All Metal Ubis Issues

[Mooselake]

After being a big fan of the All Metal Ubis I've found an issue and downgraded it from Most Wonderful to Recommended with Reservations.

I have an All Metal Ubis, purchased when it first became available in the 2014 Black Friday Sale, connected to a Beta Geared Extruder kit installed in the V2 (printed, gears in fixed position) extruder, driven with a recycled stepper replaced in an upgrade. I noticed some stepper skipping issues when it was first installed, wrote them off to the stepper (the me part of ass-u-me, as it turns out), and ignored them. While recently printing a large object in ABS where I tweaked the extrusion width to reduce curling I noticed that it was leaving gaps in the inside of the lower (not first) layers and did some investigation. Turns out that somewhere just short of 20mm of manual "high" speed (30mm/sec in Repetier-Host) it would start clicking at just under 20 mm of extrusion. Very repeatable, and not really speed sensitive (happened at higher speeds at about the same place), not really temperature sensitive (didn't test this one too hard). Hmm.

Brought it up with Printrbot, told that the wonderfullness came with a price, limited extrusion speed because of the short melt zone that reduced oozing. Interesting, as they never mentioned that one before, and once I discovered retraction oozing was never a problem with the 3.0mm ceramic Ubis, nor has it ever been a problem with the AMU. Time to investigate - this is more of a capacity (as in amount of plastic melted before it wonks out), not speed issue.

Attached is some quickly whipped off gcode that should work with pretty much any Printrbot, heats up the extruder to 200 degrees, homes and then moves the head up a bit so it won't end up in a giant plug of plastic. After that it waits for the extruder to come up to temp, extrudes 10mm, delays to let it settle down, and repeats the cycle incrementing a mm at a time. It's attached, give it a try (I'd really like to see how the Ubis 13, Ubis 13s (the cell phone model), and perhaps an E3D reacts. If you use PLA that'd be great, but pretty much any other filament (think of t-glase...) should do with a quick one line manual edit of the temperature. I only tried it with Repetier-Host, but if Cura will let you load gcode it should work with it's gcode sender. There's delays after every extrude for the so just count 10, 11, etc. If you'd be so gracious as to report back with your results (no skips anywhere, starts skipping at XXmm, along with your extruder type and anything, like temperature, you changed. Thanks!

I got skips starting around 18 to 19mm with my 2014 All Metal Ubis.

I didn't do any retraction/reprime between steps so the dreaded ooze will dribble out a bit and cause a small variation in the results.

I plan to do the usual automated test generator tomorrow to allow temp and speed variations. Today I need to do some American Heart Instructor Updates (they change the rules every 5 years, one theory is it sells a lot of several hundred instructor kits, but really it's because they're always looking for better ways to bring you back from the dead) and then do an ambulance standby at a hockey game (tends to be a physically aggressive and injury prone sport as you fans know). I'm a BLS (that's CPR and AED) and PALS (BLS plus drugs and other advanced stuff for kids) instructor, between them that's about 4 hours of updates and I've only done one of them. Tomorrow, if I don't procrastinate too long with the rather boring updates I'll start in on the enhanced tests. And if anybody wants to travel to the land of the eternal moose I'll do a CPR class or two for free...

Kirk

[RetireeJay]

Kirk, I have to ask why you are interested in extruding at 1800mm/min = 30mm/sec?
I guess you must be using 1.75mm filament if you have a recent extrusion product from Printrbot.

The area of a 1.75mm circle is 2.4mm^2. Pushing out 30mm/sec means you are extruding 72.15mm^3/sec.

But if you are using a 0.4mm diameter nozzle, laying down a trace 0.4mm in width and 0.25mm high, then that's a cross-sectional area for your trace of 0.1mm^2. If you are printing at a speed of 60mm/sec, then you are consuming only 6mm^/sec of plastic. Your test is asking for 12X more plastic than you actually need to print.

I have found my limit with T-Glase is much lower than 72mm^3/sec. I haven't pushed the latest configuration to the breaking point, but before I added the metal collar between the hobbed bolt and the E3D extruder I was limited to about 9mm^3/sec. My setup is still with 3mm filament, which has a greater stiffness to push down on the melt zone.

