Anet A8 3D Printer.

[130]

Once again I'd disagree with some of that, for example backlash in the bed "bearings" on the rails won't have any effect without slack in the belt drive because it'll still move at the selected speed.

Just had a look at my nibbler table old drawings, I varied one of the key dimensions by 0.25mm to go from too tight to a nice fit. Bear in mind that the part that fits in it was measured with a caliper with only 0.1mm accuracy and I didn't measure the cylinder in a few places to take into account the manufacturers tolerances and varying paint thickness I think that my cheap printer does ok, and it hasn't been through a close fitting part calibration process.

 

[PhillipM]

No, it won't, because your static friction is different to your dynamic friction, you don't need slack in the drive belt your entire printer and belt system is basically a huge spring, what will happen is the print head actually momentarily sits stationary whilst the force reverses direction and the tension in your sprung system builds up enough to overcome the static friction, so you'll get a small flat spot and the carriage in not quite the same place as expected.
Even worse, because the dynamic friction is generally lower - much lower if you're using bushings or cheap bearings - then head will spring forward the minute it moves and the whole thing will resonate for a while.

I take parts straight off the table as a snap fit to other off-the-shelf parts.
That's why you want consistant filament.

No adjusting, no fiddling, no printing another a little smaller, just off the bed, clips on.

 

[130]

Inertia is what it is and if the backlash is in the bearings less force is required to start the carriage moving as you generally experience it at ends of travel. The amount the "spring winds up" before travel commences will be less. But as I said lowered travel speed will reduce that effect. The "spring" will stretch less. You can do this in software.

I'm happy that you are commercially able to make commercially viable parts. However, regardless of what you say about fiddling it is recognised that materials have shrinkage and many CAD drawings are actually drawn (fiddled with) to take into account these properties. This may not be an issue for the small dimensions of the part you referenced. PLA typically shrinks about 3% and ABS about 6% (figures vary). I would expect the drawings take this into account, point being accuracy is never going to be all in the printer.

End of the day it's not a commercial exercise for everyone. As mentioned before many folk with turn a thing to do a job by eye. As long as it's appropriate for the purpose that's all that is needed

 

[PhillipM]

My parts are all printed with shrinkage already factored in, you're way out on those figures for printing though, looking at about 1% for PLA, 3% for ABS and Nylon and 1.5 for PET-G/TPU, etc.
Depends on your exact temperature of course but mine is calibrated in the slicer so it tweaks shrinkage for print temperatures.
Carbon fibre/kevlar filled shrink about half the above.

I just don't see the point in paying half price for filament and ending up printing twice as many parts with twice as much time.

 

[130]

Right so you are getting +- 0.2 for commercial use factoring shrinkage in, I assume, the drawing or gcode generation (hence they are fiddled with) And I am getting approx. +-0.25 with a variance of potentially +-1 on the measuring equipment as a non commercial exercise for fun with a printer that is almost out of the box without close calibration being done.

I've not really been put to shame there have I?

And finally I'm spending half as much on print and rarely printing twice. MOst times an idea evolves and multiple prints are not a function of inaccuracies.

 

[PhillipM]

Yes, because it's easy to measure certain dimensions in spec even on a printer, take multiple readings where it changes direction and on thickness. I'd put a tenner on it's got small flats every place an axis stops.

If you're buying the very cheapest of filament how can you know your dimensions are spot on every layer - most of the cheap filament barely meets +-0.1mm on the diameter - that's 20% extrusion changes right there.
Just in the past hour someone is asking for help on the most popular 3d print forum because he bought a bulk of 10 rolls at just over a tenner each and they vary from 1.55mm to 1.85mm on the same roll.

A structural part with that will be about as good internally as an Aero bar.

 

[130]

I think you have missed the point.

In easy rounding my tolerance is marginally more than yours with a printer that cost half as much and filament that probably cost half as much without using software that factors in shrinkage.

Secondly it can be as easy as you like to measure something but I'm not replacing the 0.1mm tolerance caliper that is sufficient. That limitation will always be what it is and hence my accuracy might well be +-1.5 or +-.35

 

[PhillipM]

And again, by the very nature of a 3d printed part from an stl, it's faceted, I could pick two dimensions on a part I've printed and measure them at one spot and they'd be bang on to 0.01mm on the calipers.
Measure somewhere trick where the kinematics or extrusion rate actually makes a difference and it's a much different story.

