Page 3 of 6 FirstFirst 123456 LastLast
Results 21 to 30 of 60

Thread: MX-Phoenix hexapod and some more

  1. #21

    Re: MX-Phoenix hexapod and some more

    So you are saying the MX28 also have much the same gear backlash as the 64?
    I don't have a MX28 to measure on, but I would expect it to be within the same ballpark based on the very similar construction, yes.

  2. #22
    Join Date
    Dec 2007
    Location
    Portland, OR
    Posts
    3,196
    Images
    952
    Rep Power
    132

    Re: MX-Phoenix hexapod and some more

    Quote Originally Posted by Zenta View Post
    You are correct Kevin!

    I thought it was about time to post a little update.
    First, a couple of days ago MX-Phoenix made her first steps! Not to brag but it look awesome, lol! I just love the fluid motion of the servos and the very quite sound of the MX's in action.

    But there was some bumps in the road before I managed to make her walk..
    At my first power-up I used a lab DC powersupply with current limited to 3,15 amps @12v. Very soon I realized there was something very wrong. The current drawn was at max and the voltage dropped to 2v or so. There had to be a short-cut. After some minutes of searching I found one faulty servo-wire!
    Attachment 6795

    As you can see, the GND and Power was crossed! I'm not sure how much reverse current the servos can handle, but I'm really glad I didn't loose a MX-106. Under normal use I'm using a LiPo batt with a 25 amp fuse, not sure if a servo handle 25 amps in revers?

    After finding the faulty wire I thought everything was fine, but it wasn't.. For some unknown reason I didn't get any respond from the servos at all. I had tried only one leg and all coxa servos before, so I knew the code was ok. Connecting one leg at the time I managed to locate the "faulty leg". At first I thought one of the servos had an ID set to 1. Actually, it was another faulty servo-wire! This one:
    Attachment 6796

    It's a bit hard to see on the picture, but the signal and GND is crossed.
    From now on, I'll always check the servo-wires to see if they are properly wired..

    After the wire-issues everything worked fine and I was able to work on the code to make it move and walk properly. I hope to post a video soon, but there is one issue I want to improve first: GEAR-BACKLASH, OMG . On the MX-64 coxa servos this is a bit annoying. I'm going to use some long coil-springs to add a constant load for reducing the backlash. I made a very fast setup and it helped a little. The spring add about 1 kg/cm of load and should be ok. At first I'll design a 3D printed holder for the spring to make sure the servos get a constant load. I'm still note sure if I'll go this route, but its worth trying.

    A question: What's your experience of the MX-28's when it comes to gear-backlash?
    My experience so far, is that the MX-106's is slightly better than the 64's.
    Yikes, this is the first incident where I've seen our cables with wires crossed. Good to hear nothing was damaged, you have my apologies about that.

    Looks like I've gotta go bark at our cable manufacturer's QA department....
    Andrew Dresner
    Principal Engineer
    Interbotix GIthub
    Interbotix Labs Google+

  3. #23
    Join Date
    Dec 2007
    Location
    Portland, OR
    Posts
    3,196
    Images
    952
    Rep Power
    132

    Re: MX-Phoenix hexapod and some more

    Quote Originally Posted by jwatte View Post
    I don't have a MX28 to measure on, but I would expect it to be within the same ballpark based on the very similar construction, yes.
    MX-28T have slightly less backlash than the MX-64/106, but not by much. 64 & 106 appear to have about the same level.

    Zenta- one thing you can experiment with is tighter PID settings on the MX-64, I tweaked these quite a bit for different situations when I was working on a recent scifi film project with some hexapods I built for them. You can start by experimenting with the P parameter alone, but be very careful with large changes from the default- test bench power supply with current limits are recommended.
    Andrew Dresner
    Principal Engineer
    Interbotix GIthub
    Interbotix Labs Google+

  4. #24
    Join Date
    Jan 2008
    Location
    Norway, Stavanger
    Posts
    784
    Images
    276
    Rep Power
    71

    Re: MX-Phoenix hexapod and some more

    Thanks for your input Andrew! I'll test adjusting the PID also one day.


    My kids did watch the Annedroids series on TV where they used PhantomX, I believe that was not the project you are referring to?


    One annoying result caused by the gear backlash is that the leg wobble a bit during the swing phase. Since the coxa servos get very little load when a leg is lifted, the play in the gear is very visible. I might go for six MX-28 one day..

    I mentioned a week ago that I would try using some coil springs for adding a little force that keep the legs more stable during the swing phase. This is one of the solutions I came up with as a springholder:
    Click image for larger version. 

Name:	KabelHolder-Left-v3_2.jpg 
Views:	41 
Size:	31.0 KB 
ID:	6800

    The springholder also work as a wire protector/guide/clamp.
    Click image for larger version. 

