Check out my favorite 3D printers, the K1 Max: amzn.to/3txQUC6 or if you are on a budget, the Ender-3 V3 SE: amzn.to/3FkCPus I hope you enjoyed this video and learned something new! For more content like this, please consider supporting me on Patreon: www.patreon.com/howtomechatronics P.S. If you are interested in the measuring tools: Force meter: amzn.to/3FEUAnS ; Digital dial indicator: amzn.to/3PDvSco
I just finished assembling the drive train of my robot. It was a pain to print 36 M2-30teeth herringbone gears @100% fill with 10 walls (perimeters are faster than fill), just for the drive-train! I advise you to get some lithium grease and apply it to your gears and shafts. You will not believe the difference it makes. Before, it ran loud and rough. A few gears even seized and had to be replaced. After application of lithium grease, it ran as smooth as a baby's butt and quiet like a whisper. I was truly shocked at the difference little grease can make. Lithium grease is the only grease I know that's safe to use on all plastics. Another trick before greasing, is to run the dry at low speed and pour powered sugar or fine salt all over the gears as they run, to lap them. As they mesh, the powdered sugar will act as an abrasive. Like sand but finer and can be dissolved in water after. Then, wash the gears to get rid of all the sugar that got embedded, then grease and assemble after drying.
Hey, thanks for the input! That's such a great idea with the fine salt. Yeah, I do believe grease will make a huge difference. As I mentioned in the video, I wanted to make a durability test, running all different types of filament for a long period of time, but I couldn't conduct the test at this moment in my studio. And the plan was also to test greasing them will affect their durability. I will leave that for some future tests. Cheers!
I'm glad for this video but measuring the efficiency of gears without loading them does not make any sense. You could use another DC motor as a generator and actually do the math or you could measure rotational acceleration of a heavy disc or something Helical gears should be more efficient under loading btw
instead of adjusting for elephant foot with bed levelling, first layer offset, or a raft, i will usually apply a chamfer to every edge that will be on the print bed. if the layer expands, it still won't expand more than the negative offset from the chamfer! Add a little brim, and you have accurate geometry on the part where you need it, and large surface area on the print bed.
Bambu slicer has an Elelphants Foot Compensation setting. Surely Creality Slicer or whatever slicer you use has one. Some slicers call it "initial layer horizontal expansion" or something relatively similar. It basically adjusts the outside dimensions for the first layer allowing it still smoosh as it should but still maintain the correct dimensions. Increasing the z offset to compensate could sacrifice bed adhesion and first layer quality.
Great analysis! For more strength print more (4-6) shells/perimeters and top and bottom layers (4-6), consider internal planar structures for additional interior wall strength. Printing hotter and slower improves layer adhesion and strength. Fixturing is important for the strength tests. As you pointed out, supporting the axle on both sides to prevent deflection and ensuring the lever force is directed directed through the center of the gear rather than offset will yield maximum strength.
@@bachaddict Yes! It can be, but as with all things, it's properties should be probed to determine appropriateness. Timing belts showcase this property effectively. The surface can still dimple (eg. grain of sand between teeth) without chipping, and the teeth can slightly bend (reducing wear), and TPU is famously wear resistant. And the softness of the material reportedly also has a side-effect of lower noise during gear operation. TPU is THE magical material hidden in plain sight -- it's almost as cheap as other filament, with incredible properties and a wonderful material for compliant hinges, functional, and decorative applications. An object printed in TPU is virtually unbreakable and will handle stresses far better than many other plastics. A TPU "vase" can be completely crumpled, and return to its original form without any sign of damage. As a composite material, it is very interesting indeed. And if that wasn't all, there are solvents that allow for the joining of TPU parts with the join being as strong (or stronger) than if it were 3D printed.
@@SeanLumly I do not agree that TPU price is comparable with other filament. I bought PLA for 8-13 euro and PETG for 16 euro, but I never saw TPU cheaper than 24 euro. That's significantly more expensive.
@@yansakovich That's fair. But certainly not exorbitantly so. It's also worth noting that while you can fine extreme deals on PLA/PETG, prices via popular channels (eg. Amazon) tend to have similar prices (at least where I'm from). This is what I was speaking to. Anyway, it's beside the point. TPU is a cheap and flexible material that warrants investigation.
A point about your test setup to determine force to destroy gears. Your pull meter has to be at right angle / perpendicular to lever arm and pull force must be applied in the same direction. Basic Physics.
I tested a gearbox made up of Helical gears for the motor shaft and Double Helical gears, all printed in PLA. The gears moved a Step Feeder system, with low torque, for about 100 hours. I did not use any lubricant, and after this time the gears showed very little wear.
