6 Days, ONE calibration (but awful cuts, so I started over)

Troubleshooting
,
1

Aloha,

On Tuesday, 4 June 2024, I began what I hoped would be a relatively smooth deployment of my Maslow 4 in the theater where I am designing and building sets for the summer musical.

Six days later, I’ve had one 7x7 calibration, but when I started a money cut (test cuts seemed fine), the circle the Maslow was cutting ended up ramped on one side. So back to first principles.

I’ve tried:

  • many different retraction thresholds from 1300 to 2100
  • lowering the fitness threshold to 0.4
  • Built a new “standard” frame this morning (my hypothesis being that my “modular” frame was too flexy)

Nothing seems to have made much of a dent, and I really want the Maslow to come through; we’ve got a build schedule, and after today it starts slipping.

I just finished another calibration; I reduced the calibration and retraction thresholds to 1300 and reduced the calibration grid size to 1500x750 and at waypoint 80, it returned 0.39.

Can anyone help me understand what I’ve missed?
Maslow-serial 202406091535.log (16.5 KB)

1 Like
2

Measure the distances between each anchor point to get top/bottom (i.e. ‘width’) and left/right (i.e. ‘height’) pairs of measurements.

Note: (you’ve probably figured this out already, but just in case) don’t bother measuring your frame dimensions as they are in fact irrelevant, it’s the anchor points that matter.

Use the average for top/bottom as ‘Machine Width’ and likewise average of left/right as ‘Machine Height’ in the Setup ‘Config’ dialog.

Do the retract-> extend thingo and remount the M4.

Now when you do the Calibration, rest your finger on the top of one of the anchor points as the cable goes slack and then retensions while the calibration is performed. If you are doing a ‘vertical’ orientation this will be either one of the bottom anchor points. If horizontal, well it moves around.

Why the ‘rest your finger’ test?

This is so you get a really good idea of how much your anchor points are moving. If this is less than 1mm from ‘relaxed’ to ‘tense’ then you should be good. Any more than that and you need to fortify/stiffen things up, whatever your frame and anchor points require.

3 Likes
3

Huh. The slack seems considerable (visible from a few feet away) during calibration—horizontal orientation—but the new frame especially seems plenty rigid. I wonder if the 3D printed anchors could have more compliance than is good for me.

4

David Negaard wrote:

Huh. The slack seems considerable (visible from a few feet away) during
calibration—horizontal orientation—but the new frame especially seems plenty
rigid. I wonder if the 3D printed anchors could have more compliance than is
good for me.

if you can see slack from a few feet away you aren’t going to get good results.

you talked about increasing the retraction force, did you increase the
calibration force as well? (no, I don’t know why the GUI doesn’t set them both)

David Lang

5

What I have observed is that the belts stay taut during the moves from waypoint to waypoint, then slacks them after it arrives? It’s done that every time I’ve run calibration, and that slack is the eyeball visible slack.

6

If we’re talking calibration. Then the 3x3 grid is super slack, and the slack gets pulled up for the measurement.
The remaining grids have a small amount of slack, before movement, which you see when they stop, and then that is pulled up for the measurement as well.

The going slack, moving, and then tightening back up before measuring, is the calibration process.

But if it’s going slack for regular movement, then that’s not right …

1 Like
7

So I changed the anchors for through-bolts and did everything I can think of to tighten things up. Also trying some higher thresholds. I was going to attach a time lapse is the first 8 waypoints of the calibration currently underway, but apparently it’s too big. Fitness for the first 8 was 0.8086 with 1800 retraction and calibration thresholds. Pass or fail, this is the last calibration of the night; I’ve been at it more than 13 hours (not counting the early morning Home Depot run).

I’ll be back in the theater at 7:00 AM Hawaiʻi time, and on the forums before that.

8

Aloha,

I’ve done a number of different iterations of frame sizing; I have taken to doing what you suggested as of a couple of days ago.

At this point, I’ve rarely had opportunity to observe normal moves; can’t get past calibration. My latest frame is built like a tank and I had high hopes for swapping out the 3D printed anchors for through-bolts, but I see little difference in symptoms: first scores (after waypoint 08) are comfortably over 0.5, then they decline until by either waypoint 48 or 80, they’re in the 3s.

