Switched BACK to Direct Drive Extruder grinds filament.

Switched BACK to a Direct Drive Extruder using an MK7 Drive Gear, Nema17, attached to RAMPS 1.4 in which the Extruder Stepper Driver is properly tuned.
Controlling with Octoprint and Slicing with Cura 2.4.0

I've used this mechanical setup before
I normally print at 30 mm/s

100 mm extrusion calibrated really accurately.
E steps are correct at 93
M503 confirms its using this value

I haven't even tried printing yet, just testing the extruder.

So the gcode that is sent when Octoprint goes to extrude is G1 E100 F1800
Which by my math feedrate of 30 (mm/s) * 60 (seconds) = 1800, right?

I try to extrude without the Hotend on and it's fine
It seems to be moving extremely fast

This is if I use the extrude button on Octoprint or if I enter the gcode in terminal

That concerns me as the gcode that Cura generates is at feedrate 1800

When I put the Hotend on, it extrudes for about 50 mm and then starts to grind the filament.

Also the Hotend isn't blocked.

Anyone have any ideas?

First of all, it shouldn't grind the filament, the motor should skip steps. Turn up the pinch roller pressure, then if the motor skips steps instead of pushing the filament, you've got a different problem to solve- what's causing the filament to jam in the hot-end?

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Just eyeballing how fast the MK7 turns compared to my previous M8 Hobbed Bolt, it made me believe it's trying to push too fast for the Hotend to melt, though maybe that's placebo. I'll try upping the pressure on the pinch roller, the indentations the MK7 left on the filament aren't as deep as the M8 Hobb. I thought that was due to them being smaller teeth, that may also be placebo. I'll see if increasing the pressure cuts better.

Printing at 30mm/s doesn't mean the extruder pushes filament at 30mm/s through the nozzle. G1 E100 F1800 is likely to fail on all printers I know.

BTW: did you tune the PID settings for your hotend?

Last night I did a redesign of the extruder design. It originally was one of those with a hinged arm held to the motor by a screw, which was unsupported on one side. I was always kind of wary about it, even when it was working successfully for me. So now it's supported and I've provided more room to pinch, I think it may have been bottoming out and not actually pinching very well. I put the original extruder back on long enough to print it.

Quote
o_lampe
Printing at 30mm/s doesn't mean the extruder pushes filament at 30mm/s through the nozzle. G1 E100 F1800 is likely to fail on all printers I know.

BTW: did you tune the PID settings for your hotend?

Doh, it's the old mm/sec vs. mm/min thing. Last time I was able to successfully extrude/retract without using a gcode directly was when I was using repetier host which allows you to set the speed and distance. I was entering 30, guess I was mis-remembering it as mm/sec when in fact it was mm/min. So i took care of that.

I get it now, even though the gcode might say the feedrate is 1800 mm/min, it may not even get up that high due to segment lengh, acceleration, and whatnot. So I guess I found the limit of how much it could push through at that speed and temperature before it built up backpressure.

I've never actually done a PID tune of anything, but I will tomorrow as well as switching to the newer extruder design.

Edited 1 time(s). Last edit at 06/23/2017 11:15AM by FA-MAS.

Yep, it was the pinch wheel tension. With the new design it puts more than enough pressure to bite heavy into the filament and it's printing now.

So even after successfully switching back to direct drive. The problem I was original hoping to resolve by doing so still exists. There's a pattern on my prints. Almost like the extrusion is pulsing. I thought it might be that in my geared setup that the Hobbed Bolt was unevenly cut. But going back to my MK7 Gear, which i've used before and I know is good, still shows the issue. Pinch wheel pressure wasn't an issue with the geared and now isn't with my new extruder design. I don't think it's a mechanical thing as far as XY movement or repeat-ability. It's a Prusa i2, but it's all tightened down and not rickety. Plus I'm only printing at 30 mm.sec. Acceleration and jerk values are low. Corners are quite sharp and layer registration appears to be good. Issue still occurs even if it's just one axis moving. Eg, if Y axis is stationary and X is drawing a straight line, it seems to draw lines like this. I also almost want to think it began when I switched from a Printrboard to RAMPS 1.4. It's a genuine Arduino and RAMPS from Ultimachine. I did already set the Stepper Driver currents. I was printing ABS back then and just thought it was because of material shrink. I'm starting to think not though.

