Trouble with swelling part corners

Hi, RepRap forums.

We have an issue with our printer which you might be able to help us with.

We have this issue where it seems like the parts' corners are getting over extruded, and we can't seem to figure out why.

Here's a picture of a 3D model and here's a picture of the result. As you can see, the corners are "wider" than the rest of the body, which they shouldn't be.

Have any of you had a similar issue and managed to fix it or found the cause?

This is printed on a custom built Kossel delta printer with Duet Wifi and E3D Volcano hotend with 0.6 mm nozzle. The parts pictured have a 29 x 26.5 mm footprint and are sliced in Simplify3D with 0.3 mm layer height. Besides this issue, which is present in every print regardless of model size and shape, the printer works fine.

The acceleration is XYZ = 1000 mm/s2 and Extruder = 75 mm/s²
The jerk speed is XYZ = 800 mm/min and Extruder = 13 mm/min

We have tried with higher values for the extruder (acc: 100, jerk: 20), but it didn't help, and we're afraid to go too high since we've already wrecked the gears in the extruder once (we have a Nema17 extruder motor and a Flex3Drive effector). We have also replaced the belts with steel reinforced belts since we thought that the old belts would rubber band due to our tall towers (1 meter), but that didn't help either.

Any ideas or tips are most welcome.

It looks like over-extrusion to me. Have you calibrated the extruder steps/mm ?

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Hi

The jerk on the extruder seems really low, 13 mm/min is only .217 mm/sec. I run a direct drive extruder and have the jerk set to 600 mm/min.

I got really terrible prints when the jerk was set too low.

Were you able to tighten the steel core belts good enough? It could be overshooting from loose belts. Others have reported, the steel core belt also needs bigger dia. pulleys to run smooth.

Thanks for all the tips!

Quote
dc42
It looks like over-extrusion to me. Have you calibrated the extruder steps/mm ?

Yes, up to several times by marking white filament, extruding a specific amount and then comparing that with actual extruded amount. We have played around with the extrusion multiplier parameter as well (lowering it) and we quickly reached a state where it didn't fill the gaps properly. Maybe we lowered it too much? We'll try fine tuning this a bit more.

Quote
SDM
Hi

The jerk on the extruder seems really low, 13 mm/min is only .217 mm/sec. I run a direct drive extruder and have the jerk set to 600 mm/min.

I got really terrible prints when the jerk was set too low.

This is interesting. We'll try experimenting with this.

Quote
o_lampe
Were you able to tighten the steel core belts good enough? It could be overshooting from loose belts. Others have reported, the steel core belt also needs bigger dia. pulleys to run smooth.

They were definitely harder to tighten than common belts, and they may not be tight enough at this point. However, we haven't noticed any difference in print quality between the old and the new belts. A different (more continuous, less discrete) way of tightening the belts is on the to-do list.

I think you need to check a few things.

You said you calibrated extruder by measuring amount extruded, this is not how you do it. You measure filament going in to the extruder not coming out. This is because at the point where it goes in its diameter is known eg 1.75mm or 3mm and so the volume of material being pushed in is then known, at the point it is extruded and comes out variations in nozzle size or die swell where filament expands as it extrudes, means it's diameter is different on every setup. So try again but if you mark 100mm on the input side of the extruder and ask for 100mm then 100mm should go in. Otherwise adjust proportionally.

As for flex3drive it needs low jerk and acceleration to work, so jerk at 5-15 and acceleration at 200 is a good starting point. The reason for this is it's 40:1 geared and needs low motor current to be able to reverse from high speed to retract. At the sort of high jerk/acceleration that a direct or low geared (like titan) extruder uses, you'll be lucky if the motor stalls and unlucky if you eventually damage the extruder.

To get more snappy retracts out of it (not usually because a slower retract is insufficient to stop strings and blobs, it often is but to speed up printing models with a lot of retracts) you can print a model with lots of retracts and gradually increase the acceleration as you print until the motor stalls or sounds like its suffering, jerk is less important with high gearing but play with it if you want in small incrementsike going from 10 to 12,14,16, 18 etc.. Only ever adjust one parameter at a time, I can't stress this enough.

Once you've done that try printing something like this [www.thingiverse.com] or just two cubes one with 10% infill the other 95%. After 6 layers or so, look at the infill if it's stringy on the low infill-cube increase esteps by 1% and keep watching, repeat if needed, when it's not stringy look at the high infill-cube if you can't see tiny gaps between the infill lines lower esteps. This way you get very precise tuning.

