Feedback request for upcoming FDM Quality Guidelines

chrislloyd · 2016-12-16T12:00:24.000Z

Hi,

Some comments on the guidelines:

1% dimentional accuracy is not good enough for me. I print loudspeaker cabinets at roughly 300mm x 200mm x 300mm scale. 1% would mean that a accuracy of 2.9mm is acceptabe. I cannot load drivers into the cabinets with this level of accuracy and would not be able to rely on prints adhering to these guidlines.

I don’t understand what a layer changing ‘seam’ means. I have received prints where there is a step in the surface and i would suggest that this is not perfect and therefore should not be allowed. I cannot sell products which have a step in the surface and i think this should be included in the guidelines. The guidelines need to have clear definitions (the photos will probably help!)

Disputes were previoulsy on the basis of prints being ‘perfect’. These guidelines seem to water that down in some areas.

I’d also like to suggest some content on infill. Infill has caused me all sorts of problems, not least because it is quite poorly defined. I asked two hubs to print identical stl’s (actually they were mirrored) and without any comms (because i was at this point completely in the dark about the concept of infill) one printed with triple wall thickness and infill of 20% on a 7mm thick enclusure wall (3mm solid - 1mm 20% - 3mm solid), the other printed with 1.2mm wall thickness and 15% infill (1.2mm solid - 4.6mm 20% - 1.2mm solid). One was 40% density, the other nearly 80% density. The difference was huge, and both were within the rules before the guideliness and would be within the rules of these guidelines. As such I think something needs to be included in the guidelines about infill. Or maybe on the ordering page.

In gereral supportive the the initiative.

Thanks

Chris

cobnut · 2016-12-16T12:05:30.000Z

Hi @chrislloyd I think we’re talking about guidelines here for the majority of print jobs, rather than rules that define what a customer must accept from a Hub in a printed object. If you need better than 1% accuracy on your prints, you can specify that with the Hub, ditto infill and perimeter settings etc.

At the end of the day, a good Hub should make sure they know what the client wants and needs and discuss the project until both sides are happy. I think these guidelines are merely to help manage both customer expectations and Hub requirements, not necessarily to say prints that meet these guidelines are always “OK”, if you see what I mean.

chrislloyd · 2016-12-16T12:17:38.000Z

Hi @cobnut Yeah that makes sense in terms of the general nature of the guidelines. And I agree that it remains down to me to specify what I need, but it is difficult to do that at first and I think some people may get put off by the difficult entry barriers. However, in the spirit of pushing the industry forward, in terms of quality and applicability of end use, I would stick with the need for dimensional accuracy and perfect finish. If I am a minority in this, then clearly they don’t need to go into the guideliness.

I guess there are two streams here, one which is focussing on low cost prototyping where these things don’t matter so much and the other is expanding into the production of finished parts. I see a bigger future for the industry in the latter stream.

cobnut · 2016-12-16T12:32:29.000Z

Hi Chris, first of all, I should point out that these guidelines are for FDM only, and FDM has specific issues that don’t apply to SLA or other print methods. If you need “perfect” quality, it’s possible that FDM isn’t the most appropriate technology (although of course, some FDM prints can be very fine indeed).

I don’t think you’re necessarily in the minority, I think you’re right that broadly speaking there are two types of print being made here; prototypes and “finished parts” (and, of course, both these categories can be split down further). I (try to remember!) to always ask customers what the print is for but even if it’s a prototype I don’t see that as a reason to compromise on quality and things like dimensional accuracy can be just as important for prototypes (after all, a prototype is often used to test fit, etc.).

Where it can be an issue is where supports are necessary. As we’ve discussed earlier in this thread, supports can often leave an uneven surface - it’s simply the nature of FDM printing; it’s virtually impossible to get as smooth a surface on top of supports as on a surface printed directly onto the bed or as a top layer. If the print is for a finished product, this could be a problem (depending upon whether the surface is actually visible (sometimes they’re parts that are built in such a way that this surface isn’t external), but for a prototype it may be no issue at all.

Perry_1 · 2016-12-16T14:28:59.000Z

@chrislloyd
Thank you for those comments. They perfectly illustrate why guidelines, setting expectations, and good solid communications are important to this issue.

