makerbot replicator accuracy

[secondsky]

Hi all, i have a question.
Do you believe that the new Replicator reaches the accuracy of
ultimaker?
The second material may be a water-soluble support material?
Thank you all!

[Luis E. Rodriguez]

"accuracy"

Not sure what you mean? I've seen beautiful prints from just about every kit machine. Layer height? Speed? Two similar but different animals. What are YOU looking for? ABS, PLA, pricepoint, speed etc? If anything being assembled and tested will eliminate assembly user error affecting prints.

Luis E. Rodriguez

Are you a Maker?

--
You received this message because you are subscribed to the Google Groups "MakerBot Operators" group.
To post to this group, send email to make...@googlegroups.com.
To unsubscribe from this group, send email to makerbot+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/makerbot?hl=en.

[Far McKon]

On Jan 19, 12:05 pm, "Luis E. Rodriguez" <lrodriguezm...@gmail.com>
wrote:

> "accuracy"
>
> Not sure what you mean?

Ditto on that. It would be fun to say "Oh, yeah. MakerBot Replicator
is ??? percent more accurate" And I'm sure Erik would want to say the
same about Ultimaker. But seriously, I don't think that would make
any sense, for either of us. AFAIK they are quite similar electro-
mechnically, I don't think one or the other has a big advantage in
motor, belt, or materials quality. That, and I am morally opposed to
nonsense and/or bad data.*

Given the same gcode, both machines should output pretty much the same
quality build. Once folks have bots in hand, I'd love to see a side-
by-side using the same GCode by someone unbiased.

Also,
As cool as mini-layer-height is, I don't think that is the best
judgement of how good (or accurate) a bot is. Repeatable prints,
general width/size accuracy, etc. Can we come up with a standard way
to measure that which isn't layer-centric? I think repeatability can
and should be part of that measure.


hack on,
- Far McKon
(Nerd Herder @ MakerBot Industries)


* For a good look at bad data: http://www.cato-at-liberty.org/how-copyright-industries-con-congress/

[James McCracken]

Yeah there's too many measures of "accurate" that are mutually exclusive.  You can get very small layers (z accuracy) at the expense of wide threads (x and y accuracy)  There's a measure of accuracy in terms of how precisely the cartesian bot can be positioned, but that accuracy is so far below the size of the threads that it doesn't mean "since I can position the axis within 0.x mm, the smallest features I can print are 0.x mm"

For me, my biggest measure of accuracy has always been z-wobble!!!  I don't care if layer 1 can be printed with 0.01mm accuracy if layers 2-10 offset by 0.1mm each from the layer before - in that case, the best piece of hardware I can print will have tolerances > 0.1 mm

There *are* kind of upper and lower limits, but they'll be very similar for all bots built with similar methods. 

[Matt Smollinger]

Not to play devil's advocate here, but while I agree that no concept
of "accuracy" should be "layer-centric", it's important to understand
that layer height is still a critical feature and probably is the most
critical as it affects x & y. These machines have their roots in rapid
proto-typing, so when comparing to a machine like a Dimension, the all
important measurement is comparison between a reference part and the
printed part. Normally in the industries I'm familiar with, you
utilize a machine like a Dimension when you have parts with extremely
long lead times that you need a stand-in while manufacturing occurs,
and/or you need to make sure your dimensions work with everything else
in the system before manufacturing occurs. That way you determine
whether or not the part needs to be modified.

Accuracy thus is determined at how well one of these machines can
produce a part that is the same as the same part out of a (well-
calibrated) CNC Mill or Laser Cutter. The X,Y, and Z calibrations are
largely the biggest issue generators here. Get the X, Y, and Z
calibrated (generally 0.01mm variance would be acceptable) and
everything else will come along. Thus, layer height / layer height
variance is one of 3 critical aspects to a machine. Get those 3
sorted, and repeatability comes along for the ride.