My parts will generally come off the printer at +-0.2mm and generally half that, for every type of plastic, every type of model, first time, every time, measured all over, even in layer voids and peaks - I'm sorry but I know how hard that is to hit, which is why I know you're not hitting that consistantly with an A8 and buying the cheapest filament you can.
Just the filament diameter variation alone on the really cheap stuff will sometimes leave you with slits in perimeters, etc.

I do use cheap filament...but only for decorative parts.

I've some Rigid TPU on the machine at the minute, I've just measured a couple of meters with caliper that's +-0.01mm - it measures 1.76mm on every reading down the first 4m that I could be bothered to check, every ~25cm or so.
I've some cheaper Real pet-g, that's 1.74 to 1.78 on what I can measure - that's an almost 5% variation in volume right there, as you go through the roll, and that's not bad stuff.
The mega cheap black amazon special PLA that I started on years ago and now I don't use for anything bar mockups goes from 1.70mm to 1.78mm in the space of two loops....

It's just not worth the hassle, my time is worth more than the difference in price even with a single failed print a roll.

 

[130]

And again you miss the point.

You make allowances for shrinkage. I make them in the drawing

I am getting the accuracy needed where it is needed and it isn't much more than your +-0.2 as near as I can measure with available kit.

I aren't expecting or needing more from the A8 as a non commercial exercise.

I haven't turned down paying work to make my time available to do this, therefore my time for this has no value.
I don't suffer from failed prints. And in theory a failed print wastes machine time and a couple of mins to clear the build plate and reset, while it's printing I do other things.

Re "Mockups" sounds like prototyping for the bin - what you called fiddling.

 

[PhillipM]

Mockups are for parts where I have no interest whether they're accurate or not, or what they look like, and get printed at the equivilent of ~300/400mm/s on a 0.4m nozzle at .2 layers - they're probably a replica of something I'm going to make in steel to see what it looks like or check for any fitting issues, or it's just a toy,etc.

How can you be accurate to +-0.2mm if you don't know whether your extrusion rate is jumping up 20%?
It's no good just measuring one dimension with a caliper, any meteorologist will tell you that, you're just measuring your high spots from both the roll and the .stl facets that you've tweaked to, you might have voids under that, you'll certainly have area's where the perimeters suddenly thin and have voids through the part, etc.
Print the same model with something softer like nylon or TPU and it'll change size, I guarantee it.
My scanner heads are 3d scanned by the firm that buys them before use - as they need the scanner head in precise locations on the roll - the entire surfaces are within that tolerance, dips between layers and all.

How on earth could you do that without consistant filament? You can't. Even if you had active filament diameter scanning and compensation in the firmware it wouldn't compensate for the pressure lag in the melt chamber.

You're better compensating for shrinkage in the slicer if you can rather than the model, then you can use the same model across different filaments and print temperatures without having to keep altering it.

 

[130]

Still missing the point I see.

As per my example many posts ago a skilled machinist will often make a part by eye where only a couple of dimensions are critical and those will be measured.

In the case of the nibble holder, two dimensions are near critical the rest aren't. Over delivery is wasted effort / resource.


Re compensating in the slicer, I agree. However, the A8 is limited in range of materials out of the box and PLA suits my needs at present.
You could argue that I would save more time using a better CAD and I would but it's not worth paying for anything for what I am using.

In the same way as most amateur welders on here aren't trying to become certified, I'm not trying to become a draftsman, 3d fabricator and 3d machine builder.
I'm getting more than 80% benefit for less than 20% effort so following pareto principle that'll do.

 

[PhillipM]

You're missing what I'm saying about the dimensions, it'd be easier to explain with a diagram but it's too late to bother atm.

Use Fusion 360 for your CAD, it's free.

 

[130]

I'll give that a looking at.

I'm not missing what you are saying about dimensions, what I'm saying is as a non-commercial exercise if the dimensions I need to fit, fit, and sides open to air are not critical the result is all it needs to be.

