Shapeways print length ultimatum (yeah you told me so Bryan :-))

[Lemosteam]

All scales WRT printing at shapeways...

   

 Disappointing.  Below is the full text and image from a Shapeways forum topic I just started.  Many of you know I have been addressing this with them for almost two years...

I am not sure what I will now do about my roof offerings or the Stillwell kit.

Fellow shop-owners,

For those designers that have models longer than 100mm offered in FUD or FED, or Fine Detail Plastics, Shapeways representatives at all management levels have sternly and recently informed me that they will not honor their stated, published accuracy capability on that materials’ page:

“ACCURACY

±0.1- 0.2 mm for every 100 mm”

As an example, if one dimension in the design is 151.38mm the part should be printed within the calculated proportional range (i.e. the ±0.1 - 0.2 mm) would become ±0.1514 – 0.3028 when proportioned from 100mm to 151.38mm.

When printed, the same measurement on the part in this example should measure between (at WORST CASE, which I have always accepted) 151.0772 (151.38-0.3028) and 151.6828 (151.38+0.3028). In this actual example, the actual part measured 150mm- over 1mm too short for its intended purpose rendering the print useless.

Sounds logical right?

This is somehow not logical to the Shapeways’ production and Customer Service team. They claim this published accuracy is a guideline, but most of us realize that when a tolerance is mentioned, it becomes more of an expectation or allowance. They vehemently disagree with this to the point they will not reprint until the part is correct, but they will only reprint once, maybe correct, maybe not, then refund.

They have reduced my printability rating on some of these designs because of customer complaints that the parts do not come close to my measured CAD models, even though the design meets and exceeds ALL of their true design guides for wall thickness, etc. I feel now that I am being punished for having my customers contact Shapeways when a part is mis-printed too short. In other cases, they would NOT be degrading my printability ratings if the part were warped, defective (print shift) or broken, all errors on their side.

They also refuse to calibrate their printers to accommodate this requirement (or at minimum shop owner expectation?)

I urge each of you or your customers to measure your 100mm+ long parts in its long axis, and compare them to your CAD, if there are critical dimensions for fit.

Your customers deserve this.

In almost ALL of my complaint cases, the printed parts were over 1mm too short rendering them USELESS to my customer, only to be offered one reprint or a refund. Don’t forget all of the customer’s time that is wasted in the complaint process.

Here is an example from a customer, the opposite end is perfectly aligned. The CAD measures intentionally long to the shell body shown here under the printed roof because by design there is a small roof overhang on the real passenger car. you can see that the printed roof (N Scale) is well over 1mm too short:



In some cases, my customers were told that this was the fault of my design, that I did not take their "guideline" into account, inherently damaging my brand and reputation IMHO.

Fortunately, some of these customer service comments and replies were forwarded to me from my customers, so this is not merely a rant, but fact.

The parts are designed specifically to snap-fit into OEM parts currently available on the market and mate to surfaces for proper look and fit.

To prevent complaints and wasted print time and materials resulting from short prints, I have even been proactively informing customer service referencing the customer’s order number and the model link, that a 100mm+ design and has been ordered and that Shapeways should be prepared for that when they print. I have been told by some reps not to bother them with this, but I will continue as long as I wish.

I have twenty designs, which I must now find a new supplier for because North America Shapeways is UTTERLY INCAPABLE of and, more importantly, UNWILLING to meet the specification above; THEIR OWN published specification.

I am extremely disappointed with this resolution because they think this problem is too small to deal with. That may be the case, as I was quoted that “millions” of other parts were just fine (likely because they were never measured, because everyone assumed that the tolerance was being met).

Please note that this DOES NOT include the Dutch Printing Office, who have been incredibly understanding, capable and helpful. In fact it was the Dutch office and their service reps that FINALLY solved the problem with one of my popular kits, after printing and tossing too many to mention prints that I had to reject personally, almost to the point of having to cancel the project. Unfortunately, I cannot request these long parts to be printed there (not sure why).

To be fairly noted, if the parts to a kit are printed at the same time, this should not affect fit between the parts, but the overall length of the assembled kit may be affected.

Maybe this is no big deal to you or your designs, but it is to me, and I assume others whom rely on printed accuracy.

At minimum we should be able to expect that the first printing fall withing the tolerances of the "guideline", no?

I am disappointed, dejected and annoyed, but I would love to hear your thoughts.

Here is the link if you have an account and would like to comment:

https://www.shapeways.com/forum/t/shapeways-ultimatum-wrt-fine-detail-plastics-print-length-100mm-issues.97356/

[Ed Kapuscinski]

Incredibly disappointing, but not surprising from a company that's on the vanguard of this stuff and which is likely experiencing significant growing pains.

I'm hoping that the issue is really just that they don't have the bandwidth and aren't scaled to deal with this type of thing.

