Robot Quad Leg Strength Tests with Un-used Over Engineering

[@PaulGouldRobotics]

In general, I'm pretty happy with 3D printed cheap ABS parts in this Cycloidal Gearbox. I'm bending steel pins. 30kg @ 12cm 360kg.cm (~35Nm). The brushless motor attached to this gearbox can only output ~10Nm

[@DerSolinski]

OK first off let me say WTF lol really going places with this design. Instead of high infill percentages try more/thick perimeter lines. Tests from a lot of people show infill pales compared to perimeter thickness. (But make sure it fuses properly) In my experience 20% gyroid infill with 3-5mm walls takes quite the beating before it goes anywhere. The added print time from the walls is IMO it worth. A interesting study would be if you fuse steel wire into the gears could strengthen them immens. Look here for reference: kzbin.info/www/bejne/jqGneGuMrb-mrZY

[@PaulGouldRobotics]

I'm using PLA now (as my good ABS printer is broken) on my Ender-3. I've been making a new smaller PCB for this design. I'll be back testing this soon. 3-5mm walls seam quite thick. I might gradually make it thicker. I think making soft feet will reduce the peak loads on the gearboxes.

[@DerSolinski]

@@PaulGouldRobotics Try to anneal the PLA in the oven, depending on your material it might works wonders. An hour around 80°C should be plenty. To find the optimum print a flat strip place it between two poles and raise the temperature in 5 min / 5°C intervals until it begins to sag then go back 10°C. But experiment with that first you will get shrinking and expansion in some dimensions that need to be accounted for. But that is also a sign that it works and recrystallised nicely. A idea would be to use a short strip of spring steel as "toes" this could also catch a lot of the initial impact of setting down the foot.

[@jamesquinn6662]

@@PaulGouldRobotics if you're using pla try baking it out post print too. Rearranges the structure to provide increased strength

[@jamesquinn6662]

@@DerSolinskiIn a design im working on I have a load cell at tge joint output which provides some active suspension

[@SergTTL]

Infill adds much less strength than thicker walls do. The stress always concentrates on the surface.

[@PaulGouldRobotics]

I'm using 2 outer layers. Should I change it to 3 or 4? Will layer height and/or nozzle dia make a difference?

[@TheMaidenOnes]

@@PaulGouldRobotics thicker nozzle+layer height = stronger. blog.prusaprinters.org/everything-about-nozzles-with-a-different-diameter/

[@PaulGouldRobotics]

@@TheMaidenOnes Interesting, I'll do some more reading and try some different heights/diameters.

[@SergTTL]

​@@PaulGouldRobotics If you are using 0.4mm nozzle, then 2 outer layers make an extremely thin wall for a part that is supposed to bear any load. Infill provides very little strength. And changing nozzle diameter or layer thickness also most likely wont produce any noticeable difference in strength. It may also even complicate things for you - you'll have to carefully tweak temperatures and printing speed in order to not weaken the part even more. Because with thicker nozzle and thicker layers it's harder to transfer enough heat into the plastic in order to melt in evenly at the optimal temperature. So the 0.4mm nozzle and 0.2mm layers should be just fine. The simple and reliable way to make the part stronger is to increase the wall thickness because the mechanical stresses concentrate on the part's surface. And, by the way, you can increase the thickness not only by doing more wall perimeters but also by increasing the extrusion width (e.g. extruding a 0.6mm line with a 0.4mm nozzle) - this increases bonding between layers by squeezing every layer on top of a previous one with a greater force. So for your part I'd recommend doing at least 3 outer layers and 150% extrusion. This should increase the strength of your part more than twice. Also you can drop infill down to around 20-40% to save some printing time and material. Also make sure you are using the optimal temperature setting for your printer/material/setting combination - this may also have a big impact on the part strength. Also, as we can see, the inner gear broke in the thinnest area of the part. So keeping the holes as far as possible away from the outer wall would be even more important. The gear should have a thick outer belt of solid material.

[@PaulGouldRobotics]

@@SergTTL Wow, Thank you for the detailed info. It will take me a while to understand the pros and cons of each printing setting. One more piece of information. The tolerances on the mating surfaces (external teeth on cycloidal gear and internal lobes on the gearbox housing) are very important. Too tight and it binds up, too loose and the back lash is unacceptable.

[@localhawk1]

A cheap cnc mill can mill POM quite well. POM is very hard to break and often used in gears. very good wear behavior, sliding POM against POM. it is not a 3D Print, but regarding wear behavior, strengh etc. I think on the long way, POM gears are a better choise and you can do this also on your own. anyway, thank you for your videos!

