Welcome! Log In Create A New Profile

Advanced

Dialling in the accuracy - 10mm cube test...smiling smiley

Posted by NumberSix 
Dialling in the accuracy - 10mm cube test...smiling smiley
August 02, 2011 01:54PM
Hi all,
Your comments and guidance would be most welcome, as I attempt to increase the accuracy of my little repstrap. I've had a few enthusiastic days now of printing various things and have been quite happy with the intial results. I did give reasonable consideration to Skeinforge (v41) settings and calibrated my firmware's Steps_per_mm 'magic number' also before printing some calibration pieces last week, but I'm now ready to return to those settings in an effort to improve print quality further.

I can provide as much detail as needed, but initially I thought a 'close-up' picture might be of more value (see pict below). I'm using Skeinforge v41, Repsnapper to send gcode. Gen6 electronics. Wade-like extruder, 3mm filament, .4mm nozzle.

Temperature control is through RepSnapper - Target temp 210, it hovers around 200. (I have a fan running to keep cold-end cold)

Basic SF settings: Carve... Layer Thickness .33, Dimension... Filament Diameter 2.836mm (average of 10 random measurements), Filament Packing Density (ratio) 1.0, Speed... Feed Rate 20.0, Flow Rate Setting 20.0, Perimeter Feed Rate over Operating Feed Rate (ratio) 0.5, Perimeter Flow Rate over Operating Flow Rate (ratio) 0.5. Travel Feed Rate 50.0.




Any observations and comments on how I can improve the quality of print is welcome. What other setting would be valuble for me to share at this point? There are quite a few.

Regards and thanks,
No.6


[numbersixreprap.blogspot.com]
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 02, 2011 03:49PM
This is too small of a part to be printing by itself unless you use a fan to cool the part. The problem is that there's not enough time for the previous layer to cool down and harden before laying the next layer down. Even if you use the Cool feature in skeinforge, the nozzle ends up heating the part either with orbit or slow down. A single-walled cube (at least 20 mm sides) is probably a better test object. You can measure the width of the filament around the fifth or higher level to make sure that its correct. The top layer on your part doesn't seem to be completely filled.
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 02, 2011 04:47PM
Is that PLA or ABS? I extrude PLA at only 175-180 max which is plenty hot enough, are you using a heated bed? and what are you fill settings? I do temp control through SF41 as I find it more accurate. and with the settings I have I have now printed 4 sets of Prusa parts and 1 set of mendel parts.

If you want I can email you my current settings in a PDF if you pm me your email address. Might give you somewhere to start from then you can continue tweaking


__________________________________________________________________________
Experimenting in 3D in New Zealand
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 02, 2011 04:59PM
The temperature depends on your hot end design, thermistor, temperature table and the plastic. The temperature setting for the old nicrhome wire design read low while the resistor in block design reads a little high. Even for PLA, the temperature can change depending on the source.

It also doesn't matter what host (repsnapper) you use or where the g-code is generated (repsnapper, reprap host or skeinforge) since the temperature regulation is being done by the firmware.
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 02, 2011 07:57PM
I looked at your blog and some of the other prints you did. I would first lower the temp if it were my printer and see what happens. My bet is the temp is 15-20C too high.

I've found the most useful test is a solid 1" cube stopped about .2" up. That way you can see how close the part is to completely filled, and lots of other things.
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 03, 2011 06:22AM
Thanks for all for great comments. Here’s some specific feedback on some of the questions raised…

Brnrd: I’ll take on board what you are saying about a 10mm cube is too small a printable item without cooling. Taking other comments on-board also, I may look for a larger simple test piece, possibly one solid and one single-wall hollow. If anyone can point me to suitable STLs of send me some via PM that would be great.

NelsonRap: I’m using 3mm PLA. I don’t have a heated bed. My test does look a bit droopy alright, so I’m going to back the heat down a bit for my next test runs. I’ll drop you a PM and would appreciate a copy of your current settings for review.
My SF “Infill Width over Thickness (ratio):” is set to 1.4.

Brnrd: I’m using a resistor/thermistor arrangment (from Mendel-parts.com) in an aluminium block as my heater. I’ve no way of checking the accuracy of the temperature it’s reaching, when I set 210DegC and it hovers around 200DegC, but from the feedback it does sound like I’m too high so I’ll back it down. It is controlling it fine though. The Heater “on” light on the Gen6 goes from solid to flickering as it approaches target temperature.

Andrew Kiehl: Tks Andrew, will back down the temperature and try a different test cube.