Now there are many factors that affect those 3 axises, the largest of
which being the electromechanical components. Printers that can do
1/128th stepping will be far more accurate than printers only capable
of 1/16th stepping. Also, gear-driven > pulley driven as pulleys
inherently stretch over time, creating variance in the drive train
(One can make the argument about gear lash on a gear driven system,
but given a calibrated system, you can estimate for that, unlike a
pulley which will change over time). Software also plays a part in its
ability to process and send the commands fast enough, and the software
running the printer being able to process those commands fast enough
and execute them. As mentioned above, layer height also affects the x/
y accuracy, but that is actually fairly easily controlled by changing
nozzle size and then reducing extruder speeds (done by magic if you're
using SF's Dimension Plugin).

So overall, while layer height may not be the only factor, I
personally consider it the most important as it indirectly affects X/Y
accuracy, whereas X/Y do not affect Z/Layer Height.

--Matt--

[ddurant]

> You can get very small layers (z accuracy) at the expense of wide threads (x and y accuracy)

 

You're saying that a print with 0.3mm layer height and 0.5mm thread width is more accurate on X./Y than a print with 0.03mm layer height and 0.5mm thread width? I'm not sure I buy that..

 

The OP asks about a how Replicator compares with Ultimaker. For speed, I don't think anybody can reasonably disagree that the UM is going to be faster, possibly far faster. For Z resolution, UM has shown it can do 0.02mm layers with virtually zero Z wobble, which is pretty darn good. Spacexula recently did 0.01mm on a MakerGear Prusa but it looked a bit Z-wobbly to me - I suspect an UM could do that or even thinner but nobody's gotten around to it yet. We don't know what the Replicator can do but the pictures of prints so far look to me to be around 0.3mm or so layer heights. Whether or not it can do better than that while keeping quality is something only MBI can answer for now.

 

I disagree with those that say this metric of Z height doesn't really matter. It (among other factors) can show how much went into the design of the machine. If two different machines both have a theoretical min layer height of 0.001mm but one can do quality prints down to 0.01mm and the other can only do quality prints down to 0.25mm, one is arguably a better machine than the other.

 

Unless a machine is designed more to be a toy than a serious bit of hardware, manufacturers should be pushing them as far as possible during design and finding ways to expand their capabilities before release. If nobody at MBI pushed a Replicator lower than 0.2mm layer heights during design while users of other machines in the same price range were doing 10x and 20x that resolution, it wouldn't speak overly well of MBI. Just IMO...

[Jordan Miller]

i completely agree with Dave here.

layer height you slice at is NOT the same as the layer height you get. I have an open call to MakerBot to send me a high res print for real measurement under the microscope, but I've never heard back yet. I've already done the measurements and blogged it for Ultimaker with a 0.5 mm nozzle...

moreover, as the layer heights get thinner and thinner, any inaccuracies in volumetric flow consistencies due to flawed hardware design become enormously more apparent and being able to reach those layer heights at all and complete a print does actually become a good metric for resolution. if the left side of the bot is 500 micron higher than the right, this will show up during a 50 micron layer height print.

the rigidity of the frame impacts this as well due to dithering effects at high speed with excess vibration.

a 500 micron nozzle moving around squashing out 50 microns of plastic at 150 mm/s is freaking impressive any way you slice it (ba da bing!).

everyone will agree a 50 micron layer height is useless if your print will take 17 days. so speed and accuracy optimized together and **carefully measured afterwards** are critical. that's the right metric IMHO.

jordan

--

You received this message because you are subscribed to the Google Groups "MakerBot Operators" group.

To view this discussion on the web visit https://groups.google.com/d/msg/makerbot/-/fWjLzH-xEq0J.

[slipshft]

While all the comments are valid, I just wanted to point out that the
accuracy of any machine should be based on what measurements it is
supposed to be reproducing, i.e. 10mm +/- 0.01mm. The only way to
really judge the accuracy of any production machine it to test the
output of said machine by measuring said output. Since every machine
is different, this would have to be done for each machine then with
sufficient data a mean could be produced.