[@PaulGouldRobotics]

I have been looking a routers for a while, I want to make one or convert my spare 3D printer. But I'm still quite happy with the 3D printed parts. The parts are stronger than I thought, even with cheap ABS plastic. I wonder how much it would cost to get them make in China/Taiwan out of AL or Engineering Plastic.I need a min of 48 per Quadruped.

[@attinyit9609]

For light machining Braun's mostly 3D printed router should be enough, if you want to get into machining metal take a look at Piotr Fox Wysocki granite cnc router or "routercnc" Stay away from Chinese PCV or miniature Al routers, as they tend to resonate right at few hundred Hz.

[@PaulGouldRobotics]

@@attinyit9609 Thanks for the information. I was thinking of a Chinese Router, maybe not anymore.

[@localhawk1]

@@PaulGouldRobotics hi, thanks for reply. with cheap cnc mill I mean something like this "Sorotec Basic-Line 0605". Kits start at around 2.600 EUR. Good for aluminum also. You can see this machine running here kzbin.info/www/bejne/gZ6yl2ythKqGaqs (german). I never had a 3D printer but a small cnc router. but if you buy two small POM pieces and rub them against each other (totally smoth, you don't even see anything) or try to break them ... I think it is a bigger difference to 3D plastic parts. anyway thanks & best regards

[@marlonlacert8133]

Put a grove on your prints, that will allow you to add a high strength Epoxy. Apply the epoxy to each side of the gear with two sheets of wax paper on the outside, and compress between a vice, as it cures.. Note: Use between two hard flat board like objects.. With vice on outside of boards.. Epoxy does cure differently under pressure. ** Also best to have lines between both sides.. In other words, think of your 3D printed part as a scaffold for supper strong epoxies. At least that is what I do, when I need any real strength.. As 3D printing, is more precises than foam cutting. And the print adds to the strength of the epoxy.. *With some epoxies, you might need to paint your print, to prevent melting...* (Or do so in thin layers. All based upon what you are printing with.) As I use ABS for all things. I found some things will melt the prints.

[@PaulGouldRobotics]

This might be to complex for me. The gears hold up well, considering their size, as they are mainly under compression and the load is shared among make teeth/lobes. The outer housing is only under tension and my guess (without further testing /research) is that they are weaker under tension than compression. How does epoxy go under tension?

[@marlonlacert8133]

@@PaulGouldRobotics , It works wonders under tension if you add fibers. Like a carbon fiber strings. Or even on big projects embed aircraft cables.. Supper strong fibers, cause load sharing. One example found that plastic epoxies with carbon fibers, are stronger than aluminum. And only a few points away from the strength of steel. Sometimes called poly composites. Note: carbon fiber string is not that costly. www.amazon.com/Kevlar-Thread-Sewing-Size-Commercial/dp/B01M2AYTS2/ref=sr_1_1_sspa?keywords=kevlar+thread&qid=1578059352&sr=8-1-spons&psc=1&spLa=ZW5jcnlwdGVkUXVhbGlmaWVyPUExSk0xQlk3VFJIWjVMJmVuY3J5cHRlZElkPUEwODM4MzI2MjgwVlYxQkJIWlBITSZlbmNyeXB0ZWRBZElkPUEwNzc4MjM0MkFMVkVKMk1WVEJKWSZ3aWRnZXROYW1lPXNwX2F0ZiZhY3Rpb249Y2xpY2tSZWRpcmVjdCZkb05vdExvZ0NsaWNrPXRydWU=

[@marlonlacert8133]

@@PaulGouldRobotics Once part is printed, all you need to do is put thread on needle and weave it into the part. Once done add a strong epoxy, press while curing. (The pressing does Three things: One it forces epoxies into the threads. Two: It give a compression preset load. Third it forces Epoxies into all points in the part, and pushes air out..) *If you have a pressure pot, then use that instead.*

[@marlonlacert8133]

@@PaulGouldRobotics Lastly, let me send you to an expert: kzbin.info/www/bejne/sGfIoYOgmKd_mac

[@PaulGouldRobotics]

@@marlonlacert8133 This give me more ideas. I would rather not use to much epoxy but carbon fibre tow seems usable and no too expensive. www.carbonfiber.com.au/category9_1.htm

[@PowerScissor]

I'm no expert in concrete or anything, but I've always thought it was called post-tensioning...because you add the tension later. Wouldn't pre-tensioning mean you add the tension before? I'm sure you know better than me...just curious if I've been calling it the wrong thing or not for post-tension concrete.

[@PaulGouldRobotics]

It is possibly both. Tension is applied after the concrete is "set" but before the next part is built on top.

[@PowerScissor]

@@PaulGouldRobotics I was curious, so I looked it up. Pre-tension slabs, the cables are tensioned, the concrete is poured and allowed to set...then the tension is released. Post-tension slabs, the concrete sets and then the tension is added. Never knew that before. So basically the same idea, just the tension is in the opposite direction depending on which was was done.

