Cranking up the print speed

[reeper 0]

Greetings to all,

Up till now I’ve tended to print slowly at lower layer heights(60mm/s, 0.04 - 0.1mm, 200 temp, 0.4mm nozzle) with some success.

I decided today that I wanted to print the automatic transmission model (http://www.thingiverse.com/thing:34778) and that I wanted to sacrifice some quality for gains in print time.

After consulting https://groups.google.com/forum/#!topic/ultimaker/3zkPAMi9z-0 I upped the temp to 220, the speed to 120mm/s and the layer height to 0.2mm for a preliminary test.

Printing one of the gears produced what appeared to be under-extrusion, ridges in the outer shell and a top skin that didn’t close up(3 solid layers, 15% infill).

I decided to increase the amount of solid bottom and top skin layers to 5 and the temp to 230. The gear ended up more or less the same, here are a few photos to illustrate the problems I’ve encountered

ridges caused by under-extrusion ?, I believe my belt tension to be decent for both the short and long belts(I have added banana tensioners)

 

5 solid layers but under extrusion appears to stop them from closing up

 

bottom skin works out fine with an initial layer printing speed of 20mm/s

 

I would greatly appreciate any advice on how to improve my results with faster prints.

 

[illuminarti 18]

You're probably pushing the limits of how much plastic you can extrude per second. With 0.2mm layers, at 120mm/s you're hitting 9.6mm³/s (= 0.2 x 120 x 0.4, assuming that your shell thickness is an exact multiple of your 0.4mm nozzle width).

That's at or above the limit of how much plastic you can squeeze through a standard nozzle - the back pressure is going to cause the extruder to slip some, and could even lead to head clogs if molten plastic gets squirted in places it shouldn't go.

You could try upping you flow rate in the slicer (or Ulticontroller) to maybe 110 or 115% to compensate for any slipping in the extruder - but beware that you're also increasing the chance of the filament grinding, or head clogging by doing this.

A better solution might be to reduce your layer height a bit to reduce the volume per second, and hence pressure.

 

[illuminarti 18]

Also, it's normal to need more solid layers to close over infill - I typically aim for at least 5 layers, but find it sometimes can take up to ten, especially with thin layers and fairly low infill percentages.

 

[reeper 0]

You're probably pushing the limits of how much plastic you can extrude per second. With 0.2mm layers, at 120mm/s you're hitting 9.6mm³/s (= 0.2 x 120 x 0.4, assuming that your shell thickness is an exact multiple of your 0.4mm nozzle width).

That's at or above the limit of how much plastic you can squeeze through a standard nozzle - the back pressure is going to cause the extruder to slip some, and could even lead to head clogs if molten plastic gets squirted in places it shouldn't go.

You could try upping you flow rate in the slicer (or Ulticontroller) to maybe 110 or 115% to compensate for any slipping in the extruder - but beware that you're also increasing the chance of the filament grinding, or head clogging by doing this.

A better solution might be to reduce your layer height a bit to reduce the volume per second, and hence pressure.

 

 

Also, it's normal to need more solid layers to close over infill - I typically aim for at least 5 layers, but find it sometimes can take up to ten, especially with thin layers and fairly low infill percentages.

 

Many thanks for the advice, for some reason I had decided that the parameters I chose were within the limits of what was possible with a 0.4mm nozzle.

I decreased the layer height to 0.18mm and the speed to 100mm/s making for a flow of 7.2mm³/s. Unfortunately this still results in visible under-extrusion both on the sides and the top skin of the gear(the skin closed with 7 solid layers but the thread is visibly thin). I checked the teeth marks on the filament a few times and saw no evidence of grinding.

The temp was at 220, perhaps I should have used 230 - 240 ?

I can lower the height and speed further but I’m starting to wonder if it isn’t a hardware issue.

 

[illuminarti 18]

Higher temp might help the plastic flow better, yes - but you're still trying to print pretty fast. So it doesn't surprise me that there's a bit of under-extrusion still. The filament won't grind necessarily, but the teeth marks get closer together, reducing your effective steps-per-e setting.

Some things you might want to check:

1) Is your extruder bolt clean, and properly assembled?

2) Assuming that you have a v2 extruder, with a spring - is the spring about 11mm long when the extruder is unloaded? Is your delrin pressure wheel in good shape - has it deformed any?

3) Does the filament slide easily in the Bowden tube?

4) Have you checked that you're using the right filament diameter value in the slicer?

5) Have you calibrated your steps-per-e setting? What is it?

I'd try dropping the layer height to 0.1mm. And see if it works better with a much lower flow rate. Then start to increase it to get a sense of how flow rate affects the extrusion amount. As I mentioned before, you can always try upping the flow rate in Cura, or via the Ulticontroller to compensate.

(I did a bunch of testing of the relationship between flow rate and under-extrusion here).

 

[gr5 2,167]

I printed the same transmission. I printed the whole thing at 240C and fast. Some parts at 100mm/sec some at 150 and some at 200mm/sec. They all showed some minor underextrusion but the transmission works great. Back then I was using a heated bed (not anymore - don't need it) and the heated bed was at 70C for most of the prints. I guess I'm not as picky as you are but the higher temp definitely helps the PLA flow like honey. Here's pics of my print:

http://www.thingiverse.com/make:32987

(looking back at it I should have leveled the bed better for more squished first layers)

After you print all the parts, read my comments at the above link to get it working nicely. It was quite a head scratcher to get it to work in all 6 gears without binding. Now it moves smooth as in Emmett's video.

