Loading ...
Sorry, an error occurred while loading the content.

Re: [diy_3d_printing_and_fabrication] B9Creator on Kickstarter

Expand Messages
  • Graham Stabler
    ... I was meaning we are not really in a position to create a standard at all :D ... Not really. If an individual layer has flat sides (not like stacked tubes
    Message 1 of 21 , Apr 9, 2013
    • 0 Attachment

      I appreciate our American's friends have their own national standards for surface finish, based on our own imperial measurements system which suited us for many a century or two ;0)

      I was meaning we are not really in a position to create a standard at all :D
       
      I see your point about stair casing but the measure of a good machine can be assessed easily by how well it does with a simple flat surface.  Horizontal or vertical or inclined.

      Not really. If an individual layer has flat sides (not like stacked tubes like FDM) then the layer thickness will not affect the surface finish so you could have 1mm layer thickness and extremely smooth sides and top to your cube (not to mention all of the vertical sides will be the same) so I don't really see that it shows enough of the whole picture, I also mentioned that the are of the horizontal surface affects the finish for the DLP printers.

      Do you think a cube with straight walls, with a sloping roof is better or perhaps one with a tiled roof  or one with an additional dome. It could get silly

      As it is a 3D printer I am not sure you can get silly because it is just a matter of drawing it though ease of measurement needs to be taken in to account, a horizontal surface, a vertical surface and and at least one sloped surface would be good although I am not sure what angle would be best, 45 degrees if the voxels are cubes but not sure otherwise.

      I assume the print from the B9 would be supported on a build structure rather than being laid straight to the brushed aluminium build platform.  

      You will have to specify this in your standard :)

      I believe honest manufacturer would be delighted to show us what they can do.

      But presumably they do not become dishonest if they can't be bothered or don't think it helpful :)

      Cheers,

      Graham



      On 8 April 2013 17:31, Graham Stabler <grezmos@...> wrote:
       

      If you want an ISO standard you might be talking to the wrong people :) 

      I don't see the reason to measure more than two faces maybe 3 if you do the base, which has a brushed aluminium finish on the B9 :D . As I said I think it it will give a poor estimate of the finish as there will be no stair casing on vertical surfaces.

      Cheers,

      Graham


      On Mon, Apr 8, 2013 at 5:26 PM, Francis Leach <francisleach@...> wrote:
       

      Graham

      I am sorry you do not understand the question but I do not understand your reply.

      A ISO standard sets controlling parameters so meaningful data can be derived by test but is does not have to be so complicated. Just print a cube and measure the surface finish on all six faces and publish. 
      Such data would be better than we have now.

      You are right about the "silly money" manufactures not wanting to play ball, as is not in their interest to have competition, but the likes of the reasonably price machines from B9 and formlab machines will soon alter the markets price profile. 


      Francis Leach



      On 8 April 2013 15:21, Graham Stabler <grezmos@...> wrote:
       


      On Mon, Apr 8, 2013 at 2:12 PM, Francis Leach <francisleach@...> wrote:

      Do you have any results that define the surface finish from any of the machines you have had experience of?

      That define it? I'm not sure quite what that means

      None of the solidprinters that I know quote any meaningful surface finish values.

      There may be a good reason for that, it is so variable with part geometry and the layer thickness is a very good indicator, more so for DLP than for FDM.
       
      It is a considerable shortfall in solidprinting machine's specification that the manufactures only quote resolution and not surface finish. I know it is a new technology but they have to catch up.

      Maybe... Going back the geometry question. The cube you imagine, when the surface finish on its top will be a function of the bed flatness but at the same time you can get poor results over large flat areas because of the way the printers apply more light to the edges, for this reason they are avoided. So the finish would depend on area and would most likely not be consistent enough accross the area for a reasonable one number value.

      On the purely vertical faces you are really looking at variation in registration and perhaps the uniformity of cure in a layer, i.e. how square the edges of a layer are. A 45 degree surface with cubic voxels will have a staircase equal to the layer thickness/voxel dimension but make the surface close but not equal to vertical and you see really tall steps much greater than the layer thickness due to the quantisation in the x,y plane. It is all very variable but you can get a good idea purely from the geometry of the pixels. The roughness is of the order of the size of the pixels.

