Increasing Accuracy - how?

apologies if this is a very dead horse - i did a quick search but found nothing


I'm looking at what we can produce at the minute - about accurate to 1mm?


What further dev's are needed to improve this?
Smaller step - how? lower geared chain?

smaller plastic exit hole - not feasible below a point, higher tempretures and pressures reuired to push it through

systematic errors - more robust frame with less deflection? impossible to eliminate though - empirical correction routines?

madd0ct0r Wrote:
-------------------------------------------------------
> I'm looking at what we can produce at the minute -
> about accurate to 1mm?
>
Accurate to about 0.1-0.125 mm depending on the machine. Mine gets 0.1. I believe Nophead's does, too. The Darwin machines, iirc, are getting about 0.125. If you are using a 0.5 mm extrusion orifice, making the positioning robot more accurate isn't going to improve your printed product quality much. Better to do what Nophead is doing and reduce your extrusion orifice down to about 0.3 mm, instead.

Has anyone actually measured the accuracy of the machines, or are those guesses? A stepper motor's accuracy is plus or minus one half step. If you micro step it to 100 micro steps per step, the accuracy is still plus or minus one half step. The more freely things move the more accuracy you have. The more power your steppers have the more accuracy have (which is a waste of energy of course).

Servo motors with encoders are far more accurate and don't require wasting energy to get the accuracy. Energy is an extremely important consideration, 'cause it will make your parts much more expensive to waste energy. Even in plastic molding, the cost of electricity is a factor in the price of the parts. So if your motors and driver boards are using more electricity, or even close to as much electricity as the melting of the plastic, you are wasting money on electricity. Also, the electricity required to keep the plastic hot is a factor as well. Insulation helps there. Very small melt area would help a lot as well.

Also, having more accuracy in the machine will help make more accurate parts. Even if you have .5mm orifice, you can still increase accuracy by having .01 mm positioning accuracy and resolution. I know that sounds weird, but it is possible to have a really accurate part with such a bad surface finish. Having the accuracy is still very useful. Of course it'd be nice to have a smaller orifice in so doing have better surface finish, but it is very important to increase positioning accuracy.

You really have to stop thinking in terms of "good enough". You need to strive for perfection if you want these machines to ever amount to anything. Even if .1 mm is ok for 3D printing, what about milling? What about 3D profile scanning? What about all sorts of other functions that these machines could have?

I expect what you call .1mm accuracy is probably .1 mm resolution. Resolution is what the minimum movement you can make is. That also has to be small. Accuracy is the amount of difference being at the location you asked for and the actual location you are at. If you ask to go to location 1.000 and sometimes your machine is at location 1.050 or .950 when it thinks it is at location 1.000, then your accuracy is plus or minus 0.050.

If you are at location 1.000, and the very next location you can get to is 1.010 than your resolution is 0.010.

In you go from 0 to 1.000 and stop. Then try to move back one step, you'll probably find out that your machine won't move. That is backlash. The location you end up at when you go from 0 to 1.000 will be different than the location you end up at when you go from 2.000 to 1.000. The difference is call back lash.

The software has no backlash compensation, no accuracy compensation, no squareness compensation, nothing to improve accuracy. Software can do a lot to fix those problems at zero cost per machine, but there is nothing like that in the software or the firmware.

That is why I am starting over from scratch, making a whole new machine. New mechanics, new electronics, new firmware, and new software.

There is just way to much thinking things are "good enough". They are not even close to "good enough".

Still... I wouldn't mind at all if you all would work on improving the extrusion and let me worry about improving machine accuracy. If you think your machines are accurate enough, fine, work on the extruder. Just don't work on them with the intent of making them "good enough". Make them perfect. Make them use less electricity, use smaller orifices, maybe even have multiple orifices each having a valve.

With plastic you will likely have limitations on orifice size. Maybe someone would like to work on making a metal extrusion head. Metal flows much more freely. Metal turns more like liquid when it melts, unlike plastic which is pretty thick when it melts.