I don't think that the original poster was looking for a true
definition of accuracy, but more of a does the Replicator print better
(or more accurate) output. I could be way off in my assumption.

[Dan Newman]

On 20 Jan 2012 , at 4:03 AM, James McCracken wrote:

> Yeah there's too many measures of "accurate" that are mutually exclusive.
> You can get very small layers (z accuracy)

That's not "accuracy". That's resolution. You can have very fine resolution
BUT poor accuracy at the same time. They are different things.

Dan

[G. Wade Johnson]

Back in engineering classes, one of my professors spent a lot of time
drilling into our heads the difference between accuracy and precision.

Resolution is close to precision.

The simple version is

Precision - decimal places
Accuracy - expected answer

I believe we are mostly discussing precision in this thread. I actually
find this pretty cool that we have desktop machines that print objects
that are mostly "right". And now we can get to the point of discussing
how precise our prints are.

G. Wade
--
That which does not kill me makes me stranger. -- Larry Wall

[ddurant]

> Back in engineering classes, one of my professors spent a lot of time
> drilling into our heads the difference between accuracy and precision.

 

I've always liked the target analogy, http://en.wikipedia.org/wiki/Accuracy_and_precision , probably because it's done in easy pictures..

[Dan Newman]

On 20 Jan 2012 , at 8:52 AM, G. Wade Johnson wrote:

> On Fri, 20 Jan 2012 08:31:22 -0800
> Dan Newman <dan.n...@mtbaldy.us> wrote:
>
>>
>> On 20 Jan 2012 , at 4:03 AM, James McCracken wrote:
>>
>>> Yeah there's too many measures of "accurate" that are mutually
>>> exclusive. You can get very small layers (z accuracy)
>>
>> That's not "accuracy". That's resolution. You can have very fine
>> resolution BUT poor accuracy at the same time. They are different
>> things.
>
> Back in engineering classes, one of my professors spent a lot of time
> drilling into our heads the difference between accuracy and precision.

Reminds me of back in the '80's designing XY plotters at HP and trying
to get the marketing droids and sales reps to understand the distinction.
Never had much luck, I have to admit.

Dan

[Dan Newman]

> the rigidity of the frame impacts this as well due to dithering effects at
> high speed with excess vibration.

I've always been curious about this aspect. For instance, the wooden
framed bots vs. those with metal frames (e.g., Up). Is being extremely
rigid a disadvantage or an advantage? Is the slightly yielding nature
of wood an advantage or not here?

> everyone will agree a 50 micron layer height is useless if your print will
> take 17 days. so speed and accuracy optimized together and **carefully
> measured afterwards** are critical.

While we can measure and check the resolution under a scope of a yoda head,
I'm less sure just how much a yoda head can really tell us about accuracy.
The high resolution ones sure look fantastic! -- no argument there.
But without a standard to measure them against, I'm unsure how to
quantitatively talk about the accuracy. As boring as it might sound, a
20 x 20 x 10mm cube might be better. At least then you can not only inspect
layers and do interlayer comparisons, but you can actually measure flatness,
squareness in all dimensions, etc. Print it with minimal width walls and
no interior and you can get a sense of overall peak-to-valley (p/v) variations
across layers which is another way to consider flatness. And, if you print
several, you can measure reproducibility.