      Photographically, the B9 creator looks a close match for them in our world of sensible machine prices but one does not really know that, without data.

      It is not as good as the Objet results I have had (although they have issues with flat surfaces too) but it FAR FAR better than anything I have seen out of an FDM.

      If you ask Shapeways for a Ra value they are of the opinion that the words "detail" and "Ultra fine" are acceptable descriptors of finish, they are clearly not, I have found that out at my cost. They publish photograph of items which are almost meanless being presented without any scale references that would enable a comparison to be made. If they specified the surface finishes of a simple flat surfaces produced by their various machines you would be able to make comparisons with other solid-printed results. 

      The B9 Creator looks very capable and I hope to be able to buy one. I believe Michael may be able to provide a set of results that would clearly show that his solidprinter is significantly better than a FDM machine and equal or even better than some commercial "silly money" machines.

      Except you still would not have any numbers for the silly money machines or the FDMs. 
       
      The point Graham makes about finishing is most relevant. Some detailed items are impossible to finish if the basic build surface finish is poor. By the time you have finished the smoothing the general area of the item you will have removed the delicate features.

      That's a good point but bare in mind that a defocussed DLP system will give you awesome Ra numbers and you will still have no delicate features.

      I think some sort of measure of printer performance would be a really good thing however I don't think a simple cube really fits the bill. There are also other issues such as feature size and accuracy that some of us engineers would be interested in but sometimes it is hard to compare the machines. An FDM can have excellent accuracy and resolution of its motion however the large filament limits the min feature to much more than its positional resolution, I think formlabs are struggling also to define their resolution for similar reasons.

      Graham




      --
      Regards

      Francis Leach





      --
      Regards

      Francis Leach


    • Francis Leach
      Graham Thank you for putting me straight. I am glad I have your opinions. Personally, I am not wishing to establish a standard test at all, I only seek a
      Message 2 of 21 , Apr 9, 2013
      Graham

      Thank you for putting me straight. I am glad I have your opinions.

      Personally, I am not wishing to establish a standard test at all, I only seek a better understanding of the surface finish of solidprinted surfaces without relying on "out of focussed, low res, long shot photographs or non-quantifiable descriptors such as "detail or ultra fine"  or Variable Voxelisation . I have previously submitted a shapeways print referred to as "detail ",   not!

      In your deliberation you have fully developed the reasons why we should not have a simple test to assess the quality of solidprinter. I am not convinced.

      The photographs submitted by The B9 creator of the statue are "honest" ( that is, clear close-ups showing every detail )  and the best I have ever seen to understand what a good print looks like. With a simple test using a surface finish meter we may be moving to an understand of what is actual a Good solid print in terms of surface finish.

      I have had many solidprints from an Objet printer with a claimed High res build of 16 micron, in a white resin which once sprayed with primer was found to be a rough as the bears bottom. 
      I believe the results from a number of kickstarter companies are now very good and better than some expensive "silly money" machines but we needs some numbers to get a measure of what "good" really is.

      No problem with that?.

      Of the "silly money" machines only the Solidscape machines has a surface finish listed of 32 -63 micro inches RMS. ( American standard ). Probably the best solidprinter results I have ever seen. Well done I say.
      On inspection of the 3DS Website and the multitude of  product brochures, surface finish performance is not listed. 
      Envisiontec do not quote a surface finish performance for any of their machines but do state that the surface finish is constant over the entire build area. What is the surface finish? I believe the Prefractory micro to be a good performer but at a price most private individuals will not afford

      Examples brochures below.

      Regards

      Francis





      On 9 April 2013 08:12, Graham Stabler <grezmos@...> wrote:
       


      I appreciate our American's friends have their own national standards for surface finish, based on our own imperial measurements system which suited us for many a century or two ;0)

      I was meaning we are not really in a position to create a standard at all :D
       
      I see your point about stair casing but the measure of a good machine can be assessed easily by how well it does with a simple flat surface.  Horizontal or vertical or inclined.