Hi there! I enjoyed the video a lot - but the backlash testing part got very messed up. There was play everywhere - well, you pointed that out yourself and that’s good. But then… the printed parts could have been made with tighter tolerances. I actually don’t know how to proceed into creating a mechanically correct testing rig to compare them. If the mechanical wisdom affirm herringbone gears have less backlash and the 3D printed part is having more backlash, then it’s clear we did something wrong, right?
Hi there and thanks for the comment! I admit that I kind of failed to provide better information on that point in this video. I guess, I got lost myself with that much printing and testing. My initial plan was to also include a comparison with different tolerances, but I didn't do it. Though, it's probably hard to make a conclusion for that matter, as every single 3D printer will have different printing tolerance, so I don't think any specific tips or numbers can be told. What I have noticed from previous tastings, which is also obvious, the tighter the tolerance, you lower the backlash but then also lower the efficiency as greater friction is introduced. Check @piconano comment out on how he used to "fine sand" the gears before greasing them. It such a cool idea and might help with the overall backlash/ efficiency and even durability performance of the gears. Cheers!
If you look at the close ups of his gears, the perimeters aren't even printed nicely. You see a lot of random "bumps" sticking out, for example in the printing sequence at 6:40. These bumps basically void any tolerance tests, so that part of the video seems pretty much meaningless to me. On a better printer and with better slicer settings you should actually get a pretty clean, smooth and consistent perimeter that would help to get better data out of these tests.
Kool!!! Forgot to talk about seams... they can ruin a big cylinder planetary print. Best is to use orca slicer and scarf seams ... manually placed to be not all syncronize seam contact.
Great video, I would like to see the process of printer calibration end to end, that would be insanely helpful, as of now I’m just offsetting everything by about 0.1mm but more accuracy would be nice. You can ignore all the comments, from „engineers” that offer their „opinion” as only contribution. They are as useful as people who only write comments on git repos without touching the code. Takes a very special person to watch 30 minutes of free education that someone put together with hours of hard work only to do an egotistical pull-up meant to show how much better they are. I swear it is like people coming to IT conferences to ask a question, with sole purpose of showing how much more they know on the topic (you rarely see them on the actual stage).
Thanks for all the work you put into this. I must point out that you have made a mistake on sounds levels. Nearly everybody (99.999999%) doesn’t know that sound is measured with a logarithmic scale because of the great variation of levels which makes a change of 3db’s twice as loud. So 50dm and 53db is twice as loud.
That is correct. On top of that, usually the sound level is recorded at either 10cm or 100cm away, so you get some comparable results and you can actually calculate the decibels you would experience at a given distance. He should probably have used either an average or peak hold setting on the mesurement device as well, as the meter in the video is jumping all over the place, so it's hard to pick just one value. When we do sound level measurements we actually move the microphone to different positions in a circle around the source too, as sound can be very directional and you would usually want to pick that characteristic as well to find the worst noise direction. Doing good, reliable and repeatable noise measurements is actually pretty tricky. The bigger labs have some fancy non echoic rooms and even fully suspended solid bunkers to cancel out most of the external noise and any reflections. You actually feel really strange in these rooms as we are just not used to the silence and audio experience you get in there.
From my experience with 3d printed gears it's always better to print with (for example: wall count 100 and Infill of 0% ) instead of (wall count 10 and infill of 100%). Depends on the gear size, the wall count must be increase to fill the whole gear even when the infill is 0%. The goal is to connect all walls (outer ,inner and hole walls) together.
I guess the slicer engine might have a significant effect on the results as well, using some variable line width slicers, like arachne, will probably help to have less voids in the print, especially at the root of the teeth where there is a missmatch in the lines. Would be interesting to see if increasing the line thickness and/or increasing the nozzle size has any significant effect as well. I would guess thicker lines and slow print speeds would actually be better to increase the strength?
Good and interesting video. To avoid the elefant foot I usually add a 0.25 chamfer to my designs. That works great. And: Please make noise measurement from a distance of 1 meter. This allows real world comparisons for others. Holding the microphone directly to the source is just wrong…
Yep, adding chamfer to the whole gear profile will do that job. And yeah, someone else also mentioned for measuring the noise I should hold the microphone at a distance from the source. Thanks for the input, cheers!
I get rid of elephants foot by dialing in a little bit of negative "Initial Horizontal Expansion" in Cura. Not sure if other slicers have this function.
Hello, when I print out the gear, the lower part of the gear is always wider than the upper part. Big or small, it doesn't matter. It's like a conical shape. I wonder where I should check. Do you have any ideas?
Thanks again! Well in this situation the problem might be my particular PETG filament. Just like shown in the video, a different brand PLA provides different characteristic, same goes for the PETG. Maybe I had a PETG filament with lower quality. :S I should have done more tests, but it would be like never ending video to produce and watch if I put everything to the test. :) I know I'm missing some things and could have been better explained and compared in this video, but it's hard to make a perfect video. :) Cheers!
