What's new
What's new

3D Printed Gears

EmGo, thank you for your insight as it seems you have a lot of gear manufacturing knowledge. Please remember you're in the additive manufacturing sub forum here though. 3D printing gears successfully is not a new thing, though it is for me. Are they equal to cut gears, likely no, but I'm trying to understand how and where those boundaries are. I'm not shipping out Etsy printed crap to customers here. This is experimental and I'd rather understand what works and why, then just assume all 3D prints are garbage and hobbing is the only way. If someone has done this comparison already, post it! It's new ground for me and I'm sure for others as well and I'd rather not keep beating the dead horse if definitive information is out there already.

I'm a very conservative guy, detrimentally at times as the world is always changing, so my mantra is to build on what's already been done as much as I can, rather than being blindly stuck in the past or assuming anything older than 'now' is worthless. I'm not printing gears for some brand new machine with zero field time. This is an established application with a working solution where I'm exploring other solutions to see what works and what doesn't to maintain the machine's current value. Even if 3D printing gears doesn't work, I can't justify investing in a hobbing machine given how little use it would see and therefore I'm still stuck in the same boat of having to rely on other shops that keep upping their prices.

We can and do make gears in-house via 4h axis VMC, and I'm working towards re-tooling to do the same manually. We can make fantastic gears this way, though it takes more spindle time and skill/labor than more dedicated automatic machines. That drives up the cost as well which makes sense for more specialized gears, but can't compete with off-the-shelf gears. This application is middle ground IMO, being less specialized but not off-the-shelf.

Should every shop with a gear application buy a hobbing machine? Who's selling them and why don't more shops do it? Is the solution to simplify outsourcing said jobs to specialized shops? Why are there not more of them and why are their rates going up and what am I supposed to do about it? My experience has been that most shops hate making parts out of phenolic due to the dust and chemicals and would not miss the business if those orders didn't come. Do they do it? Sure and you pay more for their trouble, completely regardless of the part or applications value. Is the solution to go further down the chain and find a 3rd world work force to do the "dirty work"? I'd rather not pursue that avenue on ethical reasons.
 
M.B. another thought I have here.
Is there room to make the gear thicker or wider? Doing that, drive it with a longer key in engagement. This would lesson the chance of it jumping out of the keyway from overstressing the printed gear causing it to bust from hoop stress as it did on the one posted.
Possibly, though not much more than an 1/4" extra at most. I'll have to look at the back shafts and see as those are the most confined side-to-side.
 
Be a lot of work, maybe, to put either a spline or multiple number of keyways on the shaft and duplicate that to the gears. Another thought here.
The key here is to distribute the loading across the full circurference of the shaft and gear and not just in one spot or keyway.
 
Be a lot of work, maybe, to put either a spline or multiple number of keyways on the shaft and duplicate that to the gears. Another thought here.
The key here is to distribute the loading across the full circurference of the shaft and gear and not just in one spot or keyway.
I think the design of a steel hub mating to the printed gear is important as we don't want the next problem to be the gear shearing away from the hub. A square (or woodruff in this case) key is not ideal for a plastic piece like this. We could also change the design to key the plastic gear to the shaft in a better way, but lose the reverse compatibility. If doing the gears in solid Nylon, ABS, or another filament dosn't pan out, I think the steel hub idea would be the next approach.

I like the hex drive design you posted. To make the assembly solid, easy to produce, and low profile, I wonder if we could make the hex tapered with a retaining rim attached with a ring of screws?
 
This is experimental and I'd rather understand what works and why, then just assume all 3D prints are garbage and hobbing is the only way. If someone has done this comparison already, post it! It's new ground for me and I'm sure for others as well and I'd rather not keep beating the dead horse if definitive information is out there already.
You've just demonstrated it yourself. The parts were so bad you had to face them in a lathe and they already blew the keys out where your phenolic ones never did that. You're spending dozens and dozens of hours trying to prove that some gloop squirted out of a tube is as good as reinforced material created in a controlled pressurized environment. It's not. It's never going to be.

If you want to do this, it's fine. It's your life and time. I've chased girls I had absolutely zero chance to catch, too. What the heck, the chase was fun too. But I wouldn't recommend that to people as a way to get married.

When people start in on the "let's make it out of gingerbread !" stuff, it gets to be beyond ridiculous. "I saw this teevee show where native craftsmen with a secret process were making lion spears out of carbonated elephant bones, you could pour that into a mold beat out of sandstone !" Like, c'mon.

