Timescapes - Digital Timelapse Discussion

Debugging Motion with Otto
Page 1 of 1

Author:  Biolapse [ Wed Dec 28, 2016 11:49 am ]
Post subject:  Debugging Motion with Otto

Hey folks.

I have been having quite a bit of fun with this rig which I call Otto (throwback to WALL-E)

Some of the motion was not all that awesome so I spent some time working on it. I have a few tweaks left to try, the overall result of my efforts really paid off.

On a side note, I managed to get a Compaq Presario (hahaha) win 98 machine, and I was able to successfully log into the leadshine DM432c drivers and dial them in. The Focus and 3D motors only handle .33 amps and the default setting is 1 amp.

This has been quite an undertaking and I feel it is really starting to pay off. The image below is of Otto during the testing. I feel pretty good about getting the motion accuracy I am getting. I have some test footage and I am curious to any thoughts from this community. :)


Author:  sciencelookers [ Thu Dec 29, 2016 11:05 am ]
Post subject:  Re: Debugging Motion with Otto

Chris, the rig looks totally awesome! I want one. The elevator was sort of unconventional but works really well in the space you have. Congrats on having the courage to go through with that part.

Also, nice job on making the diagnostic videos. They should be really helpful in finding the little gremlins. In the first constant speed horizontal composite, I still see a tiny amount of stutter, but no brightening and darkening in the composite. This is saying that the exact same stutter is present in both passes or (more likely) this is some sort of artifact from the editing or compressing algorithms.

The vertical run definitely lightens and darkens in the composite, indicating small differences between the two runs. If you really think parts of the rig are binding, allowing other parts to store energy like a spring, then busting loose "earthquake-style" at random times, try doing another composite of timelapsed runs but touch the suspected area with a very small vibrator between each frame. (I'd look for a real small one and avoid the humongous back-rub non-sexual ones because those big boys can really shake things up) This may unstick things so the required movement happens during each interframe interval. A good composite using this technique would verify your earthquake hypothesis. If this is the problem, try disconnecting the motor and turning the motor coupler by hand. You should be able to feel the sticking at certain points if there is any problem in addition to the motor alignment. (this is a good idea even if you don't want to put sex-toys in contact with your rig.)

When we were making the super-adaptable all metal focus motors like the one you reviewed, we had solid shaft couplers (from servocity) and we found that the alignment between the gearmotor and the input to the right angle drive was super-critical. Even the tiniest misalignment caused a lot of friction binding and growling from the gearboxes. So this is a very believable source of your binding. If you want, I can mill a block of aluminum to have about the right sized step to get rid of those round risers, which you should probably get rid of somehow. I'm offering to mill the part for free, but measuring the height of the needed step accurately enough would be the biggest problem because of the super accurate alignment needed. You will probably still need some shims to get a perfect alignment after installing the step. Mcmaster-carr sells shim sets which are very helpful in finding the exact size of shims needed. Shims with slots are most convenient because you don't need to remove the bolt completely to slide them in to check different thicknesses, but I see the minimum slot width is 7/16 which may be too big depending on your bolt size. Whether or not you want the stepped plate, a shim set would definitely help get a precision alignment. Whatever adjustments you make, I'd recommend turning the drive by hand to check for any remaining stickyness. Turn the shaft coupler with your fingers with the motor attached to get a feel for the stickyness of the entire drive train.

Another very easy remedy might be to spray some very light weight machine oil on bearings and other moving, rubbing parts. WD-40 is not a good choice, look for clock oil or some other super-low viscosity oil thats meant to stay and not evaporate. WD-40 is meant to evaporate and when it does, it leaves a relatively high viscosity residue meant to repel moisture. This tends to gum up very fine moving parts eventually.

The pause when the pan-tilt reverses direction is classic backlash coming from the orbital gear speed reducers on the SC pan-tilt head. Nice conversion by the way, and you actually have pretty good orbital gearboxes on there because the backlash is surprisingly small. Are those from stepperonline or someplace else? If someplace else, do you mind sharing where you got them? Your pre-loading the gears with the rubber band is a pretty common remedy for this and costs a heck of a lot less than a true zero backlash speed reducer like a harmonic drive. You can switch to more rubber bands or a heavier spring if you find the weight of the camera overwhelms the tension of the rubber which would allow your backlash to come back.

Do you have a belt tensioner on there other than the slots to move the motor? if so,would you mind sharing details of it?

I hope this and everything else goes well for you in the next year. Let us know what you discover regarding these very small issues with what is an awesome rig. The diagnostic techniques and fixes will undoubtedly be of use to many of us.

Author:  sciencelookers [ Thu Jan 05, 2017 11:45 am ]
Post subject:  Re: Debugging Motion with Otto

So have you fixed the alignment issue in the vertical drive yet? And did this fix your uneven vertical movement?

Looking at your test films again, the subtracting composite test of the vertical vertical movement is showing exactly the behavior you'd expect with your "earthquake theory". It doesn't shift out of registration and stay out as would happen if your motor missed steps. It goes in and out of alignment as though one pass binds up, then breaks loose ans springs back. And the binding and springing back keeps on happening over and over.

How exactly does the elevator axis work? Are there little wheels or skids where the telescoping square tubes move past each other? Is there any sort of counterweight mechanism? Have you replaced the cylindrical spacers with a block of something where the motor shaft coupler is out of alignment?

Things are so quiet here. Just trying to be helpful.

Author:  Doug K [ Thu Jan 05, 2017 1:48 pm ]
Post subject:  Re: Debugging Motion with Otto

Hey James,
I think he is using heavy duty drawer slides for the vertical bearing. Kind of looks like it. And then at the top of the rig Chris is using a linear slide with a cable of some kind attached to the vertical stage to pull it up. I could be wrong but I think that is how he has done it.

It is quiet........maybe too quiet..........its always quiet just before the Zombies attack!

Author:  Kitwn [ Fri Jan 06, 2017 5:06 pm ]
Post subject:  Re: Debugging Motion with Otto

There are more details of the rig on Chris' earlier thread...


Author:  Biolapse [ Mon Jan 09, 2017 2:07 pm ]
Post subject:  Re: Debugging Motion with Otto

Sorry folks life has been incredibly hectic the last couple weeks, James thank you very much for your thoughtful replies, the idea of the vibrator is fantastic. I believe i have everything locked in at this point, ill try to post a proper reply soon.
Thank you all!

Page 1 of 1 All times are UTC - 8 hours
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group