      Not really. If an individual layer has flat sides (not like stacked tubes like FDM) then the layer thickness will not affect the surface finish so you could have 1mm layer thickness and extremely smooth sides and top to your cube (not to mention all of the vertical sides will be the same) so I don't really see that it shows enough of the whole picture, I also mentioned that the are of the horizontal surface affects the finish for the DLP printers.

      Do you think a cube with straight walls, with a sloping roof is better or perhaps one with a tiled roof  or one with an additional dome. It could get silly

      As it is a 3D printer I am not sure you can get silly because it is just a matter of drawing it though ease of measurement needs to be taken in to account, a horizontal surface, a vertical surface and and at least one sloped surface would be good although I am not sure what angle would be best, 45 degrees if the voxels are cubes but not sure otherwise.

      I assume the print from the B9 would be supported on a build structure rather than being laid straight to the brushed aluminium build platform.  

      You will have to specify this in your standard :)

      I believe honest manufacturer would be delighted to show us what they can do.

      But presumably they do not become dishonest if they can't be bothered or don't think it helpful :)

      Cheers,

      Graham



      On 8 April 2013 17:31, Graham Stabler <grezmos@...> wrote:
       

      If you want an ISO standard you might be talking to the wrong people :) 

      I don't see the reason to measure more than two faces maybe 3 if you do the base, which has a brushed aluminium finish on the B9 :D . As I said I think it it will give a poor estimate of the finish as there will be no stair casing on vertical surfaces.

      Cheers,

      Graham


      On Mon, Apr 8, 2013 at 5:26 PM, Francis Leach <francisleach@...> wrote:
       

      Graham

      I am sorry you do not understand the question but I do not understand your reply.

      A ISO standard sets controlling parameters so meaningful data can be derived by test but is does not have to be so complicated. Just print a cube and measure the surface finish on all six faces and publish. 
      Such data would be better than we have now.

      You are right about the "silly money" manufactures not wanting to play ball, as is not in their interest to have competition, but the likes of the reasonably price machines from B9 and formlab machines will soon alter the markets price profile. 


      Francis Leach



      On 8 April 2013 15:21, Graham Stabler <grezmos@...> wrote:
       


      On Mon, Apr 8, 2013 at 2:12 PM, Francis Leach <francisleach@...> wrote:

      Do you have any results that define the surface finish from any of the machines you have had experience of?

      That define it? I'm not sure quite what that means

      None of the solidprinters that I know quote any meaningful surface finish values.

      There may be a good reason for that, it is so variable with part geometry and the layer thickness is a very good indicator, more so for DLP than for FDM.
       
      It is a considerable shortfall in solidprinting machine's specification that the manufactures only quote resolution and not surface finish. I know it is a new technology but they have to catch up.

      Maybe... Going back the geometry question. The cube you imagine, when the surface finish on its top will be a function of the bed flatness but at the same time you can get poor results over large flat areas because of the way the printers apply more light to the edges, for this reason they are avoided. So the finish would depend on area and would most likely not be consistent enough accross the area for a reasonable one number value.

      On the purely vertical faces you are really looking at variation in registration and perhaps the uniformity of cure in a layer, i.e. how square the edges of a layer are. A 45 degree surface with cubic voxels will have a staircase equal to the layer thickness/voxel dimension but make the surface close but not equal to vertical and you see really tall steps much greater than the layer thickness due to the quantisation in the x,y plane. It is all very variable but you can get a good idea purely from the geometry of the pixels. The roughness is of the order of the size of the pixels.

      Photographically, the B9 creator looks a close match for them in our world of sensible machine prices but one does not really know that, without data.

      It is not as good as the Objet results I have had (although they have issues with flat surfaces too) but it FAR FAR better than anything I have seen out of an FDM.