@@HowToMechatronics That might be an idea for new videos. Testing all the different brands of PETG in white. Or For PLA+ or Whatever. I've never seen a video like that. CNC Kitchen and alike do the same tests as you did. Unless the viewer buys the same brand or even the same batch, then there's no absolute rating. From my experience, clear (pure) PETG is a bit harder than color white of the same brand.
I think it's pretty good. Remember that a car with mass ~1t have engine that provide ~150 Nm of torgue. You could probably create ultra leightweight cars for cities with small electric motors with 20 Nm of torgue and these plastic gears.
observing the failure modes of the printed gears was interesting. it seemed like the failure was primarily happening either Wall-Wall or Wall-Infill. I wonder if minor filament under extrusion can lead to weaker wall-wall bonding. I don't like the use of the horizontal expansion setting personally. I think it is better to explicitly model allowance into the printed part before slicing.
@07:30 it's impossible for a perfect print . The key is post print processing . File the rough edges down . Àlteratively , you can increase the bottom and top later thickness and sand down the base and top by a little amount to take away any issues .
Great video! Just one thing that may caused the teeth to shear is that you have used many loops. The gears sheared at the transition of the loop and shell/infill area. Maybe a single perimeter (wall line count) and 90-100% infill would result in stronger gears.
Bearings are meant to be press fitted and are installed with presses commercially because a hand is not strong enough for the appropriate preload. So always do small hile😊
I use a resin printer to print gears for MG995 without problems, but I would advise against FDM. The resolution of the filament printer is too low to produce functional parts with such fine details.
Hmmm, that's really questionable. I guess it would work, but not sure how good. Why don't you test that out and let us know how it turned out. Cheers! :)
Sorry ,your contrivance is not stable,The result is not true.Mill is so thin and not rigid.And there is single force holder aganist your appling force..You must change test your contrivance.
It would be interesting how much more force is required to break the gears if the center shafts were kept from moving relative to each other. Both gears should have a bearing and shaft, the shaft long enough to attach a connecting plate to each to stabilize the viewing side. This way you won't get the "belt bolt" issue over time, and would make each tests far more accurate and comparison with each other.
You should try to test a "real" gear in aluminium... just to have a reference of how good are 3d printing gears. Anyway, thank you for the video I enjoyed.
Yeah, that would be a good comparison. But a durability test in such a case is mandatory. It's a pity I didn't have chance for this video to make that durability test.
If you want to test durability, reverse the ratio so that the last gear is spinning 16 (or more) times faster than the drive. High speed should wear it down faster, but you'll still probably have to run it for quite a while before failure.
Yes, that's right. I though about that, but then I also though that such a high speed will increase friction and will generate quite some heat that will affects the performance and the test results.
Obligatory spelling check comment. The spelling of the quadruped that you are referring to is "Elephant". There is a "t" at the end. It makes the "tah" sound. And the situation that you are referring to is "Elephant's Foot". Its possessive because it belongs to the elephant.
I will never trust Onshape because it's web based. They can pull the plug and there's nothing I can do. That might be OK for hobbyists but it's no way to run a business.
Cloud based is the future of softwares. Onshape just doesn’t have the longevity in the industry that Autodesk has which makes it a riskier product by perception. However the product is revolutionary in that it does not require a $4500 machine to run. So yes a risk, but you can also mitigate with data backups to a physical location. Most of these part files can be converted one to the other, if the need were to arise.
@@joshuaboardman3650 Cloud based is worthless and essentially impossible for anyone working under a strict NDA or any sort of security clearance. I agree with you cloud based is OK for hobbyists and small business, but will never be OK for the big boys. A $4500 machine could be half a month's wages for the operator, it's hardly a big deal.
ProE came out in 1988 and is still around today as Creo. Onshape is also developed by PTC the same company that created proE. Onshape is their attempt at cloud based CAD and it has some real advantages. However it doesn't really make sense for hobbiest due to limitations with the free tier and the higher cost compared to Fusion.
haha, yeah now I know as well, after reading all the comments... I watched two times the video before publishing but it was pointless, after printing and seeing and talking "teeth (tooth)"... :)
If you design a chamfer into the gear, you can avoid manual chamfering. I do that often when it’s critical. And the term is “elephant foot” not “elephant tooth”.
Your Noise levels mean nothing... 1) no one puts their ears near spinning Gears. 2) All noise measurements MUST be done at the same distance from the noise source for accuracy.
You do bring up a good point about distance for testing noise levels. But EVERY GOOD designer thinks about the end user experience. For mechanical design, noise is a very large compent for consumers.
Not defending the testing methodology, but tell the noise thing to an NVH engineer, it’s like their whole job and car companies employee more than one of them
What do you mean about that? Try the 1st hear vs the reverse gear of the majority of the cars, or high performance gearboxes with spur gears and tell me if the noise is not a thing.