[Mooselake]

From my Printrbot notebook, my feedrates have been configured at 30mm/sec since at least early 2014, although I see that the August 2012 (back in the 3.0 only days) value was 14mm/sec. No idea why I would have changed that, but it was never an issue with 2.85mm filament - I've only been using the wimpy new stuff since I installed the geared extruder some time last spring or so. I believe that my R-H has been set to 3 and 30 since it came out with the two speed extrusion options.

I'm open to suggestions. I was seeing issues printing bottom layers at 80mm/sec, 0.3 layers, 0.6mm width (iirc, computer's tied up with these instructor updates...), 14.4 mm^3/sec, or about 2.4mm/sec extruder speed (assuming no calculator mistakes, sly drool's in the basement), so the problem occurs at a much lower speed; I just haven't gotten around to figuring out what it is. I got pretty much the same results at the other speeds I tried, but never got down to 2.4mm/sec (call it 150/min), skips at the same point, still leaning towards the theory that's related to the hot end's recovery rate and the actual volume that it can melt is very low. If I can determine what that is I'll plug that number into the max extruder speed and go from there.

What if you divide my speeds by 4 or so and try the test? Inquiring minds want to know...

Kirk

[RetireeJay]

My E3D doesn't have any temperature drop at the maximum rate that I can feed filament through it. No "recovery time" is necessary. I was running tests in the stable part of this graph.
Temp Stability 9 Jan 16.PNG

I'm using 3mm filament, so for an equivalent volume per minute, my settings will be about 37% of yours. About 700mm/min would be equivalent of your 1800.

But it's well known that T-Glase is quite a bit slower than PLA or ABS. The maximum rate that I can feed reliably right now is between 200 and 300 mm/min (3 to 5 mm/sec) which is equivalent to feeding 18 to 30 cubic mm per sec. That's WAY above any practical printing rate that I expect to use.

BTW, I simplified the test to just extrude 10mm, wait a bit, and then extrude 100mm. I was watching the extrusion and I could tell if it slowed down due to the hobbed Arcol Hyena bolt losing traction (I guess my motor drive is still set just a little higher than necessary). Fortunately, only once did the hobbed bolt "dig in" and completely stall; at that time I was trying 600mm/min.

[Mooselake]

The recovery time was to allow for a difference between the melt zone temp and what the extruder read. The gearhead extruder never strips the filament, and with the size of the indentations I doubt it's an issue with any reasonable motor.

Based on your actual extrusion rate calculations I modified the test range and automated the speed steps, gcode attached. This time, after several tests, I determined that the limit was around 7mm/sec for a continuous 100mm extrusion - skips at 8, OK at 7. So using the 80% rule, a reasonable max extruder speed is about 5.5mm/sec. Never really tested the 3.0 ceramic Ubis/Wades extruder this way but I had the manual speed set at 30mm/s. I never had any problems with the Wades skipping, but while the max was set at 30 in the firmware I don't really know if I changed that max speed to 30 before or after the current gearhead.

The attached zip file has two sets of gcode, labeled with the extrusion speed limits along with the dinky basic (M$ SB) program that created them. Anybody that wants to give that a try is welcome, please report back here. If you need the speeds changed just let me know and I post an update.

Next step is to try higher temps and see if it makes a noticeable difference. I was using 200 degrees with PLA that I was printing at 180-190 before launching into this. That'll likely be in a few days. One more instructor update to go, but I'm covered for the BLS class (for a group of high school students in a CNA class) I'm helping out in at 8am tomorrow morning.

Kirk

[RetireeJay]

Here's something interesting I noticed when I was running Mooselake's experiments. Immediately after the extruder started the last extrusion, the heater turned off, even though the extrusion wasn't finished.

This is actually correct behavior, but you have to dig into the G-code documentation to figure out why.

Certain commands in Marlin are "Buffered" - namely G0, G1, G28, and G29 - G30. All other commands are "unBuffered." In this context, what they mean by "Buffered" is that any command in the "Buffered" family must run to completion before any other "Buffered" command will start. However, "unBuffered" commands will be acted upon without waiting for the active "Buffered" command to finish, i.e. simultaneously.

