What's new
What's new

Dimensional Quality in prints

Fal Grunt

Titanium
Joined
Aug 5, 2010
Location
Medina OH
I have been for some time entertaining the idea of getting a 3D printer to print a sample part for testing prior to machining. This typically means that the part needs to have an accurate dimensional quality along with good surface quality. Some for of strength is required, I have had printed parts literally fall apart in my hands, but that several years ago.

I have in the past had a friend print me some parts just for experimentation, but 90% of the time they were useless. Dimensional accuracy was very limited, surface finish was very poor.

I understand to a certain extent this can be improved depending on the settings, but I am seeing some printers claim numbers 25 micron as the smallest interval of size. Is this a readily achievable expectation? Or will it be more timely to set the part up and machine it to check geometry?

Part of the reason I am interested is development scrap. On some occasions I may machine anywhere from 1-10 test pieces for development, altering geometry to refine fit or function. While that scrap is not expensive, the time making it is. However if it takes me 10 hrs to print a part to check dimensions, then it isn't necessarily beneficial, except in circumstances when I am in no hurry to develop a part.

I have considered buying a "cheap" printer to play and learn strengths and weaknesses, but at the same time, I am hesitant to spend money on a cheap printer, rather than put the money towards a good printer.
 
I have a Projet 3600 Max and I can confirm that on the finest resolution (16 micron layers) it will easily deliver +/-0.005" per inch of accuracy and usually much better in the z-axis of the print. You're not going to get really accurate good looking parts off an FDM machine. Job it out to someone with a commercial polyjet or SLA printer or be prepared to spend at least $50K for new.
 
i would worry about plastic after some times has gone by its dimensions changed. sometimes from drying out of water or solvents or other materials in its makeup.
.
many a plastic part i have seen change dimensionally just sitting on a shelf for a few months. even metal has been known to warp after some times has gone by or upon different temperature exposures. some metal can have micro cracks or defects hard to see. usually thorough testing is needed same with equipment designed to pickup heavy loads is tested at 200 to 600% of rated load to see if it will fail under load. or tested on repeated smaller shock loads of thousands to millions of time to see if it is durable to hold up.
.
intergranular stess corrosion is where some stainless steel welded parts are exposed to water and chemicals that go into microcracks and attack the metal internally. for example welded handrail tested at 250 lbs holding but 5 years later the same test the weld just snaps suddenly with no warning.
 
Sla is usually the most accurate I've seen, just don't drop it
I have looked at Form Labs and had them send a print for me to look at. I am impressed. I haven't dropped it yet, but I haven't been able to break it with my bare hands yet either (and I have tried).

I have a Projet 3600 Max and I can confirm that on the finest resolution (16 micron layers) it will easily deliver +/-0.005" per inch of accuracy and usually much better in the z-axis of the print. You're not going to get really accurate good looking parts off an FDM machine. Job it out to someone with a commercial polyjet or SLA printer or be prepared to spend at least $50K for new.

So, just make sure I am understanding you correctly, with a print layer of .00063" your getting an average accuracy of .010"?

The one printer I was looking at advertised a 25 micron resolution, and I thought at .00098" +/- .002" would be a reasonable expectation?

For $50k, I will get another Brother to prototype on :willy_nilly:
 
i would worry about plastic after some times has gone by its dimensions changed. sometimes from drying out of water or solvents or other materials in its makeup.
.
many a plastic part i have seen change dimensionally just sitting on a shelf for a few months. even metal has been known to warp after some times has gone by or upon different temperature exposures. some metal can have micro cracks or defects hard to see. usually thorough testing is needed same with equipment designed to pickup heavy loads is tested at 200 to 600% of rated load to see if it will fail under load. or tested on repeated smaller shock loads of thousands to millions of time to see if it is durable to hold up.
.
intergranular stess corrosion is where some stainless steel welded parts are exposed to water and chemicals that go into microcracks and attack the metal internally. for example welded handrail tested at 250 lbs holding but 5 years later the same test the weld just snaps suddenly with no warning.

I too worry about the intergranular stress corrosion of my plastic 3D printed parts. Hopefully they never get welded into a hand rail.
 
So, just make sure I am understanding you correctly, with a print layer of .00063" your getting an average accuracy of .010"?

The one printer I was looking at advertised a 25 micron resolution, and I thought at .00098" +/- .002" would be a reasonable expectation?