For cases like 4GSR with a single gear from a lathe that's been out of production for fifty years, yay. Good idear. But for a production part ? That you sell to people ? You are absolutely correct, these things you are making are Etsy crap. Everything about them is worse. Every single thing.

The subject of gears seems to bring this out in people. If you want a new wall in the basement you go to Lowes, get some 2x4's and a couple sheets of construction grade 5/8 plywood and pop it up, right ? Do you cut down some hemlock bushes in the back yard and weave them into studs, then collect all the cardboard boxes in the neighborhood, put them in a pile in the basement, pee on them for a week, mash them into a pulp then set the house on fire to dry them into a slab ? This is pretty much the equivalent of half these suggestions.

Gears are actually easy and simple. No idea why people feel compelled to go off half-cocked on the subject ... but they do. Every single time. "I don't know anything about gears but .... "

Sigh.
 
True, these may be replacements to sell to others. I disagree about these being ESTY crap. All M.B. is trying to do is come up with a alternative method of manufacturing these that fit in with his business model without having to invest into specialize machines to cut gear teeth.
M.B. is not saying this is the method he is going with, just looking for suggestions to improve what he is trying to do here.
In the long run, this may not work and he may have to either setup equipment to cut gear teeth or send it outside to get done. Believe me, there is nobody in the Houston area that will touch cutting gear teeth without charging an arm and a leg AND your second born for two gears or even two dozen gears!
 
Hi again M.B.Naegle:
With regard to the hub and the transmission of forces, I have found the most robust connection to be a Torx-like connection, and if you evaluate the forces operating to strip the interference, especially when one material of the pair is comparatively soft, you'll see why I believe this.

But this, of course raises all those pesky issues of backward compatibility unless you make a hub that's keyed for the shaft and has a more torque resistant connection to the gear.

Multiple bolts behave very similarly to a Torx connection in terms of their resistance to torque loads, but of course they require a good bit of radial space and they need to be precise enough that all the bolts can participate in carrying the load.
Also if you want to load the bolts most advantageously, you really should have a flange on both sides of the gear so the bolts don't see a cantilevered load.

Will this mitigate the possibility of stripping out the connection...undoubtedly.
Will it be worth the bother...a much harder question to answer.

But if you are eyeing a hex connection, I'd eye a Torx connection instead.
It is both stronger and easier to make IMO. (at least the female socket is)



Cheers

Marcus
www.implant-mechanix.com
www.vancouverwireedm.com
 
Last edited:
I've used a PETG+10%PTFE filament that is very durable, might be a good candidate filament. But at least on the Prusa doesn't print anywhere close to the CAD geometry. I've had to adjust the CAD to get what I want.

Another I've tried is POM which is Acetal. Didn't get it to work very well on the Prusa, Parts I did make were very tough though. I didn't get very far before I ran out of time to experiment and had to make parts out of a different filament.
 
True, these may be replacements to sell to others. I disagree about these being ESTY crap. All M.B. is trying to do is come up with a alternative method of manufacturing these that fit in with his business model without having to invest into specialize machines to cut gear teeth.
M.B. is not saying this is the method he is going with, just looking for suggestions to improve what he is trying to do here.
In the long run, this may not work and he may have to either setup equipment to cut gear teeth or send it outside to get done. Believe me, there is nobody in the Houston area that will touch cutting gear teeth without charging an arm and a leg AND your second born for two gears or even two dozen gears!
I have NO doubt M.B. is and will do due diligence before selling any printed part to his customers. I am not sure if he will be able to make a part that is durable enough for his and their needs, but I think there is a decent chance. As I mentioned early on I think combining a steel hub bolted or riveted (I was envisioning a bolt circle of six or so # 8 or 10 bolts or similar sized rivets if there is room) to a sandwich the plastic teeth to a hub that would be retro-compatible with the originals. Having used both PETG and Repro One from Freeman, my impression is that the Repro One would make a better gear. Making a mold using 3-D printed gears and embedding them in Silicone (no draft needed) or Repro One (draft needed and die-like smoothness required for release) or one of their many other options coupled with what they recommend for this application (probably Repro One is not their best for this application) would certainly be an option I would explore. There is a good chance they have experience providing a pourable 2-part plastic for this application. That is because I am pretty sure a poured plastic gear made from a products that Freeman has been tweaking for industry for 3 or 4 decades is likely to be more durable than a 3-D printed gear. And producing poured gears would be many times faster than printing them.

I also agree that some dimensional tweaking will be required and I am sure M.B. has a fair bit of experience already along this line.

