Likewise - in permanent installation where the machine stays connected, it would still be best to power down the equipment between uses for safety (which for me could be weeks or longer).
2 Likes
Yea, I think that is what it is for the most part. Also an old, probably dull straight bit… I won’t bet cutting OSB much as it is not a great material. Cheap and solid, but not anything I’d use for any “real” projects.
Ron Lawrence wrote:
Yea, I think that is what it is for the most part. Also an old, probably dull
straight bit… I won’t bet cutting OSB much as it is not a great material.
Cheap and solid, but not anything I’d use for any “real” projects.
that looked to me more like the sled wasn’t sliding than the router bit not
cutting, but a dull bit could contribute to that.
David Lang
1 Like
I tried pulling it out from the wall on the bottom, and also putting a few scraps of 1/4 ply on it to see if it still did this, and it was much better, but still a bit of a tip. I’ll have to think about how I want to handle getting this frame to slide out on the bottom when in use and back up against the wall when not in use.
I did notice another thing today though… was trying to jog z when I was moving it around… Should it be doing this on z jog? I think its trying to retract all the arms and I’d think on Z jog we would not want that? I don’t think it does this when running gcode, just on jog…
1 Like
As best I can tell from the video, the upper belts (which are also higher on the Maslow’s body) angle down towards the work surface as they approach the belt end mounts.
This could be crazy, but does the belt force, angled down in this manner, tip the Maslow up onto the upper edge of the sled? Would it be better if the belts were parallel to the work surface, or even angled up slightly?
1 Like
There is certainly a component of that involved in other of my issues. I was thinking about making anchors on the top extend out another 10 cm or so to address some of that, but for this, I think when jogging z, I don’t think the arms/belts should try to move at all.
1 Like
Yup, that’s actually totally normal. Basically what you are seeing there is that while any axis is moving it will position hold on the other axis. Ideally I think that we would like it to continue position holding for like 30 seconds afterwards too, but fluidNC supports a maximum of 250ms so we might need to change some code in there.
Curious why you would want to lock the other axis for 30 seconds? does that mean after a Z change you need to wait to jog X or Y?
1 Like
It’s not that they are locked so much as position holding. If you click a jog button to move X or Y they will move, it’s just that the servos are powered up during that time and then after some waiting period they will go to sleep again.
yea that makes sense. would be good to lock without having to relock especiallly when the sled seems to shift slighlty as the lock is cycled. should I log this as an issue in github?
1 Like
Absolutely, that would be great!
The more clear that you can describe the behavior that you want to see the better.
NEW DATA:
I did another calibration last night 4x4 @ 1000x1000. I am not seeing the same results in the new code vs the calibration-simulation project, as I mentioned in the other thread. Here is my data.
My new data 4x4 points:
[{bl:1701.95, br:1715.11, tr:1731.40, tl:1766.19},{bl:1395.45, br:1408.62, tr:2113.93, tl:2145.21},{bl:995.26, br:1862.00, tr:2437.63, tl:1895.80},{bl:1299.23, br:1972.32, tr:2214.82, tl:1599.15},{bl:1541.08, br:2199.50, tr:2022.78, tl:1315.18},{bl:1838.71, br:2418.50, tr:1870.89, tl:1063.43},{bl:1988.19, br:2188.74, tr:1569.26, tl:1321.84},{bl:1715.79, br:1946.80, tr:1747.76, tl:1531.63},{bl:1465.31, br:1733.19, tr:1966.76, tl:1779.54},{bl:1249.61, br:1561.21, tr:2214.70, tl:2051.81},{bl:1542.26, br:1268.63, tr:2022.00, tl:2246.89},{bl:1717.98, br:1479.73, tr:1747.24, tl:2001.21},{bl:1936.81, br:1728.03, tr:1496.29, tl:1784.33},{bl:2179.86, br:1997.61, tr:1283.37, tl:1607.86},{bl:2406.35, br:1844.74, tr:1026.63, tl:1909.45},{bl:2183.48, br:1546.80, tr:1282.87, tl:2060.06},{bl:1991.85, br:1264.53, tr:1568.30, tl:2250.52},{bl:1841.70, br:1010.24, tr:1869.65, tl:2471.48},]
Old data which gets good fitness on sim (10000x), but not new esp3d code (same frame less points):
[{ bl:1705.23, br:1753.09, tr:1726.77, tl:1730.06}, { bl:1385.03, br:1442.87, tr:2109.66, tl:2112.47}, { bl:998.81, br:1888.74, tr:2433.00, tl:1869.06}, { bl:1868.72, br:2454.73, tr:1866.00, tl:1028.03}, { bl:2409.12, br:1891.97, tr:1022.12, tl:1868.97}, { bl:1815.98, br:1049.75, tr:1866.09, tl:2435.32},]
1 Like
Another run. This time 300x300 with 3x3 points. Stock 0.68 firmware and UI. The calculation takes a looong time, but it is finding a good fitness (80000 and counting, 0.68,) . I will try 500x500 5x5 points next. I’m going to step away and let it finish.
