KAPLAH! Anchor locator code cracked

After many many hours of pure frustration, and exhausting all other avenues, I finally cracked the infinite fitness loop code.

It was… the Belt Lengths.

So, from the Calibration video, Bar says his belts are 1720. I have a wooden frame but I first tried that, and too short to reach the corners.

Next tried 1740. Still too short, so I bumped up to 1780 and they fit, so I went with it, for months and months and months.

Wrong!

Finally, after reading a post about slack in the belts, and getting wildly low, demoralizingly low fitness results, I decided to change the one thing I’d not changed, along with entering my dimensions, 2960 x 2430 into the configuration page, and shortened the belts to 1770

There was still room for more so I went further to 1760 which took up slack in the lower belts, vertical orientation, but they still fit easily over the corner anchors. I ran the anchor locator and finally got results in the 0.45 range before the machine seemingly rebooted.

After restarting and further lowering to 1755 I was getting better results in the 0.75 range, before another spontaneous disconnect, but I was encouraged.

I dropped the belt lengths to 1750 and restarted the calibration, and it resumed at waypoint 6, ran to 8 and then rebooted.

I then dropped to 1745 which meant I had to really crank the belt end onto the anchor, and the fitness returned to the 0.1 range infinite loop. Too far.

So, back up to 1750, double checked the configuration page entries and fixed them back to 2960 x 2430, (from 3500 x 1400ish), saved, rebooted, did everything by the book, and kaplah! I was getting results in the 1.4 and climbing range on a 2000 x 1000 , 9 x 9 grid. It ran to completion, and I’m onto the next phase.

There was a slight hiccup on rhe last pass where the top left belt was slightly deflected by an obstruction, so I’ll run it again in a few days, but that was the code I needed to crack.

Now, a couple of observations.

Before all of this began, one of my belts spontaneously began to consistently stop around 12 inches before the end. I had to bump the retraction force to 2000 to overcome this, before realizing there was a piece of cobweb with sawdust trapped in it caught on the belt at that spot. Once clean, I could comfortably use 1300 for both the retraction and calibration forces, which is what I used to calibrate in the end. So check and clean your belts and arms for debris!

Next, I wish that the belt lengths had been stressed in the video. To me, this one thing above all else made the difference between sweet success and humiliating demoralizing repeated failure. As happy as I am to get beyond this point, and unless i missed the memo, it seems too simple an issue to have caused me and no doubt others that much frustration and grief. There’s clearly a narrow operating length which must be found, and I personally now cannot stress it enough.

I would appreciate if the video is redone with the changes to the firmware to show the addition of the dimensions into the configuration page in fluid NC. I’d been vaguely familiar with the page but had never paid as much attention to it as it warranted. My bad, but I don’t know if the machine would have figured it all out on its own. Perhaps one day when I’m bored, I’ll find out.

Disconnections. The machine disconnected at least once on the final attempt, but whereas previously I’d assumed it was another failed procedure and pulled the plug, the fact that an earlier disconnection resumed at waypoint 6 led me to wait, even though the maslow startup message appeared. It resumed, continued, never stopped, whatever… it was still doing its thing, and mightily, so be patient and allow it to resolve and impress, or absolutely unequivocally fail.

All of this still followed an upgrade to 4.1, hours of sanding spools to fit the new arms, and reading the forums, but all of those frustrations aside, I hope this post will give hope to those on the verge, including our friend returning to his lowrider. You had me thinking about it, trust me.

4 Likes

Yay! Glad you found a magic pathway! What project are you going to cut?

Very interesting, I don’t think this has been noted anywhere before as being a problem/solution, although there is a calculation to determine optimum belt lengths, it was more to avoid having too much belt out for esthetic reasons. Nice observations and follow up.

Excellent write up and great persistence!

Out of curiosity what version firmware is your machine on?

Dano

Yay indeed.

I’ve got a few things waiting to go. I have to make some boxes, signs and a special project I can’t name yet but, all being well, I’ll be able to post some details in a few weeks.

I’m cautiously optimistic that the cuts will go well… we’ll see, let’s hope.

2 Likes

I’m running 1.15

I looked at some of the clues, like, it originally somewhat worked back last September 2024, but then after a certain update, it never worked again.

So, after rebuilding, I rolled back to 0.84 to see if I could calibrate again, and I could.

I tried to update to 1.15 and keep the 0.84 yaml file, but that quickly failed.

I then thought that just updating the dimensions on the configuration page was the missing link, but that too failed.

All that was left was the belt lengths, as far as I could tell

Thanks.

I read as much as I could on here to make sure I’d not missed anything, and I didn’t come across the specific steps, but I’m curious to know now if this fixes it for anyone else, or if those who have had success focused their attention on ensuring accurate belt lengths.

wobo wrote:

I tried to update to 1.15 and keep the 0.84 yaml file, but that quickly failed.

yes, the file changed significantly at 1.0 and 1.9

This is the first time we’ve seen a report of changes based on how much belt was
extended.

David Lang