Edit: I'm curious if this is the issue outlined by nophead. http://hydraraptor.blogspot.com/2012/04/stepstuck.html



Edited 1 time(s). Last edit at 06/27/2017 06:34PM by FA-MAS.

Are you still using the same hotend? Have you tried to extrude manually? Maybe you feel some odd resistance? I had a direct drive extruder where the PTFE wasn't aligned with the heatbrake and this caused the filament to stutter/pulse.

Nopheads article is about slow motion/low current stuff. The only stepper that would fall in that category is the extruder stepper ( IMO 30mm/s for X/Y isn't slow )
But the pulsing happened with the geared extruder too.
You could try a RAPS128 in 1/64 stepping. Even with an 8bit CPU it is fast enough for 80mm/s retractions and runs super smooth.

Interstingly, I was just thinking about the hotend. It is the same Hotend I've used from back when I was originally using a direct drive setup. It's a Hexagon Hotend, so no PTFE or anything like that. I thought that may have been the issue when I first switched back and it was grinding filament. I extruded filament by hand and it seemed fine. I do have new nozzles and new thermistors on the way. I'll swap them both out, make sure everything's clean and do a PID just to make sure.

I just realized on the way to work this morning, one other thing I did at one point was swap to smaller pulleys, GT2 20t --> 16t. I set the steps per mm correctly. I'm going to put back on my 20t.

Yeah I read nophead's blog post and it read to me like that as standard, the A4988 stepsticks could actually be skipping due to the resistor value being incorrect thus causing the current to be too low and the off-time too short for the lower current steps. Thus is skips those steps. Guess A4988 have a mode where they will step in lower current situations that he suggests turning on. The way he identifies it is to step the motor slowly and listen. I may take a listen after I do the nozzle/thermistors/pulleys.

Edited 1 time(s). Last edit at 06/28/2017 11:07AM by FA-MAS.

I switched back to my 20t pulleys and while doing so, double checked the motion to make sure it was smooth.

So my Y axis rods are definitely hardened and chrome plated from a reputable manufacturer. The bearings are from the same manufacturer. The Y axis glides smooth.

The X axis didn't. I know one of the rods was from the same manufacturer as above, but the other rod was definitely not from the same manufacturer. It was supposedly hardened, not chrome plated for sure. The bearings in the X axis were cheap bearings too.

I found this on that rod.



Yep looks like the bearings had eaten into the rod.

Luckily I had ordered another rod and bearings from the reputable manufacturer.
Replaced them all and now the axis moves just as smooth as the Y

But, unfortunately, it didn't change the surface of the print.

I'll replace the nozzle and give the whole hotend a once over, including replacing the thermistor and doing PID.

Edit: Oh and I found out that my extruder is properly tensioned. The z0 had changed when I had it apart replacing the rods and bearings. So just a dribble of filament would come out. Now I see what you mean that the tension should be high enough that the motor skips steps. I've never actually seen that before, now I know what it looks like.

Edited 2 time(s). Last edit at 06/29/2017 06:42PM by FA-MAS.

Was doing the rebuild of the hotend and I found something else. The nozzle orifice on my original had enlarged. Probably from cleaning out some jams a while back. It's supposed to be 0.4. Replaced it and put it all back together. Running PID now.



Edit: So i drilled it out to 0.6mm and set it aside incase I ever wanted to try that size nozzle.

Edit again: It's printing the first layer now and the lines look super crisp so far.

Edited 2 time(s). Last edit at 06/30/2017 11:41AM by FA-MAS.

No change. Going to re-tune the vref on the stepper drivers. I think I may have set them too low initially, it was my first time using RAMPS.

Edited 1 time(s). Last edit at 06/30/2017 01:28PM by FA-MAS.

You said this all worked as direct drive on the Printrboard before switching to the Ramps/Mega combo. I'm assuming you've also probably updated the firmware in that process too. It's possible your reaching a limit on the Mega's cpu processing power limiting you due to all the new features that the newer firmware versions have over the older ones. You could test that by installing an older Marlin/Repetier to the same version you had on the Printrboard onto your Mega/Ramps setup.

It's actually the exact same firmware that I was using on Printrboard. All I did was change the board definition and upload it. There's literally no features installed because when I used it on Printrboard, it was a little pressed for memory. It's RCBugfix-RC7

Edited 1 time(s). Last edit at 06/30/2017 01:36PM by FA-MAS.