I’ll post updated test results once done.
Tks,
No.6


[numbersixreprap.blogspot.com]
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 03, 2011 07:58AM
I have used the Mendel circuit board spacer as a good first test shape. It is about 20x20x10 which is big enough to not need cooling and has four holes which show up a lot of problems, see examples here: [hydraraptor.blogspot.com]


[www.hydraraptor.blogspot.com]
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 03, 2011 09:30AM
Thank you sir! That will do me nicely for some test printing. Got it on Thingiverse [www.thingiverse.com]


[numbersixreprap.blogspot.com]
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 03, 2011 11:39AM
I dropped the temperature by 10C, to a Target Temperature of 200 (RepSnapper). Once up to temperature (Gen6 heater light flickering) RepSnapper displayed a current temperature of between 191 and 194.

Looks like the rediated heat from the hot-end is concentrated because of the 10mm work area. It's not a good choice of test piece at this point I think.


I also printed the 'circuit board spacer', suggested as a test piece by Nophead. It's about 23mm square, and has 4 holes. Below is a photo of this evenings print of this piece. The wall doesn't look bad, but the infill at the top looks poor. The underside was also gapped. I'll hold off on making any setting changes until folks have had a chance to review and perhaps comment.



Thanks,
No.,6


[numbersixreprap.blogspot.com]
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 03, 2011 02:59PM
So this was printed with 100% infill, linear infill, and infill and perimeter width over thickness of 1.4? Are you using any additional shells?

A better topside picture would also help. Most of it is hard to make out.

It almost looks like your extruder is being intermittent. (again, hard to tell from the pic)
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 03, 2011 03:05PM
Yes, it's very difficult to see details in the photo. The sides don't line up well even on the board spacer. 20 mm/s might still be too fast for this part. I would try printing slower (10 or lower) to give the part more time to cool down or use a fan to blow on it. You can also use the Cool plug-in and select slow down with a minimum layer time of 40 s.

BTW, you should watch how the first 3 layers go. It should be done at 100% infill. At the 3rd layer, you should be able to see if the infill is correct. Stop the print if it doesn't look right. If there are gaps, then you need to lower your filament packing density ratio below 1. If there's sufficient gap, you might be able to measure the filament width with a caliper and use the measurement to adjust the FPDR. If there's too much plastic, then you need to raise it above 1. I've never set this larger than 1 so I don't know if it will take it.

Edited 1 time(s). Last edit at 08/03/2011 03:09PM by brnrd.
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 03, 2011 03:12PM
Have you tried slower feed rates, like 25-35mm/sec?
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 03, 2011 11:22PM
To build smaller parts, one can always put G4's after each layer, e.g. G4 20000 for 20 seconds pause.


Generation 7 Electronics Teacup Firmware RepRap DIY
     
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 04, 2011 03:37AM
Traumflug Wrote:
-------------------------------------------------------
> To build smaller parts, one can always put G4's
> after each layer, e.g. G4 20000 for 20 seconds
> pause.

That would result in a big blob unless you move your nozzle away from the part and then wipe off the ooze.
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 04, 2011 11:04AM
brnrd Wrote:
-------------------------------------------------------
> Yes, it's very difficult to see details in the
> photo. The sides don't line up well even on the
> board spacer. 20 mm/s might still be too fast for
> this part. I would try printing slower (10 or
> lower) to give the part more time to cool down or
> use a fan to blow on it. You can also use the Cool
> plug-in and select slow down with a minimum layer
> time of 40 s.
>
> BTW, you should watch how the first 3 layers go.
> It should be done at 100% infill. At the 3rd
> layer, you should be able to see if the infill is
> correct. Stop the print if it doesn't look right.
> If there are gaps, then you need to lower your
> filament packing density ratio below 1. If there's
> sufficient gap, you might be able to measure the
> filament width with a caliper and use the
> measurement to adjust the FPDR. If there's too
> much plastic, then you need to raise it above 1.
> I've never set this larger than 1 so I don't know
> if it will take it.


Hi Brnrd,
It was nearly more difficult to take a decent photo of the object than to print it, but here you go... some more photos of the same object! smiling smiley I'm off to print again now, this time with slower speed. I'll only print the first few layers as you suggest and observe those closely. I’ll resist changing too many settings, and not the ones I do change, then report back!

Looking at the top of the object now it looks like it didn't in-fill the top at all (I’ll examine the layers in Skeinforge). One of the photos shows the bottom, which looks 'ok'.
My Infill settings are currently: Line… Infill Perimeter Overlap (Ratio): .15, Infill Solidity (ratio): .2, and Infill Width over Thickness (ratio): 1.4 <-- can anyone comment on these Infill settings based on what you can see in the photos?
The Solid Surface Thickness (layers): is set to 3.

