Testing Plastics

We cant access or robot, so season is almost over - we cant meet except on discord. So I have decided to start a project of my own. I am retired so I have the time. I got my printers and a lot of HIPS and some PETG and some ABS and also some Nylon 910 (taulman). 3DP has been good to us but most of the stuff was not really tested it was often hand broken or we print a thing like the Hook we use and tested it like this
Hooktest (2)
Now with all the stuff we did I cannot tell you is it barely strong enough or is it way overkill? So I decided on what is hopefully going to be the last build I am involved in that is based on guess work

I am in the processing of 3d printing (of course) a tester that is supposed to break plastic under certain circumstances. Some inspired by some youtube videos like CNC kitchen but I intend to test parts that are printed for the use in a robot and answer questions like how much can a 3d p plate made from xxx plastic of that dimension hold with a 608 bearing in it or when will a gear skip that has a 1.5modulus is double helical with a width of … How does abs/hips/petg/nylon change when I print it 10mm thick instead of 4mm etc.

At the heart of this tester is going to be a 128:1 planetary with a beveled 19:41 input so overall a 276:1 reduction. The initial print is done in HIPS and what fails will be replaced in Nylon 910. I want the cost to stay low. The power will come from up to 3 CIM motors that will drive the input bevel. something like this


Thats the input stack with the first sun on top initially the input will be on the bevel but there is allso provision for a belt an gear input from another source The 2 planetaries are 2.25mm module and the gears are 30mm wide. In my guess (and unfortunatley its a guess) it should handle the torque generated.

The Cim Motors will be run as torque motors that means I will design and build some drivers that control current and with that the torque of the motor. So the initial task will be to set a current and then measure the output by stalling the motor. So i can set the motor to stall at 10 amps if I so choose. The "sweet spot " on a Cim is at about 30 Amps where it puts out about 64 oz/in so about 4 inlb so with one motor mounted the output torque will be about 1000inlb (thereabouts give or take depending on loss)

ring on the input is 63 tooth


The Ring1 slides in the holder the holder is static and has a race for bearings where some 608s run in it like wheels tied to the axles of the planet to take off some load as the planet carriers will spin at about up to 1000 rpm

Planet carriers are also going to be used to help with allignment during assemble and in case something gets misaligned its supposed to bind and jam and stall the motor(s) before we develop shrapnel.

For assembly reasons the planet stacks come in 3 pieces


with holes for M5 bolts and also some press fits for 608 bearings As Ring 2 is 64 tooth compared to the 63 of ring one (with that and the p1=21 tooth and p2 = 22 we get the 128:1) and R1 is 63 each P2 in relation to P1 is rotated about 1.4 deg (360/256)
The Output has first a 138 tooth dual pulley on the ring itself

Which then gets support from a plate with a 3rd pulley tied to the main axle The planet holder also rides on this side on the main axle

Which then will mount to output ring - like everywhere else the bolts are countersunk with Nylock nuts embedded in the plastic. So nothing sticks out and can catch something.
Mounted to that is an output gear also running on the main shaft with bearings

with about 4in diameter and a 3.5mm module this gear will drive 2 pairs of those stacks

Which will allign and guide the output rack that hopefully will lift up to 600 lb on 1 motor like this (1 pair shown

The top gears ar 2.25mm module the bottom 3.5 the bottom ones mesh with each other the top ones with the rack the rack has a 1/2in steel 16 gauge square tubing in it as reinforcement here shown with an al. tubing on a test piece (out of al tubing at my house)

Also shows how they are concatenated and printed on an angle with custom supports to avoid ?drooping on the long inside of the square cutout or putting supports inside and its no fun cleaning out supports out of a 1/2 in by 500ish mm long square tube. (prints diagonally on my chiron)
This then is held by 2 plates like this

The channels are for aforementioned 1/2 in square steel tubing.will be held to the plastic with M5 bolts and the 8mm or (5/16) bolts working as axles for the rack and pinion will rest on said steel tubing. A 660 lb crane scale will be mounted to initially verify the output and then adding motors the intent is to scale it up until something breaks to get the data

The 138 tooth pulley will have enough torque to lift up to 250 lb per motor at 30 Amps. As we can go up to 3 motors that is why there is 3 pulleys as a properly tentioned gates HTD belt (the ones andymark is claiming to sell) is guaranteed to not skip up to about 300lb of static load and also should not break (according to the gates spec) This should answer as to how much torque can you take off a ring on a 3dp 2 stage planetary.

One of the goal is to collect some data - based on the attachments that will (hopefully) break test pieces under controlled circumstances. I hope it will help to develop some custom brackets, mounts, frame pieces, components, gearboxes that are not overkill but still strong enough for the task. Gearboxes for climbing a single robot, lifting stuff or doing a 3 robot buddy climb. Right now I am a little over 4 kg of HIPS (at $10/kg) into it have made and printed a little over 50 individual drawings and getting pretty close to putting it together.