[C855B]

Ugh. That's quite a setback. They've basically told modelers, "Don't depend on us to produce anything that fits with our other output, or with anything else."

I see their POV, however. Back when I was running 2D PostScript service bureaus a handful of clients were trying to do hi-res color separations in separate batches, expecting everything to be accurately in-register by the time everything was done. It was almost never good enough. We tried to be good customer-service doobies and re-ran a lot of time-consuming jobs, but we could not get away from the reality that the equipment and materials were simply not capable of holding that kind of calibration. Like the corner Shapeways appears to be in, it was either manage expectations and probably turn away business, or continue to lose money, either on jobs that don't quite fit the production capability and require multiple re-runs, or spending personnel, equipment and materials on ill-fated attempts to keep stuff in X calibration.

They've identified a problem area, and while their "solution" sucks for us, they're doing what they think is best for their business. Service bureaus are a very dicey business model to begin with.

To our dilemma, however... the Shapeways situation is bifurcated - they have a great user interface in their well-developed storefront, but they are falling behind the curve on the tech, which is itself not advancing as well as we had hoped. If there was an easy solution I would throw some money at it (and get back into the service bureau biz [...shudder...]), but there are no easy solutions, and no "perfect" technology yet for modelers. Very frustrating.

[C855B]

Oh... I'll mention that Shapeways has the same problem I did - market erosion from low-end producers. The "good enough" clients are flocking to cheap desktop extruders, the one area of the technology that has been making big advances in resolution and quality. In my case it was when laser printers made the jump from 300 to 600ppi. So the easy stuff has been lost to the kitchen-table folks, while the difficult jobs continue to eat at the bottom line. I closed the doors of several service bureaus after key clients decided that laser prints on high-quality paper were acceptable for their work.

It's my opinion from this experience that the main thing saving Shapeways is the variety of offered materials, which seem to be oriented towards the art and jewelry clientele and not modelers.

[Lemosteam]

Incredibly disappointing, but not surprising from a company that's on the vanguard of this stuff and which is likely experiencing significant growing pains.

I'm hoping that the issue is really just that they don't have the bandwidth and aren't scaled to deal with this type of thing.

@Ed Kapuscinski Nope, they repeatedly tell me that they can print reliably under 100mm and refuse to try to print reliably over 100mm.

I have even offered technically feasible solutions to resolve this from a sample measurement upload that would then force an automated scaling on any model being place in that machine's vat that the bounding box already shows to be over 100mm (this is data already avaliable from the model when it is uploaded)

I suggested a separate machine that would be calibrated just for these parts so they could be printed reliably, etc.

Then I was promised it would be controlled at all manufacturing facilities, and then the Stillwell debacle hit.  It took them 38 failed printings to correct the issue before i could ship models to paid customers.

Even though i am sure they just considered me an outsider trying to tell them what to do (understandable) these are good ideas that are process capable and legit but they couldn't even see that I was trying to help them.  (yes all of this is what I know and do for a living).

I tell you this, they would never even be able to come close to being an automotive supplier if they made auto parts because their process capability and repeatability is for crap.

I viewed them as my supplier and my "voice of the customer" meant nothing to them.  Sad.

@C855B, I do too, but that is simply dealt with by changing the specification so that the cuystomer understands what they are going to receive.  For instance, if that spec has asid +/- 1mm for every 100mm, I would design the roofs with a slip joint, knowing that i have to accomodate that level of variability.  So basically they have false advertised their process capability.

The way to manage this is with proper dimensional control.  Auto suppliers adjust tooling every day to keep parts within spec as they manufacture them taking hundreds of measuments per day to do so.  There is absolutely no reason why Shapeways cannot do the same.  If these auto suppliers were not doing this we would all still be driving in 70's junk with fit and finish like a bad knockoff.

They have no sense of Kazien (continuous improvement).

Someday I'll explain why this technology can never achieve what we are hoping for without a total revolution in this tech- different discussion, different day.

[nkalanaga]

Not an ideal solution, but if they can print reliably under 100mm, could you make the roof in two pieces?  I imagine the real roofs had seams in them, so it shouldn't be hard to hide one joint in the middle, and it wouldn't take any more material.

[Lemosteam]

It would not be a seam, it would have to be a 1-2 foot wide scale gap that would have to be filled.  Not impossible and it would have to have a hotwheels type tongue and mating slip joint. 

The gap would also throw off the alternating strip pattern on the Harriman roof design.

Even so, since they will not tell me what their capability is, how do I calculate the amount of gap to design into the parts?  This is why that spec is so crucial.  My entire working career revolves around design and accommodating tolerance accumulation in a design.

With their ultimatum, I cannot mathematically determine how best to integrate a design to account for their process capability. (lack thereof).