I do not mean in any way to dis 3-D printing. I really enjoy it and use it extensively for pattern making for my small foundry. My Prusa MK3 is busy as I type making a sprue/runner/riser assembly for my 12" straight edge that I am presently designing and am soon to test cast. Here is a pic of the Onshape drawing. I'll take some measurements off of the print to compare tot he drawing.

My new Prusa XL twin-head arrives Tuesday having left Prague yesterday and is now on a truck out of LA heading for me north of Seattle.

1713638785912.png

The overall length is roughly 5.75 inches. The notches in the lower fillets are intentional and required.

Denis
 
On the piece pictured in the post above I did just a few measurements to see how closely the physical print matches the drawing:
The major diameter of the Ellipse Actual 1.456 Drawn 1.466
minor diameter Actual .740 Drawn .730
runner height Actual .505 Drawn .500
riser height Actual 1.998 Drawn 1.999
runner width Actual .801 Drawn .809
This was in Jesse brand PETG with 15% infill, gyroid infill pattern, 2 perimeter shells and 5 base layers and 4 top and .2 layer height and a .4 nozzle using Prusa Slicer. PLA prints should be closer to drawn size.

Just a data point. Not claiming these results are anything special but really just a common printer, using a common slicer, and common filament type.

If I were printing gears to serve as patterns for silicone molds, I would use PLA, .25 nozzle, maybe more infill. I do not know if a grid infill would likely produce better dimensional accuracy than gyroid or yet some other pattern. And I would expect to have to do some post-print finishing like painting and sanding to get a piano finish like I do for casting patterns. I have really not done any "precision" printing as such. Or even better, I would print the gears in my resin printer which would produce a significantly nicer finish and I am quite sure better dimensional accuracy.

Denis
Added a pic of the print:IMG_0817.jpeg
 
Last edited:
You could make replacements for those phenolic gears by molding using a single point broaching the teeth in a plate on a mill, bolting it down to another plate, use epoxy/carbon fibre fill, then a cover plate. Cycle time would be faster than printing and the accuracy is determined by the mold.
I printed gears to test a starter arrangement for a BSA A65. Worked fine for proof of concept.
 
Hi All:
Have any of you ever done a cost analysis of a 3D printed gear compared to making it some other way?
I have not, but my instincts are not hopeful.

Just the cost of the filament, plus the cost of the electricity, plus the cost of the scrap you'll make, leads me to think it's not gonna pencil out, and even if it gets close, you have a weak (comparatively) crappy gear.
I haven't even considered the cost of the gadget yet, or the time it takes to make a gear

I wonder how much making the same sized gear running a gear shaper or a hobber costs by comparison.
I'm in EmGo's camp here even though I am not even remotely as experienced with gear making as he is.

Something about efficiency instead of blindly embracing new tech that :
1) takes longer
2) makes crappier gear geometry
3) makes a weaker gear.

I believe that too many young engineers and others these days, have jumped pretty uncritically on the 3D printing bandwagon, because the upfront labour is small, and then you can just let the toy noodle along in the background.
It's easy to forget that runtime and energy cost...having it "noodle along" has a real cost too.

Cheers

Marcus
www.implant-mechanix.com
www.vancouverwireedm.com
 
Hi All:
Have any of you ever done a cost analysis of a 3D printed gear compared to making it some other way?
I have not, but my instincts are not hopeful.
Just the cost of the filament, plus the cost of the electricity, plus the cost of the scrap you'll make, leads me to think it's not gonna pencil out, and even if it gets close, you have a weak (comparatively) crappy gear.
I haven't even considered the cost of the gadget yet, or the time it takes to make a gear



Cheers

Marcus
www.implant-mechanix.com
www.vancouverwireedm.com
We can debate forever whether the gear will be satisfactory for the OP. He is going to give it a try and see. I guess I will have to wait and see what he finds out.

But I do not have to wait to address the cost of filament or power. My printer consumes about 120 watts printing PETG and 80 watts printing PLA. The power difference is related to the higher melting point and the higher build plate temp needed for PETG vs PLA. And to print a 5" gear like the one shown by the OP would require about 4 to 5 hours. So, we are talking about less than a .5 Kwh. Here that would cost about 5 cents.

Filament cost based on having just sliced a comparable and the slicer tells exactly the amount of filament that would be used would be 1.323 USD.

Operator hourly wage---well it is plug and play (somewhere else).

So cost comparison to a metal gear---not a factor.

