Linear actuator using ClearPath Integrated Servo System @ 72V DC

Quick notes up front:
you can find the current version of the actuator under this link. If it should break, message me and i will update it.
Even though this project started out using ClearPath motors I eventually switched over to AASD servos. Here are a few renders of the latest version (Apr 2020):



SFU1610-700mm
AASD-20A Servos
1KW / 4NM / 2500RPM
Stroke: 580mm
Speed: 420mm/s
Cost ~500€ / pc


-------------------ORIGINAL THREAD STARTS HERE---------------------

Hey there,...

So I guess, after a few months of planning I reached a point where I should open a build-threat for my linear actuator design.

I borrowed heavily from @SeatTime , @wannabeaflyer2 and @pmvcda 's design (to say the least!). Especially @pmvcda changed the game for me, because he made me realise how much 3D printing could change the way I approached a problem. However, I didn't want to take this 3d printing thing as far as @Hoddem. I wanted the load-bearing structures to be made out of an all-aluminum construction. After all, aluminum profiles are cheap and durable

The basic design is pretty much fixed:

There are a couple of points where I deviated from previous designs and I will go through those one by one:

1. Double upper bearing (both IGUS):
I had doubts if a single upper bearing would hold the bending loads. Even though no-one ever complained about problems with a single bearing, it was literally screaming at me "this is gonna break". And since I needed the actuator to have a certain minimum length anyways, I decided to invest those 180mm for a second bearing. Even in it's fully extended position the rod and the actuator body now have an overlap of about 200mm.

I initially bought LMU bearings, but after just a few minutes the (stainless steel) rods looked like this:
...so I went with the IGUS bearings.

2. ClearPath Integrated Servo System:
I decided to use a ClearPath MCVC-3432D-RLN @ 72V DC


The biggest bummer right up front: They are expensive! I paid 2100$US for 6 of them. Then shipping and handling (380$) and german tax (400€ or 500$US). So almost 3000$ for the motors alone :-( But they have 1.1KW of power (yes, one of them!) So, definitely no concerns for being underpowered here The important part for me was the fact that those motors meant that I could completely ditch all Arduinos, break resistors, H-bridges, Sabretooths, Kangaroos, Capacitors.... all the things I only accepted gruntingly as a "necessary evil". Just serve them 72V DC and give'em a signal telling them what to do and they will go to work. Period. The signal is a speed command coming directly from Thanos' AMC1280USB.
...oh, and they're quiet! You hear nothing but a faint humming @ max RPM.

3. Two beam design:
Since the ClearPath Motors are NEMA34 size, I couldn't just position them between two continuous 60x20mm profiles. It would've worked, but it would've made the actuator very wide. I wanted the actuators to be as slim as possible near my legs, so I kept them slim around the top, but used two 80x20mm profiles at the base to house the motor between.
I didn't want to go with the "4-threaded-rods-type-of-design", as I feared it wouldn't hold the frequent torque changes.

4. All-metal load bearing construction:
3D printing opened up many interesting design options and @pmvcda and @Hoddem have shown beautiful designs using it extensively. However, I wanted the load carrying construction to be all metal (aluminum). You can get 8mm aluminum plates laser cut for a few €, so I wouldn't have to worry about print times and safety factors. Also longevity was a consideration as PLA and PET-G have their issues after a some time of use under load. So I used 3D printed parts only to keep the metal parts in place, so to speak.

Orange are the 3D-printed parts:


This is the load bearing structure:


...and the force-path:


Position feedback is through 6 AEAT-6012 12bit multi-turn magnetic encoders. They are not in the design yet. Still to go. I chose them , because I hope for good precision and the 2017 AMC1280 supports them directly


A few words of encouragement for newbies: If you read this and are thinking in the back of your heads: "I'd love to do something like that too, but do not have an engineering degree..." Don't worry! You can do it! I am living proof that all it takes is determination (and ~600€/actuator). To give you an idea of where I'm coming from: Just a few months ago I didn't even know what an H-bridge was! There was someone talking about an "H-bridge" and I had no clue,... but I saw that he was from Australia,... so the first thing that came to my mind was the Sidney Harbour Bridge!!! Seriously!!! If you can read this text, then you can built a full-on 6DOF Stewart platform.
It will be a challenge,... yes. But it will let you grow! Much more than your Highschool friends who probably actually have engineering degrees but are stuck somewhere in an office of a company doing stuff they hate or are indifferent about.

