Tuning?

[cmhunt]

Windell,

I love the watercolorbot!

When I first threaded it, I didn’t see the bit about keeping the servo in it’s initial position.  Needless to say, my first picture was a bit funky.

After rethreading it, it worked a lot better, but the carriage seems to lose some of the accuracy of it’s relative position.  Sometimes the carriage seems to stutter along.  Should I lubricate the rods a bit?

I’ve posted my second picture.  Any advice on how to tune the WCB to get it to work a little better?

Thanks!

Chris Hunt

https://www.dropbox.com/s/9v769luae9dteb6/2013-12-26%2013.47.15.jpg

[Windell Oskay]

Hi Chris,

I’m glad to hear about the (partial) success– and we’d love to help you get the rest of the way there. :)

We have generally found that when a WaterColorBot is correctly tensioned for the first time, it will need to have some additional slack removed after about an hour of use.  But after that initial stretch, it will stay tensioned correctly for months on end. So, after running for a bit, check that the rods are square to the chassis, and take out the excess slack in the cord, and that should be *most* of the fine tuning that you need.

If the cord is correctly tensioned and the rods are square, you should be able to turn the winches like knobs (with the power off), as though it were an Etch-a-sketch, or perhaps a bit more easily. The cords should be tensioned enough that they can be plucked like a string from a stringed instrument– such that they vibrate in place, rather than fall down listlessly. If it is too loose, the cord may fall out of the grooves in the bearings. On the other hand, it shouldn’t be so tight that it sounds like a guitar or violin string– if it’s too tight, that could increase friction and prevent motion.

If it’s “stuttering” — I imagine the same thing that we call cogging (jumping steps on the motors) — after checking that the cord is tensioned correctly, then please read through our troubleshooting section about other possible causes:  http://wiki.evilmadscientist.com/WaterColorBot_Troubleshooting

You should definitely not lubricate the rods unless you are finding corrosion to be a significant issue.  The bushings are self-lubricating.

In the picture, it looks like the effective brush width is increasing as you go along, suggesting that the paints are getting wetter and wetter as you go.  I’d suggest pre-wetting the paint dishes with a drop of water each, and perhaps raising the brush up slightly, so that it isn’t pushing down quite as much.

[cyclicredundancy]

I had also noticed carriage stutter/shaking in some cases, which I tried to capture in the following slow-mo video: http://www.youtube.com/watch?v=YR4lfTiA9Qg. Chris, is that similar to what you saw?

Windell should know better, but my initial guess was that it might be a consequence of trying to follow the irregularities in human drawn lines since the carriage otherwise moves super smoothly when translating with the pen up.

I am also going to see if tightening the string and increasing the motor current has any effect.

[Windell Oskay]

To clarify a bit: When we talk about “cogging,” we are referring to cases where the motors are effectively skipping big steps, and losing track of where they are supposed to be– for example if the carriage does not return to the home corner (but goes elsewhere, or only partway back) when you ask it to.  

In the video, there is some “jerky” motion, but it doesn’t look like any steps are being skipped. Rather, it just looks like it’s getting input slowly.  If you’re seeing this in RoboPaint RT  (which is Java based), you may be running up against the speed of the program on your computer. Two things you might try: Press the numbered keys on your keyboard to change the speed (1-9), slowing it down a little, and try sketching the drawing first, with the pause button selected, and then resume, to let the robot paint everything at once.  

[Author]

Posts