Plastic can get quite small though. Since small plastic fibers can be made, it should be possible to make small orifices. Friction may be the problem in making an orifice, so maybe an orifice made out of teflon or something would allow a smaller orifice. Maybe someone should try a small teflon orifice, and see if that works. Check the details of teflon first though, and also check on the safety of it, what gases it produces when it overheats. I've heard there is a plastic that produces a deadly poisonous gas when heated, so we should always be careful when heating any plastic to know if they are safe or not. I don't remember what plastic it was that produces dangerous nerve gas, so check on it first.

Tony

Ant Wrote:
-------------------------------------------------------

Tony,

I tried answering this 2-3 ways but found myself saying things like crap and bullshit about 4-5 sentences in and was pretty much into a full-fledged flame war by the time I'd responded to several paragraphs. I backed off on that, but I must say that I frankly found your post both arrogant and insulting to the many people in Reprap who have put in long, long hours getting the state of the art to the point that it is now.

By all means go build this "perfect" machine. When you've got it going come back and show us and I'll be properly amazed and worshipful. Till then....

My reaction may have something to do with the fact that it's 0400 here right now. I don't think so, though.

Forrest

Edited 2 time(s). Last edit at 12/05/2008 06:59AM by Forrest Higgs.

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

Forrest, you are my new personal hero for being able to keep cool like this.

Tony, .......nvm.

Quote

That is why I am starting over from scratch, making a whole new machine. New mechanics, new electronics, new firmware, and new software.

There is just way to much thinking things are "good enough". They are not even close to "good enough".

Still... I wouldn't mind at all if you all would work on improving the extrusion and let me worry about improving machine accuracy.

Well.. what "we" do isn't nearly good enough, but we are good enough to actually build your machine*, so that you're the supervisor of what WE actually made...
Didn't we spoke about that earlier?
I cannot reply better than Forrest did, honestly, not only because of the lack of terms, but also because the lack of kindness.
So I only repeat what's being said (iirc numerous times already)
"go ahead then, make it; but don't ask us to do so!"

Forrest, thank you! without your post up there I would have been carried away to reply in a much longer post

'sid

*that's what nearly all your other posts imply too.

PS I know, not short enough, not kind enough... I'm sorry but can't help it!

Tony,
You have some valid points, but you have expressed them in a way which is not conducive to people accepting your point of view. I hope this is because you see how important this project will become as it develops and are, therefore, very passionate about it. I want to make some big changes, myself, but the approach I am taking is to convince people with the success of my ideas, not telling them what is wrong with theirs.

Engineering is a difficult profession because it doesn't allow for perfectionism. To engineer a good solution, you need to make a tradeoff between development time, feature set, quality of features and cost. If you try to make a feature-complete system, like trying for a Von-Neuman Universal Constructor from the get-go, the project will never go anywhere; there's not enough time or money to accomplish that. If, on the other hand, you build something as simple as possible, that accomplishes its design goals, then as successive revisions develop, you can build up to the Universal Constructor over time.

Ant Wrote:
-------------------------------------------------------
> Servo motors with encoders are far more accurate
> and don't require wasting energy to get the
> accuracy.

If you check out the information on reprap2: Mendel, you'll see that servos are on the list. Also, look at the magnetic encoder board. It's designed to be used with the GM3 gearmotor. Using that, you should be able to replace the steppers without having to do any design work from scratch.

> Also, having more accuracy in the machine will
> help make more accurate parts. Even if you have
> .5mm orifice, you can still increase accuracy by
> having .01 mm positioning accuracy and resolution.
> I know that sounds weird, but it is possible to
> have a really accurate part with such a bad
> surface finish. Having the accuracy is still very
> useful. Of course it'd be nice to have a smaller
> orifice in so doing have better surface finish,
> but it is very important to increase positioning
> accuracy.

If your extruded filament is 0.5mm +/- 0.05mm, then .01mm position accuracy and resolution is lost in the size accuracy of the filament, meaning that it gives no net gain. I don't know if these numbers are accurate, but the point is the same: one accuracy can be swallowed up by another.

> You really have to stop thinking in terms of "good
> enough". You need to strive for perfection if you
> want these machines to ever amount to anything.

