Thanks. I had done all of this as well. I’ve had the glaring light problem before; although not since I moved the machine into the house from the garage.
And now I noticed that the 0,0 point seems to be off by about +/- .2mm between power on and off, which just ruined a job that I was trying to run where I had to run two cuts back to back.
There is (was?) an issue with “drift” over time. The GF needs to be power cycled every once in awhile to re-establish itself. I keep mine on most times but I do cycle it once a week and before I start doing something critical.
Right now, because of the calibration problem, I have to power cycle between every run. So, it’s not a drift problem. I ran a cut, power-cycled the machine to get it to calibrate again, and ran the second cut, which was off from the first.
I should note that I ran this same job yesterday and the day before successfully, so the “drift” between jobs has not been consistent. I needed to run it again today because I messed it up when I glued everything together.
that’s not good. Hope you can get some relief from Support.
Well, for the first time since Sunday, the machine properly calibrated when I just power-cycled it.
I really need to wait before hitting enter on my posts. It started to calibrate and then stuck after moving the head out from under the camera.
I believe this is a bug in the cloud.
After running each ‘motion’ file, the device resets its position to 0,0, and it lets the cloud keep track of where the head is located.
When you see “Scanning Material”, the device receives a ‘motion’ file from the cloud telling it where to move to so it can take the thickness measurement. After it is done, the cloud sends another ‘motion’ file to send it home*.
This is where the potential bug lies. The return motion sends it back to -1,1, instead of 0,0 (there is an extra “step” for each axis), but the device resets its internal position counter to 0,0 after the move.
Every scan (print) cycle results in an additional position error on both X and Y axes of .01875 mm. Doesn’t sound like much, I know. But if you never turn your unit off, it will accumulate over time.
It also could be that there is some hardware/firmware issue that I haven’t located yet that they are accounting for by adding the extra steps (I hope not).
*This is different when it sends a ‘print’ file. In this case, the device keeps track of where it is, and returns home based on its own tracking of its position.
12 Likes
Mate… your posts are like CSI-Glowforge.
They are an incredible read, thank you
7 Likes
I had an interesting conversation with a coworker today that got me thinking about these intermittent presumably-WiFi problems many people have been reporting. We were talking about a subject that had nothing in common with this except for radio interference. He mentioned that he was getting frustrated something that I was also frustrated with, until he discovered that the problem only manifested when he had his phone in his pocket. Given that, and a number of news stories recently about various devices in the home interfering with each other, I’m wondering if it might not be completely insane to try troubleshooting calibration failures by moving other wireless devices, especially mobile ones, away from your work area. I would normally dismiss this as techno-illiterate voodoo, but the increasingly complex nature of interaction between all of these things means my formerly rock-solid understanding of how the world works that would lead me to say “that’s ridiculous” is on shakier ground.
3 Likes
Something definitely seems different within the last hour. I let it sit while it was “stuck” and it eventually completed the calibration sequence, but I estimate that it took over 2:30 minutes to do so. I power cycled it again, and it stuck again after initially moving the head over to the left–after moving it under the camera–and it sat there for approximately 100 seconds (by my count in my head).
I ran two back-to-back jobs successfully without having to power-cycle (for the first time since Sunday). I then power-cycled it again and actually used a stop watch to time:
00:00 Turn on machine
00:24 Clicks start
00:31 Light flashes
00:41 Gantry moves head under camera
01:03 Gantry moves head to the left
01:23 Gantry moves to the back of the machine (this was the quickest it has been since my calibration problem started)
I have seen these types of issues in the past. Not often, but they can happen. I have my GF at least 15 feet from any other transmission source (except my mobile phones when I’m carrying them), and it’s only about 25 feet from my core access point; which is on the other side of a single sheet of drywall on the unfinished side of my basement. I even turned off the repeaters I have in my house at one point to make sure.
1 Like
Wow. That’s actually a lot. Think about how many times that happens to someone doing a bunch of jobs and then even worse doesn’t turn it off for a week so it’s hundreds of times .01875mm.
I guess I’ll start power cycling it daily now.
This could be one of those royal PITA things to fix. Consider some dev sees this behavior of incorrect positioning manifested but doesn’t know why so in some other part of the app doing something else they decide to adjust for the drift by coding some trick routine that is dependent on the drift happening. Then if you simply fixed the “go to -1,1” to “go to 0,0” that would then cause the other app area’s adjustment to be wonky and that will now be wrong and good luck digging it out.
sheesh
1 Like
If Scott can find the bug with no access to the source code I can’t understand why the GF staff didn’t find and fix this years ago. Some bugs can be hard to find but one that happens every build cycle should be easy to squash.
3 Likes
I’m sorry to hear you’re still having trouble, and I apologize for the late reply. Thanks for taking the time to do so much troubleshooting and keep us updated on the results. I’m looking closer at the issue before recommending next steps. I’ll follow up when I have more information.
@vee, I have been able to run 3 jobs successfully since yesterday. I plan on doing more testing late today and tomorrow to see if I can figure out more of what has been happening. I will keep updating this thread with information when I have time.
We don’t know they didn’t find it.
But as I noted, it may be far harder to fix than just change the parameter to go back to 0,0. They’ll need to check the rest of the code to see if anyone built on top of the machine’s behavior of going to -1,0. If anyone did that, then fixing the original issue is going to break other things and that’s where it gets hard & will require a ton of regression testing.
1 Like
Thanks for your patience. All the information you’ve provided has been very helpful as we looked into it. It looks like your unit is experiencing an issue that we can’t resolve remotely. I want you to have a reliable unit, so I’m recommending we replace this one. I’ll be in touch via email to sort out the details. I’m so sorry for the bad news.