For $50k, I will get another Brother to prototype on :willy_nilly:

Everything in the land of 3D printing is geometry dependent. On small prismatic parts less than 1" cube I usually see +/- 0.001" on the x-y and +/-0.0004 in the z. As soon as you add off axis features and wildly varying cross sections that all goes out the window and you are back at factory spec.
 
Form Labs seems to have a nice selection of resins with different properties you should look at. We get a few parts here, and if you match the resin to the job, they function well.
 
Form Labs got back to me with details on the print that I received:

Rigid is a glass filled resin -- this gives it a number of improved properties including low creep (low deformation over time) and high stiffness.
The material is great for printing thin walls and features! Look for the Formlabs logo on the part for an example.
Volume: 20.54 ml
Layer Height: 100µ
Material cost standard resin: $4.12 USD
Printing time 1 piece: 3 hr 9 min
We print 8 Brackets on 1 build platform and this takes us about 10 hours 2 minutes
1L standard resin: 48.6 Brackets
 
Before you buy a Form Labs printer make them send you a "strip and ship" part or better yet with all of the supports still attached. There's more labor in making parts off those machines presentable than most people expect.
 
...
For $50k, I will get another Brother to prototype on :willy_nilly:

Just for an idea so we can all know the money side:
TKassoc,
What was the installed and running on the floor cost of your Projet?
Consumables per pound?

Fal Grunt,
I see 3 hr 9 min and $4.12 material for a quote.
I don't see the one piece price, what is this in dollars?

As far as resolution goes when moving 2 or more axis I have 1/20 of a micron resolution grinders. They won't make forms anywhere near this number.
Since most fast moving systems will lead/lag by +/-10 counts 20 times the base resolution is a good bet.
Under "nice" conditions you can get 2-4 counts but start moving both or turning corners and.....

If the parts can be cut on "another Brother" and it can run in the background not using up your time that might not be a bad option.
Bob
 
Fal Grunt,
I see 3 hr 9 min and $4.12 material for a quote.
I don't see the one piece price, what is this in dollars?

If the parts can be cut on "another Brother" and it can run in the background not using up your time that might not be a bad option.
Bob

The whole idea originally was to save run time on the Brother. I am not so busy as to not have that time, but I often get frustrated doing development work. Looking for an easier way. Machine the first side, cut soft jaws, cut the second side, machine soft jaws, cut the third side ...... make changes to the part that hopefully mean I do not have to recut jaws...

They did not provide a "cost" as delivered, it was a free sample you can request off their website. I may send them a part file to quote just for giggles. $4.12 is a LOT of steel... probably 4 pieces worth depending on type.

In particular, the Form Labs machine, costs around $5k, plus freight, etc etc.

To me I think this boils down to, printing has not reached the quality necessary, the cost is still high, the trade off really only being setup times. With a printer I can print a part in one go, where machining may require multiple setups. Cost to operate (labor) the printer is cheaper, as that 3 hours is unattended run time, where as running the part through on the machine is a labor expense.

If business grows over the next few years (might not given the "buzz") I think a more viable option may be another spindle for development.

Good comments as well Bob concerning the lead/lag, gives me a good perspective.
 
We are forced to use plastic printed parts in our facility for test pieces. The biggest driver was engineering who just had to have 6-32 threads, and stupid radius tolerances for wire routing. In other words all the dumb things engineers do on a drawing because they have no concept of machining. The costs of the printed part far outweigh starting with a good drawing, machining out of Delrin or aluminum, and then doing the final work (or just build a good designed part to start with). Oh yeah, the printed parts of any size larger than a bar of soap warp, and hold a tolerance of 0.01", whoop tee doo. We have lost the manufacturing game and may never get it back.
 
The whole idea originally was to save run time on the Brother. I am not so busy as to not have that time, but I often get frustrated doing development work. Looking for an easier way. Machine the first side, cut soft jaws, cut the second side, machine soft jaws, cut the third side ...... make changes to the part that hopefully mean I do not have to recut jaws...

I know some people are 3D printing the soft jaws. Might be worth looking at.
 
Call up Forecast 3D and ask if they'll do a small sample print for you. I've found them easy to work with, and if you explain the function of your part, and which tolerances are the most critical to you, they'll work with you to ensure you nail your tolerances. They print a wide range of materials, too.
 