Since you're going to use that kind of language, I don't mind flat out saying: you're wrong. If you read what the designers of reprap have said, they don't care if the machine is perfect on the first shot. That's not what this is about. They want the machines to do their intended job (isn't that what all engineers care about?). The point of reprap, at this stage, is not to be a high accuracy 3D printer, it's to replicate its own parts. Which it has done. Which makes it "perfect" from the standpoint of achieving its goals.

> Even if .1 mm is ok for 3D printing, what about
> milling? What about 3D profile scanning? What
> about all sorts of other functions that these
> machines could have?

The reprap project doesn't have the resources to make reprap all that it could be. So what they did is made the project open source, then got it just far enough that it became useful, particularly useful in its own development. Now, the contribution base has expanded to many times the original size, based on all the early adopters who think that something could be done better, both you and me included. Now that reprap works and can print useful things, like its own parts, you and I can contribute our ideas on how to make reprap better!

> I expect what you call .1mm accuracy is probably
> .1 mm resolution. Resolution is what the minimum
> movement you can make is. That also has to be
> small. Accuracy is the amount of difference being
> at the location you asked for and the actual
> location you are at. If you ask to go to location
> 1.000 and sometimes your machine is at location
> 1.050 or .950 when it thinks it is at location
> 1.000, then your accuracy is plus or minus 0.050.
>
>
> If you are at location 1.000, and the very next
> location you can get to is 1.010 than your
> resolution is 0.010.

Try not to insult the PhD and his team who have put this thing together. I think it's safe to assume that they know the difference between accuracy and resolution. Of course, what's ironic here is that accuracy doesn't matter on a 3D printer. Precision is far more important than accuracy; that is, repeatability matters much more than whether you get to exactly 200.0mm from the edge of the manufacturing plane.

> In you go from 0 to 1.000 and stop. Then try to
> move back one step, you'll probably find out that
> your machine won't move. That is backlash. The
> location you end up at when you go from 0 to 1.000
> will be different than the location you end up at
> when you go from 2.000 to 1.000. The difference is
> call back lash.
>
> The software has no backlash compensation, no
> accuracy compensation, no squareness compensation,
> nothing to improve accuracy. Software can do a lot
> to fix those problems at zero cost per machine,
> but there is nothing like that in the software or
> the firmware.

This is the point of the evolutionary approach to the project. If that's a problem, then build a reprap, fix the problem, and submit the changes, they'll be disseminated in the next revision and everyone with a reprap will have the opportunity to use the added benefits.

> That is why I am starting over from scratch,
> making a whole new machine. New mechanics, new
> electronics, new firmware, and new software.
>
> There is just way to much thinking things are
> "good enough". They are not even close to "good
> enough".

This is engineering; good enough is anything that gets the job done. Reprap fulfills its design goals, therefore it is good enough. This is a key difference between science and engineering: if you're an engineer, you improve the part that you can, but don't start over from nothing; it's a waste of time and resources. If the mechanics don't do what you need, then hack something together that does. I'm pretty sure that the electronics are mostly just fine. If you need more I/O, then use a bigger controller, like maybe an LPC2378 or LPC2468--now that's a lot of I/O--but don't scrap the motor drivers or magnetic encoders that are already there. There's no reason to give up a good design because you think yours is better, there are a lot of keen minds that spent a lot of time on this project. If you replace one thing at time, maybe you'll discover why it's done the way it is. If not, your substitution will probably make it into the next version.

> Still... I wouldn't mind at all if you all would
> work on improving the extrusion and let me worry
> about improving machine accuracy. If you think
> your machines are accurate enough, fine, work on
> the extruder. Just don't work on them with the
> intent of making them "good enough". Make them
> perfect. Make them use less electricity, use
> smaller orifices, maybe even have multiple
> orifices each having a valve.

It's this paragraph that made me call you arrogant. No one cares what I think. At the same time, people care deeply about whether any given idea I put forth is worthwhile, and whether the idea succeeds. Don't try to tell people what they should and shouldn't work on. Present an idea, if people think it is worthwhile, they will work on it. If you want the credit, or no one is willing to work on your idea, then make it work and then, and only then, present it back here as a finished solution. If your ideas and your implementation are worthwhile, they will be included in the next revision.