It is my understanding that one of your prints did in fact fall below the current rules of 3dhubs.

Prints are supposed to have a minimum of 20% infill on 3dhubs.

In the case of walls, or shells as we call it in FDM printing, the rule I believe is 3 shells. So in both cases, the hub did both OK.

As for the layer changing seam, this is an important thing to point out as a difference between FDM and other forms of printing.

Primarily, what should be explained to the customer is they should have some expectation of seams.
It was mentioned in the previous thread that one of the common complaints from customers is that they were expecting the part to be fully smooth. That is, they were unaware that FDM is printed via hot threaded plastic drooling from a hot nozzle in tiny oval threads a layer at a time, resulting in noticeable layer changes. Thus, an explanation that there are in fact noticeable layers on the print is needed.

As for expecting “perfect prints” from such a technology, I think it depends on what one might define as perfect. I would never do an FDM print for a customer that was asking for “perfection.” In fact, I am not sure I would do an SLA print for someone that had an expectation of “perfection.” This is an unrealistic expectation in almost all disciplines of manufacturing, but definitely something that may be unattainable as you might define it, in 3d printing.

If you expect your print to have no shrinkage, no layers, no retraction marks, etc., then I suggest that you are a perfect prospect for the pre-order guidelines being discussed in advance of an order, so that your expectations can be brought inline with your perceptions of what FDM can deliver.

This is where a good solid relationship with a singular hub can be beneficial to you. Working with the same printer, same material, same brand, same color, same software, etc. can allow you to develop models that best match your expectations. That is, if your models need to be at perfect tolerances, you can then work with that hub to get to exact results. IF the PLA always shrinks at say .5%, you can always make your models .5% bigger, and get good fits.

I am not saying your expectations are too high, just that most customers need to be informed that FDM is the lowest cost, fastest solution for 3d printing, and as such, should be aware that if they need perfection, they should be notified of other options, that may take longer and cost more.

cobnut · 2016-12-16T14:48:34.000Z

Sorry, @Perry_1 I really, really, hate to do this, but the pedant in me simply can’t let it lie… If printing results in a 0.5% shrink, i.e. a 100mm wide part becomes 99.5mm when printed, then the model will need to be 0.5025% larger, not 0.5% larger. Sorry, I’ll go away now…

chrislloyd · 2016-12-16T14:59:09.000Z

Hi,

Yeah I appreciate ‘perfect’ is not a completely helpful term in this sense. Although the reason I used it is because it is used in the FAQ of 3D hubs quite freely and used in terms of what would happen in the event of a dispute between printer and customer. 3D hubs have an expectation that a print will be perfect, otherwise a refund is available. As such, it needs to be dealt with in the guidelines or removed from the FAQ.

Understand on seams (these were layers in my mind) I’m slowly getting up to speed with the terminology. My concern was for steps in the seams, where the print is misaligned.

I would love to move to other techniques, but realistically, at my price point i’m only interested in FDM.

Perry_1 · 2016-12-16T15:17:35.000Z

It would be nice if 3dhubs did not have to do this. This should really be handled by the hub. I know this is not happening, and this is why 3dhubs is trying to fix the customer expectation every time, because the number of complaints is too high, which costs them and the hub revenues when resolved. I know I am a prolific communicator on every order, and have even had customers exasperatedly declare to “just print it!” This is worth it, every time. I have had good luck with my reviews, and almost always it is about typing a lot up front!

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

Support marring is a complicated issue, and needs a lot of explanation. For a small thin overhang that requires a little support, the print should show little marring. For a larger, wider, longer, flatter section that requires support, I tell the customer they should expect a very rough, marred surface.

This is because on a such a supported area, the surface is mostly bridged across the support material, and will sag minutely as it is printed across that area. The more of this that occurs, the less perfect that surface will be.