      If you ask Shapeways for a Ra value they are of the opinion that the words "detail" and "Ultra fine" are acceptable descriptors of finish, they are clearly not, I have found that out at my cost. They publish photograph of items which are almost meanless being presented without any scale references that would enable a comparison to be made. If they specified the surface finishes of a simple flat surfaces produced by their various machines you would be able to make comparisons with other solid-printed results. 

      The B9 Creator looks very capable and I hope to be able to buy one. I believe Michael may be able to provide a set of results that would clearly show that his solidprinter is significantly better than a FDM machine and equal or even better than some commercial "silly money" machines.

      Except you still would not have any numbers for the silly money machines or the FDMs. 
       
      The point Graham makes about finishing is most relevant. Some detailed items are impossible to finish if the basic build surface finish is poor. By the time you have finished the smoothing the general area of the item you will have removed the delicate features.

      That's a good point but bare in mind that a defocussed DLP system will give you awesome Ra numbers and you will still have no delicate features.

      I think some sort of measure of printer performance would be a really good thing however I don't think a simple cube really fits the bill. There are also other issues such as feature size and accuracy that some of us engineers would be interested in but sometimes it is hard to compare the machines. An FDM can have excellent accuracy and resolution of its motion however the large filament limits the min feature to much more than its positional resolution, I think formlabs are struggling also to define their resolution for similar reasons.

      Graham




      --
      Regards

      Francis Leach





      --
      Regards

      Francis Leach





      --
      Regards

      Francis Leach

    • Graham Stabler
      ... I m not against a simple test nor against some way of being able to see the results only the exact means by which you propose to do it. Personally I feel
      Message 3 of 21 , Apr 9, 2013
      • 0 Attachment

        In your deliberation you have fully developed the reasons why we should not have a simple test to assess the quality of solidprinter. I am not convinced.

        I'm not against a simple test nor against some way of being able to see the results only the exact means by which you propose to do it. Personally I feel able to assess the quality well enough for my needs from a half decent pic, preferably with scale.

        I've been wondering how to make an image that really shows the finish and came to the conclusion that a crossection with scale would be rather good. This can be done using a saw and wet and dry paper and a basic microscope/macrosystem with reticule or scale so you can see how large the features are. I'm just thinking of this as something anyone could do, the part needs to be chosen carefully again though.

        I have had many solidprints from an Objet printer with a claimed High res build of 16 micron, in a white resin which once sprayed with primer was found to be a rough as the bears bottom. 
        I believe the results from a number of kickstarter companies are now very good and better than some expensive "silly money" machines but we needs some numbers to get a measure of what "good" really is. 

        No problem with that?.

        I'm not really as interested in finish as I am in accuracy but it would be nice.

        Cheers,

        Graham
      • bobgarrish
        Accuracy is #1 for me right now, because we re not quite there on accuracy yet and the finish is an afterthought on a part that doesn t fit (engineering bias).
        Message 4 of 21 , Apr 10, 2013
        • 0 Attachment
          Accuracy is #1 for me right now, because we're not quite there on accuracy yet and the finish is an afterthought on a part that doesn't fit (engineering bias). I have seen some test objects designed to test accuracy / resolution and they seem well designed.

          That being said, the finish on FDM parts is unacceptable for anything that's going to be molded or truly functional (ie: a cleanly meshing gear) so there are limits.

          I think it would be easy enough to test surface finish, given a meter and a standard part. A collection of solids would likely do, actually: a cube on flat, a dodecahedron, a tetrahedron, and the top and bottom parts of a sphere. I'm thinking the features that matter would be flats, vertical surfaces, near-flats (10 degrees), near verts (80 degrees), spherical surfaces, and underhanging versions of all of the above.

          If you took a cube, left one side vertical, did a very shallow and a very steep ramp off two other sides, and bulged the remaining side out like a sphere then you'd have a pretty good tester. Then mirror it across the plane on which it sits so you have the underhanging versions of all of those.

          I might draw up such a thing later on...for right now the focus is on getting a printer made and printing, and then doing dimensional testing. Accuracy will probably be a series of cube-ish objects with holes and cylindrical protrusions: easy things to measure with micrometers and pin gauges.