I suspect that complexity and cost for improving accuracy will be limiting factors. I hope I'm wrong. Please go prove me wrong. But don't ignore the work that's already been done. It's worth a lot and it wasn't easy to do.

> With plastic you will likely have limitations on
> orifice size. Maybe someone would like to work on
> making a metal extrusion head. Metal flows much
> more freely. Metal turns more like liquid when it
> melts, unlike plastic which is pretty thick when
> it melts.

Metal is very difficult to work with in this format, mostly due to the temperatures involved and the flow involved. Check out this site [staff.bath.ac.uk] for information on how they've done with metal extrusion. I think the problem is that melted metal has much too low a viscosity.

> Plastic can get quite small though. Since small
> plastic fibers can be made, it should be possible
> to make small orifices. Friction may be the
> problem in making an orifice, so maybe an orifice
> made out of teflon or something would allow a
> smaller orifice. Maybe someone should try a small
> teflon orifice, and see if that works. Check the
> details of teflon first though, and also check on
> the safety of it, what gases it produces when it
> overheats. I've heard there is a plastic that
> produces a deadly poisonous gas when heated, so we
> should always be careful when heating any plastic
> to know if they are safe or not. I don't remember
> what plastic it was that produces dangerous nerve
> gas, so check on it first.
>
> Tony

Plastics are divided into two groups: thermosets and thermoplastics. Thermosets tend to break down when heated, releasing the aforementioned toxic fumes, and becoming significantly less useful. Thermoplastics can be safely melted without fumes. Reprap uses only thermoplastics for extrusion material.

Again, the one point I want to stress is that this project has been developed over a number of years by a lot of very intelligent, people who have a lot invested in this project. It would not have the form it currently has if they didn't believe that this is an important step on the road to their end goals. Right now, they're trying for proliferation: having the printer print its own parts is important in getting more of them out there. The next gen can probably be printed using the current printer, which means many more people will get access to it when it becomes available.

Tony, I hope you manage to improve reprap. Any improvement is good. Try not to throw away what's been done. That's a mistake that I have made a few times, and it has cost me a lot. I lost three months of development time to doing things my own way rather than reusing something that was available to me once. That resulted in my spending a lot of time on the phone to an upset customer, which slowed things down even more. Learn from my mistakes.

I must urge you again to reuse ANYTHING you can. It will make progress much faster, and you will enjoy it a lot more.

Regards,
Brendan

Edited 1 time(s). Last edit at 12/05/2008 03:22PM by Annirak.

cptwinder Wrote:
-------------------------------------------------------
> Forrest, you are my new personal hero for being
> able to keep cool like this.
>
Naah, I think Annirak takes the patience and forbearance crown in this matter. I tried to respond to the post that way but couldn't manage.

Actually been thinking about it - perhaps I ought to try dropping the orifice size as far as humanely possible (0.1mm? 0.05?) and simply use a filament of that diameter - such a small piece of plastic would require less energy to heat as a bonus.

GAH - I'm going to have to build a REPRAP aren't I?
the question now is wether it'll live in the kitchen or the bedroom.
either way 'er indoors will not be imprssed.

Hi madd0ct0r,

... i've used dispensers with 0.06mm inner diameter for low-viscouse fluids - it's only possible for very short tubes and high pressures.

I don't think you can extrude molten plastic with that sizes ...

Viktor

madd0ct0r Wrote:
-------------------------------------------------------
> Actually been thinking about it - perhaps I ought
> to try dropping the orifice size as far as
> humanely possible (0.1mm? 0.05?) and simply use a
> filament of that diameter - such a small piece of
> plastic would require less energy to heat as a
> bonus.
>
If you are here in the states, you will be able, in theory, to get down to about a #97 gauge drill bit (0.15 mm). A #80 (0.343 mm), however, is the smallest that I've been able to find for sale regularly. #80's are a pain to get aligned in my Dremel. It's like working with a coarse hair. A #76 (0.5 mm) is scary enough to work with. They break easily.