Additionally, support material across long surfaces also means that section of the print is being printed on less stable terra firma, and is thus going to be subject to a little more “wiggle” as the print is occurring, resulting is much messier surfaces on long, wide, flat surfaces held up by support materials.
---------------------------------------

As for warping, shrinking, and corner curling with high shrinkage materials such as ABS, the customer should always be told this is a possibility or expectation. I always communicate this to a customer who has a print with large, long, flat models. I don’t have the issue usually, but I tell the customer it can happen, because I do not want to do a 10 hour print that the customer is expecting the next day, without them understanding that it COULD happen.

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

Infill expectations I usually do not have to communicate. But it should be stated by 3dhubs, as it is unclear to many hubs what the minimum infill is supposed to be. I see many hubs that say they print 15%, but 20% is the minimum for 3dhubs, I believe.

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

It would be easy for a large ABS print to shrink outside this tolerance of 1%, and I mention it as a possibility when ordering a print in ABS. ABS is a material used for injection molding, because it shrinks away from the mold so well!

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

Additional expectations should be communicated to the customer when they order, by the 3d hub. Sometimes I can see the use of the model right away, and can see there will be no problem. A bust of Ceaser is all good with a little shrinkage. A phone case could easily shrink to the point of not being usable with ABS. Small holes in the objects are going to be the wrong size, almost always! So I always discuss these tolerances when I see them in a model.
A 10mm hole in one model, and a 10mm post that goes in that hole on the other model; I can almost always be sure there is going to be an issue. We are printing in a smashed hot thread with FDM, and it is going to smash outwards a little. If this is the case, I need to discuss it with the customer.

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

I always type a lot before I accept an order. It takes time, but it is always worth it. It is all about looking at the model, material, etc., and discerning the use of the final print, or asking questions when that is not discernable. I ALWAYS want the customer to get what they want, not necessarily what they think they want.

I just did a quick review of some of my comments with customers, these are real:

-the posts on the thingiverse raspberry pi case will be very weak, and will likely break off if they are not careful

-some ABS orders should be printed in PLA

-gold PLA will not look like gold, it will look like plastic.

-get the drill bits out, because it will not work without it

-support material is going to leave your model unrecognizable, lol

-white will look better than black

-using it for cosplay, and going to paint it grey, it should be printed in grey, so that when it chips, it will be less noticable

-I cannot accept the laptop stand you designed, because it is so weak it will simply collapse

I am never shy about adding costs to the order for things the customer needs, like stronger infill, more shells, or better resolution. I offer .4mm layer height, because some customers just want a quick, cheap prototype. But I get orders from customers at that layer height, and I have to tell them not to do it! And I have had 3 orders where I told the customer to get it done with SLA, even though it will cost more and possibly take longer.

This is all key to a good experience for both parties. I have even told customers to get the order in PLA, and if it did not work out, I would reprint it again in ABS for free, to get them to switch away from a shrink prone disaster. I have never had a customer get it reprinted.

Perry_1 · 2016-12-16T18:34:19.000Z

@cobnut

Thanks for that, that is pretty good! I would have printed that part, and it would have been off by 25% of the width of the hole in my nozzle, on each side!

cobnut · 2016-12-16T18:44:37.000Z

Sorry, I had an attack of Friday afternoon mischief As it happens I’m doing an order now where the customer needs very accurate dimensions. One print had an X of 117mm and it came out (in ABS) as 116.68mm. Pretty happy with that but an going to reprint at 100.4% to see if I can get closer.

Perry_1 · 2016-12-16T18:49:07.000Z

I did not know remember that in the FAQ, and sure enough, you therefore have every right to expect it. It really is a bad term to use for complaint resolution. Wow, that needs to be a different phrase.

As for your price point, that is exactly the key. Realistic expectations for the technology, and thus realistic expectations for the price.

There are more perfect ways to 3d print, and less perfect ways to 3d print, but there are probably no perfect ways to 3d print.

But sometimes the customer is looking for that perfect price point, in which it is likely FDM will be more perfect!

Jackson_Moody · 2016-12-16T22:59:06.000Z

Great! That makes perfect sense! Thanks!

Odense3DPrint · 2016-12-18T11:08:29.000Z

@Robin3D

also that is on an ultimaker… but maybe 3dhubs does not care what printer is in use?

