3D printing (Was: New early P&R steel.....)
On Thu, Dec 19, 2019 at 01:50 PM, Dennis Storzek wrote:
Thanks Tom, very informative. This proves to me that Shapeways is just maxed out and still found wanting. It's no longer an issue of resolution, but designing around the "wax tracks". It seems the SLA process is better suited to our parts. Compare these pix to the pix of the D&RGW coal hopper Eric presented a week or so ago.I believe Corey Bonsall, who does the D&RGW and Utah Coal Route gons, uses a Form 2 "upside down" SLA printer where the part is built from the bottom up as it's lifted, layer by layer, out of the resin. For best results parts need to be oriented at an angle and parts of any complexity require a literal forest of supports. Shapeways had a similar process, Hi Definition Acrylate, but they dropped it because their trimmers lacked the skill to trim such parts quickly and without damage. In this field you can't have low prices, low wages and high skill levels across the board. For a while they offered HDA parts untrimmed but they've dropped that as well. If you _can_ design around the wax track problem the 3D Systems' multi-jet modeling process (which Shapeways calls Fine Detail Plastic) is faster and much less expensive than any SLA. For large parts, like passenger car sides, the Form 2 and other small SLA printers won't work at all. Here are three images (front, back and detail) of a pair of car sides I got from Shapeways on Monday. These are images of passenger car parts, but they're shown for the technique and not the product. On the PCL and some other forums any discussion would involve the parts and not how they were made. I'm not ready for that discussion yet! http://www.pullmanproject.com/3989A-1.jpg http://www.pullmanproject.com/3989A-2.jpg http://www.pullmanproject.com/3989A-3.jpg Those are flat with no sidewall detail. Shapeways Smoothest Fine Detail Plastic process (16 micron layers) shows minimal stairstepping on sloped surfaces. That encouraged me to design some Pullman blind and solarium ends with full rivet detail on the end sill and in the upper ends. Both regions are sloped. I also included the handrail mounting flanges and bolts on the door frame, a vertical surface on the print. Here's a photo showing four different ends They're castings, but the masters were printed: http://www.pullmanproject.com/EndsC.jpg Here's what the sidewall detail is supposed to look like: http://www.pullmanproject.com/EndsD.JPG And what it does look like: (Best I could do without a macro lens.) http://www.pullmanproject.com/EndsG.JPG There is a wax track below the detail, but in this case it's not objectionable. Since it's the only such detail in the area, and the wax track is short and hidden under the handrail, it's not at all noticeable. You need to pick your battles. Tom Madden
|
|
Tony Thompson
Tom Maddenwrote (in part): In this field you can't have low prices, low wages and high skill levels across the board. This reminds of what printers used to say to publishers, who naturally wanted the very best deal in every dimension: "Quality, speed, price. Take your pick." Tony Thompson
|
|
Dennis Storzek
On Thu, Dec 19, 2019 at 03:05 PM, Tom Madden wrote:
http://www.pullmanproject.com/EndsC.jpgThere is still a problem with banding, very visible on the end doors of the lower two ends.. Not a problem is used for resin masters, but not ready for prime time without a lot of sanding. Cory's Utah Coal Route GON shows the angular layers typical of SLA; again they can be sanded out, but good luck if there are a multitude of rivets. I sometimes wonder if these are machine issues due to a lack of rigidity; the machines seem to be built like computer printers rather than machine tools. Dennis Storzek
|
|
Corey Bonsall
Hi Dennis,
I've had my Form 2 apart enough to clean and keep running, and you've hit the nail on the head. The guts are derived from paper printers, and the forces needed to break the last cured layer off of the tray window are enough that when the build plate returns the part back to the tray, the alignment is slightly off. If I pause the printer, the build plate moves up even higher in the Z direction, and I end up with an even more noticeable layer. That being said, if I make sure to use a proper primer and gloss paint base coat, most of those lines go away. My examples of the UCR cars did not have primer coat, as they were tests to see if the Model Master acrylic would adhere to the resin as well without the primer. Adhesion is still good, but as you noticed, the print lines are more pronounced. Corey Bonsall
|
|
dalemuir2@...