Name:	KabelHolder-Left-v3_1.jpg 
Views:	35 
Size:	20.9 KB 
ID:	6801

    Another picture of all six of them printed and mounted:

    Click image for larger version. 

Name:	MX_Phoenix_CoilSpring_1.jpg 
Views:	70 
Size:	103.0 KB 
ID:	6802

    This spring solution made it better and sort of damped the backlash effect. I'm still not 100% satisfied, but it's good enough. Currently I'm working on the code. I'll post some more info about that later.

    Merry Christmas to you all!
    Kåre Halvorsen aka Zenta
    ---------------------------------
    Zenta's YouTube channel
    Zenta's Blog
    Zenta's Instagram

  5. #25
    Join Date
    Jan 2008
    Location
    Norway, Stavanger
    Posts
    784
    Images
    276
    Rep Power
    71

    Re: MX-Phoenix hexapod and some more

    Oh, I forgot to mention that I've just connected and wired up the foot-switches. They work ok but I might have to add an extra spring since I've had one case where one switch didn't release properly.

    Just took some very poor pictures in my tiny workshop floor after walking on to an obstacle.

    Click image for larger version. 

Name:	MX_Phoenix_28.jpg 
Views:	63 
Size:	134.3 KB 
ID:	6803
    Click image for larger version. 

Name:	MX_Phoenix_29.jpg 
Views:	45 
Size:	123.3 KB 
ID:	6804

    I'll take some better pictures later.

    I also need to work a lot more on the TA code. Compared to using load readings on my PhantomX MKI and using a simple foot switch proved a switch to be a lot more reliable and faster.

    Having fun!
    Kåre Halvorsen aka Zenta
    ---------------------------------
    Zenta's YouTube channel
    Zenta's Blog
    Zenta's Instagram

  6. #26

    Re: MX-Phoenix hexapod and some more

    Since the coxa servos get very little load when a leg is lifted, the play in the gear is very visible.
    I was going to say you should try a torsion spring, but it seems you already did :-)
    Especially helpful might be a torsion spring that pre-loads the joint in a way that counter-acts gravity, and thus saves battery life and generates less heat in the servos.
    I did something very similar on a Mech Warfare conversion of a Phantom Quad (AX12 based) a while back.

  7. #27
    Join Date
    Jan 2008
    Location
    Norway, Stavanger
    Posts
    784
    Images
    276
    Rep Power
    71

    Re: MX-Phoenix hexapod and some more

    Thanks for your feedback Jwatte!
    Quote Originally Posted by jwatte View Post
    Especially helpful might be a torsion spring that pre-loads the joint in a way that counter-acts gravity, and thus saves battery life and generates less heat in the servos.
    Seriously? I tried to tell her that but the MX-106's on her femur became really offended by that suggestion.
    It might saves some battery, but the heat isn't a problem so far.
    Kåre Halvorsen aka Zenta
    ---------------------------------
    Zenta's YouTube channel
    Zenta's Blog
    Zenta's Instagram

  8. #28

    Re: MX-Phoenix hexapod and some more

    the heat isn't a problem so far
    Glad to hear that!
    If the main problem to solve is "it looks bad when it raises the leg," then loading it in either direction will "fix" that, but loading it counter gravity has additional benefits that I'd rather take than throw away :-)

  9. #29
    Join Date
    Jan 2008
    Location
    Norway, Stavanger
    Posts
    784
    Images
    276
    Rep Power
    71

    Re: MX-Phoenix hexapod and some more

    Quote Originally Posted by jwatte View Post
    If the main problem to solve is "it looks bad when it raises the leg,"
    Thanks but that's not the problem. The "problem" is mostly caused by the backlash from the coxa servos. They make the legs wobble a bit when lifted.
    Kåre Halvorsen aka Zenta
    ---------------------------------
    Zenta's YouTube channel
    Zenta's Blog
    Zenta's Instagram

  10. #30

    Re: MX-Phoenix hexapod and some more

    Ah, sideways wobble! Yeah, a small spring seems better for that.

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Similar Threads

  1. Question(s) Phoenix Hexapod
    By nimda in forum Humanoids, Walkers & Crawlers
    Replies: 24
    Last Post: 04-06-2016, 11:40 AM
  2. KURT< HELP!!!!!!!!! Phantom X Phoenix code to run on my hexapod.
    By gregorio_626 in forum Projects For Request and/or Sale
    Replies: 8
    Last Post: 01-26-2016, 05:51 PM
  3. Sale: Black Phoenix Hexapod (complete)
    By BlackDogCpt in forum Garage Sale
    Replies: 4
    Last Post: 02-11-2015, 10:10 AM
  4. For Sale: Complete Phoenix Hexapod
    By Tyberius in forum Garage Sale
    Replies: 4
    Last Post: 01-06-2010, 10:28 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •