Hitman

Show off your prints!!!
charles.yates
Posts: 59
Joined: Sat Dec 19, 2015 9:12 pm

Re: Hitman

Post by charles.yates » Sun Feb 28, 2016 9:04 pm

ednisley wrote:Now you're close to the spirit of the thing!

You adjust the Z Offset to put the nozzle at the correct mechanical Z position, set the actual filament diameter, adjust the Extrusion Multiplier so the extruded thread matches the slicer's expectation, then dink with your models to get the desired clearances based on an accurate printer setup.
Unsurprisingly it's dropped to around .35 - this is inline with the reduction in the multiplier anyway. All cubes were roughly consistent anyway, though the near side didn't stick so well during the skirt part - first layer of both had a small gap, but the other 3 were fine. Will have a crack at levelling the bed. Neat idea anyway :).

So, I need to put it back to .9 and work around the discrepancy in the model then?

User avatar
ednisley
Posts: 1073
Joined: Fri Apr 11, 2014 5:34 pm
Location: Halfway up the Hudson
Contact:

Re: Hitman

Post by ednisley » Sun Feb 28, 2016 10:41 pm

charles.yates wrote:So, I need to put it back to .9
No, you need to adjust the Extrusion Multiplier so the gooey plastic squished onto the previous layers produces the width that the slicer expects it to have. That's the whole point of the exercise: adjusting the machine to produce the correct results.

Using default values should get you close, but then you tune for best picture using the actual results.

Read through Jules' startup guide again; all this is documented in very specific detail.
and work around the discrepancy in the model then
In round numbers, you can expect a tolerance around ±0.2 mm on larger structures. That sounds like a lot when you're talking about a 2 mm wall, but that's just how this 3D printing process works. You're stacking layers of molten goo, not machining slivers off a solid, and the printer simply cannot hold tighter tolerances. You must design your model with those tolerances and the production process in mind, just as you would for a milling machine or a lathe.

But you start by adjusting the machine to produce consistent, measurable results.

charles.yates
Posts: 59
Joined: Sat Dec 19, 2015 9:12 pm

Re: Hitman

Post by charles.yates » Sun Feb 28, 2016 11:33 pm

OK - It's just that I seem to be getting a different opinion from jsc :). Tuning the extrusion multiplier based on a thicker wall is OK?

So, I'm good with putting it back to the lower value (as my prints worked fine) - and I *think*, but could be wrong, that they're turning out fine with the same model with 0.9 - which suits me just fine if that is how others are doing it.

It's not a particularly complex print, and I have many prints of it now :) - all at different settings :) - but it is taking my attention and I think I should focus on that first. Sorry if I'm coming across as a tad distracted.

For anyone interested, an untuned prototype of the customisation tool is available here:

http://openjscad.org/#http://www.gemini ... uino.jscad

I think it's a tad more interesting than trying to deal with my monkey at the controls approach to printing :), but I will return to tuning properly when I have more time.

This was a very, very early prototype. Things have got a lot slicker since then :).
2016-02-25 17.24.35.jpg
Will open another thread to discuss it in more detail, as it's kinda fun... and may even be useful. Currently more like:
2016-02-28 23.29.52.jpg

User avatar
ednisley
Posts: 1073
Joined: Fri Apr 11, 2014 5:34 pm
Location: Halfway up the Hudson
Contact:

Re: Hitman

Post by ednisley » Mon Feb 29, 2016 1:32 am

charles.yates wrote:Tuning the extrusion multiplier based on a thicker wall is OK?
No, it's not.

That's why I keep talking about using thinwalls squares. We're both telling you to do exactly the same thing: follow Jules' startup guide.

Do exactly what it says, rather than asking if something else will work just as well. It might, it might not, but following a known-good procedure will get you where you need to be, faster and easier, than doing anything else.

It's really that simple.

jsc
Posts: 1864
Joined: Thu Apr 10, 2014 4:00 am

Re: Hitman

Post by jsc » Mon Feb 29, 2016 4:22 am

I don't think there's any disagreement between me and Ed. It's his hard won experience that I rely on. I happened to write up a small portion of it as a step-by-step procedure, then Jules took that and incorporated it into the M2 Beginner's Bible (which you should totally follow, step by step, skipping nothing, exactly as written, lest you be led astray). Once you are in a known good state, with good extrusion multiplier, z-offset, and a level bed, and only then, may you begin mucking about with other settings, in the knowledge that you are now on solid foundations.

For calibrating holes and slot/tab widths (after having completed the above), you can either adjust the model, or rely on S3D's layer compensation ability if most of the interesting dimensions are in the X/Y plane.

charles.yates
Posts: 59
Joined: Sat Dec 19, 2015 9:12 pm

Re: Hitman

Post by charles.yates » Mon Feb 29, 2016 8:50 am

Just to reiterate - my thin wall test is checking out at 0.4mm (+/- 0.01), and the height is 2mm as expected.

As I explained, I reduced it further because a 2mm thick wall in my model was coming out too thick, and my reading of Ed's reply to that was that was the right thing to do. jsc replied that I shouldn't do that. I confirmed that the reduction I had made gave me a corresponding reduction in the thin wall test (which I fully expected), I then suggested I should go back to the state with the 0.4 +/ 0.01 state to which Ed said that I shouldn't :).

Really, like I said in my last post - I will address the configuration again once I have more time to focus on it properly. For now, I really want to focus on the model and the software which is running on the Arduino.

jsc's earlier comment about multiple perimeters would suggest that there will be a discrepancy between the 2mm thick wall in the model and the printed one, hence, I will make allowances for that in the model. I will parametrise all the fuzzy factors in the jscad model, and will adjust those if necessary when I have more time to more thoroughly investigate the tuning of the machine.

Disappointed though :) - really thought the project was more interesting than the tuning :).

User avatar
ednisley
Posts: 1073
Joined: Fri Apr 11, 2014 5:34 pm
Location: Halfway up the Hudson
Contact:

Re: Hitman

Post by ednisley » Mon Feb 29, 2016 4:10 pm

charles.yates wrote:because a 2mm thick wall in my model was coming out too thick
It's "too thick" only in the sense that you're expecting tighter tolerances from the 3D printing process than it will deliver. You can certainly find a set of machine parameters that will produce exactly the correct measurements for one particular section of one particular object, but you cannot expect those same settings to produce exactly the same accuracy for all sections of all models thereafter.

With the mechanical and extrusion parameters set to produce accurate thinwall squares, tedious though that process and those boxes may be, you're in a position to get consistent results for all models. Then, knowing how the process works and what tolerances to expect, you can design your parts to fit correctly.

You may find that specific parts require different slicing parameters for different sections, but those parameters all depend on the accuracy of the machine setup.

Designing the parts first, without considering the tolerances, then sequentially tuning the machine to optimize each part as you print it, will, as you've discovered, require a "monkey at the controls approach to printing", take far longer than you expect, and create a pile of scrap plastic.

At least for me, I'd rather do the tuning once, then spend the rest of the year cranking out parts that fit and work as expected. The designs all assume that ±0.20 mm tolerance I mentioned. To get tighter tolerances, I'm willing to ream holes, mill surfaces, and hand-finish edges: all operations that can't be done on a 3D printer.
really thought the project was more interesting than the tuning :).
It is, once you understand both what proper machine setup does and what it cannot do. With those limits in mind, you can get on making everything else you bought the printer to build... and that's what it's all about.

jsc
Posts: 1864
Joined: Thu Apr 10, 2014 4:00 am

Re: Hitman

Post by jsc » Mon Feb 29, 2016 9:15 pm

Hm. Going back and rereading all of Ed's responses in this thread, I think there is some misunderstanding here. I certainly didn't understand him to be saying what you think he is saying. Perhaps the confusion stems from his use of the term "thinwall boxes"; when he says that, he is talking one perimeter thin (as I am). Post your current settings and we can see if anything else looks amiss (if using Simplify3D, File->Export FFF, zip it up so you can attach it).

Nice work on the model, by the way. I love 3D printing for custom enclosures: https://github.com/jinschoi/parking/blo ... re/box.stl

charles.yates
Posts: 59
Joined: Sat Dec 19, 2015 9:12 pm

Re: Hitman

Post by charles.yates » Mon Feb 29, 2016 9:32 pm

ednisley wrote:At least for me, I'd rather do the tuning once, then spend the rest of the year cranking out parts that fit and work as expected. The designs all assume that ±0.20 mm tolerance I mentioned. To get tighter tolerances, I'm willing to ream holes, mill surfaces, and hand-finish edges: all operations that can't be done on a 3D printer.
Good - the .2mm value was what I found online, through manual experimentation and now confirmed by yourself. The model here takes that into account and is working just great - in light of this conversation, I am a bit concerned that it may not work for others, but hey, that's fine :) - hopefully, I'll get some feedback down the line.

Cheers for that.

charles.yates
Posts: 59
Joined: Sat Dec 19, 2015 9:12 pm

Re: Hitman

Post by charles.yates » Mon Feb 29, 2016 9:41 pm

jsc wrote:Hm. Going back and rereading all of Ed's responses in this thread, I think there is some misunderstanding here. I certainly didn't understand him to be saying what you think he is saying. Perhaps the confusion stems from his use of the term "thinwall boxes"; when he says that, he is talking one perimeter thin (as I am). Post your current settings and we can see if anything else looks amiss (if using Simplify3D, File->Export FFF, zip it up so you can attach it).
Agreed on the misunderstanding - probably my bad for not phrasing what I meant about the 2mm wall clearly enough and just mixing it up with the general calibration conversation.

When I have the time and inclination to review the calibration properly, I will send you my settings (currently slic3r - never know - may change to s3d, but happy enough with my mostly open source tool chain for now). Not much point doing it now, as I really won't be inclined to make any changes immediately.

Many thanks for the offer and I will probably take you up on it some day soon though :).
Nice work on the model, by the way. I love 3D printing for custom enclosures: https://github.com/jinschoi/parking/blo ... re/box.stl
Cheers - it seems to be working well, and I'm currently putting together some other stuff using the jscad model (I have a couple of shields which can really benefit from this, and some kinda hokey/dumb ideas for the touch screen device).

Yours looks nice - what's it for?

Post Reply