 

[MattF]

I'll give that a looking at.

I'm not missing what you are saying about dimensions, what I'm saying is as a non-commercial exercise if the dimensions I need to fit, fit, and sides open to air are not critical the result is all it needs to be.

It's the same scenario as with a lot of us on here. Some are professionals who need everything spot on with no fuss, whereas some of us are tinkerers, amateurs etc. who may not always need close to zero error. You & Phillip have different requirements is all, and you're both using a system/setup which best fits either requirements or needs.

Not a point worth disagreeing over chaps, as your requirements/needs are far apart.

 

[Wendelspanswick]

Still missing the point I see.

As per my example many posts ago a skilled machinist will often make a part by eye where only a couple of dimensions are critical and those will be measured.

In the case of the nibble holder, two dimensions are near critical the rest aren't. Over delivery is wasted effort / resource.


Re compensating in the slicer, I agree. However, the A8 is limited in range of materials out of the box and PLA suits my needs at present.
You could argue that I would save more time using a better CAD and I would but it's not worth paying for anything for what I am using.

In the same way as most amateur welders on here aren't trying to become certified, I'm not trying to become a draftsman, 3d fabricator and 3d machine builder.
I'm getting more than 80% benefit for less than 20% effort so following pareto principle that'll do.

My point about cheap filament also applies to welding.
Sure you can weld with cheap unbranded MMA rods but if you have a poor batch it will be a struggle even for a skilled welder.
Buy some decent rods that are consistent and with known characteristics and it will make life so much more easier for yourself, you get repeatability and a knowledge of how they react in different scenarios.
The point I'm trying to stress is that to make life easier for beginners you need consistency to build experience, to minimise the complications in a process that has lots and lots of variables, your not going to get that if your using cheap filaments that have no guarantee of consistency.

 

[mrsbruce]

My point about cheap filament also applies to welding.
Sure you can weld with cheap unbranded MMA rods but if you have a poor batch it will be a struggle even for a skilled welder.
Buy some decent rods that are consistent and with known characteristics and it will make life so much more easier for yourself, you get repeatability and a knowledge of how they react in different scenarios.
The point I'm trying to stress is that to make life easier for beginners you need consistency to build experience, to minimise the complications in a process that has lots and lots of variables, your not going to get that if your using cheap filaments that have no guarantee of consistency.

That is it exactly! My cad and printer knowledge was zero to start with (cad still limited), so throwing in cheap pla problems meant that layers would be uneven, or the extruder would clog due to a slighter thicker bit of filament being shredded by the gear.

Lots of little problems getting in the way. Ok, a great learning experience, but a very unnecessarily steep learning curve.

 

[130]

It's the same scenario as with a lot of us on here. Some are professionals who need everything spot on with no fuss, whereas some of us are tinkerers, amateurs etc. who may not always need close to zero error. You & Phillip have different requirements is all, and you're both using a system/setup which best fits either requirements or needs.

Not a point worth disagreeing over chaps, as your requirements/needs are far apart.

however, you're saying what I've been saying

 

[MattF]

however, you're saying what I've been saying

The difference is, I know when I'm merely expressing my personal opinion, and when I'm unnecessarily arguing the toss over it. Easier to just voice ones opinion once, and then let the reader decide which advice/course to follow. We'll all always have slightly differing opinions, results, needs & expectations, because everyones circumstances are specific to them.

 

[ukracer]

Ok chaps....not sure if this proves my point (personal) but 38 posts into the thread and i know almost nothing about the A8 than i did before. Lol

I am not complaining as all info is gratefully received. Please bear in mind text is hard to interpret and lacks emotion and i dont want my thread locked if its deemed people are falling out.

Thanks for the info....i need some time to process it.

 

[ukracer]

They can, but that doesn't stop you getting it from stick slip from alignment/rigidity/cheap bearings, etc.

No, you're right, my printer was an incredibly expensive....250 quid.
And I print roll scanning heads for steel mills with it that need to be +-0.2mm on all holes, slots and overall dimensions, these are 140 x 80 x 80 parts. I hit under +-0.1 usually, at 25c

What printer is it you have mate?