Rostock MAX METAL - First Build (Updates: Ugly Boat Machine)

[Jimustanguitar]

I was just skimming, so please forgive the random comment.

Flip your extruder drive gear over. As long as the motor shaft is long enough, that should solve several of the issues that you encountered.


Great build! I'll dig in a little deeper later.

[Eaglezsoar]

bradjshannon, because I am basically lazy and know CAD about as well as I understand brain surgery, could you post the STL files
for your TNUTS and wire guides. Thanks and this is one hell of a good thread.

[bradjshannon]

I was just skimming, so please forgive the random comment.

Flip your extruder drive gear over. As long as the motor shaft is long enough, that should solve several of the issues that you encountered.


Great build! I'll dig in a little deeper later.

Random comments very welcome! Which issues do you think that would help? I'll give it a try tonight.

could you post the STL files
for your TNUTS and wire guides.

Of course! The t-slot nuts are here on Thingiverse, and also here:

Drop-in_T_Nut_for_80_20_brand_10-Series_Extrusion.zip

(201.32 KiB) Downloaded 403 times

[Eaglezsoar]

Thanks, got the file and it prints fine.

[bradjshannon]

I'm having calibration issues. Probably because I'm doing it all wrong.

My understanding is that, with the dc42 firmware calibration, you can correct for just about every physical error in the build with some software parameter. So I didn't try to set the endstops in any particular way, or to level the bed, or to square the towers perfectly.

Here's what I tried yesterday:

1) I set the Z probe offset to 0 in config.g
2) Removed any/all H parameters from the G30 commands in bed.g
3) Reboot. Home. Auto-cal.

Now my goal is to figure out the Z value at each calibration point where a piece of paper catches under the nozzle. That's what I want the printer to see as Z=0. But instead, the FSRs trigger below that height.

I made a table of where the FSRs trigger (e.g., what the printer thinks is Z=0), and where a piece of paper catches nicely under the nozzle (e.g., what the printer should see as Z=0):

Code: Select all

point	trigger Z	paper-catch Z	difference
p0		 0.18		 0.22				0.04
p1		-0.38		-0.24				0.14
p2		 0.33		 0.55				0.22
p3		-0.36		-0.16				0.20
p4		 0.58		 0.86				0.28
p5		-0.20		 0.18				0.38
p6		 0.96		 1.14				0.18
p7		 0.65		 0.85				0.20
p8		 0.35		 0.57				0.22
p9		 0.42		 0.60				0.18
p10		0.85		 1.15				0.30
p11		0.85		 1.13				0.28
p12		0.05		 0.15				0.10

At P0, which is right next to the X tower, the FSRs trigger at Z=0.18, and a piece of paper catches nicely at Z=0.22. So, in theory, I want the printer to believe it's at Z=-0.04 when the FSR triggers at P0.

First of all, these difference (offset) values don't make intuitive sense. You'll see what I mean if you look at the data as a "heat map":
[img]http://ldhk.org/img/Screenshot_4_7_16__ ... B70A06.png[/img]

I circled the FSR locations. This doesn't make sense to me. The FSRs should trigger most easily when the nozzle is directly above them, at P0, P2, and P4. Only appears to be the case for P0. The FSRs should be harder to trigger as the nozzle moves farther from them, at P1, P3, and P5. Only appears to be the case for P5.

I'd also expect the calibration points nearest each other to have similar offsets, or at least change in the same direction. For example, if the glass bed were simply convex or concave like a contact lens, there would be a ring pattern like a target. And if the "center" of the convexity were some other place on the bed, I'd expect a ring pattern radiating out from there.

Instead, I get weird FSR trigger heights that suggest a very strange glass bed.

Am I thinking about this incorrectly?

[Eaglezsoar]

This sounds like a question for our resident expert in FSRs, Duet and so many things that I could not type them all.........MHackney! "aka Einstein of the forum"
And that is meant as a complement for Michael.

[bradjshannon]

Progress, maybe.

I spent an hour loosening and then properly tightening the frame, working my way around the bottom then the top. Brian (of TrickLaser) did wonderfully precise work. I think this will help with calibrating nicely.