Guess thats what happens if a retired old dude is locked up in a house with some filament and 3D printers lol. So far I got about 2 and a half weeks into it.

If it works I intend to share are drawings on my grab cad site - maybe someone wants to print it in Onyx lol. Also all test results will be published here - and maybe on grabcad or my website.

So far everything is printed on my Anycubic Chiron with a Volcano and a .8mm nozzle. All prints are printed solid here is a slice pic of one of the gears


So as close to injection molded as you can come with 3DP

Of course comments welcome

13 Likes

Your work never ceases to impress. If you’re going to be designing your own motor driver, that’s quite an ambitious task. I’m curious to learn more about that once you get there. I think you’ll find that peak torque on a CIM is well above 30 Amps, but perhaps you are designing to keep your load low enough to keep your current draw reasonable.

You are correct - peak draw is way above that. But the peak efficiency according to the char (seet spot is at 29.something A) so above 40 efficiency starts to drop and 30A already creates quite some heat at longer periodes especially if you get close to stalling its approaching about 360 watts. Plus at 30 amps you can get away with 12 gauge wire if you keep the length under 6 feet if you go to 40 amps you drop down to 3-4 feet That is why I will fuse each motor at 40A and its the reason why FRC does the same a 30x30 robot will keep wire length under 4 feet hence the 40 A max fuses and breakers. So I rather run 3 motors at 30A each than one at 90 now you are creating over 1000 watt of heat - enough to heat a room and you need something in the neighborhood of a 4 gauge wire

Now the motor controller is very simple. It will have 3 H bridges for the 3 motors. The ground will run through the ballast resistor of a 100 amps ammeter so at 90 Amps (3 motors) you will have less than 1/2 V drop over that resistor but still measurable. Now you can amplify that with an opamp - very simple. Now you got 2 ways to go and I havent made up my mind yet. You can generate to PWM either from a 555 - and voltage control the duty cycle of the pwm signal by mixing a pot and the feedback from the amplified voltage from the shunting resistor (couple of miliohms) or you read that and the pot into 2 analog ports of an arduino and then do a program to control the motors. The 555 way is cheaper. But I am researching some cheap arduino boards. I got a printer board lying here but I am hesitant of using a $100 board for that. I could get and use a PI too. The advantage going the PI/arduino road is for one flexibility and then you can also connect a computer or safe data to a sd card or something and get a graph off of it. With the 555/opamp solution you just get the data off of the Ammeter or the settings on the pot . So I probably go the arduino road - and yeah I really should learn arduino stuff anyway. So the firmware then will have some settings like 1,2 or 3 motors maybe determined by some jumpers or a switch which limits the current to 30,60 or 90 amps. Not all motors will be exactly the same but it will be close enough so it is tolerable especially as the CIMs are very predictable from CIM to CIM under 40 amps current/torque wise. The error will be lets say you set it to 90 amps and 1 motor runs at 31 A, one at 30 and one at 29 the efficiency difference between them will be less than 1% so you get the expected torque ± 1% I expect bigger variations from mechanical things like plastic changing size due to increased/decreased humidity in the air or that humidity messing with some lubrication or not.

Either way the goal is if I set the output to 30A Iwill get close to the same torque all the time. And As I said I got a 660 lb crane scale and can recalibrate.

Another advantage of going the programming road is I can home the machine more easily and work soft limit switches or positioning switches like if I want to know how much weight can something hold if I want less than 3mm deflection over a given length. Or repeadetly hit a piece with lets say 100 lb or something like that .

I big determining factor is going to be if and when the kids decide to go back to work on our community outreach robot as then most my time will be taken by that plus life comes up with all kinda unexpected things

At the current levels you are working at, the physical placement of the components becomes important.

Related to this, how and where the high current paths are run and where they are connected can make the difference between a circuit that behaves properly or one that oscillates on each switching event and possibly melting the switching devices.

Lastly, it would be best to install some appropriately chosen reservoir capacitors and bypass capacitors as close to each H-bridge as possible. This detail is rarely shown in the simplified schematics of a motor driver circuit that one commonly finds on the internet.

Edited to add: you will also need something like an IRS2183 or IRS21834 to drive your upper MOSFET in each of your H-bridges. There are through-hole and surface mount versions of both chips. I can give you guidance on the order codes if necessary. I can also give some guidance on how you would use them in your circuit. I am using the IRS2183 in a brushless-DC motor controller.

I intend to use some mosfets on the H bridge. I am the “proud” owner of a copious amount of FQP50N06 60A mosfets I have used in the past. To provide the proper voltage in the past I have used 2N3904 in an open collector mode with a pull up resistor to boost a PIs output from 3.3/5V to 12 and supply ample voltage to use the Mosfet as a switch. Yeah I know but this is a low budget operation. I want to get it done and am on a fixed income so what is onhand is important both in tools and components. As diodes for the back current I got a load of 1N4001 in my drawers here, They served me well in similar applications. As for capacitors - we will see that is usually a trial and error thing to strike the proper balance but I got quite a few - probably not the right one but that is how it always goes.

