Announcement

Collapse
No announcement yet.

3D printing accurancy and limitations - whats possible?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • 3D printing accurancy and limitations - whats possible?

    Many of us who make things either have or are thinking about 3D printers, enough that I didn't think it OT. A resent presentation at TSME (Toronto Society of Model Engineers) got me thinking more about it. Theoretical for the moment, I'd like to better understand just how accurate a 3D printer could be and what the major challenges are

    By one account if you spent $4000 you'd realistically be able to get .005" thou accuracy. Probably many have the reaction I did, .005" doesn't seem very good and $4000 to get that accuracy? Yeah right. If I want to print a small gear or something to be lost wax cast, being difficult to get to even .005" makes it less compelling

    Accurate linear motion would seem an obvious challenge for low cost machines. Precision movement in three axis to an exact degree is highly doubtful with a few hundred dollars in AL extrusions. But lets put that aside with the assumption we can figure it out - accurate ground or scraped frame, linear bearings, accurate screws etc. While the dollars can add up, its stuff we know about and can do; we know how to largely move the limitation on accuracy from linear motion.

    That leaves the extrusion (temp, orifice dia, feed rate, and ???) and the material itself and possibly software and whatever else I don't know about. If you spend a bunch you can get a ruby .2mm nozzle, approx .008". jewels for watchmaking are often smaller. Smaller = more accuracy, I think. If you're able to squirt out a few thou wide bit of plastic, and have motion control to say a thou, can you get to .001" accuracy?

    I understand print times get very high if striving for accuracy, but that's more a decision than limitation imo. So if you didn't care about print time, and built an accurate and precise motion system (or possibly even used a cnc mill), what's the next constraint and where could take this insofar as accuracy is concerned.
    Last edited by Mcgyver; 04-20-2018, 08:57 AM.
    in Toronto Ontario - where are you?

  • #2
    I don't think the inherent accuracy of the mechanical axes is the limiting factor. Aluminum extrusions are extremely consistent in dimension, and stepper motors with XL timing belts are very accurate and have almost no backlash.

    The fact that you are melting and squirting plastic seems to me to be the most difficult part to control with long term consistency. Not to mention that the plastic itself shrinks as it cools.

    Comment


    • #3
      Within reason you can calibrate the machine/software to whatever tolerance you like, putting aside mechanical limitations - steppers. drive screws etc

      Its pretty accurate but not stunning, there are a home systems and pro systems - the laser ones that work in a liquid are pretty good, then there is powder deposition and loads more.
      If it does'nt fit, hit it.
      https://ddmetalproducts.co.uk
      http://www.davekearley.co.uk

      Comment


      • #4
        One inexpensive and easy way to explore aspects of these questions would be to order parts from a company like Shapeways. This diverges a bit from your question of machine specifics, but there is nothing like parts in hand, and on your desk.

        The test parts could specifically exercise features and tolerances you care about, in various orientations, for both precision and accuracy.

        Comment


        • #5
          I think it might be about what is "affordably" possible as you allude to in the first post with $4000 number. 0.005" isn't all that bad when you start stacking up tolerances even with subtractive machining and affordability. The typical el-cheapo ballscrew for example can have upto 0.003" backlash. Pay more get a better subtractive machine with more mass, more rigidity, and better lead screws and things change dramatically.

          I would think the practical limit is determine by your media management device. In the case of a subtractive machine its the spindle. In a 3D printer the limit is the print head, but that is misleading. It could be an extruder or it could be a laser. For an extruder the limit is the ability to manage a smaller and smaller amount of material accurately. For a laser sintering machine the limit is two fold. The particle size, and the resolution or focal size of the laser.

          As you can see in "theory" the resolution of a 3D part could be quite good. In practice the machine that could meet my desired expectations is something I probably couldn't afford if I took out a second mortgage on my house.
          *** I always wanted a welding stinger that looked like the north end of a south bound chicken. Often my welds look like somebody pointed the wrong end of a chicken at the joint and squeezed until something came out. Might as well look the part.

          Comment


          • #6
            I am in the process of getting acquainted with my first 3D printer. The first thing I can say is that I did not pay even 10% or the $4000 price mentioned. But I probably have around $400 in the printer and some accessories that I have purchased in the past two or three months. My second observation is that although I have not put a micrometer to the parts that I have printed, I believe I had no trouble making that 0.005" accuracy figure on my first printed parts. And that was done with the default 0.4mm nozzle. My Cetus printer came with three nozzles: 0.2mm, 0.4mm, and 0.6mm and replacement nozzles are priced at about $12: that includes the 0.2mm size. I know that there are more expensive 3D printers out there, but I have to question the numbers that the OP used. Perhaps he is talking about a larger one, mine has a print volume that is about a 7" cube. I am sure I am going to want a larger one sooner or later.