I feel like (hope) there’s something straightforward to address this, but I’ve tried everything relating to calibration that I’ve found in the forums, and I’m missing what I’m missing.

I want to mahalo everyone who’s chimed in, though; I trust that hopefully sooner rather than later, something will work.

1 Like
9

Is there any chance that you could get a video? I’m having a hard time picturing what is going on and a video could help us to spot something

10

This is a time lapse of the 3x3 portion of last night’s calibration attempt.

1 Like
11

Also, quick question: How can I extend the full length of the belt so that I can make sure it’s wound cleanly on the spool?

12

What does the terminal print out when you click the test button.

1 Like
13

Unfortunately that looks 100% great

This is either due to flex somewhere in the frame or due to an issue somewhere in the math. If we want to see if it’s in the math a couple folks have created alternative versions of the math in excel or online, I think that those could be worth testing.

It looks like the measurements from the log above are:

CLBM:[{bl:2129.00, br:2127.59, tr:2115.33, tl:2078.02},{bl:2087.06, br:2076.88, tr:2172.06, tl:2130.62},{bl:1930.70, br:2234.40, tr:2325.79, tl:1984.65},{bl:1989.60, br:2282.35, tr:2272.49, tl:1926.23},{bl:2047.91, br:2334.22, tr:2224.97, tl:1870.77},{bl:2195.05, br:2182.49, tr:2063.73, tl:2026.87},{bl:2348.07, br:2034.62, tr:1907.39, tl:2187.63},{bl:2295.82, br:1976.38, tr:1966.84, tl:2232.43},{bl:2248.23, br:1920.92, tr:2023.74, tl:2281.03},{bl:2204.41, br:1868.25, tr:2083.24, tl:2331.74},{bl:2039.64, br:2019.22, tr:2228.01, tl:2191.22},{bl:1879.17, br:2182.23, tr:2378.46, tl:2047.68},{bl:1724.19, br:2349.30, tr:2533.10, tl:1911.85},{bl:1781.46, br:2390.42, tr:2483.77, tl:1846.41},{bl:1841.25, br:2436.50, tr:2436.78, tl:1783.46},{bl:1903.85, br:2484.79, tr:2392.54, tl:1723.39},{bl:1970.46, br:2538.19, tr:2351.16, tl:1666.30},{bl:2104.97, br:2388.76, tr:2183.03, tl:1818.49},{bl:2249.20, br:2238.85, tr:2018.16, tl:1977.76},{bl:2399.23, br:2095.03, tr:1857.72, tl:2141.41},{bl:2553.15, br:1958.59, tr:1702.82, tl:2309.32},{bl:2505.39, br:1893.38, tr:1762.09, tl:2348.79},{bl:2458.32, br:1830.63, tr:1822.89, tl:2392.57},{bl:2414.00, br:1770.71, tr:1886.42, tl:2438.70},{bl:2372.53, br:1713.78, tr:1953.06, tl:2488.13},{bl:2333.80, br:1657.92, tr:2020.04, tl:2540.86},{bl:2162.68, br:1813.07, tr:2152.04, tl:2396.94},{bl:1994.43, br:1977.26, tr:2292.84, tl:2250.94},{bl:1830.03, br:2144.83, tr:2436.84, tl:2111.52},{bl:1670.50, br:2315.92, tr:2585.65, tl:1980.04},{bl:1517.42, br:2486.33, tr:2740.10, tl:1858.20},{bl:1573.01, br:2523.34, tr:2694.87, tl:1785.85},{bl:1633.23, br:2562.91, tr:2648.26, tl:1715.52},{bl:1697.04, br:2605.12, tr:2604.21, tl:1647.62},{bl:1763.44, br:2651.00, tr:2562.87, tl:1582.33},{bl:1833.62, br:2699.85, tr:2524.34, tl:1519.93},{bl:1905.60, br:2749.62, tr:2488.64, tl:1460.98},{bl:2026.41, br:2593.60, tr:2315.09, tl:1617.60},{bl:2159.28, br:2442.06, tr:2143.92, tl:1777.68},{bl:2301.26, br:2295.86, tr:1975.67, tl:1939.30},{bl:2449.16, br:2156.00, tr:1811.22, tl:2108.62},{bl:2603