adarcher's build/troubleshooting

[adarcher]

Thanks guys.

geneb, I take it your g32 doesn't complain about too many errors with a heated bed? I can see my effector dip (slowly) into the bed with every try if the bed is on--then it errors out. Anything you do differently than stock or made sure of when putting things together?

[geneb]

That's weird. I've never had any issue running the G32 before a print starts on any of my Duet-equipped printers, and that's three v1s (two 0.8.5 boards and one DuetWifi) and my Artemis. Your Artemis is equipped with FSRs, correct?

g.

[wepollock]

In re calibrating with a hot bed I did not get good results; I have best results calibrating with a hot and clean nozzle (say 150) and a cold bed. I have the offset so that a .20 mm gauge moves smoothly under the hot nozzle w/o friction, the temperature is too low to allow filament to leak out of the hot end. There are two reasons I can think of as to why you would not want to calibrate with a hot bed.


1) most FSRs are not designed to function at 60-100 Celsius, and placing them in direct contact with heater elements may be a problem http://reprap.org/wiki/FSR

Another issue with FSR's in http://reprap.org/wiki/FSR where the bed might be lifting up on the opposite side of probe

2) Although there is a big change in temperature, the glass is not all that thick, and glass has a relatively low thermal expansion. A hotend would have about three times as much expansion than would glass, and plastic has 10 - 20+ times as much thermal expansion.


Measure and(vs) Feel;

a) How precise and accurate do you need to be? Or is better to be consistent and repeatable? The metal feeler gauge is going to give me a uniform measurement in comparison to paper which might not give you meaningful gap measurement. How thick is paper, and how thick is what kind of paper.

b) beyond measuring; how does the first perimeter (or the brim) and first layer look/feel. I am looking for a rounded brim and a smooth as silk first layer. If I have a rough spot on the first layer I am too close. To me, How it looks and feels is more important a factor than how it measures. I am more the measuring kind, but in this case I think look and feel is more meaningful.

[adarcher]

That's weird. I've never had any issue running the G32 before a print starts on any of my Duet-equipped printers, and that's three v1s (two 0.8.5 boards and one DuetWifi) and my Artemis. Your Artemis is equipped with FSRs, correct?

g.

yup, all stock except I added some springs to the bed clip bolts the other day so I could eventually use my pei bed from my V2, which is a little taller. This was a problem before I did that.

Wonder if I have the FSR wires wrapped around a power line or something, or the wires under the bed in the "wrong" spots.

As for the friction in the bowden, I do have a somewhat sharp bed from the extruder and the end of the whip the enters the tower. I'll look into minimizing that this weekend.

@w, thanks for the tips.

[adarcher]

Here’s a comparison pic of different settings, left to right:
210, 6mm retraction 50mm/s
210-200, bottom to top gradient
190
210, 3.2mm 45mm/s

They all look the same to me.

I can see the extruder nob retract, but it’s almost as if it isn’t from all the strings—includingbto one on top at the end.

I’ve tried repitier and marlin and reprap gcode flavors.

These are currently reprap and using the newest stable firmware.

Noticed that after the print finishes, the move speed is stuck at some very low value. It won’t change till I start a new print and the nozzle starts to actually print. This means the hiking post print and the first move down are really slow.

[geneb]

For the very first line in your start and end g-code blocks, add this:

M203 Z30000
That will set your max Z travel high enough that it won't crawl from place to place.

Then make sure you've got Cura set for RepRap g-code and in the slicing profile you're using set the max x/y travel speed to 400mm/sec and the max z travel speed to 500mm/sec. Make sure you've got your z-hop height set to 1mm (or more) and z hope when retracted set.

g.

[wepollock]

In re tuning for stringing, suggest you change one variable.
Also keep a log of all changes.

Test series one
EG keep temperature at 190
Set Zhop to 1mm
Retract Speed 45
Zhop when retracted check
Then
Retract distance 2
Retract distance 3 (better or worse)
Retract distance 3.2 (better or worse)
Retract distance 3.8 (better or worse)
Retract distance 4.4 (better or worse)
Retract distance 5

Test Series 2
Best case from retract distance above
Temperature 190
Temperature 185
Temperature 180 (check your material spec first)
Temperature 195
Temperature 200
Temperature 205
Temperature 210


Below you can see how I am tuning simplify3d.. I have 23 changes so far, and I can roll back to a profile if I make a change that is worse than before.

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

[adarcher]

I’ll look at that gcode for the crawl fix.

As for the tip of changing one thing, already doing that.

I’m showing you that where you’d usually see a difference, you’re not.

Won’t get a real chance to test till Sunday.

[seedjar]

adarcher, how did you manage to get the pins out of your KK connectors? I've managed to get pins out of and back in to other Molex connectors but these ones have me stumped.
Thanks,
~Joe

[geneb]

The little silver tab that shows up through the square hole in the connector face can be pushed in a bit to release the socket.

g.

