Cable Tips Tricks, Close-ups for C and C'

[Blake Hament]

Hello Dr. Forsslund,

First off, many thanks for all your support thus far. You've really launched a great project here, I'm looking forward to adding some contributions soon.

I'm reflecting forces with the device and I feel comfortable now with most of the code, electronics, and mechanisms. I would like to get the device to feel less "mushy" and have stiffer response. I've added user interface to change k for modeled spring forces. With very high k, certainly it is a stronger response, but there is still some slipping and force lost in transmission.

I think my weak point now is probably the steel cabling for transmission between the capstans and rotating bodies. I tried to slow down and zoom in on frames from the youtube videos to understand how you ran the cables. I think I mostly got it right, but I am not confident in my cabling for body C and C'. Could you please post some pictures of close-up views of the cabling for body C and C'?

In general, the cables seem to lose tension more and more over time, then start slipping and overlapping. I'm getting better at the cabling, but please do share if you have any good tips/tricks for working with the steel cable. Do you find best performance with ~5 coils around the capstans? I get less overlapping with fewer coils, but maybe there is some negative tradeoff?

Thank you and best regards,

Blake

[Jonas Forsslund]

Hi Blake,

Thanks for the kind words. 

Stiffness is much an effect of k-value in the simulation, and to crank it up without causing vibrations/unstability the update rate has to be high. 

There should not be any slipping - if you have slipping (I assume you mean that the cable is not tight enough so the motor can spin freely a bit) you also lose position, and forces will be rendered incorrectly and virtual object not shown where the manipulandum is in reality. 

Cabling is by far the trickiest part of assembly but also very difficult to video record (for me at least). It is almost a handiwork craft. Can you show pictures or video from your work? Did you get the cables suggested or some other? I have 4 turns on the C'-C and 5 on the other pulleys. Then tension needs to be tight! On C'-C you can pluck the cable and hear the tone. It should be like the highest frequencies on a guitar (maybe I should measure..?). And you may need to tune it / tighten it after a while. Regarding overlap: try to avoid it by loosing tension a little bit and move back/forth or pushing aside the cable with a sharp tool. 

Mechanically design wise this is obviously an area for improvement, especially if is to be used for a longer period of time. For research and short-term use it should work very well though.

Jonas

--
You received this message because you are subscribed to the Google Groups "WoodenHaptics users and developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to woodenhaptic...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/woodenhaptics/1893a204-ba7c-4acd-bd87-fa3c1730c308%40googlegroups.com.

[Blake Hament]

Good tip about the update rate, I did increase the rate already, that helped drastically. I definitely need to increase tension. Usually, after I tighten, the tension is released again within a few hours and I have to tighten more. After a few tightenings, the cable tensioning bolt and block have moved too far and the whole cable needs to be reset.

Please see attached for some pics, a video, and a (very crude, sorry!) drawing showing how I am cabling body c and c'. I am using same steel cable that you suggested in WH BOM. It's very hard to find the cable and turnbuckle that small in the US, so I ordered from TechniCable.

I think maybe I have assembled some of the inner, smaller C cylinder body incorrectly. Currently I thread the steel cable through a cut in the inner cylinder, then I have to put a ferrule at both ends where it leaves the wood so that the cable does not slide through either direction. Then I attach more cable to either ferrule and wrap that cable to the C' body and connect with turnbuckle. I included a drawing, hopefully it is somewhat helpful in describing this.

Any closeup image or drawing of this C-C' cabling would be great for me to compare, thank you!

Attachments: https://www.dropbox.com/sh/dzl9sane247spft/AADyo3Wp1kw7qb59N4IdQHH-a?dl=0

--

Blake Hament

321-693-5294

B.S. Physics, Duke University '14

M.Ed. Secondary Science, UNLV '17

Ph.D. Mechanical Engineering Candidate, UNLV

DASL: Robotics Research