Enza3D · 2016-12-19T19:33:43.000Z

I do want to add that guaranteeing dimensional accuracy is a bit of a point of concern for me. I am not concerned by my ability to meet it, I am concerned by customers potentially misinterpreting what that means. Guaranteeing dimensional accuracy does not make any difference if the part is poorly designed; i.e. the design of the part does not take into account tolerancing for mating parts and such. This is the classic example of making a 2 mm hole for a 2 mm rod; even if the hole is perfect, the rod will never fit into that hole, because the hole does not take into account dimensional variations that will occur in the rod. It will need to be made explicitly clear to customers that this guaranteed dimensional accuracy is based off of the model they provide, not what they intend it for. If a part is provided that is 2"x2"x2", then it is only the Hubs responsibility to make sure that the print is within your set dimensional variation.

My other concern with this is that it’s a royal pain in the butt to get dimensions from STL files as that is not really their intention; STL files just contain information about the geometry of the surface but contain no data about dimensions. If 3D Hubs is going to implement dimensional accuracy requirements, either the customer needs to provide the dimensions (such as via a drawing or by filling out a form), or Hubs needs to require uploads of parametric models instead of/along with STLs. I do not want to be held responsible for a dimension being wrong because I more or less need to guess as to what the dimension should be. Just my 2 cents on it.

PepCo_Parker · 2016-12-19T19:48:42.000Z

This raises a good point, a customer may ask for money back on a part that is according to them dimensionally inaccurate as it doesn’t fit into other parts, when in reality the issue may be the part design itself. This would be very hard for 3D hubs to gauge when siding with disputes. Almost impossible I would think in fact.

Enza3D · 2016-12-19T19:57:38.000Z

If the issue is with the part design, then it is the fault of the customer if the final print does not work as expected. Tolerancing is a core tenet of mechanical design; if you brought a poorly designed part to any prototyping shop, it’s on you if things don’t work once manufacturing is done. If a print is within the allowable accuracy range (1 mm or 1%) then it’s not the Hubs fault.

Shapeways and any other larger print shop is very much what you see is what you get; as long as they deem a part printable, they will print it but that’s it. If it was printed how it was supposed to, they do not care if the design itself was incorrect.

PepCo_Parker · 2016-12-19T20:11:35.000Z

Yes but let’s say two parts were designed to fit each other, but we’re designed improper via a hole or gap being a mm too small or too large. Diagnosing these issues before printing is nearly impossible, and the customer could argue they don’t fit due to tolerances of the hub being poor. Both of these sides of argument are hard to prove unless you are there in person. With the way 3D hubs handles refunds I could forsee it being an issue

Enza3D · 2016-12-19T20:54:15.000Z

That improper design will show up in the model, and that is why I suggested also requiring either a form/drawings or parametric models. If the feature in question is in the model, there is no argument about who’s at fault and parametric models display all this information with a couple clicks. It is not the job of a Hub to comb through a model and find problems, that is the job of the designer to prevent in the first place. If a feature is designed to be 10.5 mm wide, and the Hub produces it with a 10.1 mm width it’s correct by 3D Hubs definition. I know each Hub handles this differently, but this is the reason my Hub does not guarantee pieces will fit together unless we designed them. It is on the customer to make sure the design is correct, we just make what they give us. If we design it, it is designed with appropriate tolerancing for any mating features and I have no issue guaranteeing it will work because I made it and know what I am doing and the tendencies of my printers.

This is also why I think implementing a guideline like this will cause problems, and I agree with you there.

Odense3DPrint · 2016-12-19T21:54:10.000Z

@Robin3D has an “official” test piece been made?

i want to be on this “wagon” since it will indeed ensure somewhat a minimum level of print quality

Also how will you verify this? i could imagine that the hub print the test bit an use a caliper and photo document that it can be printed right.

Just like that you have to print a marvin when you create a hub

And while at it i think that there should be an option to reverify this… ie when hub rebuilds printer etc

Joerg_4 · 2016-12-20T07:28:17.000Z

Good points, you are absolutely right!

Hopefully one of the 3D Hubs officials will reply to your post, may be @Robin3D

Cheers,

Joerg