Right now I am leaning towards and arduino and have 2 outputs for the pwm one for forward and one for reverse So i just drive each pair of the bridge that way

Have not done something like this in this millenium but used to do it all the time. Just some catching up on a lot of new stuff that came out. First motor I ever made a driver for was making a solid state relay for a friend for the wipers in his car when I was in 7th grade and that was in 71 lol. That was the good old times when you rolled your own drivers for motors including steppers. Did a lot of that in the 70s and 80s. Stopped doing electronics and robotics in the early 90s when I started to get full time into writing software - back then mostly for stock day trading.

So yeah as always apreciate any suggestions I can get

The problem is the drive circuitry for the upper MOSFETs. The chips I was recommending take care of that for you very nicely and are just over $2 in small quantities from DigiKey.

You can get away with a pull up resistor to turn on the MOSFETs but it can be prone to oscillations and sometimes, device destruction.

It is best if the reservoir capacitor is a low ESR type but it needs to be big enough. The decoupling cap can be a film or ceramic type.

This sounds like a fun project.

1 Like

I will get some

Why spend all the time and effort to design your own controller when you probably already have a COTS solution available? Seems to me if your goal is to test the different types of plastics then that should be the focus rather than designing a motor controller. The Talon SRX and SPARK MAX both natively support current limits (effectively torque limits). And if by chance you don’t have one of those available, the difference in cost between buying one premade and the components to make one yourself is more than made up for by the fact that you can use the COTS one on your robot next year.

Very simple. Yes there is a cots solution and that is a talon SRX run in CAN mode. Now I don’t have any they are close to 100 bucks a piece - I would need 3. And then I need a computer/arduino/PI that does Can bus. Now it already dies at 300 bucks for 3 Talon SRXs or similar. A victor SPX might work too on paper only I never used one so IDK for sure - cheaper but still a budget buster.

And then… I like to build stuff. You know I could shell out 1000s of $$$$ (if I had it and buy a completed tester or multiple ones. But then for starters I don’t have that kind of $$$ lying around and there is no fun in spending $$$ especially if you dont have a lot - and then the question is can you 3DP and build from scratch a tester that helps you get better in 3DP. And on top of that everything that is fun is still closed or limited due to covid and it beats watching TV. So I rather potentially waste a 100 bucks and learn something than spend lots of $$$ and learn less or next to nothing and I already know how to hand over $$$ feels like I have been doing it all life long till it hurts

1 Like

Will you need to reverse the motors at all? If you do not, you can use just a single MOSFET (open-drain configuration) per motor instead of an H-bridge (4 MOSFETs) per motor. The gate drive circuit for the open-drain MOSFET is much simpler than for the upper MOSFETs in an H-bridge and you would not need any gate driver chips.

Yeah I have too as what goes one way to put a load on something has to go the other to reset for the next test unless I put a crank on lol with a 276:1 reduction lol. Hence the H bridge, Theoretically I could replace the bottom Mosfets with a Double Throw retail but that would not be that good of an idea lol I ordered IRS2183PBF in a dip package as I will do it first on a solder breadboard

That makes sense. You would need one of the chips per half-bridge, 2 per H-bridge.

You could probably put an automotive grade DPDT relay between the single MOSFET and the motor to change directions but they are still $10-$15 each.

exactly and you get all kinds of noise from arcing etc at 30 amps or so

Well soon I hope. Today I had to fix an oversight. I neglected to reliably mark the parts and with a twist of 1.4 deg it was pretty much impossible to keep them apart


So now I numbered them in cad and reprinted them
Also ordered some Arduino stuff and get my feet wet with this. So it will take a bit which is fine - more research necessary.

In your application, you can probably stop the motors then change direction so there is no current flowing through the relay contacts as they switch. That should eliminate the arcing and prevent burning up the contacts.

That might be a possibility


Well today I am pressing in nuts and bearings

All holes for nuts are tapered by 2-3 deg depending on depth and have a fillet on top so you get it started by lining it up with your hand and pressing down and turning a bolt in and "sucking it down the taper So the nut stays put and is properly aligned to take the bolt in the final assembly

1 Like

Would you like a new keyboard? Seriously. DM me your address and I’ll have the great rainforest in the sky send you one. This is really bugging me man. It distracts from your posting. I bet somewhere, students are making book on how much to lick a key or something.

21 Likes

Martin, I would contribute to your Patreon if you set one up. Or buy stuff off your Amazon wishlist.

5 Likes

One on the keyboard. Itr sticks sometimes I bang it lol I dont see much wrong but I have real difficulties seeing hence I drive rarely and only in real good visibility and definitely not in the dark or when its raining etc. Currently I am working and printing in real cramped quarters.

If you want to contribute I got a patreon on my youtube. But an amazon card would probably be better as patreon takes a cut plus tax etc.

1 Like

PM me an address. I am going to have a better (and cleaner) keyboard shipped to you.

10 Likes