.44, br:2023.94, tr:1651.73, tl:2275.42},{bl:2761.81, br:1901.15, tr:1498.63, tl:2447.36},{bl:2714.02, br:1829.17, tr:1561.75, tl:2482.39},{bl:2667.36, br:1759.19, tr:1624.35, tl:2521.66},{bl:2623.26, br:1691.61, tr:1690.15, tl:2563.84},{bl:2581.76, br:1626.55, tr:1759.48, tl:2607.52},{bl:2543.18, br:1564.50, tr:1827.79, tl:2653.02},{bl:2507.29, br:1505.65, tr:1904.57, tl:2702.63},{bl:2459.42, br:1467.06, tr:1990.35, tl:2738.83},{bl:2283.81, br:1628.43, tr:2109.73, tl:2587.44},{bl:2110.44, br:1791.21, tr:2233.33, tl:2439.46},{bl:1939.59, br:1957.13, tr:2367.68, tl:2297.26},{bl:1772.19, br:2128.06, tr:2507.46, tl:2161.97},{bl:1609.11, br:2299.71, tr:2654.05, tl:2034.98},{bl:1452.11, br:2471.05, tr:2803.34, tl:1917.99},{bl:1303.11, br:2644.14, tr:2956.62, tl:1812.79},{bl:1360.64, br:2674.80, tr:2913.63, tl:1733.92},{bl:1425.19, br:2708.38, tr:2866.79, tl:1656.53},{bl:1492.00, br:2744.86, tr:2822.31, tl:1581.10},{bl:1562.94, br:2785.22, tr:2780.26, tl:1507.51},{bl:1637.37, br:2827.43, tr:2740.79, tl:1436.40},{bl:1712.76, br:2872.57, tr:2703.98, tl:1367.96},{bl:1791.36, br:2920.39, tr:2669.96, tl:1302.77},{bl:1872.10, br:2971.59, tr:2638.84, tl:1241.16},{bl:1974.63, br:2817.73, tr:2461.81, tl:1394.29},{bl:2092.14, br:2664.82, tr:2286.38, tl:1551.30},{bl:2218.31, br:2516.58, tr:2113.00, tl:1713.79},{bl:2355.88, br:2373.88, tr:1942.20, tl:1880.46},{bl:2498.50, br:2237.82, tr:1774.87, tl:2050.08},{bl:2646.94, br:2109.66, tr:1611.92, tl:2219.81},{bl:2799.35, br:1990.97, tr:1455.01, tl:2391.94},{bl:2956.73, br:1883.52, tr:1306.07, tl:2566.69},{bl:2911.27, br:1805.34, tr:1371.90, tl:2596.94},{bl:2864.42, br:1728.78, tr:1437.96, tl:2630.79},{bl:2819.95, br:1654.08, tr:1508.21, tl:2665.72},{bl:2777.90, br:1581.37, tr:1579.70, tl:2706.01},{bl:2738.43, br:1511.01, tr:1653.68, tl:2748.09},{bl:2701.64, br:1443.32, tr:1728.17, tl:2792.37},{bl:2667.67, br:1378.67, tr:1807.44, tl:2839.98},{bl:2636.68, br:1317.55, tr:1884.88, tl:2890.62},]

The easiest way to do that is to enter some made up and large values for your frame size in the configuration window and then you should be able to keep pulling the belts all the way out.

I think that this is a great suggestion

1 Like
14

It lists firmware version as 0.76 and indicates motor, encoder, and magnet detected on all arms.

1 Like
15

Here is the online calibration thing:

https://m4cal.etskinner.com/

I’m playing around with it and your data now. it looks like your lower right data points are the ones that are a little funky looking

16

New here, correct me if I’m wrong, but isn’t the frame supposed to be positioned vertically at a slight angle instead of lying on the ground?

17

And the index version should print too?

18

Both work! There is a Vertical or Horizontal calibration.

2 Likes
19

Peter wrote:

New here, correct me if I’m wrong, but isn’t the frame supposed to be
positioned vertically at a slight angle instead of lying on the ground?

Both are supported, and it may work slightly better on the ground (if you can do
anchors into a concrete floor, the stiffness is worth a lot)

David Lang

1 Like
20

Thanks for the clarification.