Re: Shapeways issues.
I have had inconsistent results from Shapeways printing long (205mm) bridge girder parts. Parts I received were off by 1mm or more. This makes it impossible to assemble parts that must have consistent dimensions. Imagine printing car sides, roof, and floor with convenient locating pins only to discover that nothing fits.
The following illustrates the issues.
The model is a seven track underpass with two independent single track bridges and a five track bridge. There are a total of six full height girders and four "split" girders consisting of a top and bottom part. The floor is sandwiched between the top and bottom part. Total part count is 14. (Split girders are two parts each.)
One of the four split girder top parts was about 1mm too short, so the locating pins didn't line up. All four split girder bottom parts were useable.
Two of the six full girders were about 1mm too short.
In summary, 3 of 14 parts (about 21%) were unusable. All 3 unusable parts were duplicates (same 3D model) of parts that came out near perfect. Anyone with a little knowledge of process control would recognize immediately that Shapeways' process is out of control. Shapeways just dismissed the issue claiming that the parts meet their specifications for Fine Detail Plastic. They won't refund or re-print.
I had to re-order (and pay for) the defective part to finish the project. The re-ordered parts were about 0.5mm too short, but I was able to make them work.
Because of this, I can't add this model to my Shapeways store.
The following is an excerpt from an email exchange I had with the Shapeways Quality Control person.
Begin quote from Shapeways QC person The print resolutions are as follows: End quote
To me, the frustrating part is that Shapeways can produce accurate parts, but they won't. They seem to be more interested in cranking out parts a cheaply as possible.
Dale Muir Geneva, IL
From: main@RealSTMFC.groups.io [mailto:main@RealSTMFC.groups.io] On Behalf Of Tom Madden via Groups.Io
Sent: Thursday, December 19, 2019 5:05 PM To: main@RealSTMFC.groups.io Subject: [RealSTMFC] 3D printing (Was: New early P&R steel.....)
[Edited Message Follows] On Thu, Dec 19, 2019 at 01:50 PM, Dennis Storzek wrote: Thanks Tom, very informative. This proves to me that Shapeways is just maxed out and still found wanting. It's no longer an issue of resolution, but designing around the "wax tracks". It seems the SLA process is better suited to our parts. Compare these pix to the pix of the D&RGW coal hopper Eric presented a week or so ago. I believe Corey Bonsall, who does the D&RGW and Utah Coal Route gons, uses a Form 2 "upside down" SLA printer where the part is built from the bottom up as it's lifted, layer by layer, out of the resin. For best results parts need to be oriented at an angle and parts of any complexity require a literal forest of supports. Shapeways had a similar process, Hi Definition Acrylate, but they dropped it because their trimmers lacked the skill to trim such parts quickly and without damage. In this field you can't have low prices, low wages and high skill levels across the board. For a while they offered HDA parts untrimmed but they've dropped that as well.
|
|
I gave up on Shapeways a while ago, mostly due to their exorbitant pricing. Try https://print.all3dp.com/ the next time you need some printing done. I found the results to be quite satisfactory and the prices to be about a third of what Shapeways was asking for the parts I needed printed. Admittedly, my prints were not nearly to the level of precision these are, but it might be worth a try. I'm looking at them for bolsters and end beams for a narrow gauge tender I'm working on. they will be printed in casting wax and cast in brass. the total cost will be about $45 for all four parts.