          After that, I'm sure the industry will innovate on how the tests are done such that they can scam them a bit. A lot like how CNC router manufacturers quote their accuracy as the resolution of the motion system when the sum of the mechanical errors is thousands of times larger.
        • Francis Leach
          Bob You completely get it, thank you
          Message 5 of 21 , Apr 10, 2013
          • 0 Attachment

            Bob

            You completely get it, thank you

            On 10 Apr 2013 17:34, "bobgarrish" <bobgarrish@...> wrote:
             

            Accuracy is #1 for me right now, because we're not quite there on accuracy yet and the finish is an afterthought on a part that doesn't fit (engineering bias). I have seen some test objects designed to test accuracy / resolution and they seem well designed.

            That being said, the finish on FDM parts is unacceptable for anything that's going to be molded or truly functional (ie: a cleanly meshing gear) so there are limits.

            I think it would be easy enough to test surface finish, given a meter and a standard part. A collection of solids would likely do, actually: a cube on flat, a dodecahedron, a tetrahedron, and the top and bottom parts of a sphere. I'm thinking the features that matter would be flats, vertical surfaces, near-flats (10 degrees), near verts (80 degrees), spherical surfaces, and underhanging versions of all of the above.

            If you took a cube, left one side vertical, did a very shallow and a very steep ramp off two other sides, and bulged the remaining side out like a sphere then you'd have a pretty good tester. Then mirror it across the plane on which it sits so you have the underhanging versions of all of those.

            I might draw up such a thing later on...for right now the focus is on getting a printer made and printing, and then doing dimensional testing. Accuracy will probably be a series of cube-ish objects with holes and cylindrical protrusions: easy things to measure with micrometers and pin gauges.

            After that, I'm sure the industry will innovate on how the tests are done such that they can scam them a bit. A lot like how CNC router manufacturers quote their accuracy as the resolution of the motion system when the sum of the mechanical errors is thousands of times larger.

          • Ben Mahony
            In fareness I think Graham and Michael get it as well Ben Mahony. ** Bob You completely get it, thank you
            Message 6 of 21 , Apr 10, 2013
            • 0 Attachment

              In fareness I think Graham and Michael get it as well

              Ben Mahony.

               

              Bob

              You completely get it, thank you

              On 10 Apr 2013 17:34, "bobgarrish" <bobgarrish@...> wrote:
               

              Accuracy is #1 for me right now, because we're not quite there on accuracy yet and the finish is an afterthought on a part that doesn't fit (engineering bias). I have seen some test objects designed to test accuracy / resolution and they seem well designed.

              That being said, the finish on FDM parts is unacceptable for anything that's going to be molded or truly functional (ie: a cleanly meshing gear) so there are limits.

              I think it would be easy enough to test surface finish, given a meter and a standard part. A collection of solids would likely do, actually: a cube on flat, a dodecahedron, a tetrahedron, and the top and bottom parts of a sphere. I'm thinking the features that matter would be flats, vertical surfaces, near-flats (10 degrees), near verts (80 degrees), spherical surfaces, and underhanging versions of all of the above.

              If you took a cube, left one side vertical, did a very shallow and a very steep ramp off two other sides, and bulged the remaining side out like a sphere then you'd have a pretty good tester. Then mirror it across the plane on which it sits so you have the underhanging versions of all of those.

              I might draw up such a thing later on...for right now the focus is on getting a printer made and printing, and then doing dimensional testing. Accuracy will probably be a series of cube-ish objects with holes and cylindrical protrusions: easy things to measure with micrometers and pin gauges.

              After that, I'm sure the industry will innovate on how the tests are done such that they can scam them a bit. A lot like how CNC router manufacturers quote their accuracy as the resolution of the motion system when the sum of the mechanical errors is thousands of times larger.

            • Francis Leach
              absolutely ... -- Regards *Francis Leach*
              Message 7 of 21 , Apr 10, 2013
              • 0 Attachment
                absolutely


                On 10 April 2013 20:07, Ben Mahony <ben.mahony@...> wrote:
                 

                In fareness I think Graham and Michael get it as well

                Ben Mahony.