[adarcher]

adarcher, how did you manage to get the pins out of your KK connectors? I've managed to get pins out of and back in to other Molex connectors but these ones have me stumped.
Thanks,
~Joe

What geneb said, I've used a few things, including the fine tip of a ball point pen. Just put a slight tension on the wire and when you press that tab, it'll slide out a little. That's all there is to it

[adarcher]

Someone want to just run my gcode and see what they get?

I wiped out cura, got cure 3.2 (looks like there's nice new features), and set it up fresh.

I downloaded the string test and sliced it.

I first printed it and it came out similar (actually worse, the z seam looks gnarly). Then I redid the extruder tuning like in the manual and tried again. Not sure about the difference, but the Z travel went crazy fast after that. I got hopeful since something different happened, but the same results. Thick web between the poles.

[geneb]

I'm printing this part right now. It's not so much a "stringing test" as it is a "let's see how much heat we can dump into two little plastic posts" test.

When I ran the test I was using Atomic Gunmetal Gray, which is my go-to material of late. That part prints /very/ quickly - there's no time for the previous layer to cool down and that's contributing quite a bit to the stringing issue. You can't print tiny things like that fast - there's just no time for the part layer to cool enough before the next layer is put down. I'm printing what I use for retract calibration - a pair of 25mm cylinders about 25-30mm apart. I'm printing this using the SE300 PLA Default profile from github @ 195C. When it's done, I'll take a photo of it and post the gcode here.

Can you attach the STL file if your test part? I'd like to try slicing it myself.

Also, you want a G32 in your start code for calibration, not G30.
[EDIT]
Here's the image (it's unresized):

Here's the gcode:

A300_retract_cylinders.gcode

(2.11 MiB) Downloaded 235 times

I stopped the print as it got the point across.

Give it a shot and let me know how it works!


g.

[adarcher]

I can't do G32, the FRS' don't register well at temperature... if that's a problem, I wouldn't mind some tips on fixing that issue too. I was going to opena support ticket for that after I finished the stringing.

I'll run your gcode now and update this.

The model I used was the one linked previously by wepollock: https://www.thingiverse.com/thing:2640671\

OK, I did a few mm of your code and it does string less, but still has a thin web between the two disks and a lot in the infill. You've got a point about the other model, it stays hotter than the larger disks.

I'll try the other one at half speed, I'd like to be able to print small items like that.

[geneb]

Ok, if you're seeing stringing with my gcode, you need to edit the file and drop the temperature 5C. Save it and re-run it. If it's still stringing, drop it by another 5. The lines to update are the M104 and M109 commands.

I won't be able to try that "tiny columns" file until tomorrow.

g.

[adarcher]

Cool thanks.

I'll re-slice mine as default as well, was using fine since that's what I expect to be using most of the time.

Same thing. I'm beginning to think I'm getting some kind of heat creep and retraction wont do anything because it's stretching inside the heat sink or heat break.

I just disconnected my effector and held it upside down. Extruded 10mm, it continued to squeeze out filament seconds after the nob on the eztruder stopped. Is this the bowden binding the filament inside or something. I've never had this problem with my V2, even before I switched to a e3dv6.

On that note, is the SE300 made by SeeMeCNC? I can't find info on it outside of this site.

[geneb]

I suspect that you were seeing the effects of the bowden tube pressure being relieved - I still think there's something wobbly with your temperature reading - those IR thermometers are fairly useless when it comes to 3D printing. I've always checked with the thermocouple wedged into the "corner" made by the face of the heater block and the side of the nozzle.

The SE300 is built entirely in-house by SeeMeCNC.

We WILL figure this out, it just make take a while because of the iterative nature and the fact we're so far apart geographically. I can't swing by your house and threaten the Artemis in person.

Have you tried reprinting the gcode I uploaded, but at a lower temp?

g.

[adarcher]

I hear ya, I was toying with the idea of shipping it back to be looked at

I tried your gcode again at a lower temp and it leaved thinner strings than the smaller model I used still--mind you I've never had to resort to a hair drier like other Artemis users are using, as seen in the FB group. I ended up scaling the string test I linked to to 200% X/Y to get a mid way print between yours and mine. I didn't like waiting so long for all the first layers and infill to see if stringing continued to be trouble.

I found the reason why my hot end was so mucked up even though it never blobbed. The strings often waff around if the layer fan is on, and get caught on the nozzle/heater block.

I got a 6" ptfe tube I had laying around and used that with the effector held in my hand. If I had 3 hands I would have recorded it having the same pressure problem. I even tried holding it straight up with the tube as straight as I could make it. I compared it to my e3dv6 on my v2 and prusa and those both stopped immediately but would ooze a little if I left it sit with no retraction. I've got an extra beatup e3dv6 as well, and I might try that after reducing the top down to the 10mm diameter.

I haven't tried a different slicer yet, I'll do that this weekend.

Thanks for the continued support on this.

[geneb]

Man, this is downright weird. I swear you've got a haunted hot end.

What are Artemis users using a hairdryer for?

What brand of filament are you using?

g.