Given that my nozzle is .4mm I’m wondering how that might impact such settings as Carve… Layer Thickness (currently 0.33) and Perimeter Width over Thickness (ratio) (currently 1.6)?





In the prints you see I currently have a Speed Perimeter Feed and Flow over Operating Feed and Flow Rates (ratio) set to .5. I thought this might help outer quality by slowing down when printing perimeter. I’m happy to set it back to “1” for testing if it’s not helpful at this point.

(In the bottom shot, the line across the bottom is a gap in the blue tape surface.)

I’ll print some more and see how I get on.
Tks to all for now.
No.6


[numbersixreprap.blogspot.com]
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 04, 2011 08:58PM
It looks to me like your extruder can't keep up at 20mm/s, but seems to work at 10mm/s. (why your perimeter looks ok, but the infill is incomplete)

Try this:

First, set feed and flow rate to 10, then set Perimeter Flow Rate over Operating Flow Rate (ratio) to 1

Second, Set fill to 100% and Infill Perimeter Overlap (Ratio) to 0. Partial fill will tell you next to nothing about your feed/flow relationship

Third, set both perimeter and infill width over thickness to the same value. 1.3 or 1.4

Fourth, set travel feed rate to 30

Reprint, and let us see what it looks like smiling smiley

I realize it's really slow for now. As you narrow down what is causing the issues, you can address them individually.
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 05, 2011 12:21PM
Hi again,
Another test print and photos...
Note: Anyone scrolling down to this point, the 10mm cube is too big a challenge for now so instead I’m testing using a "circuit board spacer" reprap component, off Thingiverse. It’s roughly 24mm square. I’m only printing the first 8 layers or so, and filling it solid for test purposes. Speed down to 10mm/s all round. Fill 100%. 3mm PLA, .4mm nozzle. Heating to 200DegC set by Repsnapper, all other settings via Skeinforge v41.

Infill width over Thickness 1.4
Perimeter Width over Thickness 1.6 (Sorry Andrew – only figured which setting you were saying make the same as Infill Width over Thickness after I had printed and photographed. May change it later).
Speed settings:
Feed/Flow = 10
Perimeter Feed Rate over Operating Feed Rate (ratio) = 1.
Perimeter Flow Rate over Operating Flow Rate (ratio)= 1.
Travel Feed Rate = 30mm/s

Here are photos of the latest test print using the settings above, taken from different angles. (5mm squares on the grid paper).

Top


Bottom


Edge and Top Face


Top Face

Edge-on

Feedback on any other fine-tuning I might consider at this point would be welcome. Keep in mind this is very close-up of a very small component, and in general I’m happy with it, but am also ok with trying some further fine-tuning. I just don’t have the experience to judge piece.

It is becomming evident that my home-brew extruder doesn't like going fast so I may have to put it back "up on blocks" one of these weekends, and do a bit of rework. That may help speed things up, but I would like to know I have given a fair go at setting things up when printing slowly before I step on the gas! smiling smiley


Tks,
No.6


[numbersixreprap.blogspot.com]
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 05, 2011 01:06PM
That most recent picture of the top looks like an extruder problem to me - the alternating fat/thin parts within a single thread really aren't what you want to see. On my machine (which isn't a reprap) that means I forgot to close the filament tensioner or that I need to floss the drive teeth. I'd carefully go over the filament mechanism again and make sure it's all tensioned/aligned/flossed..

It still looks a little hot to me, too/
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 05, 2011 01:34PM
There are large gaps between perimeter and loops so they're not touching each other and the infill looks uneven.

When you extrude in air, do you get a round filament with a fairly uniform diameter? If you haven't checked this, then I recommend doing this. When you do this, you might want to calibrate your extruder at the same time. Do this by putting a mark (you might have to use tape since your feed is black) on the filament say >50 mm up from the entrance of the extruder, measure the distance to the mark, extrude 50 mm at a conservative speed like 120 mm/min, and then measure again. Multiply the measured distance by 2 and divide by 100. Enter this number as the filament packing density ratio in the Dimensions plug-in.

Also, did you set the retract distance to a number a little over 1 mm and the restart distance to 0? Retraction speed should be around 15 mm/s.
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 05, 2011 01:38PM
I never use W/T less than 1.5, except for support, as it makes weak objects because the filament is more round and thus has less contact with the layer below.