DO IT!!!

Nice work . Would have loved to use Clear Path servos, but after getting some quotes, the cost to Australia was even more expensive then yours . Look forward to seeing it all in motion. Note. The DC motors that I use are also quiet, but ballscrews do create some noise during long strokes.

Hi Dirty,
Firstly; Great looking design, and finally someone is using Thanos AMC1280USB, I have one (bare-bones) maybe you can offer me some tips in using the AMC1280USB.
Secondly; do you mind saying what you choose for your lead (pitch) on your a lead screw?
So again Great Job and I cant wait to see it in action!

Regards Jerry.

I have used 1280s for some time and currently have two in my rig. An old one and the new one.

Hi SeatTime, Sorry I must have missed the use of the 1280's in your rig, Ive been waiting for a current "up to date manual", in the hope that I can decipher what it all means. (not my area of understanding) Have had a look at Thanos's videos but way over my head, if I had a manual as mentioned, with pretty pictures and some wiring diagrams I might have a chance.

So; can you say what you use for a lead screw, lead, Ive asked in the forum but had NO response, maybe you can help!

Regards Jerry.

5mm.

Hi SeatTime, and thank you for the reply, I will work around 5mm and see what happens.

Thanks again
Jerry.

I am also using 16mm leadscrews with a 5mm pitch. AFAIK they come only in 5 or 10mm Pitch. Since the 5mm version already give me an actuation speed of around 330mm/s (which was way fast enough for me as a flight simmer) @ 4000RPM , I didn't even consider any other.

I can't tell you anything about using the AMC1280 (yet!). So far, I have one on my shelf, but haven't used it in any way. All I know is that it should be able to do the job and if others have figured out how to use it, I will too . A few months ago, I wanted to read into this thing, but there wasn't even a manual around. Maybe there is one now? ...Thanos? ...Anyone? A real manual with wiring diagrams n' stuff sure would be more than just 'nice-to-have'.

I am currently waiting for those laser cut aluminium parts to arrive and am refining the 3D printed parts to finish the first actuator maybe within the next month or so. As soon as I have a working prototype I will try to hook it up with the AMC1280 and see how that goes...

Speaking of 3D printing... PLA does not hold up well over time and PET-G feels kinda too flexible still. I guess I will use carbon fibre infused Nylon (Matterhackers "NylonX") for the printed parts. Never used it before though. In fact I had to order myself a new printer (Prusa i3 MK3) because upgrading my previous to an all metal hot-end would've been wasted money.

I will publish all plans and files (Autodesk Fusion360) here as soon as I think the design works.

Dirty

looking good

what is the cost of each actuator ?

Depends a lot on where you are ordering from. Shipping can make quite a difference. I'd say US and UK are probably the places where you have the best chances to get the parts you need at a reasonable price.

With the ClearPath motors I think I will end up somewhere in the region of 700€ (~850$US) per actuator. With simpler brushed DC motors probably somewhere around 500€ (~600$US). I heard people talk about a full 6DOF rig could be built for 4-5000$US but I find that hard to believe. I'm expecting to end up significantly higher than that when done. I have set aside a budget for this and so far it looks good.

And keep in mind, when you are building for example a 2DOF seat-mover, you can get away with a much less expensive design. This design got expensive when I insisted on having speeds of over 300mm/s with a payload of 150KG. The premise was: "It must never be underpowered". That means it should be able to handle even peak demand. If you are willing to accept a design that will perform well over let's say 98% of the operating envelope, then you can definitely built something around 400€ (~500$US) per actuator.

The clearpath motor connect same as on old version controller:

https://www.dropbox.com/sh/ejp77w6z15wmprl/AABe88mvk4pLthVy21Y5XV5da?dl=0

http://motionsim.freeforums.net/thread/61/faq-teknic-clearpath-integrated-sermovotors



The setting in the lcd menu is pwm 8bit for the correct outputs.

You can still use SSI 12 sensors to improve the resolution of positioning to 4096 positions

Thanks
Thanos

Thank you, Thanos, for the links I found some of them already, but not all of them. I noticed that there are all kinds of resources for the AMC1280 out there somewhere, but they are scattered all over the place. Might be not a big issue for you (or other professionals) who know what they're doing, but for a newbie like myself a real manual would help a lot!

