I found that my frame was flexing, resulting in a measurement that was 12mm to 15mm shorter than ‘at rest’. The flex was both in the plane of the work area and twisting of the arm caused by the lever arm of the plywood motor mounts. A line stretched between the motor mounts at rest sagged when doing the measurement.
I used a piece of unistrut for a top beam and made motor mounts that extended along that to counteract the twist from the cantilever of the motor. That eliminated the problem for me.
Just a thought - I noticed in the pictures above the 1 motor wire routes over the fan. I would reroute this if you haven’t already. Try running it without the fan. The fact you are getting the same results is actually a good thing. My linkage is on the way. I ordered Logans’ kit like you. As soon as I can I’ll post results. I believe this will be after the holiday.
Has anyone tested or run both linkage kits? I did the manual calibration with no linkage I believe my sled is low as well but my simulated cuts are in tolerance.
So far because I can’t get the darned machine even running properly I haven’t had to run the fan do it’s off and certainly not making interference.
The linkage kit seems fine but again without the machine running properly I haven’t been able to utilize the linkage kit
With the steel cable measurement I was able to do it without sag and I checked for frame flex when I pulled tension and it’s not flexing.
for you to be getting different measurements from the chain to the steel cable
means that either
- the frame is flexing when it pulls the chain tight
- you are loosing encoder steps (but loosing the same number every time doesn’t
sound sane) - the chain you have has a lot of slop in it (not likely)
- the software is using the wrong value for the encoders.
- some other software bug
one thing I just thought of.
during calibration, after the chain is pulled tight, the system backs one motor
off 10mm to give you slack to detach the chain. That sounds like about the
amount that your right motor ends up off.
This could even account for you being a bit low.
@bar after backing off the chain by 10mm to allow it to be more easily detached, does the calibration ever move the motor back the other direction by 10mm (or otherwise account for this shift?)
I only did the measurement between motors using the chain and ground control program once. The number seemed very wrong so I stuck with my manual measurement. The process seemed to work fine, the chain was pulled taught using the software and when it was done it did give slack back in the chain like it’s supposed to. I can’t possibly imagine that everyone out there is using wrong numbers given by this sequence, something must be wrong with my hardware.
Like I said when I carefully put up a steel cable between motors no slack and measure it I get 2882.24mm when I do it with a steel tape measure trying to have no slack I get 2881.70mm which is very close to my other measurement. Then when I do it with the system pulling the chain taught it’s 2859.69mm??? I feel like that’s so far off from my manual measurements I can’t believe it.
Oh and I put up a reference post behind each end of the frame and it’s not flexing any measurable amount
the change to move things back 10mm to detach the chain is new
try doing the distance between the motors a few times and see if you always get
the same results or if the results vary.
If they vary, then it could be lost encoder pulses, if it’s the same distance
each time, something is odd for it to be off by that much.
David Lang
I just measured my distance-between-motors with a steel tape to be 3020mm. The value recorded during the automatic measurement is 3008.17mm. Haven’t gone to the next obvious step of manually changing the value and finishing the calibration, though. I’ve been satisfied with the 0.5mm accuracy I’ve measured in my projects so far.
I also put arrows on the sprockets and ran the chain calibration. The arrows were straight up at the start of the procedure (no pictures). After the first movement, the arrows were symmetric - left and right. After the ‘Finish’ movement, they were still symmetric - left and right.
I find that 0,0 is 1/8" below and 1/8" to the right of the center of my sheet, but that could be within the tolerance used attaching the sheet to the frame 
.
1 Like
In your possible problems list #4 says software encoder value. I’ve never changed mine from default and it is 8148 encoder steps per revolution in GC vers 0.97 What is the value supposed to be?
How accurate do you think you were with the steel tape measurement?
I tried to be very careful in my manual measurements and I think my tolerance of measurement was within 1/8 inch. My manual measurement is 22.5mm (~7/8inch) different than the value recorded during the machine automatic measurement
how much chain does ground control say it fed out for each of the motors in the auto chain feed out calibration?
I didn’t try to go sub-mm, but the 3020mm is good to the mm. I mounted the sheet by myself on the vertical frame, and, well, close enough was good enough for the sheet location. It’s really only there to attach work stock to, anyway. To set up a job I attack the stock in the general middle, then drive the sled to the point I want to start from.
I get that the sled doesn’t end up in the exact center of your sheet and that’s fine but those measurements between motors manual vs auto measure are still 12mm different. That seems like more than your error would be measuring with a steel tape don’t you think?
As a brief review, because I purchased two of these kits I can try swapping parts.
First I swapped out the motors and motor wires and ran the auto chain length feed out. Sprockets both start pointing vertical but are asymmetric after feedout by about 10 degrees. Also made sure nothing is near anything that would make electrical interference.
Next I swapped out the Arduino and motor controller board combo and ran auto chain feedout and still asymmetric sprockets after feedout. Another weird piece of the puzzle is that this asymmetry error seems to be totally repeatable.
This is making me insane I am about to cut my losses and fix it with a hammer…
Grasping at straws here - right motor to MP1, left to MP3…
I think you meant to say right motor to MP3, left motor to MP1. The way you said it is the stock setup.
Either way I tried that but when you do that it moves the right motor when it thinks it’s moving the left motor. The problem there is that the auto chain length calibration feed out is commanded to move the left motor clockwise to feed chain toward the center of the frame and the right motor is commanded to move counter clockwise to feed chain toward center. When you swap the motor wires on the motor control board and hit auto chain calibration they both fall right off the sprocket
8148 is the correct value for the stock motors. I wasn’t trying to list the
probable problems (I tried that earlier), I was trying to list all possible
causes for error, however unlikely/impossible 
thanks for your efforts, I’m just totally stuck
don’t panic yet, give @bar a chance to look at things here.
The fact that the motor is consistantly off by the same amount is significant.
Speculating here, I am thinking that there is a bug where the backing off is
taking place before the final measurement is determined, and the backing off
never gets ‘undone’
Well, I was just assuring that you’re connected to the proper connectors, and of course you are. I’m sorry that I wasn’t clear.
You did give me an idea, though. With the motors reversed (and the chains off), what will the arrows do? Will the 10 degree difference be on the right motor or the left one? Will it be off in the same direction as before or not?
You might be getting tired of all these suggestions - call a halt when yuy’ve had ‘enough’