[www.3dreplicators.com]

[www.arizonasilhouette.com]

Smaller than that, you have to get a special drill press and special bits. It's a whole other technology.

You can get 0.02" (0.5 mm) microrod filament from New Image Plastics if you want.

[www.plasticweldingrod.com]

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

In theory, you could use monofilament "invisible thread" as stock. You can get that down to 0.1mm, so you wouldn't need to extrude, just place and weld. It tends to be nylon, though, which isn't easy to melt. It's also likely to be more expensive by weight than larger diameter filament.

One idea I toyed with at one point was a two-stage extruder. Take the 3mm stock and extrude it down to 0.5mm on the desktop, then feed that to the head on the RepRap itself. The idea would be to reduce the weight of the head, which can now be much smaller since it's not pushing 3mm plastic through a 0.5mm hole. Instead, it just needs to heat it up enough to bond it to the previous layer. Basically, you move all the "heavy lifting" off the platform. It over-complicates things, though, because now you have to deal with spooling the 0.5mm filament. More trouble than it's worth.

sorry I'll be thinking out loud...

there was this university link with prototyping machines once.. they made i.a. a webbing machine that uses thread to make 3d pieces.
Using monofilament and a similar technique paired with some applied heat one could fuse the woven structure to a solid.

...finished

I'm getting a price on that half millimeter filament.

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

Hmm, I wonder if you could do something similar to the laminated paper fabricator using sheets of plastic mesh instead.

You can get 0.040 outside diameter tubing, for use in hole poppers. A hole popper is a electric discharge machine which burns holes in steel. The hole is for water flowing through the tube. Inside diameter of that tube would probably be around 0.020. I did a quick search, and I found a ceramic guide for a hole popper which is .3 mm, so the tubing for that would have a outside diameter of .3mm and inside diameter would probably be half that or so.

Water will obviously flow through tubing of that size at around 100 psi or so. We can easily put several thousand PSI on, without even straining ourselves. For example, use a .050 inch piston and put 2 lbs of pressure on it and you have 1000 psi. By the way, if you have a dowel pin or something stuck and can't get it out, you can drill a hole in it, fill it with oil, and put another pin in and pound on it with a hammer and the hydrolic pressure pushes the dowel pin out. Interesting trick I've learned.

I don't mean to insult people, but I'm just pointing out that some are going about things the wrong way. If a person takes insult from criticism, that doesn't mean people should not try to critic their work. I'm not a social expert, I try to improve, but people often take things I say the wrong way. It is not all my fault, it is also their fault for listening badly. Listen better, and I'll try to say things better.

As for accuracy, Do you want printed models that look digital? Or would you rather have printed models that look artistic? Higher resolution creates more artistic look. It reduces digital error. For example, if you make a glass, with high resolution it would have circles of plastic. If you polish it, you'll get a nice looking glass. With low resolution it has a steppy look. If you polish it, you still get a polished steppy look, so fight with it a while till you get it to look ok. If you print a square with a slight angle on the side, that angle will have huge steps that you won't be able to polish out. Well, you can polish it, but you have to be artistic in your polishing, polishing more in some spots and less in others. With higher resolution though, those steps are much smaller.

Also, who is making a machine that is only a 3D printer? You want a machine that is useless for any other purpose? Or do you want a machine that can be used to do other things? What about scanning, put a part there and scan it so you can make more. Wouldn't that be nice? That'd require more resolution, so you don't loose so much when you make the new part.

I am in process of making a machine, and am having no difficulty at all getting accuracy. The difficulty I'm having is speed. I'm having speed difficulty, 'cause the $1 motor I bought is limited on power. The motor, with no friction, can do 80 lbs of force at 200 inches per minute, but that is with no friction. With ball screws my problems would be solved, but I'm not going to use ball screws.

