have a discussion where we can set up Smoothieware to be perfectly dialed in for Rostock Max 2.

Was that 0.35 or 35.0? I've made the switch to a Smoothie as well with the last bits for the GLCD here this weekend.

It was 0.35. On the small print it gave me a perfect cube. However on a large print I dropped it to about 0.20, due to jerky movement over a large print. Polygonhell mentioned that there is a mathematical way we can get a precise number for junction-deviation for Rostock.

As per my conversation on smoothiboard IRC:
"[22:54] <wolfmanjm> plasma_: smoothie does not use jerk it has a different algorithm. If your corners are round you probably need to increase you acceleration.
[22:54] <wolfmanjm> BTW the default is 0.05
[22:55] <wolfmanjm> and has no relationship to whatever repteier uses"
[23:24] <wolfmanjm> I'd start with 0.05 and increase as needed

So I will try to experiment more with the acceleration while reverting back to Junction-deviation of 0.05. I'll post my results as I try to print some cubes with various acceleration adjustments.

[626Pilot]

I would support that as-well. I'm sitting on two smoothieboards with GLCD's that I intended initially to upgrade the v2 and one for a kossel style build. After changing the v2 to 400 step motors and seeing the visual improvements, I'm not seeing a huge benefit to making the switch.

Comparing Smoothie to Marlin/Repetier is like comparing a modern cell phone game to an old Pong TV console from 1980.

If Pong is your heart's desire, and you don't mind feeding it D cell batteries, you can just have Pong. But if you want a new feature, something that won't run on that 1980 processor (which is essentially what RAMBo, RAMPS, etc. runs on), you're out of luck.

For someone running a farm of 3D printers, and they already have the calibration nailed and don't need support for any special features, maybe RAMPS is better. For anyone else, Smoothie can present a much more tempting choice than something that runs on a RAMBo. Smoothieboards cost less, and are far more future-proof. SeeMeCNC's adherence to the RAMBo as a platform is supposedly about how reliable they've tended to be. I have been using a Smoothieboard 5X for about half a year, put it through the kind of torture only a developer can inflict, and it's still kicking. I think that as Smoothieboards prove themselves more and more in the field, SeeMeCNC will eventually switch. Marlin and Repetier already look like they're standing still, with all the better controllers that are coming out these days. I put my heuristic delta calibration code in front of both firmwares' authors and neither had anything to say to me about it, even though it's better than what either of them has out there. I honestly don't think either of them can shoehorn that code to run on those old processors that run like cold molasses and have only 4K RAM.