Yes. I double checked. With 0 in Compensation was 0 in X 5mm in Y but reentered Compensation from previous and way off. Is it possible that my board is faulty?! I never dreamed it would be this hard to get reasonable results.
can you post your groundcontrol.ini file? Getting a 900mm square to come out 1888mm wide is really surprising.
Sorry that wasn’t the square. That was the Maslow benchmark file that has the 4-100mm corner squares and should be 1905X and 900Y
So its -17mmX and +13 mmY
Oh, good. So the chain compensation are “experimental” at this time, but it helped me get to good results. There’s an issue with using it during calibration, that’s why I suggested the steps I did.
So, what I did to get to where I am (which I think is pretty good) was to manually increase the rotation radius to around 138 or 139. Try upping that value to see if you get improvement.
Should I put chain compensation back to 0 first?
And again Thank you sooooo much!
No, keep them there… just try to up the rotational radius.
No guarantees… just the steps I used.
My opinion is that we shouldn’t be calibrating the rotational radius. We should be able to come up with an accurate measurement (as a community) and just use that value and calibrate other things that are less fixed… chain wear and chain sag.
I’m back on shift now and won’t be home until Friday evening but I’ll update as soon as I can.
Good luck… I’ll try to keep an eye out for the update and not let you go 21 hours without a response. I’m very interested in trying to get chain compensation to work.
1 Like
I have done a lot with trying to dial in my Maslow and would like to help but I am having trouble following what you are doing.
Have you read this Post by Bar to make sure you are measuring your benchmark correctly?
I think it would be easiest for us to help if we could visualize your results.
To start I would suggest you:
- Wipe EEPROM and delete your groundcontrol.ini
- Run the full calibration
- Post your groundcontrol.ini so we can see your values after a calibration
- If your groundcontrol.ini values look as to be expected then run a benchmark
- Post your benchmark results in a similar fashion as above
- We will suggest a change based on your values
- Copy groundcontrol.ini and name it groundcontrol.ini.bak
- Make changes suggested
- Run another benchmark and post for us to compare with the original benchmark
- If the results did not get better. Revert back to groundcontrol.ini.bak
- Repeat steps 6-10
2 Likes
Yes I have refered to this post numerous times, even book marked it. LOL
Me too at this point! 
Very good suggestion to start over. I was thinking the same thing since I discovered yesterday that the tape I was using was off by 2mm when I stretched it over my steel meter stick. Now I don’t know which measurements were done with my good tape and which with the bad. 
A couple questions: 1)when getting motors to measure distance between motors should I use left chain on left measure and right chain on right measure because last time I used the same chain for both? 2) when doing triangular calibration cuts (which was suggested to do twice) should I put the numbers in, hit enter values then go back in and run it again or should I put the numbers in NOT hit enter values but instead hit cut test pattern again then enter the new measured values and hit enter values?
Do you also think its a good idea to start from scratch @madgrizzle?
Personally, I’d try tweaking the rotational radius higher a little to see if accuracy improves before I start all over. It worked for me.
As I understand, right now you have a distance between motors entered based upon a tape measure that you think is off by 2mm. I’d consider fixing that value by rerunning the chain compensation routine. You should use the left chain to measure from left motor to right motor and the right chain to measure from the right motor to left motor. This gives you compensation values for each chain.
put the numbers in and hit “Enter Values”. It will “iterate” over different values to come up with the best ones to use. After done, go back to Actions->Advanced->Run Triangular Test Cut Pattern and it will bring you back to where you can do another set of cuts.
One more thing… don’t do a calibration with chain compensation factors. All calibrations should be done with those values equal to 0. That’s why I suggested manually upping the rotational radius… can’t do it with chain compensation factors enabled.
@madgrizzle may have a better understanding of what your machine is doing since he has been helping you out from the beginning. I am coming in late and reading posts that makes it seem like you have never had a solid baseline and are getting frustrated from a trial and error approach.
If it were me, I would run through the calibration as is. You should be able to get decent accuracy by doing this. If we find your accuracy is way off after doing the calibration as intended then I would think there is something off in your build or the values you are inputting.
Same goes for checking chain tolerances. I don’t think this is a necessary step for you at the moment. I don’t believe anyone has actual data that shows this has helped accuracy yet. I think it has actually done the opposite. However, once you get to a point where you are comfortable with the accuracy, then we can play with chain tolerances to get more data so bar can get a larger sample to help figure out how to make the chain tolerances work.
For me, from what I remember, running the Triangulation Test Cut Pattern more than once without resetting my ini file gave me incorrect chain sag and rotation radius values (almost like the numbers compounded). Similar to the numbers you posted before
I would expect to see a rotational radius around 135-140 and a chain sag 25 - 30
Those values came about, I believe, because he performed calibration with chain compensation values entered… which (for unknown reasons) you can’t do.
I sensed from early on he had run out of things to do to calibrate, so I suggested the chain compensation values as an option. I currently use them to “tweak” my calibration and am getting better results than without.
I’ll try to post my values tonight when I’m home.
Awesome! I think this would help a lot of people.
I don’t doubt that what you did helped with your accuracy. I’m just wondering if you accidentally got there or if we could duplicate your process and get better accuracy then with the automated calibration. Thats why I thought if we could get @xrok1 at a baseline then we would know which “tweaks” are actually helping.
When you say chain compensation values are you referring to the chainsagcorrection value or the rightchaintolerance and leftchaintolerance values?
- I would love to see what your ini values were after running the full calibration
- What your benchmark was after the initial calibration
- What adjustments you made and the ini values after making adjustments
- What your benchmark was after the adjustments
For example:
If you run calibration and you get a benchmark error average of 3
Rotational radius of 135
Chainsagcorrection of 28.5
Then you run the triangulation test cut pattern and get a benchmark error average of 6
Rotational radius of 123
Chainsagcorrection of 52.8
Then you up your rotational radius to 140 and get a benchmark average of .3
Rotational radius of 140
Chainsagcorrection of 52.8
If others could repeat that success, then were on to something.
Edit: these are not actual numbers or the process used. Just a “template” or my opinion on how we should post our results and processes.
I tend to agree with @ScrumdyBum on this one. I’m kinda lost in the process at this point too. Maybe I could reset and try both directions with guidance from each of you. Hell at this point I might as well be an experiment so help others and hopefully the overall process. Lol
1 Like
Myself, @Dustcloud, and @arnoldcp were all in the same boat at right about the same time. We kept testing documenting and sharing with the team in hopes someone with more experiance could shed some light. It kinda fell to the side as I started doing projects that didnt quite need the most accurate cuts but I am only off by .05 in in the Y so I have been working that into my designs. I’m glad you are still trying because I’m excited to try and finally nail this down. Maybe when @madgrizzle posts his results we will have another Avenue to try.
1 Like
I’ll post what I have and recall tomorrow…
@xrok1, don’t feel that you’re abandoning me if you start from scratch… chain compensation is highly experimental and is a bit difficult to work with. Maybe I got lucky… don’t know.