Cheers,
Dan

[G. Wade Johnson]

Agreed. Except my professor gave four possibilities for the target. I
reproduced them here years ago.

http://anomaly.org/wade/blog/2006/01/accuracy_and_precision.html

G. Wade
--
All things are possible, given enough time and caffeine.
-- Danny Hoover

[Luis E. Rodriguez]

Oh well hell I've been doing that bottom right bit for years! :)

While not to belittle education, I switched from EE to Graphic Design in college :) but I think what people print and under what settings is so wrought with variables it would be tough set some standards. Ultimaker does some amazing things with a .5 nozzle, using 3mm PLA. Now I will say the bowden/retraction/stringiness bothers me a bit but the speed of that monster seems to overcome it and when they print at microns those strings are wispy little hairs. I also think most people just don't give a rats ass about the race to .0000000001 layer height. If you are bothered by layers you should choose another printing technology other than FDM, i.e. powder-based or better yet UV Resin. We will all get a slap in the face when a DIY UV Resin printer comes out and blame we are left holding a reel of archaic ABS or PLA! :) 

But my meandering point is that some folks are ok with .3/.4 layer height because it really doesn't matter for the function of that particular piece. I can agree that the use of a high-res setting is useful sometimes. Mold-making or to minimize post-processing. An Engineer and his project is going to have different views on it than an artist or kid.

In this case above if its accurate but not precise than would that be akin to when we get a layer shift? I've printed BEAUTIFUL objects but came back to see it shifted in an x/y axis but finished the model. Cupcake or early ToM modding adventures. :)

So to contradict myself I ALWAYS thought when were "whipping it out" we were referring to speed and layer height. I'd like to throw in retraction/reversal because stepper extruders has made it possible to pretend we are dimension printers. SO say something like, if necessary how small of a layer height, and at what speed and how clean can it come out, can I print X. What I think is funny is usually we are comparing whistles or yoda heads!

I think the competition and trash talk is fun to watch between open-source users. I suppose give credit where credit is due but if it's open-source and all originating from RepRap which was inspired by Dimension printer tech/patents than what are we arguing about? It's funny how the latest printers have hard to find BoM's, its not THAT easy to build your own these days, the users help with that!

Luis E. Rodriguez

Are you a Maker?

--
You received this message because you are subscribed to the Google Groups "MakerBot Operators" group.

[Eric Palmer]

I've drilled the concept of accuracy and precision in the heads of middle school lids on the FLL robotics team. The concept allowed us to slow more missions. Get the repeatability (precision) up (more repeatable) and then change the targeting and the accuracy can be improved. 

I think for a solid 3D printer we need both.

I love the chart on Wikipedia.  That says it all. But middle school kids we used a red bulls eye.

--
You received this message because you are subscribed to the Google Groups "MakerBot Operators" group.

To post to this group, send email to make...@googlegroups.com.
To unsubscribe from this group, send email to makerbot+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/makerbot?hl=en.

--
http://www.thingiverse.com/DaddyOh
Join the 3DPrinter Revolution
http://makerbot.com

[ddurant]

> But my meandering point is that some folks are ok with .3/.4 layer height

> because it really doesn't matter for the function of that particular piece

 

I think that's changing.. As machines speed up and can easily & repeatedly do higher resolutions, the expected norm is going to be faster and with higher resolution.. This is the way tech has always gone.

 

If popular machine A and popular machine B cost about the same money and print the same part but machine A can do it at 5x the resolution in the same amount of time, isn't it safe to assume that "the public" will grow towards expecting the better quality output?

 

Why, unless you're explicitly trying to show that you've printed something in 3d, would you ever want layers to be easily visible? Is that ever better than having them be so small that they're not easily visible?

 

> I also think most people just don't give a rats ass about the race

> to .0000000001 layer height

 

I agree and said something similar around here just this morning - it's really no more than bragging rights at this point (which, again, doesn't minimize the great work Spacexula did - it's quite a feat!).

 

That said, I think there's a level you have to get to before you can really say that.. If a new desktop printer could only get down to 2.5mm layer height and they said "it's so awesome that resolution doesn't matter!" we'd laugh our asses off at them. If they could do 0.05 or so layers then said that, it'd be a different story.

 

I've got desk toys here in my office that I printed on my Cupcake and ones I printed on my newer machine. When people pick up something printed on the new machine, they ask if it was 3d printed. When they pick up something printed on the Cupcake, they don't. Companies that don't think a lot about stuff like that are going to have an increasingly hard time keeping up...