Sure, design requires making sacrifices, but we don't have to go that far. I'm willing to sacrifice a little speed, but not till I've exhausted all possibility of keeping it. Speed is important when milling steel. I like to mill hardened steel at 200 inches per minute, 'cause it is often the ideal cutting speed for the cutters. Helps the cutters stay sharp longer. Still, it isn't to bad to slow it down and slow the spindle down as well. To be honest, I've never worked on a CNC that can cut at 200 inches per minute, but I did cut 100 inches per minute with a 5/8 diameter cutter and it cut really nice (although the CNC was horribly inaccurate at that speed). I won't cry if I have to drop to 50 inches per minute, but I'm going to try for better.

Don't give up on things so easily. That's what I'm trying to say. Don't say it is "good enough" until it really is good enough. This argument was started from someone saying there is no need to do better, but there is a desperate need to do better. It isn't even close to good enough yet. I'm not saying the people who worked on things were stupid, rather it is the comment that it is good enough and we don't need to do better that is wrong.

Tony

Just get on with it.

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

So, Tony
How would you go about achieveing higher accuracy then?
Any hints?
i am not intrested in milling steel but higher accuracy is always intresting.
Per

Hi Tony,
well accuracy is nice, but you must face the fact, that there IS a limit in printing plastics, and we're not too far away.
Nevermind, keep on trying

oh just one thing:

Quote

Do you want printed models that look digital? Or would you rather have printed models that look artistic?

Well I personally like the printed look (in most cases i.e. nopheads results or the metalabs) and don't mix smooth with artistic please
I love the paintings from Roy Liechtenstein, and they are not smooth and "digital" and artistic (you wouldn' say liechtenstein wasn't an artist, would you?)


'sid

Tony,

I think many people at first think that a lot of things can be done better. There is a lot of work to be done for to create a 3D printer, so in a sense you're right. But the funny thing I noticed is that I started learning about the reasons things were done simpler than you would expect. You start to understand where the trade-offs come from. I'm sure you could contribute to the project with a little perfectionism, but don't expect that you'll be churning out new improvements at the rate that we're doing it. So, sure, go ahead and improve things. And you can even try to teach us what we're doing wrong, but maybe we're not doing this wrong, we just accept that everything doesn't need to be perfect from the start.

I liked to learn about the fact that steppers become more accurate when you run them with more power. When you get to a point where you change movement speeds, that would be a good point to increase power. Also, software compensation for backlash/play would be nice. I think I use positioning and precision interchangably, and I shouldn't, so your lecturing was useful to somebody

Besides the fact that you could've been a bit kinder, we should perhaps tolerate some critisism. Just not saying that our thing is rubbish would be appropriate!

B.t.w. there is a post processing station, I don't know the mechanism yet, but it could improve the surface accuracy. Perhaps we can even make one of these things ourselves?

[www.stratasys.com]

I've been using a mixture of water and acetone and I could rub some of the ABS prints to get a more smooth surface. Post processing shouldn't require too much manual labor though... but perhaps to make masters for a mould, it would be acceptable. The finishing station seems to require no manual labor. Anyone know more about this?

---

Regards,

Erik de Bruijn
[Ultimaker.com] - [blog.erikdebruijn.nl]

I've always seen that there is a constant trade off between speed and resolution in terms of how smooth the steps look (ie roughly akin the pixelisation of the object), but I don't think it's usuaully the accuracy in placement that is holding us back. If you halve the exdruded filament size, then the amount of times the extruder has to go back and forth to create a set volume goes up. This isn't an issue that arises because the printer is somehow poorly designed, it's just a natural effect you'll always get.

Increasing the level of control in the actual placement of the head through the xyz axis seems pretty simple. Designing a gear system so that a stepper motor making one rotation would be geared down to a half rotation on a second gear doesn't seem like that big a leap in either mechanical design or software and is already done for the extruder motor so the idea is obviously already there. In fact I'd guess that this solution will end up the way that we get it done if control over head placement becomes the actual issue. But odds are problems like rigidity and stability will become more dominant at these scales than any problem with actual control from the steppers.

The solution in the end to the accuracy vs speed problem will probably end up being somehow making two extruder heads work together on one object or something invented by one of the people here that are way smarter than me. A dual head which could both extrude with a small filament for high accuracy, but get the speed up by building two parts at the same time seems like mostly a problem for software (as well as getting the extruded filaments to bond properly). Although this issue is a while off into the future.