So,..... I finally got the aluminum parts I was waiting for I got them from www.laserteileonline.de and paid about 140€(180$) for a total of 18 parts.


They were surprisingly good! I mean, I read they were going to be precise to within a few 1/1000 of a millimeter, but only when I held them in my hands did I realise what that really meant. The representative said, that there could be problems because the hole diameter (5mm) is smaller than the thickness of the material (8mm) but that was absolutely within limits.

Here you can see that the 5mm part is perfectly smooth but the 8mm part is a bit rough where the laser exits the material.


I ran a FE-analysis before ordering them to see it they would have sufficient safety-factor. At a load of 1000N (my nominal load per actuator) the parts were well within limits:
Safety factor 2.94:


Displacement max 0.012mm:


To put this in perspective: When all 6 actuators apply their nominal force simultaneously the platform will accelerate upwards at a "gentle" 3.5 G for a few milliseconds. And the safety factor is still 2.94. So theoretically, the aluminium parts could withstand a constant 10G load on the platform. Whatever parts are gonna fail on these actuators, the aluminum parts won't be one of them.

Dirty

I made some progress on the position sensing. I will use a cheap GT2 belt attached to the slider that connects the ball nut with the push-rod.

The belt will run through an idler pulley on each end of the actuator...



...then through the sensor-drive...


...which looks like this on the inside:


All pulleys and idlers run on ball bearings, so they are pretty much inaudibly quiet, something that surprised me very much. I expected a noisy "BRRRRRRRR" when in motion. A full stroke results in about 10 turns of the Hall encoder. At 12bit resolution (4096 steps per turn) that should result in about 40000steps per stroke (~350mm).
All-in-all, I expect a (theoretical) sub-millimeter precision on the position sensing.

The accessories that are needed are mounted onto the actuator using slot-nuts and M5 countersunk screws:


The end-stop switches are simple OMRON SS-5GL2T mounted into a 3D-printed part with M2 Cylinder-head screws. The plastic part itself is mounted with two M5 slot-nuts which has the advantage, that it will let you adjust the position of the switch seamlessly anytime. The switch is activated when the "slider" moves past it.



I started preparing the wiring for the motors. Signal is provided through six Molex Minifit JR8 connectors. I chose cheap CAT5 Network cable because it had exactly 8 strands of wire, is well shielded and has the right wire gauge.


Molex Crimp-tools can be very expensive, but I found this Japanese tool, that would do the job just fine



...and because I won't need those cables for a few months and I like everything nice and tidy, I have packed them neatly and stowed them away.




Next is going to be finding the right way to connect the upper and lower aluminum beam. Screwing, clamping, glueing, welding, soldering... or a combination thereof...?!?



Dirty

how have you attached the different size profiles? I guess what I'm asking is what is transfering the load from the top to the bottom structure. as its illustrated the top is floating and the load will be transfered throught the bolts or whatever you attach the top to the bottom with.

if you terminate the larger profile and extend the aluminum plate the load is actively carried. I know this resembles other designs but it is the only way to carry the load on the lower profile rather than on the connection. look at how a header or lentle is properly placed in a structure to carry the load. floating the structure will place a lateral load on the connection rather than a direct load to your lower profile. I'm not saying it can't be done it just seems a bit overcomplicated.

That is exactly what I am asking myself right now

You are 100% right! I had the same thoughts and I could not have said it better!

Unfortunately that occurred to me only after I ordered (and paid!) the aluminum parts :-(

Initially I even found my "sliding overlap design" rather elegant, because I thought it would give me the option to adapt the actuator in length somewhat when needed. But over those past weeks the realisation crept in that those lateral loads would exert so much shear force on the connection that it would almost certainly rule out a "sliding overlap design". Hence, I considered glueing, welding, soldering --> "fixed overlap design".

So in a nutshell: My current design elegantly combines the disadvantages of those two variants. Not adjustable and huge loads. The "direct load" variant you suggested would've been WAY better!

...and to be honest,... the longer I think about it, the less I care for those extra 50€ (60$).

Good welds might hold. Maybe a shop close could do one for testing. Just a thought