Denis
 
Last edited:
" So, we are talking about less than a .5 Kwh. Here that would cost about 5 cents." Not in California- thats about $.25.

The point is, printing is great and you could print utensils for every meal, but the picnic pack at the store sells forks and knives for less than a penny each.... so you can print them, lets get real here. Use the tech when it makes sense to do so........
 
" So, we are talking about less than a .5 Kwh. Here that would cost about 5 cents." Not in California- thats about $.25.

The point is, printing is great and you could print utensils for every meal, but the picnic pack at the store sells forks and knives for less than a penny each.... so you can print them, lets get real here. Use the tech when it makes sense to do so........
I would say the point of the post you are quoting was to answer the objection that filament and power costs for printing a part somehow argued against printing. I was simply pointing out that that was far from the case. Material cost, power to run a lathe, hobber and key seater and material cost as well as labor cost to make a steel gear might easily be 50 to 100 times greater than printing. So economics do not argue against, but in reality argue strongly for printed gears.

Unresolved is whether said printed gear will perform adequately. My understanding is the OP wants to find out. So, let's do what we can to help him. He stands to gain some knowledge from trying out the idea and we will learn something in the process as well.

Denis
 
What happens to the original gears when you do these destructive tests on them? Do they fail or do they just shrug them off?

Why not model in a flexure element between the hub and the gear teeth to help absorb the shock of the sudden shocks and the foreign objects through the mesh? Just because you are trying to replace a traditional element like a keyed gear, doesn't mean you have to use the traditional design.

That being said I'm using the PA6-GF (Bambu's) in a item right now and am impressed with it. I like the PAHT-CF better, but not the additional 50% price increase.
 
dgfoster said:
Unresolved is whether said printed gear will perform adequately. My understanding is the OP wants to find out. So, let's do what we can to help him. He stands to gain some knowledge from trying out the idea and we will learn something in the process as well.
Actually, no. It's been more than a hundred years since people would build a bridge by eye then walk fifty horses and wagons on it in the hopes it would not fall down. Outside of Florida, that is. And Missouri, where people are not 'creepy' but they do seem to be a bit behind the times, engineering-wise.

Since I know Bellingham is modern - you have a Trader Joe's and everything - I'll assume you have access to the web and about 500 sites that have these things called "strength of materials" charts. They show all kinds of properties determined by controlled testing. A query on any search engine will give you a large number of returns for things like "tensile strength phenolic" or "ultimate tensile PETG" and so on. (I just chose that material because you used it as an example, not that I have any recomendation). You will find a lot of papers exploring these subjects. They go into great detail and show things far beyond simple strength figures. I find them useful.

Anyway, for our readers' convenience, here are a couple of tables ... (No I did not cherry-pick, these are average tables chosen from somewhere in the middle on both materials. No need to cheat, the middle numbers show the facts without even trying to bend reality).

props_phenolic.jpg

props_petg.jpg

and a quick conversion from the hated metric to sensible units :

conversion.jpg

You will notice that even in the weaker direction, average ol' off-the-shelf phenolic is a little more than twice as strong as heated-plate PETG. That's not even going into notch sensitivity and other factors (one paper somewhere mentioned that PETG did not like sliding, which is sort of a problem with gears). So, using modern methods, we can say with a good degree of confidence that a printed gear will be roughly half as strong as one made of phenolic.

Moving right along to accuracy ... here are your findings : (I added the error figures, for convenience - also remember the example was about a 1" part. Whether the difference is larger on larger parts we don't know)

(( I bet the forum software will butcher this table, sorry))

Code:
major diameter of the Ellipse Actual 1.456 Drawn 1.466  ->  .01"
minor diameter Actual .740 Drawn .730                  ->  .01"
runner height Actual .505 Drawn .500                   ->  .005"
riser height Actual 1.998 Drawn 1.999                  ->  .001"
runner width Actual .801 Drawn .809                    ->  .008"

Just a data point. Not claiming these results are anything 
special but really just a common printer, using a common 
slicer, and common filament type.

Here is a chart of AGMA tolerances. For reference for people not acquainted with Old AGMA tolerance classes, back before the airhead Denis Gimpert turned AGMA into the world's laughingstock, classes went from 1 to 15, with 1 being something a caveman could bang out of an old tree trunk with a stone axe, and 15 being a part for the SR-71. For a shop guy, just about any gear cutting machine is capable of AGMA 9. Capabilities by process

Gear-Tolerance-5b7f22cca57f3.jpg

An AGMA 9 pair of gears feels nice rolling together. That'd be like a nice automotive transmission gear rolling with its mate. AGMA 8 would feel maybe a little grinchy in spots, like you'd expect from a Nova or a Pacer. Not trash like a Yugo but not a Lincoln, either. AGMA 12 thru 14 feels like running your hands over a teenage C cup, exceptionally nice and even makes your heart pound a little.