                 

                Bob

                You completely get it, thank you

                On 10 Apr 2013 17:34, "bobgarrish" <bobgarrish@...> wrote:
                 

                Accuracy is #1 for me right now, because we're not quite there on accuracy yet and the finish is an afterthought on a part that doesn't fit (engineering bias). I have seen some test objects designed to test accuracy / resolution and they seem well designed.

                That being said, the finish on FDM parts is unacceptable for anything that's going to be molded or truly functional (ie: a cleanly meshing gear) so there are limits.

                I think it would be easy enough to test surface finish, given a meter and a standard part. A collection of solids would likely do, actually: a cube on flat, a dodecahedron, a tetrahedron, and the top and bottom parts of a sphere. I'm thinking the features that matter would be flats, vertical surfaces, near-flats (10 degrees), near verts (80 degrees), spherical surfaces, and underhanging versions of all of the above.

                If you took a cube, left one side vertical, did a very shallow and a very steep ramp off two other sides, and bulged the remaining side out like a sphere then you'd have a pretty good tester. Then mirror it across the plane on which it sits so you have the underhanging versions of all of those.

                I might draw up such a thing later on...for right now the focus is on getting a printer made and printing, and then doing dimensional testing. Accuracy will probably be a series of cube-ish objects with holes and cylindrical protrusions: easy things to measure with micrometers and pin gauges.

                After that, I'm sure the industry will innovate on how the tests are done such that they can scam them a bit. A lot like how CNC router manufacturers quote their accuracy as the resolution of the motion system when the sum of the mechanical errors is thousands of times larger.




                --
                Regards

                Francis Leach

              • Francis Leach
                Michael is a positive honest manufacturer ... -- Regards *Francis Leach*
                Message 8 of 21 , Apr 10, 2013
                • 0 Attachment
                  Michael is a positive honest manufacturer 


                  On 10 April 2013 20:15, Francis Leach <francisleach@...> wrote:
                  absolutely


                  On 10 April 2013 20:07, Ben Mahony <ben.mahony@...> wrote:
                   

                  In fareness I think Graham and Michael get it as well

                  Ben Mahony.

                   

                  Bob

                  You completely get it, thank you

                  On 10 Apr 2013 17:34, "bobgarrish" <bobgarrish@...> wrote:
                   

                  Accuracy is #1 for me right now, because we're not quite there on accuracy yet and the finish is an afterthought on a part that doesn't fit (engineering bias). I have seen some test objects designed to test accuracy / resolution and they seem well designed.

                  That being said, the finish on FDM parts is unacceptable for anything that's going to be molded or truly functional (ie: a cleanly meshing gear) so there are limits.

                  I think it would be easy enough to test surface finish, given a meter and a standard part. A collection of solids would likely do, actually: a cube on flat, a dodecahedron, a tetrahedron, and the top and bottom parts of a sphere. I'm thinking the features that matter would be flats, vertical surfaces, near-flats (10 degrees), near verts (80 degrees), spherical surfaces, and underhanging versions of all of the above.

                  If you took a cube, left one side vertical, did a very shallow and a very steep ramp off two other sides, and bulged the remaining side out like a sphere then you'd have a pretty good tester. Then mirror it across the plane on which it sits so you have the underhanging versions of all of those.

                  I might draw up such a thing later on...for right now the focus is on getting a printer made and printing, and then doing dimensional testing. Accuracy will probably be a series of cube-ish objects with holes and cylindrical protrusions: easy things to measure with micrometers and pin gauges.

                  After that, I'm sure the industry will innovate on how the tests are done such that they can scam them a bit. A lot like how CNC router manufacturers quote their accuracy as the resolution of the motion system when the sum of the mechanical errors is thousands of times larger.




                  --
                  Regards

                  Francis Leach




                  --
                  Regards

                  Francis Leach

                Your message has been successfully submitted and would be delivered to recipients shortly.