Hi Peter,
Milli seems to be evolving through multiple personalities since it’s inception way back at the beginning of this thread. A robot is so cool. I’ll be following this thread with great interest.
Jayne
Hi Peter,
Milli seems to be evolving through multiple personalities since it’s inception way back at the beginning of this thread. A robot is so cool. I’ll be following this thread with great interest.
Jayne
Could use hydraulic (oil) bearings instead, that would be plenty stiff and air bearings are mostly useful for high surface speeds and/or to avoid transmission of vibration which are two features that won't really be an issue for you. In the same vein you could use a magnetic bearing, stiffness is higher than an air bearing and also a programmable feature, but the most involved type in terms of design.
Allowable size tolerances on the oil bearing would be looser to.
Still, given your application it would seem a bushing would suffice and a rolling bearing would excel. In our robots that we build at work we are using a lot of cheap plastic bearings actually, but we have a complex pick and place situation of light objects, 99% of the bearing load is just the robot dynamics and not the material being moved. Plus we design iterate and rebuild all of the robots every year since we don't sell the equipment to the clients and we're still in startup mode.
Our investors, which includes two major SCARA and industrial robotics suppliers, favor large rolling element bearings. They withstand the abuse of customers very well and for some reason no engineers ever want to admit they crashed the $65k 6-axis robot into something when they were dicking around...
Hi Strawb - Oil recovery on hydraulic bearings would be a pain. I admit I have crashed robots, CNC mills and manual laths, probably every piece of equipment I have used I've crashed at some point I think I'll stay with the HUB bearings for now just have to find a good supplier. Peter
Hi All - dropped into a truck parts place today and looked at hub spiders for the base bearing. They are big! plus two bearing companies are looking for a suitable HUB3 type for me. So once those are sorted I can commence designing the Milli-IRB. The rest is fairly straight forward... Peter
Hey Peter,
Ever used food grade oil for tapping? I'm thinking colza oil(rapeseed oil), really popular here instead of sunflower oil but since its used in machinery too, I got curious if I could use it for tapping. I have a 1.4404 stainless steel plates and need to tap M4 threads. It's a part of a custom humidifier and since I'll be breathing in whatever is on the surface I'd love to avoid anything that's not biocompatible. The tap is a 3 piece.
Hi Ard - Nope always used tapping fluids for stainless. We call 1.4404 S316 here. Only one way to find out - try it. Peter
I'm thinking I should have made an extra hole in the plate just to try things out. Anyway looking forward to the robot. Have you decided on what servo's to use on the axes? A harmonic+servo combo or a dedicated bldc drive? I've been looking to add 4th and 5th axis and found this https://www.aliexpress.com/item/1005...63996440%22%7D
It's about $200 cheaper than a harmonic+80st servo I was considering before.
Hi Ard - re lub. I used to run a draw bench for making aluminium tubes and it used a combo of peanut oil and pigs fat. Horrible stuff when it went rancid the tank had about 600kg of the stuff... The bench was 22m long and the motor gearbox was the size of a small SUV... I have no idea about motors yet. I have to sort the bearings. I usually think through the design and try to solve the hard issues first. Motors are easy so way down the list. Looking at these for the arm bearings. But they are heavy. The first version will use as many stock easy to get parts as possible.
nice diy robot
Evening All - I have spoken to my resin supplier and they are happy to do a tensile modulus test on the CF sample. Excellent. Waiting on a drawing of the coupon size so I can get into it. Peter
Morning all - I have the drawing and finish specs for the test coupons. They are a bit different to usual for composites. I asked them some questions and they usually test neat resins so I shall prep a coupon how its usually done for composites. They wanted it polished to G6000 and I said nope and why? I can see that small defects in neat resin coupons would be an issue but not on a CF coupon... So many things to get done. This will come to the top of the list soon. On a commercial project for a welding robot installation I have been speaking to the people at Octopuz very good off line programming software....Peter
Hey Peter, I finally found a vendor that sells all three types of slewing rings I mentioned. https://lyyjindustry.en.alibaba.com/
For reference the stiffest YRT series eg.YRT80 costs $160, YRT200 $560.
Thanks Ard - I want to cast the CF test coupon next. Now I have a suitable cutter for composites... Peter
Hey Peter,
Will you be using direct drives for each joint or regular servo's with belts?