For these parts, let's choose AGMA 8 ... not great but not trash. Easily achieved by any conventional process --

AGMA+Classifaction.jpg

Using composite measurements - that's functional testing, run the part against a master gear, measure how far off it is in use - Lessee, max tooth to tooth allowed .001", total over the whole part, average of .002" (TTE is "tooth to tooth error" and TCE is "total composite error").

Roughly speaking using the given figures for Denis' part less than 1", measured sizes are off by four to five times what is allowed in an AGMA 8 gear. Phenolic cuts nicely, it's is quite easy to achieve 8 or 9 or even better. In practice, 8 is if you throw the part onto the mandrel from across the room, you set the machine up in ten minutes and didn't check anything. You get 9 if the bores are snug and the faces parallel and you indicate the mandrel and cutter and so on. 9 could be considered an average nice gear. The chart above is for one grade lower.

Does this fulfill Mr Naegle's need ? I dunno, his product, his decision. But what we can say, easily, without going through days and days of guessing, is that printed parts will be about half as strong and one-fourth to one-fifth as accurate as garden variety, common as dirt, cut teeth.


A few references so you can verify I didn't make this shit up ...







(Not trying to swamp anyone with data but it's very easy to get numbers, took maybe ten minutes. Printer material is a little harder because the results are swamped with home hobbyist stuff "PETG is stronger than PLA" ... well that's a big help. But there's also a lot of research on printed plastic behavior, just have to look a little harder.

One thing that's apparent is, printed plastic is a lot more variable. Things like platen heat, nozzle size, layer thickness, exact material, supplier, all seem to make a difference. Phenolic is easy - it is what it is. Reliability, I wonder if that has any value ? hmmm. )
 
Last edited:
After reading the above post I know one thing for certain. The old word limit per post that PM used to have has either been eliminated or greatly increased! So, I will be brief.

I agree that printed gears are likely to show variability between machines and between filament manufacturers even if one uses the same type of filament. However, once a user chooses a machine and a chooses a reputable filament vendor like Prusa or Jesse, vendors I use, a huge portion of the variability is eliminated.

It is also true that accuracy of printed parts will be expected to be less than that of metal parts. However precision of a printed will be considerably better than initial accuracy. I did not post above the fact that I repeated, with some minor changes, the print on the same machine and using same filament above and the results showed variability in dimensions of less than .1%. So, as mentioned in one of my first posts in this thread, after a test print or two, a maker will tighten his variability and come closer to nominal dimensions by taking into account observed shrinkage factors which should be consistent from then on.

There can be more factors than you cited that could adversely affect printed part strength like layer-to-layer adhesion problems. If trying to maximize strength, printer parameters can be tweaked to improve those properties just like accuracy and precision can be adjusted if needed.

Not addressed is the question as to what how pourable two-part plastics might perform. My best guess based on experience with printed parts and with plastic tooling from Freeman Supply is that poured plastic parts made in molds of two-part silicone or one of their several flexible high-quality polyurethane molding materials (which I have also used) would provide parts that are considerably tighter in dimensional consistency and more durable than printed parts. In addition, a batch of fifty could be made in a single shift by a person who would need to attend to the process only for mixing and pouring and then setting aside the mold to cure for a couple hours (maybe use a warmer to further accelerate turnover) while doing other jobs.

And, finally, as modern as Bellingham is having full internet access and a Trader Joe's, I could not find anything as to whether the proposed gears would work well enough for M.B. in the machine in question. That remains unresolved. ;-)

Denis
 
Nope. Doesn't work. this is the PLA gear that I printed last November for the lathe in my sons welding shop. He has used it quite a bit plus has single pointed many threads since then. But if he does crash it and strip out the gear, he has a spare that I printed for another $2 in material cost. 1713921605677.png
 
Nope. Doesn't work. this is the PLA gear that I printed last November for the lathe in my sons welding shop. He has used it quite a bit plus has single pointed many threads since then. But if he does crash it and strip out the gear, he has a spare that I printed for another $2 in material cost. View attachment 437194
Looks like it is doing the job nicely. Did you use any special parameters for this print like increasing infill and perimeters? I guess it has a keyway?

Denis
 








 
Back
Top