toggle quoted messageShow quoted text
HTH Bill Lugg
On 12/20/19 7:59 AM, dalemuir2@comcast.net wrote:
|
|
Dennis Storzek
On Thu, Dec 19, 2019 at 03:05 PM, Tom Madden wrote:
I believe Corey Bonsall, who does the D&RGW and Utah Coal Route gons, uses a Form 2 "upside down" SLA printer where the part is built from the bottom up as it's lifted, layer by layer, out of the resin. For best results parts need to be oriented at an angle and parts of any complexity require a literal forest of supports.Now that we have some folks with real hands-on experience in this discussion, can anyone tell me why SLA parts are oriented at an angle? Is it an attempt to change the angle of the overhangs so they don't need supports, or simply to provide more room for the supports? Back to Tom's Shapeways Smoothest Fine Detail Plastic parts, is this a fused deposition process? One would think that some angular overhang would be possible, where each layer would project out less than half the width of the filament being deposited, and therefore be self supporting, like the overhanging bricks in fancy brickwork. If the underside of those handrail brackets would have projected from the vertical surface at a 45 deg. angle, could they have been built without the wax support and it's attendant track? Dennis Storzek
|
|
Tom Madden
On Fri, Dec 20, 2019 at 08:30 PM, Dennis Storzek wrote:
..., can anyone tell me why SLA parts are oriented at an angle? Is it an attempt to change the angle of the overhangs so they don't need supports, or simply to provide more room for the supports? Standard, "top down" SLA parts don't need to be angled - each layer is created atop the previous one. On "bottom up" SLA printers like the Form 2 each layer is created against the surface of the resin tray and has to be broken free of the tray surface after each layer is printed. Angling the part minimizes the area of the part in contact with the tray so it's always that joint that gives way and not the part as the build plate rises after each layer is printed. The CLIP bottom up printer uses a permeable tray and a process that keeps a layer of oxygen (?) between the tray and the part so it doesn't attach to the tray, but they (so far) have chosen not to serve the low end market. Back to Tom's Shapeways Smoothest Fine Detail Plastic parts, is this a fused deposition process? One would think that some angular overhang would be possible, where each layer would project out less than half the width of the filament being deposited, and therefore be self supporting, like the overhanging bricks in fancy brickwork. If the underside of those handrail brackets would have projected from the vertical surface at a 45 deg. angle, could they have been built without the wax support and it's attendant track?Fine detail is a jet printing process. I believe the jet printers Shapeways is using are 600 x 600 DPI X & Y, and at 16 microns per layer the Z is 1600 DPI. (3D Systems has 750 x 750 DPI machines with 13 micron layer capability, and 1600 x 900 but only for 32 micron layers.) In this process each individual dot/voxel has to have something under it so the wax is unavoidable. With standard SLA you can have unsupported overhangs of half the laser spot diameter from layer to layer but those layers are thicker - like 0.002". Tom Madden
|
|
Tom Madden
Fine detail is a jet printing process. I believe the jet printers Shapeways is using are 600 x 600 DPI X & Y, and at 16 microns per layer the Z is 1600 DPI. (3D Systems has 750 x 750 DPI machines with 13 micron layer capability, and 1600 x 900 but only for 32 micron layers.)If these don't sound like very high resolution printers, remember that there are two jet nozzels per pixel - you can place either wax or body material at each location. Tom Madden
|
|
I would like to offer to shed some light on the Jet printing Process, as it seams there is a genuine interest by this group to further understand the process. At work I manage and run both a 3D Systems Projet 3600 and Projet 5500 printers. These are what are know as ‘Polyjet' or ‘Multijet' 3d printers and is what Shapeways uses for their Fine Detail processes. The 3600 has a build area of about 8" x 12” inches and the 5500 16” x 20”.