CLBM:[{bl:1701.36, br:1721.41, tr:1732.31, tl:1760.87},{bl:1606.42, br:1626.25, tr:1831.36, tl:1858.66},{bl:1488.68, br:1756.80, tr:1943.43, tl:1751.23},{bl:1591.00, br:1843.45, tr:1850.40, tl:1647.32},{bl:1694.76, br:1929.88, tr:1765.27, tl:1551.05},{bl:1796.87, br:1814.41, tr:1641.24, tl:1671.23},{bl:1700.13, br:1719.59, tr:1732.45, tl:1761.00},{bl:1606.20, br:1626.29, tr:1831.35, tl:1858.62},{bl:1733.03, br:1508.69, tr:1725.07, tl:1971.34},{bl:1821.26, br:1609.30, tr:1619.69, tl:1879.59},{bl:1911.08, br:1709.51, tr:1521.84, tl:1795.89},]
EDIT: finished:
Fitness: 0.6927636266951753 in 107200
Calibration complete
Calibration values:
Fitness: 0.6927636266951753
Maslow_tlX: -61.5
Maslow_tlY: 2226.1
Maslow_trX: 2615.6
Maslow_trY: 2223.2
Maslow_blX: 0.0
Maslow_blY: 0.0
Maslow_brX: 2614.3
Maslow_brY: 0.0
These values have been automatically saved for you.
You MUST restart your machine for them to take effect...I know that is annoying, it's getting fixed ASAP.
1 Like
As threatened, here is a 500x500 @ 5x5 (back to bad fitness) Fitness: 0.3500917713064714 in 3600
cal-04-13-c2.txt (9.1 KB)
As a sanity check I measured the distance between my bottom left and bottom right anchors (assumed to be on the same line, y=0) and its pretty much right on 7’ 11 3/4"’ which should be 2432mm but even my “good” fitness number for lower right x is 2614. I’m pretty sure even my tape measuring skills are not that far off… even a cm or so I’d believe, but not that much. I am not sure how else I can help at this point. The discussions around this are very interesting to me but I don’t have the math to help much beyond gathering data and making inane observations 
So did it get to 0.68 and then get worse again?
It looks like it got better after that, ending at 0.693. The lower fitness was for the next run on a 500x500 @ 5x5
2 Likes
Today a 4x4 @ 400mm square
When run with all the points, it gets a .467… fitness. when I put it in the calibration UI and remove point 2 (bottom), it gets up to 50.5 or so then starts going back down from there. Might point out something in the algorithm? (it kept going down from 50, also at that fitness the bottom side length – brx – is still over 200mm off from my measurement from bl anchor to br)
Please let me know if you want me to try anything else. I will see if I can spend more time staring at the algorithm some more and see if I see anything.
CLBM:[{bl:1695.98, br:1710.55, tr:1738.50, tl:1768.10},{bl:1569.54, br:1586.77, tr:1874.14, tl:1902.02},{bl:1410.82, br:1762.69, tr:2020.23, tl:1765.05},{bl:1505.19, br:1835.45, tr:1935.47, tl:1667.07},{bl:1599.61, br:1910.91, tr:1856.39, tl:1574.56},{bl:1696.06, br:1990.59, tr:1783.74, tl:1487.95},{bl:1780.12, br:1886.20, tr:1671.51, tl:1593.68},{bl:1688.95, br:1803.47, tr:1748.85, tl:1674.72},{bl:1601.59, br:1722.57, tr:1832.52, tl:1762.12},{bl:1514.31, br:1645.73, tr:1921.93, tl:1855.06},{bl:1627.88, br:1531.71, tr:1827.66, tl:1950.03},{bl:1706.84, br:1616.62, tr:1733.63, tl:1861.76},{bl:1789.70, br:1704.52, tr:1644.84, tl:1779.21},{bl:1873.07, br:1792.14, tr:1562.61, tl:1703.15},{bl:1976.42, br:1704.08, tr:1457.73, tl:1815.94},{bl:1896.27, br:1610.70, tr:1545.64, tl:1887.48},{bl:1816.41, br:1518.91, tr:1639.68, tl:1965.39},{bl:1743.05, br:1428.68, tr:1738.87, tl:2049.18},]
In the CLBM line, I see 11 sets of belt lengths given, though a 3x3 should only need give 9 points. What are the two extra needed for?
So I finally got my Frame built tonight. Anchor points and Pins are in. Next is Calibration. I know the website isn’t 100% current with the Calibration process. So I want to make sure I am doing this right. Am I correct to say we not longer update the Yaml file for the Frame size it’s in the Fluid Config Items tab? My Belts are fully retracted, So I am going to have to extend all, and connect them to the 4 corners. Once that is done, do I take Slack? Or just press the Calibration button, and It will automatically take up Slack?
Just trying to keep up with the Updates Versions to make sure I am getting this correct.