Experience enables you to learn to do this on the fly at the start of printing with a new filament or after a change to your machine.

Edited 1 time(s). Last edit at 12/20/2017 04:58PM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Thanks for the reply and your general tuning tips, DjDemonD.

Sorry for the poor phrasing before, but we're more than aware that we have to measure the filament on the way into the extruder and not out of it. What we did was to take a measuring caliper and measure from the effector to a certain distance (e.g. 20 mm) up along the filament prior to extrusion and place a mark there as accurately as possible, then extrude that same amount and then see the difference. Our conclusion has several times been that we're spot on with our extrusion steps/mm. If we'd measured the outgoing filament, things would've been so off that nothing would work at all!

For clarification: are your 5-15 jerk values mm/s or mm/min? I assume you're talking about extruder values only here? Would you say that extruder acceleration at 200 mm/s² is low? We have never increased ours beyond 100 mm/s².

What you say about retraction here is fine, but is this relevant to the bulging in the corners?

This is our printer, by the way: [www.norwegiancreations.com]

We have already printed two similar cubes, one with default movement speed values and one with a 20% speed override (the exact same gcode). In the slow print the problem with the corners disappeared, as seen in this picture, so we have a theory that it's either the belts that are not tightened enough, that it's something related to the extruder which deccelerates to slowly when approaching corners relative to the effector, or that it is due to our E3D Volcano hot-end, which has a large chamber and thus might kind of "low-pass" (i.e. filter out the high frequencies of) the output of the extruder somewhat without compensation (such as the M572 command).

Additional tips are much appreciated!

Thanks for the clarification on calibration, and apologies for the mix-up. However try to use more than 20mm you are potentially looking for just a few percent either way so use 100mm it's much easier to see.

If you want to save filament take the hotend off, and just pass the filament through the extruder. If doing this extrude 200mm this gives way more accuracy to the result. This is a test worth doing anyway, just to rule out your hotend as a problem. With the hotend attached the amount extruded should be the same as it is with no hotend, so long as you extrude slowly and with sufficient temperature.

With any delta the belts are critical, definitely replace them, and tension them evenly.I am still searching for the best belts, steel work great until the steel fatigues and snaps, the cheapy ebay belts I'm never very convinced by, I can't find a supplier in the UK of genuine gates 2gt powergrip open belt.

As for the Jerk that's mm/min you don't need the extruder to start from 0. The acceleration is the crucial bit. As for whether 200 is low thats subjective. If you own a titan extruder you run 3000 accel. and I have run 9000 accel so 200 seems ridiculously low, I mention retraction to try to explain why you need these low accel values, you can't ask a stepper motor running at 400 rpm to suddenly change direction for 0.2 seconds then start running forward again, without decelerating it gently, the accelerating it gently. I'd start with accel at 100, then increase it in 50's as you print a tower and chose the settings that look best, if the whole tower looks like crap its not the acceleration setting causing your problem.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Thanks again, DjDemonD.

The reason for 20 mm was that our caliper isn't much longer, but I totally see your point and next time we can use a measuring tape or something instead. The hotend-off trick seems handy.

The belts we got were these: [www.ebay.com], which you might call "cheapy eBay belts". I think having a pretty tall printer like ours (1 meter tall towers) introduces/amplifies this issue since the elasticity in the belts increases due to the larger span. We thought this would be fixed by replacing the standard belts with steel rod belts, but we didn't take into account that they were more difficult to tighten. And if in addition to this the reinforcement might snap, these belts seem like a poor choice.

The reason we haven't gone beyond 100 mm/s² is that we wrecked the gears in the extruder once at that acceleration. After this happened we've stayed at 50 and 75 mm/s², so we're a bit hesitant to increase it further.

I started with those belts (same size printer) which do work if you stick to 30 tooth pulleys but then you lose resolution. Much smaller pulleys and you get issues. I was using 16t pulleys and my belts reinf9crment snapped in just one place, in the position which sits near the motor pulley when then effector is near 0,0,0. It caused very odd leaning parts which straightened up after 55mm as the belts higher up were not snapped.

One proposed solution is to use 30 tooth pulleys top and bottom and drive then from a closed belt with 2:1 reduction.

Youll get better performance immediately with fibre reinforced belts. I'm also told wide belts help like 9mm but haven't tried them myself.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

I found it easier to measure 150mm up the filament, call for 100mm to be sent through the extruder, then measure the distance to the mark. 150 - remaining length = amount used. It's easier to find the mark, especially if you're over-extruding.