In Mooselake's G-code, the last Extrude command, G1 E25 is a Buffered command, but the following M104 S0 command is unBuffered. So this means that the extruder will start to extrude 25mm of filament, but then immediately the heater will be turned off! It starts to cool down very quickly, even while that last 25mm of filament is extruding (or in my final experiment 100mm - which takes quite a while). The solution is to include a final G4 pause command before ending the program. (In fact the command turning off the heater is redundant; Repetier - and I presume all UI's - will turn off the heaters when the G-Code program ends.)

Note that this "Buffering" is not (quite) the same as the buffer (notice the lowercase "b") for controlling the flow of commands from the computer to the Printrboard. The Printrboard can store quite a few commands in its queue, but eventually it fills up its buffer of waiting commands, and then the communication protocol between the Printrboard and the computer tells the computer to wait until there's space before sending a new command. It's like there are two separate "gates" - one (input) gate controlling the commands coming in from the computer, and another (output) gate controlling commands sent to the motors and heaters. The "corral" between those gates is the communication buffer. The rule about finishing one Buffered command before starting the next one is part of the output gate's logic.

[ahaer]

I just fired up my printrbot after stepping way for a long time during which I upgraded to makers kit 1405 with the new hot end and noticed the same clicking issue...

I didn't realize that there were internal changes to the hot end. I didn't do any formal testing but was able to eliminate most of the skipping by upping the current to the extruder to about 95% of full power. Before my extruder motor was never warm, now its moderately warm to the touch.

I don't remember the speeds that I was trying but I was also using a .3mm layer height to reduce time. Obviously if I use smaller layers I will use less filament but I'm trying to get a reasonable fast print without being insane quality. Maybe if I use small layers I can run faster and get good speed AND good quality

[Mooselake]

I've already tried increasing the extruder motor current with no effect.

It's quite likely that the issue has to do with the small melt zone on the AMU, unknown whether or not the 13 or 13s is affected.

RJ noticed that this really doesn't affect normal printing a whole lot, as the extrusion rate while printing isn't all that high. I initially noticed it while loading and unloading filament in the extruder, where a fairly lengthy retract and extrusion is needed. I saw it while printing, but only for wide internal layer extrusions, and a slowdown there won't add much time. From the extruder entry to nozzle is 80mm; I've been doing 100mm manual moves when swapping filaments. Guess I'll need to live with this taking lots longer.

I was going to try to work up a spreadsheet for extrusion rates vs layer thickness and extrusion widths, but Libre Office has a bug with Intel 5500 Graphics, which are part of my Skylake 5200U CPR (listed in their database, still open). Guess it's about time to order up that i5-6400 (along with MB and memory) and fix the Moosecave desktop.

Kirk

[Mooselake]

Is it necroposting if you revisit your own topic?

I recently received one of Matt's Thingybot Delta Pros with a 1.75mm E3D hot end, and dug out (and reworked it to work with mm^3/sec directly) the extruder rate test gcode generator. Fired it up with the TBD. Fixed a math bug after disappointing results. Wow! The E3D started occasionally skipping at an extrusion rate of 20 cubic mm/sec, noticeably at 22. This isn't quite a direct comparision since I boosted the temperature to 220 (over the 205 I've been using with the included Makergeeks black PLA), but it qualifies as pretty darn close.

Using a conservative 16 mm^3/sec (or is that cmm/s?), 120mm/s max with Slic3r's autospeed the favorite Celtic Coaster test print printed in 3/4 the time it did on the TBD using the 8 mm^3/100mm/s max settings carried over from the KCSP/AMU. WIth a 0.3mm layer/0.5mm extrusion width it'll hit a (calculated) extruder limited 107mm/sec print speed. YMMV...

I've been keeping TBD comments in the Thingybot forum since it's not a Printrbot, but there's considerable interest in E3Ds here. I've attached the extruder speed test gcode (it's not very big) compiled for 2 cmm/s steps from 14 to 24 cmm/s, just in case somebody else want's to test it. As usual the generator source is in M$ Small Basic, feel free to rewrite it in Python or whatever if you don't like that. No warranty, no returns, bug reports welcome.

My intent is to return the KCSP back to 3mm wades, along with some other mods that have been sitting around (beefier Y stepper, MIC6 print surface, new stand, etc.), when it works up in the todo list. When that (finally) happens I'll try to remember to add what the wades/3.0 original Ubis will do.

Kirk