[wepollock]

The "haunted hot end" might be spot on. It could be filament path issue. If that is the case then;

I would check the Bowden tube,

1) did you secure the lanyards on? they are going to stop bends and kinks near the push fitting, they are going to reduce friction
2) is the bowden tube seated; the bowden tube extends into the push fitting - well into the PTC adapter
3) make sure the bowden tube is also well seated into the push fitting near the extruder motor
4) physically check to see that the bowden tube is not crimped in any way
5) I would remove the tube from the cable sleeve as well and cut a small bit off the tube ends making sure that there are no crimps. Personally, I would score a cut with my electrician's scissors much the same way I would strip wire.

A) Once this was done I would go into the Duet web interface and heat the hot end to 210 or even 215
B) Then I would do a series of extrudes and retracts using the duet web interface; starting in small increments and then going to large increments say 100 (this will allow you to monitor the smoothness of the filament path). If you get clicking on a large extrude you can look to tuning extrusion factor - you can try the same with declining temperatures.

NOTE: I have seen PLA brands that are happy at 180-185 (hatchbox) and others that are happy at 230+ (makergeek)

C) you could even replace the whole tube.

G also mentioned that debugging is "iterative"
I would add another word and that is "baseline"..
A baseline is critical to an interactive problem solving process.


1) If you modified the printer in any way that differs from the factory, you no longer have a debugging baseline. ANY changes that were made to the physical device need to be rolled back to the factory configuration. Why? Because the factory configuration works w/o stringing or issue its been tested. ROLL BACK THE PRINTER TO THE FACTORY CONFIGURATION. Adding complexity into a system be they springs or PEI just makes any debugging much harder than it needs to be. Beyond that, there is absolutely no need to modify the Artemis in any way.

2) Two things were mentioned that have been ignored A) don't level the bed when it is heated B) test another slicer. Item A is not negotiable, simply because FSR's just don't work under heat. Item B - that was a broad stroke "change" that will give you lots of information and can easily be rolled back. I fell off my chair that item B has not been done, simply because it would have eliminated cura settings and configuration issues in one broad stroke - if you get the same result with another slicer you can check off slicer as the issue.

This is from an engineering white paper.

the real power of an iterative approach is found in identifying the correct starting point for your problem and then solving that particular portion of the problem first, so that the solution can be used to build a solid foundation for solving the rest of the problem.

vs.

Dartboard, where variables and parameters are are constantly changing and there is no starting point.

It could be a "haunted debugging process"

What is the starting point? - here is a few in sequence
1) Physical things
2) Slicer
3) Filament
4) Parameters



There are posts on bowden path which go back years.... here is a good one.
http://forum.seemecnc.com/viewtopic.php?t=4261

[adarcher]

Understood about baseline. Being an engineer myself, I work a lot with trouble shooting things iteratively. Part of why I don't really want to deal with it at home

Modifications to the machine only include the build plate springs and a quieter fan up top. I'm going to keep the system in stock shape as much as possible. The e3dv6 option I mentioned would be a swappable part.

I've taken apart the hot end and checked the internal ptfe as well. It was pretty firm into the heat break though not to the point that I thought it was binding the filament... I'll look at cutting a piece of extra ptfe and replacing that next actually. That might be it, who knows. I marked the filament once and watched it extrude/retract near the hot end entrance, it really didn't seem to look like it was moving still after extrusion stopped.

[adarcher]

I knew I had a while before I could fiddle again, so I ordered some cap tubing.

The little insert of the hot end was actually kinked:

Got my hopes up too early though... I replaced it with a short piece of cap, and then replaced the whole bowden tube with some. The filament glides through so nice! But, alas, not my stringing culprit.

Adding my current test gcode. It's a 2x of the other one in the x/y to get a little more movement before hopping.

** edit
Anyone have a good way to loosen a terminal screw? The ones attaching the heater to the PCB from the block are really really tight in there.

[geneb]

If you're printing with the default SE300 Cura profiles, it's unlikely that you've got a retraction issue.

I'd look at how hot you think you're printing vs how hot you're actually printing - test using a thermocouple.

I'll try your gcode sample and let you know how it goes.

[edit]
I cut back the temp from 205 to 198 and I was getting angel hair strings. I suspect had I cut it back to 190, it would've stopped that as well. If you can point me to the STL for that particular print, I'd be happy to slice it and trying printing it here at 190.

g.

[geneb]

Try printing this file. It's currently printing on my machine without any stringing at all. I was getting some at 198, so I backed it off to 190.

Stop trying to tear your hot end apart until you're positive it's the hot end. (chances are that it's not the issue)

A300_string-test-minimal.gcode

(3.1 MiB) Downloaded 127 times

g.

[adarcher]

Shoot, my file was wrong. It's the same model as before, scaled XY 200%

I had modified it from the duet web interface to 185... no change in strings.

I don't have a thermal couple, not sure how to even use one in this case to test things.

I put in a request to get a replacement hotend kit so I can test that.

I'll put my hotend back together in the mean time and try your gcode. I couldn't get the screws loose to attach a different one.

Thanks for the continued support.