Shaking the frame around a bit loosened the belt on the Z tower... and I screwed up the belt ends trying to just pull them tighter. I ended up having to splice some leftover G2 belt and figured out a way to keep the splice out of the way. That was a waste of an hour.

That led me to messing with the endstops, because the Z tower carriage now has a little metal belt clamp on the tails (overkill on my part, maybe, but this particular belt is a real jerk).

Messing with the endstops led to editing the homedelta.g file to drop the carriages down a little more, because the Z endstop has to be pretty high...

And I had to change the config.g and homedelta.g to remove previous calibration stuff while making sure the nozzle doesn't smash into the print bed.

Okay, looking good after a calibration attempt... 0.280 deviation, about the same as yesterday. Plugged new endstop values and such into config.g.

Test print onto glue sticked glass... nozzle too low. Extruder can't push anything out. I tried to pause and raise the Z but that didn't work, of course...

Oh, great. Why isn't the X stepper motor working ? Why is it responding to some commands some of the time?

It took me an hour and a half to notice the endstop light on the Duet board was flickering. Found a loose connector on the frame immediately.

Finally, all set to try another test print, maybe at the correct Z height. Calibrating... oh, hello:

Calibrated 6 factors using 13 points, deviation before 6.844 after 0.099

I did a z-probe at the center of the bed, which triggered the FSRs at Z=0.70. Based on the paper test, I want Z=0.86 to be my new Z=0. I'm going to make that happen with G92 and try a print.

(I removed the homing command from my slicer's start-gcode, so this new Z setting will stick, so to speak)

[Jimustanguitar]

I was just skimming, so please forgive the random comment.

Flip your extruder drive gear over. As long as the motor shaft is long enough, that should solve several of the issues that you encountered.


Great build! I'll dig in a little deeper later.

Random comments very welcome! Which issues do you think that would help? I'll give it a try tonight.

You mentioned having to make your adapter plate for the extruder thicker for the teeth in the drive gear to line up with the filament path and idler bearing. Then you had to cut out a channel to reach the grub screw. If you flip the gear over the screw should be accessible, and as long as the motor shaft is long enough you ought to be able to slide the drive gear into alignment with the filament path.

The opening in the clear cover is actually for you to reach in with an allen wrench. MHackney has as awesome picture of it.
[img]http://mhackney.zenfolio.com/img/s9/v87 ... 4738-3.jpg[/img]

[bradjshannon]

I was just skimming, so please forgive the random comment.

Flip your extruder drive gear over. As long as the motor shaft is long enough, that should solve several of the issues that you encountered.


Great build! I'll dig in a little deeper later.

Random comments very welcome! Which issues do you think that would help? I'll give it a try tonight.

You mentioned having to make your adapter plate for the extruder thicker for the teeth in the drive gear to line up with the filament path and idler bearing. Then you had to cut out a channel to reach the grub screw. If you flip the gear over the screw should be accessible, and as long as the motor shaft is long enough you ought to be able to slide the drive gear into alignment with the filament path.

The opening in the clear cover is actually for you to reach in with an allen wrench. MHackney has as awesome picture of it.

Oh, of course! I'll try that next time I'm messing with the extruder. Fortunately, that's the one part of this build that hasn't given me trouble, yet.


I'm printing things! Pretty good looking things! Clearly need better cooling to get nice overhangs, but this is totally usable:
[img]http://ldhk.org/img/Screenshot_4_8_16__ ... B84415.png[/img]

First layer wasn't perfectly level, but within +/- 0.15mm, I'd guess. And with a 0.3mm first layer height, it worked out fine. The rest was printed with a layer height of 0.12mm:
[img]http://ldhk.org/img/Screenshot_4_8_16__ ... B844BD.png[/img]

Bit of an "elephant's foot" on this part, but it recovers by layer 3 or so -- I'll just file off the sharp edge around the bottom. Checking dimensional accuracy now.

[bradjshannon]

Several more updates to post, but primarily troubleshooting-related.

There are many outstanding problems with (my use of) this machine. For one, when the heated bed (Onyx rev 7) is re-heating during a print, the hot end can drop from a stable 200C to as low as 183C. The fans also slow down dramatically during these periods, which leaves a banded finish on my prints.