 

[reeper 0]

Higher temp might help the plastic flow better, yes - but you're still trying to print pretty fast. So it doesn't surprise me that there's a bit of under-extrusion still. The filament won't grind necessarily, but the teeth marks get closer together, reducing your effective steps-per-e setting.

Some things you might want to check:

1) Is your extruder bolt clean, and properly assembled?

2) Assuming that you have a v2 extruder, with a spring - is the spring about 11mm long when the extruder is unloaded? Is your delrin pressure wheel in good shape - has it deformed any?

3) Does the filament slide easily in the Bowden tube?

4) Have you checked that you're using the right filament diameter value in the slicer?

5) Have you calibrated your steps-per-e setting? What is it?

I'd try dropping the layer height to 0.1mm. And see if it works better with a much lower flow rate. Then start to increase it to get a sense of how flow rate affects the extrusion amount. As I mentioned before, you can always try upping the flow rate in Cura, or via the Ulticontroller to compensate.

(I did a bunch of testing of the relationship between flow rate and under-extrusion here).

 

1) The bolt was clean and seemingly well positioned

2) I have the v2 extruder drive and the spring was around 12.7mm long when the drive was disengaged. I tightened it as you suggest and it did make a significant improvement

3) Slides fine

4) I had measured an average diameter of 2.8mm for the filament when I first started using it, a quick check showed that it had increased to 2.9 so I updated the parameter. I knew that this could potentially worsen the issue but I didn't want to falsely compensate for the under-extrusion(although I guess past the flow limit the system takes strain with minimal flow gain)

5) Not calibrated(bad, I know), I use the 836 value I've seen associated with the v3 bolt and the v2 extruder drive.

I managed to reduce the issue somewhat, in spite of the diameter increase, by tightening the extruder drive and upping the temp to 240. I can now print with a speed of 110mm/s, 240 degrees and 0.18m layer height resulting in only a slightly visible under-extrusion. Thanks again for your help

Has anyone achieved better results with the stock nozzle ?

 

I printed the same transmission. I printed the whole thing at 240C and fast. Some parts at 100mm/sec some at 150 and some at 200mm/sec. They all showed some minor underextrusion but the transmission works great. Back then I was using a heated bed (not anymore - don't need it) and the heated bed was at 70C for most of the prints. I guess I'm not as picky as you are but the higher temp definitely helps the PLA flow like honey. Here's pics of my print:

http://www.thingiverse.com/make:32987

(looking back at it I should have leveled the bed better for more squished first layers)

After you print all the parts, read my comments at the above link to get it working nicely. It was quite a head scratcher to get it to work in all 6 gears without binding. Now it moves smooth as in Emmett's video.

 

Did you achieve those speeds with a larger diameter nozzle ?

I wish I had a heated bed at this point, these larger parts are incredibly difficult to remove from the bed. I think the squished initial layers I've seen on my gears will interfere with their alignment(I leave a gap between the nozzle and bed large enough to push a sheet of paper between with a little friction)

Thank you for the pointers on the transmission assembly, hopefully I'll have a positive result to report once I've printed all the parts.

 

[gr5 2,167]

Did you achieve those speeds with a larger diameter nozzle ?

 

Nope. But I think you care a little more about the quality than I did. Now that your spring is a little tighter I think we probably will get similar results at a given speed.

 

I think the squished initial layers I've seen on my gears will interefere with their alignmen

 

I doubt it. I had some warping in my gears and they mesh perfectly.

I recommend you print the pins all facing the same way and you reduce their width and height by a scale factor of 0.9 (but leave the long direction scaled at 1X). I did not do this and regret it. I did probably 4 minutes filing per pin!

Recently I printed out Emmett's "heart gears" and I printed them at 90% (but left the length of the pins at 100%) and they were just about perfect and didn't need any filing. Emmett designed the heart gears and also this transmission.

 

[reeper 0]

Another query:

I’ve printed the anulus with the above mentioned settings(240 degrees, 110mm/s, 0.18mm layer height) and ended up with the following result:

 

The inset text is not so inset unfortunately, could the rippling be due to over tightened belts since I have banana tensioners installed ?

 

Edit:

 

 

I recommend you print the pins all facing the same way and you reduce their width and height by a scale factor of 0.9 (but leave the long direction scaled at 1X). I did not do this and regret it. I did probably 4 minutes filing per pin!

 

Recently I printed out Emmett's "heart gears" and I printed them at 90% (but left the length of the pins at 100%) and they were just about perfect and didn't need any filing. Emmett designed the heart gears and also this transmission.

 

 

Odd, I printed the heart gears and found the pins to be fine if only a bit too long. The gears were a little loose and I have yet to print shorter pins to force them to fit more snugly.

 

I printed a set of pins for the transmission, pushed a medium pin into one of the idler gears and it seems to fit fine, the idler rotates quite freely. I don’t have the carrier printed yet to test the length though

 

[gr5 2,167]

The inset text is not so inset unfortunately

 

yes. Oh well. That would be a good design change.

 

could the rippling be due to over tightened belts since I have banana tensioners installed ?

 

Not sure. The rippling could be "ringing" caused by each line segment but it seems unlikely as the change in velocity is small at each vertex. More likely the rippling is caused by the infill pattern overlapping too much. The latest Cura calculates infill overlap differently and it's kind of a problem. Best fixed by making the "skin" .8mm so that the infill is less likely to bump out the skin.