(1) 3D printing in either plastic or metal will not have the geometric accuracy of a machined part. My machine will print a 6-32 hole but it will be utterly meaningless for verification purposes.

(2) The demo parts you get from the printer manufacturer are about like the program a demo jock runs at a trade show—optimized to the max. Unless you make them print your part file.

(3) 3D printing is unbeatable for producing difficult part contours for visual evaluation of appearance, and likewise for patterns and coreboxes for sand casting, or at least the masters. Minor distortion in sand cast parts is a fact of life and the methods for dealing with it in the machine shop are over 100 years old and well established, so the errors at the print stage are inconsequential.

If I have to see how a machined part is going to look I'll just look at it onscreen along with a model of its mating parts.
 
Call up Forecast 3D and ask if they'll do a small sample print for you. I've found them easy to work with, and if you explain the function of your part, and which tolerances are the most critical to you, they'll work with you to ensure you nail your tolerances. They print a wide range of materials, too.

I spent some time on their website, thanks for the reference. May definitely be an option for larger, higher complexity parts, for which machining prototypes may be cost prohibitive.

(1) 3D printing in either plastic or metal will not have the geometric accuracy of a machined part. My machine will print a 6-32 hole but it will be utterly meaningless for verification purposes.

(2) The demo parts you get from the printer manufacturer are about like the program a demo jock runs at a trade show—optimized to the max. Unless you make them print your part file.

(3) 3D printing is unbeatable for producing difficult part contours for visual evaluation of appearance, and likewise for patterns and coreboxes for sand casting, or at least the masters. Minor distortion in sand cast parts is a fact of life and the methods for dealing with it in the machine shop are over 100 years old and well established, so the errors at the print stage are inconsequential.

If I have to see how a machined part is going to look I'll just look at it onscreen along with a model of its mating parts.

Unfortunately 99% of parts I machine don’t have a solid model. I make a model from prints typically. This also means I don’t have a mating part to verify fit/function. The work I do on the industrial side, Tool & Die etc, is much easier because I have prints to go off of.

The majority of other items I make, especially products I sell, fit into something that I have no prints, models, and in some cases do not have an example. Some are brand new, some are over 100 years old. So usually it takes some “tweaking” to find a middle ground that works for most.
 
The majority of other items I make, especially products I sell, fit into something that I have no prints, models, and in some cases do not have an example. Some are brand new, some are over 100 years old. So usually it takes some “tweaking” to find a middle ground that works for most.

Well, that is horse of a different color. Parts over a hundred years old may have been made under less than controlled conditions, in which case it may be impossible to offer items to "fit" them.

For example, we broach parts to fit male automotive steering splines, most of which were originally intended for pinch-clamped assembly and consequently had a wider production tolerance than comparable industrial splines. Since almost all of them are—or were—proprietary designs it is usually impossible to obtain original specs, and if you could, you shouldn't trust them. What we do is ask the client to provide three examples of the male spline taken from different production years of that car, and then we measure them to establish an expected variation, making the broach(es)big enough on the pitch diameter to assemble with the largest example. Others in the same business will simply make a wild guess and offer grotesquely oversized—and thereby possibly dangerous—parts. That's what you have to go through to offer good "ancient" retrofits.

Depending on the type of retrofit you are dealing with, and how critical the mechanical fit, 3D printing probably isn't the answer. You'd need a sample of the actual physical object to model accurately, at which point you might as well just go ahead and machine your part to fit...along with a stock of extras, maybe semifinished for future use.
 
You'll have a hard time coming up with high accuracy parts from fused filament fabrication (FFF) or fused deposition modeling (FDM). As most have pointed out SLA/DLP/UV Masking or similar resin-based printing methods would garnish the best results. The downside to most resin-based printing is the underlying polymer is based on a form of acrylic, thus more fragile (I've broken my fair share). In terms of my own work, 3d printing was never the end product, but a way to get to it. I know what my printer can put out in terms of dimensional accuracy so I, when really needed of course, create a model with compensation built in. However, most of the usefulness of the 3D print is in general fit and feel.

And honestly, you don't need to spend a huge (relatively speaking) amount on a printer. I play around with a resin-based one that only costed around $500 which has produced more accurate prints than the Form Labs printer at my work.
 








 
Back
Top