My PSU is a Mean Well 24V 350W tuned down to 20.5V.

I'm assuming that the power supply isn't beefy enough to supply the entire system with the current it wants if everything draws current simultaneously. I've already bought a second, identical PSU to use only for the bed. But if there is another way to solve this issue, please let me know.

edit/update: hot end drops as low as 160C... extruder can't push cold filament... extruder strips filament... print fails

[Polygonhell]

the onyx alone will pull >300W at that voltage, so yes I'd guess insufficient PSU.
Assuming you aren't using Bang Bang control for the bed, in which case this will make no difference, you can try turning down PID Drive MAX for the bed, it ought to reduce the worst case loading, I'd try dropping it 50-100 from where it is and see if it helps.

[bradjshannon]

the onyx alone will pull >300W at that voltage, so yes I'd guess insufficient PSU.
Assuming you aren't using Bang Bang control for the bed, in which case this will make no difference, you can try turning down PID Drive MAX for the bed, it ought to reduce the worst case loading, I'd try dropping it 50-100 from where it is and see if it helps.

I'm using bang-bang now, but thanks. New PSU arrives in a couple days.

[bradjshannon]

What the heck is going on, here:

[img]https://i.imgur.com/7yUUkM0.png[/img]

slic3r is finding new and exciting ways to screw up even the simplest prints

[bradjshannon]

slic3r extrusion settings' effect on toolpath:

[img]https://i.imgur.com/tlgKRHi.png[/img]

[img]https://i.imgur.com/S1Hl7at.png[/img]

[mhackney]

Which is why I don't use Slic3r. I found all sorts of odd inexplicable paths in my gcode. This one takes the cake though!

[bradjshannon]

Which is why I don't use Slic3r. I found all sorts of odd inexplicable paths in my gcode. This one takes the cake though!

slic3r is really weird. But I think I can make its weird work for me, as long as I carefully inspect every bit of the toolpaths. There are some things that slic3r is really great for (infill every N layers) but wow, it takes every chance it gets to ruin a print.

~~~~~~~~

So, now I'm reading old threads and realizing I forgot to put a precision shim in my dual-bearing V wheels. When I tighten the nylock nut holding the wheel assembly onto the trick trucks, the wheels bind up. I've been very careful not to tighten them to the point of binding so far.

Should I put those shims in?

If yes, do I put one shim in, for each side of the V wheel?

Code: Select all

bearing - shim - wheel - shim - bearing

or

Code: Select all

bearing - wheel - shim - bearing

Will that make it so I don't have to be as careful with the nylock nuts?

[ccavanaugh]

Which is why I don't use Slic3r. I found all sorts of odd inexplicable paths in my gcode. This one takes the cake though!

slic3r is really weird. But I think I can make its weird work for me, as long as I carefully inspect every bit of the toolpaths. There are some things that slic3r is really great for (infill every N layers) but wow, it takes every chance it gets to ruin a print.

~~~~~~~~

So, now I'm reading old threads and realizing I forgot to put a precision shim in my dual-bearing V wheels. When I tighten the nylock nut holding the wheel assembly onto the trick trucks, the wheels bind up. I've been very careful not to tighten them to the point of binding so far.

Should I put those shims in?

If yes, do I put one shim in, for each side of the V wheel?

Code: Select all

bearing - shim - wheel - shim - bearing

or

Code: Select all

bearing - wheel - shim - bearing

Will that make it so I don't have to be as careful with the nylock nuts?

Shims are absolutely needed... One shim per wheel. (#2)

[bradjshannon]

Shims are absolutely needed... One shim per wheel. (#2)

Well, that was a chore to do. And now I feel kind of silly for leaving them out. So obviously necessary. And print quality improved immediately -- banding effects along z axis dramatically reduced or eliminated.

RETRACTION/SPIDERWEBS
I spent all day yesterday printing retraction setting samples, trying to get that dialed in.

After an hour of tweaking & re-printing over and over without noticeable change in the product, I printed a sample without retraction at all and realized that there were fundamental slicing issues behind what I thought were retraction issues. Sigh.

Fixed slicing issues (slic3r ugh). Started retraction tuning again.

