how much can the initial values be off and still get it working
what if you put the results of one run in as the initial values of the second
run (same data)
what if you do more iterations (same data)
what if you change the outlier threshold (make it really large to include
everything, or more aggressive to only include the best data), there are a
couple of boxes to help you visualize what you are keeping and excluding as you
tinker with this.
how much do the results move, do they seem to be more or less accurate
(compared to what your working machine has or what you measure with a tape
measure)
Ah! The new version still has those points internally, it’s just not sending them to the computer (because the computer doesn’t do the calculations anymore).
We should totally be able to bring those back for debugging
#995 Does this index.html.gz (141.5 KB) firmware.bin (1.9 MB) @bar#991 also needs approval
Update startup Z/Z-home handling, persist Z-stop home across restart, and sync with Maslow-Main#991
Which stops reset of Z Stop happening automatically regardless of actual Z height, and fixes loss of Z home across restarts (except for faults or power interruption)
Hello, I’m a bit confused and I’m not sure which version to use to ensure stable operation. I’m currently on version 1.20 and, after applying tension, if I try to move it directly to the home position, it shows a position error on one of the motors; however, if I apply tension and move the machine slightly in any direction, it then allows me to move to the home position without showing an error. When I’m extending or retracting the belts, if I press stop, the machine freezes and I have to switch the power off and on again. Whenever I switch the power off, it loses the Z position. It usually displays a series of endless errors on start-up. I’ve tried several times to find anchors and it doesn’t work; I have to enter my own measurements. Today, the Z shot upwards when I was searching for the Z home position. I can’t remember which version worked best for me. My question is, is 1.21 more stable? Or which previous version is more reliable? I can’t remember if it was 1.15 or 1.16.
I was on the road before, so will deal with your issues a little deeper.
If you haven’t got a good Find Anchors result, then it might be having trouble getting to the extremes on your layout. Is your home position in one of the corners?
Pressing stop while extending or retracting is not a normal operation, so the Maslow will not know where it is, which is why you have to power off, retract, and extend again
V1.21 I think the Z problem was fixed in V1.21, if not definitely in the next Version.
My starting position is in the centre. I never had any problems calibrating in older versions, and I can see that the reels work well with a retraction force of 600. It’s always possible there’s a hardware issue, but I’m seeing a lot of inconsistencies in the software—which is why I’m asking about the different versions, so I can identify the most suitable update and then figure out where the problem lies.
In versions 0,.. it was possible to use the stop when retracting and extending; it’s actually a feature I find quite normal, I’ve often noticed when retracting that a strap is tangled up in something and I’ve stopped it to free it without causing any damage; it often happens that when retracting the straps, one of them doesn’t retract on the first attempt, so what you do is press ‘extend all’ and with a slight movement of the strap that had got stuck, it comes free, At that point, I would press ‘stop’ so as not to have to extend the straps that were already fully retracted, as they don’t return to this state until all of them have reached the maximum position; after that, I would press ‘retract’ and simply deal with the strap that required attention. The ‘stop’ function has always been a source of confusion
It’s very difficult for me to test firmware, as I usually have to carry out the tests directly on the job. There are issues you get used to and learn to live with, which is why I’m asking about previous versions – I can’t remember which one was the most reliable, even with minor bugs, but I need stability.
Once you have a good calibration (now Find Anchors) you should not need to run Find Anchors again, so long as your Frame doesn’t change, and you preserve the settings in your yaml file.
Version 1.15 was reasonable for configuration I think.
I’ve done few with various starting frame sizes and haven’t had any complete failures. Where it has been off initially (low fitness but still more than 4.5) the second run gets it close to original values. The actual frame has not been modified, just the values in yaml file.
Release probably Monday so if people have problems they won’t be frustrated over the weekend.
I would like to replace a generic ‘fitness’ score with something more
understanable, like a real error estimage. I also think we should set it to
through out outliers
I think a button pusher should be included in all kits purchased, it must be frustrating to build the Maslow and then have to source a button pusher to use it, especially if you don’t have access to a 3D printer.