[orionfield]

@Lemosteam I feel your pain! Its just so hard to find a good vendors these days. After being screwed over financially by GoEngineer, I thought that Shapeways would be a good temporary alternative until I could buy my own printer, which in turn would be a stopgap to complete the funding and development process for a proprietary printer we're working on designing. I though for a moment of Giving 3D hub a chance, but their pricing is through the roof! I honestly don't know what to do.

[nkalanaga]

To eliminate the gap, one could make the two halves a little long, and trim the length to fit, but I don't see any way to get the roof texture in a two-piece roof.  That would be a problem even if the required length could be determined.

[Maletrain]

Maybe we need to go back to printing the roof like an old extrusion that will be cut to fit two printed end pieces.  At least that way, the symmetry could be maintained, and the fit could be assured.

I don't have any real insight into 3-D printing technology.  So, I am wondering what the real problem is.  With regular computer laser and ink jet printers available today, the fit to a regular 8.5" x 11" page does not seem to be anything like +/- 1 mm for every 100 mm (=4").  So, does the 3-D print material shrink after printing, or what?

[nkalanaga]

Good question.  Since some of the material is ordinary plastics, ABS and similar, it probably does shrink as it cools, and the shrinkage would increase with the size of the product.  I don't know what other materials are used for 3D printing, so can't say anything about them.

On the other hand, injection molding has the same problem, and manufacturers get around that every day.

[Maletrain]

On the other hand, injection molding has the same problem, and manufacturers get around that every day.

Yes predictable shrinkage should not take anything more than a calibration process to fix.  But random shrinkage would be a problem, if that occurs.  Dots of plastic added in various manners/time intervals might make shrinkage less uniform than injection molding.  Just speculating, since we don't yet have a response from anybody who really knows.

[nkalanaga]

Maletrain:  True, printing random shapes will result in random shrinkage.  On the other hand, if a manufacturer is going to sell the service of printing random shapes, they need to be able to tell the purchaser/designer what to expect.  That would probably be a computational nightmare, but until they can, I wouldn't want to buy the first printing of an object.

And even if they can't predict the shrinkage up-front, they should at least be able to make it consistent!  The same material, in the same shape, should shrink the same way each time.  It may not be the way the designer wants it, but as long as it's consistent, the part can be redesigned to get it right.  Inconsistent shrinkage is unacceptable in a "mass production" process.

[bbussey]

Yeah, there are limitations and they don't want to truly take responsibility.  So I work around their shortcomings.  With a passenger car roof for example, I would render in two halves slightly over length, and the modeler would have to splice them together after shortening them to a combined length that fits the model.  What I'm doing with the acid tank is having the etchings (which are constant) dictate the joins between the Shapeways parts.  For example, the length of the FXD body core is around .050" shorter than the width of the etched body wrapper.  The FXD ends have a key that inserts into the end of the FXD body core, but they fit flush against the etched body wrapper for a seamless fit. So the key between the two FXD parts keeps the dome ends in the proper orientation, and the etched wrapper keeps everything flush.  FXD works for the gondola kits and the White Towers because the entire body is FXD, so if it is off, all the parts are off proportionally, and the difference is insignificant in the areas that matter (coupler box interior, bolster pin hole, etcetera).  With the KV Caboose, I didn't finalize the etchings until I had good FXD masters.

Even with the other contractor I used, there could be issues, as the ProJet 3000 machines can get out of register fairly easily.  That contractor was great and worked to get everything right, but it was tough to get everything calibrated precisely.  What I ended up doing with them is outputting a 2" x2" by 1/8" plate (X/Y/Z), measuring with calipers to get the scale factor, and compensating by rescaling the STL file.  But that only worked because the jobs were being produced on the same machine in the same orientation every time. 

The Shapeways FXD work great as detail parts.  It can be feasible for the larger parts as well, but you have to get creative and eliminate the fluctuation of the part sizes as a factor.

[Maletrain]

When I used the word "random", I was not thinking about what you interpreted it to mean.  If the issue is only that the machine is asked to make various parts in various shapes at various orientations to the machine x-y-z axes, and shrinkage is uniform in its machine-oriented x-y-z components, then that is just a machine calibration issue.

On the other hand, when I was talking about random shrinkage effects, I was suspecting that the material itself may not always shrink the same way each time an object is printed the same way on the same machine.  Because the material is laid-down as tiny bits that stick together with lots of intersticial spaces, it may not shrink the same and as uniformly as it would when the same material is injected into a mold to form a solid piece.  Maybe that really isn't a factor.  But, if it is, then I can understand how the printer could not guarantee a specific size by calibrating the machine for a specific shape - there would be some unpredictable amount of shrinkage that differs from print-to-print.  But, so far, nobody who really knows has said that is a real problem.  Unless it is, then it seems to me the fault lies with the inability of the printer vendor to calibrate the printers and maintain that calibration.