[@PaulGouldRobotics]

@@PowerScissor I will have to test to see how much stronger it is.

[@AdityaMehendale]

@@PowerScissor Nope. The tension is in the exact same direction. The cable is always loaded in tension and the concrete always in compression. The difference is that in the first case you need a large external jig to tension the cable/rebar BEFORE you pour the concrete (when it has zero compression-strength) whereas with the second, the tension is applied in situ with nuts or face-plates after the cement has set. The "pre" in pre-tensioning is intended as "before the train rides over the bridge".

[@stefanguiton]

Great update on your testing process! Keep up the good work!

[@PaulGouldRobotics]

Thanks

[@Personnenenparle]

Use concentric infill or just print with like 20 walls. Using 110% gilament flow can also help with strength

[@PaulGouldRobotics]

I'm using 6 wall layers now and it is much better. Thanks

[@0redthunder0]

Could you try steel pipe to increase the second moment of area?

[@jpjay1584]

an SLA printer will have stronger results

[@AdityaMehendale]

The youTuber _GearDownForWhat_ kzbin.info/www/bejne/npupqpyhgLWfmqc printed ring-gears for the actual differential of his pickup-truck, with interesting mixed results. Indeed filament containing chopped carbon fiber will help, but beware that it wears out nozzles and only adds tensile strength to the horizontal directions and is fairly anisotropic. Another KZbinr (I forget who) has actually printed press-brake anvils, so the compressive strength is apparently not an issue. Good luck!

[@AdityaMehendale]

Here: kzbin.info/www/bejne/raTbd4uKgMaLmqs :)

[@PaulGouldRobotics]

I actually work in the press brake industry. The punches (top part) are under compression but the die (bottom part) are under both tension and compression. I agree, plastic looks good for high compression forces. I'm printing a new test jig now, results in a few days.

[@gerokatseros]

Cycloidal Gearbox has less backlash than panetary gearbox? Wonderful construction my friend

[@PaulGouldRobotics]

In theory it can have zero backlash where an involute gear assembly requires some by design. I will find a whitepaper on the explanation why that is.

[@gerokatseros]

@@PaulGouldRobotics thank you for your quick reply ... very interested in your constructions

[@lochlanbogaers3291]

could you just fix one end of the femur to something and then add weight to the opposing end till it fails not very scientific but should be enough to prove that the femur can handle the forces involved and for the parts in the cycloidal gearbox perhaps a different filament might be better something that has a higher tensile and compression strength, this is awesome work and because of this started building my own cyclo drive 78mm outer diameter 30:1 reduction onto the third iteration now

[@PaulGouldRobotics]

I've actually started using PLA for this design, mainly because my ABS printer is broken and I'm stuck with my Ender-3. 3D printed parts, with the right settings, are stronger than I thought. What are you making with your drive? Are you using two or three disks?

[@lochlanbogaers3291]

@@PaulGouldRobotics sorry for the late reply im using two disks in mine, and im developing them for a quadruped robot my brushless motors are only 500 watt hobby king motors so the entire robot is considerably smaller and after seeing other people problems with weight and such im primarily building it from carbon fiber and aluminium tube

[@Thefreakyfreek]

if you need i can macine parts from aluminium plastick or steel for you small parts only like the gears and the the crankshaft i can make from steel including the excenters on the crankshaft jsut pm me

[@PaulGouldRobotics]

Hi Thefreakyfreek. That is super nice for you to offer to make some small parts for me. I'm trying to design my quadruped without the need of special machined parts. I have some magnetic gearbox designs I'm working on. I may need some parts for that in the future.Cheers Paul

[@Thefreakyfreek]

@@PaulGouldRobotics I'm happy to help I have a small cnc mill and a small manual lathe which can pritty much do all materials in low volumes

[@kevinyang8801]

Great work as always! I wonder if you could have other KZbinrs to collaborate, to make you components out of Delrin, Phenolic, or Bronze etc., so you may be able to tackle much heavier load? And for the material of the steel pins, maybe hardened pin can work better? Just some ideas :D

[@PaulGouldRobotics]

Thanks. I'm pretty happy with the load handling of the printed parts. My goal is to design the quad so that anyone can make one at home without any special tools or exotic material. That's why I'm working on an Actuator controller based around the Arduino.

[@tacpreppers4906]

Hi Paul, great series... thanks for making this open source, I've downloaded and started printing the thigh and cycloids using Taulman Alloy 910 Nylon, crazy strong, So much so my file almost skates across the surface of this stuff. You have to try it, i think it's the answer to all your strength and rigidity requirements. I printed using a Creality CR-10S, bed at 60C, nozzle at 265C, 30mm print speed, 90% infill, 0.2mm resolution using Cura. Perfect recipe for amazingly strong parts. I challenge you to break it... my bet is you'll need thicker pins e.g. 5mm. Looking forward to the next vid!