toggle quoted messageShow quoted text
The print head on these printers has two rows of jets, one for the model or body material as Tom called it and another for the wax or support material. The print head on the Projet 5500 has 2400 jets across a width of 8”. This works out to 300DPI. To achieve higher resolution the print head shifts sideways by a pixel width and makes two passes for each layer to achieve 600DPI. For 750DPI it makes three passes for each layer and shifts both left and right of the first pass by 0.66 of a pixel width. I may have the math wrong for 750 DPI but I hope you get the idea. After each layer is deposited at what ever DPI selected, the head has a spinning roller that pushes or forces the deposited resin down into the previous layer. Behind the roller is a knife blade that slices the resin layer to the choose layer thickness. The final step is the head has a UV light that turns on and cures the layer. The 5500 head is only half the width of the build plate and must make a shift half the build plate width to print the entire 16” width. The 3600 print head is the full width of the build plate. Part print orientation is critical to get better printed parts. The head moves in the x axis. Thus if you want to orient your part along or across the print direction you need to pick which side of your part aligns with the X axis in your CAD file. I always design models as flat kits with the show side facing up on the build plate so no wax is used to support the critical faces of the part. To answer an earlier question as to the rigidity of these machines. They are built like machine tools. They have heavy frames, use servo motors(not stepper motors) and linear guide way bearings just like the CNC mills we have at work also. The Projet 5500 weighs in at 3000lbs. As far as process control. There really isn’t any. You load the part into the software, Orient it the way you want and press the print button. The only human involvement is the support removal. The wax is melted off in an oven, and then the last of it is removed in an ultrasonic cleaner. The big issue is warpage caused when the part is in the oven. I no longer remove wax for model parts in the oven. I remove it with a hobby knife and a dissolve the last of it with alcohol. Thus avoiding warpage. Unfortunately Shapways doesn’t offer manual removal of support material. If your part is not symmetrical, or has thin sections and thick sections there is a good chance it will warp in the support removal oven. You end up with bent parts or dimensional issues. I hope this explains things a bit. Ryan Mendell
|
|
Dennis Storzek
On Fri, Dec 20, 2019 at 06:57 AM, Corey Bonsall wrote:
That being said, if I make sure to use a proper primer and gloss paint base coat, most of those lines go away.Thanks for the further informed discussion on the 3D printers. I have a further thought for Corey, or anyone else dealing with an SLA model: It's clear that if a prime coat will hide the diagonal layer lines, they can't be very latge steps, and only show because they are an aberration on what is a smooth surface. I wonder if grit blasting with 220 grit abrasive would remove them, or more correctly break up the surface enough that the eye can't pick them out. I know that many of the resin kit builders grit blast resin kits to improve paint adhesion with no ill effect to the surface detail. Dennis Storzek
|
|
Dennis good point you brought up about the grit blasting. I do grit blast all my 3D printed parts or sand flat areas of parts with the sanders I sell trough National Scale Car. I developed the sanders specifically for this purpose.
Even surfaces that don’t have wax still have lines on them. These lines are left by the roller and knife of the jet printing head. The roller and knife need to be serviced at least a couple times a year to keep them in top condition. This is done by a 3D Systems technician that we have to pay and come in to do it. It’s not cheap, they charge upwards of $2000 a day for servicing the machine plus parts. Another source of part quality is shrinkage. This can be controlled with axis print compensation. Basically an offset put into the printer setup to correct for part shrinkage in x , y and z. But if parts are printed on different printers the compensation might be a bit off between them and parts could come out different lengths. This may have been the case with the girders in one of the previous posts. Also the longer the part the more possibility of length discrepancy do to shrinkage. Ryan Mendell
|
|
radiodial868
Ryan, thanks for this very good explanation. I reread it 3 times. I've been monitoring 3D printing with interest for about 3 years now, looking to move past having shapeways print my small detail parts and hoping commercial technology will eventually be able to do more complex car bodies. In the meantime, I keep watching for a home printer that can do those small detail parts. Speed is not important, but finish and ease of maintenance is.