Hi Ard - Been offline for a while. My elderly mother passed away and I've been supporting the family and making arrangements etc. Shes in a better place now. Milli-IRB will use ballscrews for the main axis and I hope to use tendons like in our arms for the outer axes (biomimicry) Servos will be used with axis position sensors.. I have a nephew who is a mechatronic eng and he's keen to be involved so two heads are better then one, plus the forum input should be good.... But Frankie is the next machine. I have been doodling in down times and have resolved a few things for Frankenrouter so will be concentrating on that first...when I get back next week some time... not sure when the dust will settle...Just figured out my passwords to forums and emails and catching up at the moment...Peter
Hi All - back at my desk, will sort out the 100's of emails and get back to my routines. No new thoughts On Mill-IRB other then I'd like to cold cast most of the parts... so need to cast the test coupons for modulus testing asap...Peter
Ah - the Kiwis have done it again.... BIG CF rockets... Perhaps Milli should be a Tape layer....Peter
Hi All - An excellent article on a CF helicopter part. This is where I want to go to with my machine parts. Peter
https://www.compositesworld.com/arti...rse-helicopter
Hey Peter,
For all of your simulations, you use 1000N, where did this number come from? Is this based on a torque or high speed machining?
Based on your number I went with 1200N as my spec for a rotating swivel head, but I'm having space issues regarding the C axis actuator, that make the headstock look just ridiculous so I'm secretly hoping a high speed application does not need 1000N for machining...
Hi Ard - The 1000N is a nominal number. Machine design is based on stiffness. So a target axis stiffness is set. A modern VMC will have an axis stiffness of say 150N/um. So if I apply 150N to the axis the tool will deflect 0.001mm. If you look at various machine parts say ballscrews they have a stiffness spec in N/um as well. So the applied load in FE does not matter as the model is linear (or should be) so you can apply 100N or 10,000N then calculate the deflection then convert that to N/um to see if it is what you want. See the attached image for typical machine stiffnesses. This is a very old data, machines have become much stiffer since FE has improved the design phase. I also use models with a forced deflection, this saves doing the normalisation calc. I apply 0.001mm to the tool and the FE will calculate the force required to deflect it this far eg 55N if my target is 100N/um then I have to fix this...
Actual cutting loads are a whole different story. It has two components 1) the torque required to pass a single cutting edge through the material. Called the chip load or chip thickness. This is the torque required to make a single chip. Then there is the force required to move the tool/machine through the material. This has to do with the metal removal rate MRR. Power equals force x velocity so the machine translational power depends on how much material you want to remove in unit time. There are heaps of torque and force calculators out there for tooling.
Tool forces are all predicated on the fact that the machine has enough stiffness to be able to cut the chip thickness and not deflect through that action, if it does deflect it either bounces (chatter) or it rubs vs cutting. Tool suppliers specify target chip thicknesses for their tools. Max chip thickness has to do with the gullet size of the tool so the chip will clear the tool and exit & the tool stiffness. Small tools deflect more then big tools...
So if we have an aluminium job and we have a chip thickness (I prefer to say chip thickness as this is more descriptive then chip load) of say 0.005" or 0.127mm (see table attached) if the tool deflects 0.060mm then it will not work. So the question is what is the minimum torque required to produce the chip and the force required to move the tool through that chip. Once you know these you can multiply these up to get to your MRR required.
Makino publish a lot of stuff on machining loads& their machine designs
https://www.radical-departures.net/a...sert%20cost%29.
This article quotes 1000Nm torque required for Ti6-4 at high MRR. But I've machined Ti6-4 with hand mills and HSS tooling so not sure on minimum requirements.
So to answer your Q another way, look up what the tooling load calculators say and use those numbers or pick a static target stiffness for the machine and design to that. Small manual mills are about 2N/um, the stiffest machine I could find was 650N/um
https://www.mmsonline.com/articles/a...w-world-record D=63mm 4F, 1130ipm, 0.424" DOC 295HP A7075 material... 1000 cub inches per min MRR
So find a machine that does what you want and look at the ballscrew sizes and motor sizes and this will give you an idea on the forces required, cross reference these to tooling load calculators and you will be close. Most machine makers don't publish their machine stiffness figures. Some publish their axes forces.
Hermle C22 quote 4500N on all axis.. but most of that is for accelerations. It can accel at 0.9g...
You mention HSM strategies and these can create issues in maintaining constant chip thickness (Ct) at corners and arcs. If the controller is optimised on maintaining a constant velocity then the Ct will be wrong in corners. There are some controllers that adjust for that out there.
Hope this helps. I have not found a simple one doc answer for all of this. Maybe I do a Phd on it... Peter