The math calculates chain length from the real-world top of the sprocket (it
calculates what angle the chain is and so knows how much is ‘used up’ between
the top of the sprocket and where the chain leaves the sprocket).
Since this can be as shallow as 10 degrees from horizontal, and the arm is 45
degrees from horizontal, there are lots of places on the workpiece where the
chain is not in contact with the sled at ‘stright up relative to the angled
motor’
I believe this includes the center of the work area.
So after an evening of testing and head scratching I will try to summarize my total confusion haha.
I did not go through the entire machine dimension calibration. I am only doing the automatic chain calibration feed out.
Step 1: I selected Actions -> Calibrate Chains Length - Automatic and got one tooth on each sprocket pointed 12:00 high perpendicular to the floor not to the motor body. I marked this sprocket with sharpie and also put a piece of yellow tape for the pictures.
Step 2: I hooked the first link of each chain onto that vertical sprocket and started the feed out. I carefully watched and made sure neither of the motors skipped a chain link or wrapped chain on itself.
Step 3: After both fed out I took a picture of the ground control screen which said what each chain measured out 1648.90mm (left chain) and 1648.89mm (right chain)
Step 4: I tediously counted how many links were fed out starting with the link that was resting on the tape marked sprocket tooth that was originally straight up 12:00 high and ending with the last link that gets hooked to the linkage kit with the cotter pin. I did this for both sides and counted the same amount for each side 261 links. This led me to believe everything is the same, the computer says it fed out the same amount and the actual counting of chain links were the same amount.
Step 5: I took a picture of each sprocket that was originally pointed straight up 12:00 high and were now both pointed at an angle toward center to see if they were “mirrored” positions of eachother. They are not mirror images.
It may be a little difficult to tell from the picture but on the right motor the marked tooth is angled 36° off of vertical. On the left motor the marked tooth is angled 46° off vertical
When I hook up the sled and say return to center the center of the bit is 1/2" below the actual center of the wood board.
If anyone has any advice please help.
For what it’s worth - I used the top motor screw as 12 o’clock. Drilled an extra 5/16 hole in my sled on right of the router to mark it’s movements with a pencil. After running calibration, I drew an outline around my sled. I then tested different movements and “Homed”. My sled always goes back to the same outline. Hope this is helpful.
That sounds as if you may be missing steps from the encoder wires collecting interference
That isn’t a big difference, but it’s enough to count.
I believe that the calibration step lets you feed out additional chain. I would suggest that you disconnect the chain and feed out 1000mm on each side, possibly getting a video of it to see if there are any hiccups as it’s feeding. See if the two sides move consistently.
It looks like though logic to move to 0,0 at the end of chain calibration in the full calibration sequence, it wasn’t added to the end of the the automatic calibration on the Actions page. I’ll add an issue to GC to capture it, GC issue #460.
@wjcarman, I’ve created PR#460 to address this. If you like, you could try this version of GC to see whether that moves to the correct position with the arrows as expected when the ‘Finish’ button is pressed.
Thanks so much for your thoughts and efforts.
After I took my pictures I hooked my sled back up and hit actions ->return to center. This moves the sled and the numbers in ground control say X0, Y0. The x seems to measure within 1/16" of center of my 4x8 workpiece but this is where the Y measures 1/2" low.
Thanks for identifying a hidden issue.
You could get that half inch of Y by editing the value of ‘Motor Offset Height in MM’ on the Maslow Settings page. Add 12.7 should move it up.
There was a spate of discussion earlier this year about whether that dimension should be measured from the center of the motor shaft or from the top of the sprocket between teeth. Can’t remember the consensus, but 12.7mm is in the ballpark of that amount.
To be honest I’m still a little confused as to what issue I helped identify.
I realize I can just change the motor offset height but I didn’t think I was that far off in my measurement to begin with so it didn’t seem like the right thing to do
I am still kind of new to this and I tried to click your link for PR#460 where you hyperlinked the words “this version” but it takes me to a page https://github.com/blurfl/GroundControl/tree/issue-%23460 and I’m not quite sure what to do when I’m there to download a new version of ground control. It seems like it’s all of the code and subroutines that make up the gui but I don’t know what to download.
I routed all motor wires away where they should not have any electrical interference and I re-ran the auto chain length calibration feed out. It produced the exact same result as shown above in the photos. Ground control says the same amount of chain is fed out on both but the initial vertical tooth on the left motor is rotated over ~10 degrees more than the right motor (not mirror images of one another as they should be)
Because I bought two maslow kits I have two other brand new unused motors. I decided to swap them in and try the whole thing again… same exact result. I’m totally stumped. I still never figured out what hidden issue blurfl thinks I identified. When I hit home after the chain lengths are fed out and then I re mount the sled it’s still 1/2" low from the center of the workpiece where it’s supposed to be (XY 0,0) I realize I can just fudge the location by changing the motor offset height but I carefully rechecked that measurement to the center of the motors and i’m certainly not off by 1/2 inch.
Use the green ‘Clone or download’ button to download the .zip of all the files and unzip it. To run it, open a command line session in the GroundControl folder and depending on which platform you’re on, use the command ‘python main.py’ or ‘kivy main.py’ to open the program.
That said, please hold off for now, I need to make sure that my change hasn’t introduced a new problem .
If the measurement is supposed to be to the top of the sprocket, that is about a half inch more. Thinking about how the measurement is used, the top of the sprocket better represents a point on the chain path than the center of the motor does. I’ll put some tape arrows on my sprockets tomorrow and repeat the sequence here, maybe I’ll see something that gives a clue.
I held off on downloading your code as you requested.
I realize that that the measurement to the top of the sprocket is another 1/2 inch but the instructions within ground control say to measure to center of the motor
As a note, I just did the calibrate machine dimensions action to measure the distance between motor centers using the chain (and it gets pulled taught) and it gave a value of 2859.69mm
To check that value I used a steel cable (1/16" diameter) and captured the length from center to center of the motors. I then used some fancy equipment at my work which has pins at known distances apart on a large machined metal plate and it says the length of cable I made was 2882.24mm That is a tremendous amount different and thus I don’t trust the chain auto measurement at all.
Has anyone else done verification measurements like this?
If you are loosing encoder pulses, that would account for this error, and being off that far would be a problem.
Since you have two kits, swap out components (motors, cables, interface board) and see if you can find a bad component.
either measuring between the motors, or running both motors the same distance and seeing if the sprockets match should be enough to see what’s happening.
.