RJ Dial Burlingame, CA
|
|
Bill Welch
Here is a used 3D Systems Projet 3600. https://www.ebay.com/i/264277258999?chn=ps&norover=1&mkevt=1&mkrid=711-117182-37290-0&mkcid=2&itemid=264277258999&targetid=593772166493&device=c&mktype=pla&googleloc=9012145&poi=&campaignid=2086169716&mkgroupid=76147899766&rlsatarget=aud-412677883135:pla-593772166493&abcId=1141016&merchantid=6296724&gclid=EAIaIQobChMImd65nN_J5gIVTvDACh3RMgOnEAQYASABEgIM5_D_BwE
Based on the used price alone, can only assume it will be awhile before we will see one for home use to turn out quality parts. Plus depending on where we live we might have to pay to fly in the techincian Ryan mentions plus the $2K for the routine maintenance. We can dream however. Wondering if there is any video showing the 3D Systems Projet 3600 or Projet 5500 printer in action? Thank you Ryan for taking time to explain the complexities. Bill Welch
|
|
I will take a video but I am off work for two weeks so it will have to wait
toggle quoted messageShow quoted text
On Dec 22, 2019, at 12:24 PM, Bill Welch <fgexbill@...> wrote:
|
|
Bill Welch
No worries, see you at THE Beach Ryan.
Bill Welch
|
|
Thanks for the pix of the actual prints. So infrequently see the real results. Agree that in HO the striations (stepping) are not objectionable. As you move up to S scale, I am tempted to think from your very helpful pix that one could live with them or use some of Ryan's National Scale Car sanders to get at them. You play the cards you are dealt or go without as they say.
Chris Rooney
|
|
steve_wintner
Thank you! That explains what happened to some parts I ordered - I didn't understand the surface striations I got. Not big but visible. Now I do. Well... May need to try again.
Steve
|
|
Dennis Storzek
On Sun, Dec 22, 2019 at 09:24 AM, Bill Welch wrote:
Based on the used price alone, can only assume it will be awhile before we will see one for home use to turn out quality parts. Plus depending on where we live we might have to pay to fly in the techincian Ryan mentions plus the $2K for the routine maintenance. We can dream however.Bill, it depends what you want to do. While the Prohet machines seem to be the only ones that have a large enough build area to do complete HO scale passenger car sides and roofs, the SLA machines, such as the Form 2 that Corey used for the UCR gons are considerably cheaper, and seem to do decent work, as attested to by Corey's pictures. And intriguingly, there is the little Anycube Photon, at 1/10 the price of the Form 2 machines. Also a photo cured resin system, but not SLA because it uses an array of UV emitters rather than a scanning laser to cure the resin, it actually claims thinner layer resolution than the Form 2 which would help with curved and angular surfaces. Not a big build area, but could be very useful for doors and ends. Typical of the hobby grade machines, the web site is very lean on technical specifications, but this review has what I was looking for: anycubic-photon-review Of course it remains to be seen if such a lightly built machine can keep the layers aligned, but we'll never know unless someone tries one for our type of parts. If I wasn't still working, I'd give one a try. Dennis Storzek
|
|
dalemuir2@...
Hi Ryan, It's great to hear from a subject matter expert that actually works in this industry . Thank you for your excellent explanation.
For reference, here is a quote from both Projet 3600 and Projet 5500 specs: Accuracy (typical) ±0.001-0.002 inch per inch (0.025-0.05 mm per 25.4 mm) of part dimension. Accuracy may vary depending on build parameters, part geometry and size, part orientation, and post-processing. End quote That means accuracy should be within approximately 0.1% to 0.2% For reference, Shapeways specification for Fine Detail Plastic is way looser: ±0.3- 0.7 mm for every 100 mm
Would you please answer these questions?