Looking at the builds that are currently being done, accuracy seems pretty good to me, the objective of the project is to build a cheap 3d printer that is also a partial von-neuman machine, it seems to be going well thus far.

Edited 1 time(s). Last edit at 01/15/2009 11:01PM by Peter Hillier.

How about making the hole with a laser?

... i've been working with micromachining lasers some time and have a 50Watt fiber-laser with a spot-diameter down to 5 microns at home, so laser-'drilling' of really small holes is possible, but you have some limitations with materials and a common laser would cost above 30000 USD, so not really suitable when you didn't have someone around with this equipment

Viktor

About taking forever to print a high resolution object, I believe there's a fairly simple solution. Use two extruders, but not simultaneously. The high-resolution extruder prints the outer shell and the low-resolution extruder fills. Should make for a much faster solution than printing the entire object at high-resolution, depending on the surface area / volume ratio. Assuming the higher resolution is a multiple of the lower one, it shouldn't be difficult to tesselate pixels either, for lack of a better phrase.

Of course this depends on being able to switch between multiple extruders.

-Hypoon

One problem could be that if you extrude only a very small amount of material (a very thin wire) it hardly contains any heat. It can be hot, but might not contain enough Joules of energy to really weld properly with the layer below it. I've been using very fine settings myself (~ 0.3 orifice and 0.22 layer height) and delamination becomes an issue. Though this might be solved with a heated chamber or bed (or a small hot air feed, as I built but haven't tried yet).

---

Regards,

Erik de Bruijn
[Ultimaker.com] - [blog.erikdebruijn.nl]

I don't have my reprap yet, but still I don't understand what you mean, Erik. How does the amount of material being extruded affect the thermal energy it contains? If anything, I would think a small amount of material would get hotter. (By the way: heat is the *flow* of thermal energy, typically from one object to another)

-Hypoon

I think this is what's going on, but I need to read some papers and Ed's thesis.

A tiny plastic filament L long, radius r will volume
V = L*Cross_Section_Area = L*Pi*r^2,
and will have a surface Area
A = L*2*Pi*r.

[en.wikipedia.org]

For a given chunk of volume dV, it will lose heat with direct proportion to its area, but the ratio of its area to its volume is
A/V = 2/r

Which means small things cool faster than large things. A tiny metal pellet cools faster than a bowling ball and so on.

The small radius extrusion laying on a sublayer probably cools faster from its air-exposed top compared to a thicker radius extrusion. In order to weld with its sublayer, it needs to remelt that plastic, and if it is cooler and cooling faster, it can't dump enough heat into that sublayer plastic to reheat up to welding temp.

(My physics is very rusty, and I need to read the papers.)

Hi Sebastien,

My physics aren't very sophisticated and mostly based on introspective thought and observations rather than learning theory. Your explanation of surface to volume ratio is indeed what I see as the rationale behind it (surface being to the power of two and volume to the power of three, which is why babies need help keeping warm, or at least need more energy to stay warm).

I think the existing polymer structure needs to heat up enough to be able to bond to the newly extruded molten polymer and that the time that both are in a molten state (or at least above the glass transition point). If you extrude faster and move proportionally faster in the X-Y plane, layers take a shorter time to finish and the object has less time to cool towards the ambient temperature (stays hotter). This is probably why I started to see beter inter-layer adhesion as soon as I got the Bowden extruder up and running and could do very rapid X-Y translations, even for the infill. This obviously helps keep movement accurate at higher speeds, since there are less low-frequency vibrations (but probably more higher frequency vibrations).

---

Regards,

Erik de Bruijn
[Ultimaker.com] - [blog.erikdebruijn.nl]

Ok, that makes a lot more sense. I remember seeing something rather innovative. I believe it was nophead that created a heated bed to help keep the previous layers warm. This way, the extruded plastic does not need to be as hot to bring the layers to welding temperature. Perhaps that's really the trick to finely working with certain materials.

-Hypoon