Ok, so when I fixed the vref (looks like I had it way underset) I also swapped the extruder driver with the Z axis one. And it seems to be resolved. The only thing I see now is it looks like the infill pushes out the perimeters a bit. And I probably need to do a fine tune on the extrusion. And maybe do a temperature tower to see the best temp for this material.

Here's a picture of the test I've done. Don't worry about the 1st layer. The oldest being on bottom and the newest on top.



Edited 2 time(s). Last edit at 06/30/2017 02:18PM by FA-MAS.

Problem solved, I'd say.
The top part on the pic shows a bit of warping, maybe from setting nozzle a bit too high?

Oh, that's because I pulled it off the bed while it was hot.

Whoa, replaced the thermistors. Had to switch the definition in my firmware. Redid PID. Had to re-level the bed (because I took it off to replace it's thermistor), and lastly am running a print of the 5mm_steps block and dialing in the extrusion. Its very close now and the walls are absolutely perfect.

I'm starting to suspect these 0.4 nozzles I bought are actually 0.35. At least Cura's set for a 0.4 line width and I get gaps in the solid infill. The actual lines drawn measured 0.35. Measurement taken from a top layer, so I wouldn't expect much of a squish for it to be wider. Doesn't look to be under extruded, actually really nice lines. Any way other than a gauge to test?

Insert drill bits.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

You could extrude in free air and measure the diameter.

Quote
o_lampe
You could extrude in free air and measure the diameter.

I had actually just thought of this. There's noticeable die swell when it leaves the nozzle. It could give an indication of the nozzle.

I did do a single wall test by increasing the steps per mm and the lines thickened. I'mean probably just underextruding.

So looks like I was probably just under extruding. I calibrated it and it's all good. Dealing with another issue though, though I don't know if it's really an issue.

I print a 50mm x 50mm cube and the X an Y are slighly different.
The Y is perfect. Considering material shrink. It's 49.92 (actually 50.02 before it cooled, measured on the plate)
The X is smaller. It's 49.85 (49.97 before it cooled, also measured on the plate)

I had done some work on my Y axis to create a perfectly straight belt path. You know the Prusa i2 had it going over some idler, then the motor then an idler on the other end.

I created these:
[www.thingiverse.com]
[www.thingiverse.com]

And used this:
[www.thingiverse.com]

Using machined pulleys on both ends

My X axis is similarly setup

They both feel equally tensioned. I even printed a beefier idler mount for my X so I could really tension it and make sure it was equal to the Y.

But when I did the Y axis, I didn't have enough belt so I bought more belt, it was from a different seller.
So my X and Y are from different sellers, but both say they're GT2 2mm pitch.
Could a difference there cause?

Is this even something I should worry about? The difference is almost 0.1 mm. Is that so minuscule to not worry about it?

Edited 1 time(s). Last edit at 07/12/2017 10:54AM by FA-MAS.

Be carfull not to over tighten the x axis belt, you may cause the smooth rods to bend.
Your T2 belts from different suppliers should not cause any worry, if your x and y printed dimensions are slightly different you can alter the steps/mm to adjust them.

My tensioner applies pressure directly on the ends of the X axis smooth rods, doing it this way it can't apply pressure on the X ends, and thus that pressure doesn't translate to the Z bearings/rods.
I was under the impression that you use the calculated steps per mm (80 steps per mm for GT2 20t) and if it's not quite right, it's probably mechanical.

Checked out my belts and even though they appear to have the same tooth profile, they are different.

This is my original belt, both used for X and Y. After I did work on Y, only X had this belt. It has labeling to make me believe it's a legit Gates GT2 belt.



It's thicker, I'm curious if I'm not able to tension it to what it needs to obtain a completely accurate 50mm move. Cause of you know, plastic parts.

Here's what I have on my Y currently.



It's thinner and looks like I can really tension it.

I still have a length of the thinner belt. Put it on my X but it's too thin to stay in the toothed profile on my carriage. Think I might thicken the ends up a bit by putting some contact cement on the back?

Edited 2 time(s). Last edit at 07/13/2017 10:40AM by FA-MAS.

An update, I got new belt. More than enough to replace both axis with the same belt. I was able to tension the X better and now both axis are within .03 difference.