1. Is part orientation a factor in process control? 2. I've had multiple copies of the same 205mm long model made in Smoothest Fine Detail Plastic vary in length -0.5mm to about -1.0mm, even from the same order. Yet, about 60% of the same part are the correct length. I've been stonewalled by Shapeways' Quality Control person , saying that the lengths are within specification. So the question to you is "Are these dimensional inconsistencies a limitation of the machine, or could Shapeways do something better?" 3. If I interpret the specs correctly, my 205mm long part should be within ±0.2mm to ±0.4mm of 205mm, or between 204.6mm and 205.4mm. A difference of 0.8mm would mean the parts might not fit another printed part or a part from another source. 0.8mm is almost 3 inches in HO scale. My question then is "How can such 3D parts be used to create masters or molds?" "How can 3D printed parts be used for low volume parts instead of injection molded parts?" 4. Finally, as you probably know, Formlabs is coming out with the Form 3L late next year. I didn't see any specification on accuracy for the Form3 on their web site. In your opinion, would a Low Force Stereolithography (LFS) machine like the Form3 give more accuracy?
Thank you for taking the time to answer these questions. My hope was to be able to create low volume kits for items I need and share these with others. However, I can't do that with large parts due to the accuracy limitations with large parts, such as a passenger car or bridge girder.
Dale Muir Geneva, IL
From: main@RealSTMFC.groups.io [mailto:main@RealSTMFC.groups.io] On Behalf Of Ryan Mendell
Sent: Saturday, December 21, 2019 12:26 PM To: main@realstmfc.groups.io Subject: Re: [RealSTMFC] 3D printing (Was: New early P&R steel.....)
I would like to offer to shed some light on the Jet printing Process, as it seams there is a genuine interest by this group to further understand the process. At work I manage and run both a 3D Systems Projet 3600 and Projet 5500 printers. These are what are know as ‘Polyjet' or ‘Multijet' 3d printers and is what Shapeways uses for their Fine Detail processes. The 3600 has a build area of about 8" x 12” inches and the 5500 16” x 20”.
The print head on these printers has two rows of jets, one for the model or body material as Tom called it and another for the wax or support material. The print head on the Projet 5500 has 2400 jets across a width of 8”. This works out to 300DPI. To achieve higher resolution the print head shifts sideways by a pixel width and makes two passes for each layer to achieve 600DPI. For 750DPI it makes three passes for each layer and shifts both left and right of the first pass by 0.66 of a pixel width. I may have the math wrong for 750 DPI but I hope you get the idea.
After each layer is deposited at what ever DPI selected, the head has a spinning roller that pushes or forces the deposited resin down into the previous layer. Behind the roller is a knife blade that slices the resin layer to the choose layer thickness. The final step is the head has a UV light that turns on and cures the layer. The 5500 head is only half the width of the build plate and must make a shift half the build plate width to print the entire 16” width. The 3600 print head is the full width of the build plate.
Part print orientation is critical to get better printed parts. The head moves in the x axis. Thus if you want to orient your part along or across the print direction you need to pick which side of your part aligns with the X axis in your CAD file. I always design models as flat kits with the show side facing up on the build plate so no wax is used to support the critical faces of the part.
To answer an earlier question as to the rigidity of these machines. They are built like machine tools. They have heavy frames, use servo motors(not stepper motors) and linear guide way bearings just like the CNC mills we have at work also. The Projet 5500 weighs in at 3000lbs.
As far as process control. There really isn’t any. You load the part into the software, Orient it the way you want and press the print button. The only human involvement is the support removal. The wax is melted off in an oven, and then the last of it is removed in an ultrasonic cleaner. The big issue is warpage caused when the part is in the oven. I no longer remove wax for model parts in the oven. I remove it with a hobby knife and a dissolve the last of it with alcohol. Thus avoiding warpage. Unfortunately Shapways doesn’t offer manual removal of support material. If your part is not symmetrical, or has thin sections and thick sections there is a good chance it will warp in the support removal oven. You end up with bent parts or dimensional issues.
I hope this explains things a bit.
Ryan Mendell
On Dec 21, 2019, at 1:25 AM, Tom Madden via Groups.Io <pullmanboss@...> wrote:
If these don't sound like very high resolution printers, remember that there are two jet nozzels per pixel - you can place either wax or body material at each location.
|
|