            I do not look at 3D printing as a super accurate way to make parts. But then, neither is casting. But people have made and used cast parts for thousands of years now. Look at your lathe, milling machine, drill press, or almost any machines in your shop. At least 99.9% of these machines will be made from castings. Some of the surfaces of those cast parts have been machined, ground, scraped, etc. in order to obtain the needed accuracy. But then there may be cast parts that have had no additional operations performed on them. Some castings are used as they come out of the mold, sometimes with little or no cleanup.

            I see no reason why we can not view 3D printed parts as being just as acceptable as a raw casting. It is all about knowing the limitations of what you are doing and how those limitations can be incorporated into a project or what additional steps may be needed before doing so. I see things like 3D printers being made with 3D printed parts. These 3D printed parts are not used for the rails or to construct a stepper motor. They are used for things like custom corner connectors, mounts for things like the extruder head, braces, covers, etc. They can also be used for things like timing pulleys.

            Timing pulleys are an interesting use for 3D printing. One of these pulleys will have many teeth and the belt will bridge around half of them in a typical use. So there is an averaging effect going on there. The belt's position will be determined not by a single tooth, but by the average of the position determined by that number of teeth that are engaged at any particular instant. And if there is a small high spot on one of those pulley teeth, then the rubber-like belt will likely allow that to produce a small, local deformation in the belt with the result that such a small high spot will not have a lot of influence on the position of the belt. Many of the other 3D parts are used in a manner which provides a similar, averaging of the position of the other parts that they come in contact with. A corner bracket may hold two or three metal rails together. If there are any local high spots, those spots will often be sheared off or crushed down in the assembly process. So a surface that has 0.005" irregularities may provide 0.002" positional accuracy. These things can be considered while designing an assembly with 3D printed parts.

            Side story: one of my first 3D printed parts was a stand for a TV remote. I had a TV remote with a broken latch on the battery cover. That cover would not stay closed and I had been using tape to hold it. I made some fast measurements with a caliper and did a 3D design. I printed it and, almost beyond any expectations that I had for it, it fit with a nice friction fit, on the first try. Now, with a 0.005" gap it would fall off. And with 0.005" of interference I doubt that it would go on the remote without a lot of force and probably some damage. But it works with a nice, friction fit. It goes on easily and it stays on in use.





            So I ask, if you were machining (milling) this part, just what accuracy would you need to shoot for in order to achieve this kind of fit. I don't think that 0.005" would suffice.

            All in all, I think that 3D printing will have many uses in my shop.



            Originally posted by andywander View Post
            I don't think the inherent accuracy of the mechanical axes is the limiting factor. Aluminum extrusions are extremely consistent in dimension, and stepper motors with XL timing belts are very accurate and have almost no backlash.

            The fact that you are melting and squirting plastic seems to me to be the most difficult part to control with long term consistency. Not to mention that the plastic itself shrinks as it cools.
            Paul A.
            SE Texas

            And if you look REAL close at an analog signal,
            You will find that it has discrete steps.

            Comment


            • #7
              Paul, you said "with .005 gap it would fall off." You also said it was a nice friction fit, which means that it IS an interference fit. My experience is that it was probably a rough surface, so there is a lot of deformation of the surface. Have you tried to map where it's actually touching? Is the tape acting as a compressible shim? Is it just touching at 3 spots?

              You also ask what accuracy I'd need if I were to machine this holder. I'd probably design it differently so that the remote is touching at both sides as well as at the high spot on the back. Not knowing how much that plastic deforms, I don't know what the interference fit would call for.

              Dan
              P.S. My first instinct would have been to fix or replace the broken cover.
              At the end of the project, there is a profound difference between spare parts and left over parts.

              Location: SF East Bay.

              Comment


              • #8
                afaik you're not getting into a ballscrew equipped machine for <1000, maybe more. The low cost ones are belt drive. Even with micro stepping that would seem a limitation, i.e. resolution on the stepper. My initial sense of it is that in striving for accuracy you'd probably want to use ballscrews, or a V thread with an Evan-nut (no backlash & are there minimal load)....reason being the increased resolution.

                Paul, apparently getting to .005 isn't that easy. As for treating it like a casting that has to be machined to tolerance after the fact, ok, but that lobs off a huge amount of the appeal. If I could get to .001 tolerance so small parts could be made in a finished condition I would find that compelling.

                Anyway, what I was hoping to get at is what are the limitations? hard stops? challenges? what makes .005 to be considered about as good as it gets? How far could you push the accuracy - anyone done so? attempts and strategies at fine tuning?
                Last edited by Mcgyver; 04-20-2018, 05:41 PM.
                in Toronto Ontario - where are you?

                Comment


                • #9
                  I will simplify all of this for you. If you purchase a 3D printer with the mindset of a machinist, it will be garbage to you.
                  They do not replace machine tools, they do not offer the surface finish and tolerances of machining.
                  Throw away all of your machining notions, and approach a 3d printer as something that will make you useable plastic parts where high quality surface finish and tight tolerances aren’t required. They are far more useful than you realize.