Same deal -- a couple hours of tweaking settings with little/no change in quality. No matter what I do, I get spiderwebs. Very very fine spiderwebs. Finally thought to turn OFF my fan: it's blowing close to the nozzle and possibly forcing the nozzle to draw a lot more current than normal to hold the set temp. The spiderwebs could be ooze caused by the tip of the nozzle being a different temp from the heater block, or something like that.

Totally turned off the fan, and the print quality improved. Still spiderwebs here and there, but dramatically less.

VIBRATION
My downstairs neighbor complained recently of weird "engine revving noises" late at night -- definitely the sound of stepper motors, transmitted through the floor.

First I tried to dampen the vibrations with various forms of padding -- mouse pads, folded thick fabric. It helped a lot but not enough.

Next, I thought to print ccavanaugh's printer feet in ninjaflex. So I got ninjaflex, learned to print with it (UGH EXTRUDER NIGHTMARE), and made them. They help a lot, but not enough.

Okay, maybe sorbothane feet instead of ninjaflex/TPU? I ordered some, but they're not coming for a few more days.

In the meantime, I wanted to try something else: suspension. Hanging the printer from something instead of resting it on something. Just a different way of absorbing vibrations.

So, uh, I braided three lines of 1.75mm ninjaflex filament into a rope, drilled a hole in my loft bed, ran a loop of the ninjaflex-rope through the hole, and hung my printer from it. The whole time I was worrying, this is not going to work, this is going to break my printer, this is going to make the print quality terrible somehow.

NOPE, it's great, almost completely silent, and prints are better than before. I mean, they weren't great before, so I don't know if hanging the printer has had a detrimental effect. But it's no more detrimental than me being the operator.

I'll post some pics shortly.

[bradjshannon]

My first 3d printer fire! Hooray!

heat bed -> PSU connector started smoking, smell of burning nylon

Guess it's time to swap in that heavier-gauge wire I picked up the other day...

[Polygonhell]

If the connector was burning it was a bad/loose connection at the connector itself, nothing to do with the wire gauge.

[bradjshannon]

If the connector was burning it was a bad/loose connection at the connector itself, nothing to do with the wire gauge.

That makes sense. I saw it spark a couple times as if it were loose, but couldn't tell once I powered everything down. I'll recrimp and see how it does tonight.

[nebbian]

What the heck is going on, here:

[img]https://i.imgur.com/7yUUkM0.png[/img]

slic3r is finding new and exciting ways to screw up even the simplest prints

Welcome to Slic3r. This catches everyone out.

180% is relative to the layer height. Not something you'd expect. So if you think you can just change layer height and get the same extrusion widths, nope. Try changing layer height and you'll suddenly get vastly different extrusion widths.

This is why I specify the extrusion width for all entries, in terms of mm. That way Slic3r can't come up with some strange formula that makes no sense. And I can change layer height without screwing everything else up.

[bradjshannon]

What the heck is going on, here:

[img]https://i.imgur.com/7yUUkM0.png[/img]

slic3r is finding new and exciting ways to screw up even the simplest prints

Welcome to Slic3r. This catches everyone out.

180% is relative to the layer height. Not something you'd expect. So if you think you can just change layer height and get the same extrusion widths, nope. Try changing layer height and you'll suddenly get vastly different extrusion widths.

This is why I specify the extrusion width for all entries, in terms of mm. That way Slic3r can't come up with some strange formula that makes no sense. And I can change layer height without screwing everything else up.

Ahaha, no matter how many times I read "layer height" in the descriptive text, it never clicked. I always thought "nozzle size" in my head. I was wondering why yesterday, after HOURS tinkering with other settings, when I switched to explicit (mm) extrusion widths everything sliced and printed nicely.

The great thing about slic3r is how you can control so many settings, but the terrible thing about slic3r is that you can control so many settings.

[bradjshannon]

Now I want to fix this banding issue:

[img]https://i.imgur.com/b7apUAD.png[/img]

[bradjshannon]

Banding issue caused 100% by Onyx heat bed thermal expansion. Solution unknown.

I do not have very good feelings about this product at the moment. If it's going to cause banding like that while simply holding at 50C, I don't see how it's even going to be usable.