[www.hydraraptor.blogspot.com]
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 05, 2011 06:25PM
Definitely some form of extruder trouble. My guess is not enough force pushing the filament against the hobbed bolt, or a slightly eccentric hobbing job.
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 06, 2011 09:20AM
David – Thanks for your observations. I’m learning to read the signs and peoples explanations of their observations are invaluable. I have ‘reasonable’ pressure on the filament idler bearing, but I’m beginning to question the consistency with which I hobbed the drive spindle. It was my first effort. I think I’ll make a new one. I’ll also take on-board your observation about it still looking too hot. I presume the general ‘melted’ look makes you say this?

Brnrd – I had the Infill Perimeter Overlap (Ratio) set to 0, (for test purposes, as suggested by Andrew). I would have expected the fill loop to just butt up against the perimeter, but as you observed there is a gap. Anyone know what might cause that gap if the Infill Perimeter Overlap is 0? (I can obviously set a reasonable overlap when I need to, but I’m interested in what might be causing the gap. Mechanical inaccuracies/backlash on the turn perhaps (I’ve no idea)? Might not be worth worrying about if I just set an overlap.

The filament extrudes nicely in air. Nice and round and has a consistent diameter or about .43mm. I drilled the brass nozzle hole by hand with a .4mm drill. See photo of filament.



I have calibrated and set the E0_STEPS_PER_MM value in the firmware for the Gen6, and left the Dimension Filament Packing Density Ratio set to 1. I believe this is equivalent to the calibration you describe, and a value of 1 ok for PLA, but I’m open to correction. I’ll re-calibrate once I remake the hobbed shaft.

In answer to your other question, I have the ‘retract distance’ set to 2mm and a speed of 30mm/sec (is this too fast?). I have the Restart Distance set to 0.

Nophead – With my Width over Thickness setting of 1.4, I printed some novelty items last evening also, and a small protrusion on one of the printed objects just snapped clean off, giving credence to your suggestion of weaker bonding with lower W/T Ratios. As a matter of practice do you set both Perimeter Width over Thickness and Infill Width over Thickness to the same values?

Andrew – I’m with you on your guess at this point! smiling smiley At least I have a functioning printer which will allow me to build a new and better extruder assembly. I’ll cut a new hobbed bolt also.

Thanks to all for now. I’ll park the calibration exercise at this point but will return to it once I have a better running extruder. I’ve learned quite a bit about Skeinforge settings in the process, which I can put to use when I resume calibration tests.

One last thing… I spend so long sqinting a very small printed objects over the last few days I decided to print these to help! smiling smiley They are of good Reprap stock!... [www.thingiverse.com]



Regards,
No.6


[numbersixreprap.blogspot.com]
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 06, 2011 10:13AM
Quote

As a matter of practice do you set both Perimeter Width over Thickness and Infill Width over Thickness to the same values?

Yes I always keep them the same and very rarely use anything other than 1.5.


[www.hydraraptor.blogspot.com]
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 06, 2011 12:20PM
> Brnrd – I had the Infill Perimeter Overlap
> (Ratio) set to 0, (for test purposes, as suggested
> by Andrew). I would have expected the fill loop to
> just butt up against the perimeter, but as you
> observed there is a gap. Anyone know what might
> cause that gap if the Infill Perimeter Overlap is
> 0?

Its likely your flow settings are still a little off. so the perimeter and infill is somewhat more narrow than it 'thinks' it is.

Once you fix your extruder issue, it will be easy enough to properly adjust the flow rate.

Appropriate width over thickness depends on a lot of stuff, and there is no strictly correct answer. I re-read the post and saw you are using a layer thickness of .33, so a 1.5 w/t is probably more appropriate for your settings.

As the layer height approaches the nozzle diameter, the w/t needs to be decreased. Otherwise, the filament is not stretched when it comes out of the nozzle (and therefore can't bridge gaps.) I haven't noticed any significant strength difference from it though.
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 06, 2011 03:30PM
Andrew Diehl Wrote:
-------------------------------------------------------
> > Brnrd – I had the Infill Perimeter Overlap
> > (Ratio) set to 0, (for test purposes, as
> suggested
> > by Andrew). I would have expected the fill loop
> to
> > just butt up against the perimeter, but as you
> > observed there is a gap. Anyone know what might
> > cause that gap if the Infill Perimeter Overlap
> is
> > 0?
>
> Its likely your flow settings are still a little
> off. so the perimeter and infill is somewhat more
> narrow than it 'thinks' it is.
>
> Once you fix your extruder issue, it will be easy
> enough to properly adjust the flow rate.
>

I agree with Andrew that it means that your flow rate is still a little off. But I doesn't necessarily mean that you have an extruder issue. The extruded filament in air doesn't look as bad as in the print. I think the infill in the print might be irregular because of the same reason. Since the flow rate is lower than ideal, the fill doesn't make contact with the perimeter and so it gets dragged back when the head reverses direction. In those areas, you see thicker infill threads.