                  Sent from my iPhone using Tapatalk Pro

                  Comment


                  • #10
                    Originally posted by RB211 View Post
                    They are far more useful than you realize.
                    o
                    I don;t question whether its useful, I question the technicality and accuracy limitations. you're a pilot; I'm asking for an understanding around lift and payload and wing area and power....not whether a plane is useful.
                    in Toronto Ontario - where are you?

                    Comment


                    • #11
                      You need to say which axis you want the accuracy on. On my FDM delta 3d printer I can adjust the Z axis in .01mm increments. With a .3mm nozzle, the normal extrusion width is .5mm which means that 1mm features are probably the smallest that can be accurately printed.

                      Module 1 gears and GT2-3mm timing pulleys print quite nicely.

                      Comment


                      • #12
                        I think there is too much focus on accuracy. My grandsons $300 3d printer printed a 1” cube right out of the box to an accuracy of better than 0.005”. One the factors is that the forces are extremely low compared to machining so very little deflection except from acceleration and deceleration. The deflections can not be sensed by the human eye or hand so most people don’t get that the printer frames and rails are not stiff enough. Stiffening many printers is really easy and produces amazingly better results.

                        The really important parameter is repeatabilty, just like machining. Your clapped out Bridgeport or lathe can do accurate work if you measure a part, adjust 0.003” on the dial and it is now that distance away from the previous position. Because of the very light forces the printers can be very repeatable.

                        Comment


                        • #13
                          Datapoint:

                          Got some parts printed for a client a while back. About 3" diameter x 4" high, mostly hollow... nearly impossible to machine the parts, unless you make the big one in several pieces and glue it up, as I did for the very first piece.

                          Anyhow, got them in ABS from a guy up by Chicago. Accuracy is about 0.0025" overall vs the 3D model in diameter, about the same in the length. Call it roughly 0.001" per inch. I do not have info on the machines he uses, no clue what he has.

                          Surface is rough, but not too bad. Very very similar to the parts I got about 30 years ago that were done by the previous method of laser hardening the plastic out of a liquid. Except that I got these parts made of a material that was good enough to use them as test parts in the application.
                          2730

                          Keep eye on ball.
                          Hashim Khan

                          Everything not impossible is compulsory

                          Comment


                          • #14
                            Dan, No the tape is not there for friction or as any kind of filler. It was holding the label on until I could get around to applying some more adhesive, which has been done since that photo was taken. I removed the tape and it still fits just fine. The front is a flat surface and as far as I can tell, it is in contact there except for the slight recess where the label sits. The label does not touch it. I can not see any gaps on the sides and can not slide a 0.001" shim in there. The back is not a complete match for the arc of the holder but has two lobes which are in contact with holder. The broken battery cover wants to stick out and adds some more friction, but the fit is OK with it removed. I would think that any deformation is in the range of 0.001" to 0.002" or so.

                            The only place where I can take any measurements would be between the two sides. Outside measure of the remote = 1.771" to 1.775" and inside measure of the stand = 1.771" to 1.774". Both seem to have a slight variation and both are a little wider at the front.

                            It really is a very good fit. Now, what percentage of that is due to careful planning and what is due to luck I can not really say. As I said, I was really surprised at the accuracy of the fit. This was only the second or third thing that I printed.

                            The point is, no matter how you cut it, we are talking about a surface that is accurate or even to a good bit better than the 0.005" mentioned in the original post.



                            Originally posted by danlb View Post
                            Paul, you said "with .005 gap it would fall off." You also said it was a nice friction fit, which means that it IS an interference fit. My experience is that it was probably a rough surface, so there is a lot of deformation of the surface. Have you tried to map where it's actually touching? Is the tape acting as a compressible shim? Is it just touching at 3 spots?

                            You also ask what accuracy I'd need if I were to machine this holder. I'd probably design it differently so that the remote is touching at both sides as well as at the high spot on the back. Not knowing how much that plastic deforms, I don't know what the interference fit would call for.

                            Dan
                            P.S. My first instinct would have been to fix or replace the broken cover.
                            Paul A.
                            SE Texas

                            And if you look REAL close at an analog signal,
                            You will find that it has discrete steps.

                            Comment


                            • #15
                              That is kind of what I was trying to say. You have to work with the limitations of the process you are using, not against them. If you haven't tried 3D printing yet, don't write it off so easily.



                              Originally posted by RB211 View Post
                              I will simplify all of this for you. If you purchase a 3D printer with the mindset of a machinist, it will be garbage to you.
                              They do not replace machine tools, they do not offer the surface finish and tolerances of machining.
                              Throw away all of your machining notions, and approach a 3d printer as something that will make you useable plastic parts where high quality surface finish and tight tolerances aren’t required. They are far more useful than you realize.


                              Sent from my iPhone using Tapatalk Pro
                              Paul A.
                              SE Texas

                              And if you look REAL close at an analog signal,
                              You will find that it has discrete steps.

                              Comment

                              Working...
                              X