Finished my build and frame. Now trying to calibrate, but I am having issues. I have updated to latest firmware (0.67) and latest yaml and index files. When I try to calibrate, it doesnt seem to release the top 2 belts far enough. The anchors JUST reach the anchor points and the sled is not centered on the 4x8 board. And then, when I click calibrate, it moves to the bottom as is should, but then when it moves to the left, it slides all the way off the board and fall into the frame. I have built the frame to exact specs on the website. It feels as though some of the info in the YAML file is not right? Frame size? Calibration size? Any ideas?
I think that this is your culprit. Try increasing that a bit and you should be good.
There should be a comfortable amount (several inches) of slack when those are extended.
So I measured anchor point to anchor point and converted to mm. I had to change the values to 2946.4 and 2425.7 respectively. And now is seems to be calibrating properly. Why is this different than what comes in the updated YAML file if I am using the default wooden frame? Shouldnât this be set in the YAML file for the default frame?
That is compared to 2924.3 as the value that was already in the .yaml file?
That sounds like normal variation between our frames to me, theyâre hand made out of 2x4s so a small difference like that seems reasonable.
I guess I mean the 2425.7 number was orignally 2000. Seems like a big difference. But anyway, I post results when calibration is finished. Thanks Bar.
1 Like
Now one thing I have noticed, is during the intial calibration process. It first moves to the bottom, then to the left. When it moves to the left, it lets out way too much slack on a couple belts which causes jams. I know in your video it says you may have to tend the belts some. But this is something I think could be fixed in software. It shouldnât let out that much or that fast during the initial movement left during calibration.
I agree, I think that we can improve it. Itâs a hard problem to solve though because there is a bit of a chicken and the egg thing where we canât know the right amount of belt to extend because we donât know the anchor point locations yetâŚbut we donât know the anchor point locations because we havenât run calibration. I think that there is some clever solution that we havenât come up with yet.
would callibrating in a spiral work, starting in the middle?
if the next location is close by, there is no need to spool out a lot of belt, and after the first little round (4points?), maybe a preliminary callibration can be made before the rest of the locations are visited
2 Likes
Thatâs an interesting idea. I wonder how the total distance traveled would work out vs the zig-zag. It could be an improvement
Bar wrote:
I agree, I think that we can improve it. Itâs a hard problem to solve though because there is a bit of a chicken and the egg thing where we canât know the right amount of belt to extend because we donât know the anchor point locations yetâŚbut we donât know the anchor point locations because we havenât run calibration. I think that there is some clever solution that we havenât come up with yet.
I think you know which belts may extend, can you put those belts into the same
mode as the âextend beltsâ button does, where they feed out belt as needed,
keeping just a tiny amount of slack and when that slack is pulled, feeding out
more?
David Lang
1 Like
Thatâs exactly how it works now 
The problem is that we can count on a human to keep pulling as fast as the belt is extending. As the machine moves from location to location the belt is pulled out more slowly than it is un-spooling leading to extra slack building up.
Got everything calibrated and working now. But I did notice at the end of the calibration, there was a warning that said something like âWarning, fitness values too low. Do not use these valuesâ. But everything seems to be working. Is this something I should be concerned about?
1 Like
Bar wrote:
Thatâs exactly how it works now
The problem is that we can count on a human to keep pulling as fast as the belt is extending. As the machine moves from location to location the belt is pulled out more slowly than it is un-spooling leading to extra slack building up.
shouldnât it stop unspooling when there isnât tension on it?
David Lang
Also, I keep getting and Error MSG for âTop right axis exceeded 1mmâ But its usually exceeded by .06 or less. Should I re-calibrate?
1 Like
Yes, I would be very careful using those calibration values.
Basically it didnât get a very good fix on where the anchor points are so your belts could end up being too tight or too lose while moving which could break something or lead to a belt getting sucked into the gears.
Iâm working on a fix right now which might fix that.
This one is pretty normal. Thatâs because the PID controller needs to be tuned, but we havenât had a chance to do that yet
Ok thank you for the quick responses.
1 Like
So I upgraded to firmware v0.68 and am having issues calibrating. In the attached pic you can see my settings. My frame is built to exact spec listed in the âframe videoâ and documentation. My anchor points are 116 inches apart on the long axis and 95.5 inches apart on the short axis. But when I enter these settings and try to calibrate in the vertical orientation, the belts tighten with the sled too low on the spoil board. So when its starts to cailbrate, on the first move down, it moves too low and falls off the spoil board.
And why do all the new files have the frame size and 29xx by 2000 something? 2000 isnt even close to the specs for the frame that you guys give documentation for. If the frame is 8x10 the numbers should be similar to mine above, correct?
And Bar, I think you may need to post and video showing your exact calibration steps with the new firmware. Including how you measure between the anchor points and which values go where. You need to be very specific, I think. If its working on your end, but most people are having issues, something is getting lost in translation.
1 Like
So I thought about this for a minute and I was like hmmmâŚyou are right
So I busted out a tape measure and checked and it looks good.
Then I was confused for a while and finally measured the length of my supposedly 8â 2x4s and they are 82 inches long.
I am 100% certain that I did not cut them, but I canât find any reference online to 82 inches being a standard length of 2x4. Iâd like to go over to Home Depot and see if they are selling them listed that length.
Either way, you are 100% correct. My frame is decidedly smaller than specified.
The good news is that you should be able to put in the dimensions of your machine for any size frame and have it work, so I donât think that is the issue.
To fix this I would try shrinking the numbers that you have starting out for Machine Width and Machine Height to keep the router on the board.
Another option could be to run a really small calibration grid (something like 3x3 points and 300x300mm) near the center of the board to get a rough estimate of the anchor points and then a larger calibration using those numbers as the starting point. Thatâs how Iâm planning to make the calibration process work pretty soon.
This is top of my todo list. I have a couple more firmware improvements that I want to make before filming the video so that it isnât out of date immediately, but I 100% agree and Iâm working on it.
This one is pretty out of date at this point:
Will changing the numbers of the machine size effect cuts in the future since it thinks the machine is smaller than what it really is. I want to be able to cut the full size of 4 x 8 sheet of plywood all the way to the edges if needed. Or are you saying to shrink the size for the calibration and then change it back to the full size after calibration?
And to answer your question a full 8 foot 2x4 should be 96 inches, no matter where you buy it from.