Try lowering the filament packing density ratio to a number less than 1 like 0.8 and see what happens.
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 08, 2011 06:25AM
NumberSix Wrote:
-------------------------------------------------------
> David – Thanks for your observations. I’m
> learning to read the signs and peoples
> explanations of their observations are invaluable.
> I have ‘reasonable’ pressure on the filament
> idler bearing, but I’m beginning to question the
> consistency with which I hobbed the drive spindle.
> It was my first effort. I think I’ll make a new
> one. I’ll also take on-board your observation
> about it still looking too hot. I presume the
> general ‘melted’ look makes you say this?

Yes on the melted look. That said, black plastic is notoriously hard to photograph so if you think your temperature is already pretty good, ignore my comment that it still looks a bit high..

Have you tried *gently* holding your fingers on the filament as it goes into the drive while it's printing? If such a thing is possible (and safe!!) on your printer, you might try doing that and seeing if you can feel the filament stop going into the drive when the drive is still turning.

> The filament extrudes nicely in air. Nice and
> round and has a consistent diameter or about
> .43mm. I drilled the brass nozzle hole by hand
> with a .4mm drill. See photo of filament.

I'm not sure that's as valuable of a test as it might seem. If the filament is slipping, it could still produce an even-width noodle out of the hot end.

Some (hand-crafted?) gcode that just draws a single layer grid on the platform might be a better test. Maybe make a very thin (in height) object that would cover most of your print bed, set Base Layers = 1 in skeinforge Raft then edit the object out of the gcode so you're just left with the raft. Having something like that might give you a better idea of how the extruder is behaving. I like this idea a lot better than the finger-on-filament idea.
Re: Dialling in the accuracy - 10mm cube test...smiling smiley
August 14, 2011 11:33AM
Hi all,
Update: I removed the extruder drive shaft and re-hobbed it cleaner and deeper (M3x0.5 tap). I tightened the springs on the idler bearing also once re-assembled, and tigntened some other cold-end motor mounting bolts that had come slightly loose. The whole assembly grabbed and fed the filament much better. Deffinetly no slippage now.

I adjusted some Skeinforge setttings also before re-commencing test printing.
Speed – Feed and Flow Rates 15mm/s, With over Thickness (both) 1.5, Layer Thickness .35mm.

Looking at thinks now, I think my layer thickness should be reduced, given my nozzle hole is .4mm.

Extruder retraction changed from 1mm to 2mm and retraction speed from 20 to 30mm/sec. Cleaned up blobs a lot!

I re-checked my firmware flow setting (E0_STEPS_PER_MM) and increased it slightly – now set to 540.

I’ve lowered the temperature from 200Dec C to 195Deg C. That’s its Target Temperature set in RepSnapper and it hovers at an actual of about 185. (I don’t have a digital thermometer so I’m going on look/feel of extruded plastic. When I lowered the target even more the stepper had trouble pushing the filament, and started skipping steps. I could increase power slightly to the stepper, but I think that would risk increasing hot-end pressure – it’s a balancing act.

I’ve removed the blue tape from my Perspex platform and the PLA took to the Perspex very nicely. It separated easily when cold and gave a good finish to the base of every print, visible in photos.

Some new test prints: (Ignore my little number stickers - they just are reference numbers for my own notes.)
Circuit board spacer (my square test piece) [www.thingiverse.com]
Underside - started a bit too close to the Perspex, but it gave an interesting finish.


Topside



LM8UU Linear bearing holder [www.thingiverse.com]
Bottom


General view of LM8UU holder



Endstop bracket [www.thingiverse.com]
Top


Underside


Side


The infill on what should be a solid top surfaces needs improvement. It may be my layer height for my given nozzle diam (.35mm for a .4mm nozzle). I’m not sure. I’m going to change parameters slowly and continue my observations. A general remark is that these photos are very close-up and stepping back the items I’m printing are robust and very useable.
(oh… there’s a great bunch of test objects here and some interesting discussion with it also, which I must return to: [www.thingiverse.com]

Late update: In removing the extruder earlier today to make some change, I snapped off the wire from the heat resistor. I don’t have a spare resistor so that puts me ‘dead in the water’! sad smiley

I’ll have to order one from somewhere tomorrow.

Thanks to all for observations and feedback. Hope all the photos and detail isn’t too monotonous folks. It’s been invaluable to me.

Regards,
No.6


[numbersixreprap.blogspot.com]
Sorry, only registered users may post in this forum.

Click here to login