[@PaulGouldRobotics]

I have a "strong version" with 5mm pins but with bearings. I will have to start with Nylon it seems like it will solve a lot of my problems. Do you preheat/bake your Nylon before you print with it? I may also try 3mm drill blanks for the pins. I should be able to buy them in bulk.

[@tacpreppers4906]

@@PaulGouldRobotics No preheating at all, just make sure it's good and dry. I simply opened your STLs in Cura, set the print quality to draft (0.2mm - best result), 90% infill, 265C print temp, 60C bed and the key is the speed, 30mm seems to be optimal, oh and i forgot to mention, no fan for the entire print, best results. Alloy 910 is super easy to print with and the results are simply remarkable. I'm not kidding, this stuff is insanely strong and quite rigid. Looking forward to hearing the results, should you choose to try it out. cheers p.s. quick review and examples of strength kzbin.info/www/bejne/bpDTpZJmaqekj80

[@andrsam3682]

You may try some engineering plastics - PET, Nylon or similar

[@PaulGouldRobotics]

The final design will be made of some material like the one you suggested or made from Al in China. My current aim is to design it well enough so that the material type doesn't matter and anyone, with any 3D printer, can make it.

[@Robonza]

Keep up the great work. I love your design. If you can get 3mm drill blanks, they are impossible to bend. Spring steel is another idea, it will add compliance and bend back after. Also consider carbon fibre filament. I don''t know much about it but its supposed to be stronger.

[@PaulGouldRobotics]

I like the idea of Spring steel. But I was also amazed that the silver steel pins bend before the rest of the gearbox was destroyed.

[@Robonza]

Me too, 3mm piano wire should be easy to get hold of.

[@Robonza]

Oh yeah, Nylon laser cuts really well. You might need to stack it to get a good gear at the right thickness. Cuts very fast too

[@quentinmunch3700]

Maybe less damage with a planetary gear drive ? 🤔

[@PaulGouldRobotics]

I've been looking at some of Gear Down For What videos. His small gearboxes break at 18Nm. His bigger ones are much much more. With the OpenTorque Actuator Test (developed by Gabrael Levine) kzbin.info/www/bejne/bJ26Y4x9hrZ8h5Y they are getting 2.58kg/m (258kg/cm) or ~25Nm but that is limited to Temp/Current. It is also a bit bigger. I like the Cycloidal Gearbox because the backlash can be very low and it's different.

[@gedr7664]

are you making progress on this?

[@PaulGouldRobotics]

I think the standard 3D printed will be good enough. I've been designing a new, smaller PCB for this leg design. I'll be back to testing this soon.

[@aretgh]

try polymaker CoPa (insane strong and easy to print nylon (I print 360mm dia gear on open printer) it not cheap but it super strong but for other project I spend sooo long to make 3dprinted cycloidal reducer and it broke every time......so I cnc it from aluminum and now its working no problem!!!!!!(in your case 3d printed almost worked ok so I think cnc pa6 or pa6.6 plastic will work endless

[@PaulGouldRobotics]

I've tried cheap Nylon but the warping was huge. CoPa looks good and easier to use. It is expensive (5x ABS), not good for prototyping but as I only need 1kg for the final quad, I'll reprint the final design in CoPa or something similar. I could get the Cycloidal gear machined in Al or plastic from China/Taiwan for a few dollars per gear. I need at least 48 gears per quad. What are you making with Cycloidal gears? How much torque are you putting though it?

[@aretgh]

Paul Gould 250-500 watt mid drive e bike (motor and reduser 500-600gramm weight(super light!!))

[@Robonza]

I think for nylon you probably need to put your printer in a heated chamber. The guy above is right, Nylon is awesome, you just need to work out how to print it. All your parts would benefit.

[@PaulGouldRobotics]

@@aretgh I currently use the ML5010 (600W) motor and with the 25:1 Cycloidal gearbox it weighs ~500g. The main reason for the Cycloidal gearbox is the very low backlash

[@PaulGouldRobotics]

@@aretgh Most e-bike motors are 2-3Kg. Where did you see the 500g one?

[@uthvfyrekbnm6008]

I think you should to test composite filament

[@PaulGouldRobotics]

My printers are pretty cheap and don't like fancy filaments.

[@uthvfyrekbnm6008]

@@PaulGouldRoboticsMy ender 3 is good. I thing it not a problem. nowadays exist PLA with carbon fibre embedded

[@PaulGouldRobotics]

@@uthvfyrekbnm6008 I have a standard ender 3 which prints PLA well.