Perfect. Thx!
1 Like
Iām 99.9% sure that this is the issue with FluidNC crashing if it loses the wifi connection while cutting. Iām digging into what is going on there right now. Sorry!
2 Likes
Hey @ronlawrence3
Do you know when the heartbeat changed from log_info to something else? I have old code on my Mac (havenāt fetched in a while) and was just looking through this to figure out a way to disable the heartbeat at least for the serial output.
My machine has 0.75 or 0.74. I can connect to with Universal Gcode Sender as a GRBL (it fails to connect as FluidNC due to it expecting > 3.3.0 or similar for version and Maslow reporting 0.7.4ā¦)
Anyway, it is able to connect enough to receive the machine output but I havenāt been able to send anything, and I was wondering if it were due to the heartbeat message per an older comment by @dlang. And per @dlang yesterday, I was able to use the terminal/command area in FluidNC and enter some of the commands referenced above ($EXT and $ALL) and they worked as expected and reported back in the webui as well as the serial port to UGS.
I need to update my repositories and see if it works any better. and if spoofing the version will allow it to connect with the FluidNC profile.
This is great! Super psyched.
1 Like
I think it was 0.74, but I could be wrong. The latest definitely does not have info heartbeat messages though.
1 Like
Ok! Progress update.
I installed 0.75, and it no longer has the heartbeat showing in the UGS console.
However, I think something wrong with my UGS install as Iām not able to type anything into the console. Iām happy to blame Java.
So I searched for some alternatives - several that use python that look like they have potential, but I found https://software.openbuilds.com that had a Mac OS installer.
Quick install, connect over serial and I was able to enter in $ALL to retract all, with appropriate response from the Maslow sitting next to me.
Will try it out tomorrow I think! Added some macros. Will have to take a closer look at the UI when Iām connected to a frame.
2 Likes
Hey - you can connect over the network too!
Connect to your maslow wifi, āScan Networkā type in your IP subnets (192.168.0ā¦)
It scans your network and then Maslow @ 192.168.0.1 shows up:
You can type in javascript instead of g-code in the macros, I wonder if there is enough capability to handle the calibration processā¦ ha!
2 Likes
I use this on linux for another small CNC I have and I really like it.
1 Like
It would certainly be possible to fork this and add watch the messages and perform the maslow calibration calculations in it. GitHub - OpenBuilds/OpenBuilds-CONTROL: OpenBuilds CONTROL: Download from https://software.openbuilds.com and GitHub - OpenBuilds/OpenBuilds-CAM: Online CNC CAM System
1 Like
These macros look like a great solution to an idea I had for one-click jogging to common non-0,0 home locations.
Way better than my original idea of creating gcode to do it for me and having those files clog the list.
1 Like
Beautiful! Openbuilds control is nice-tons of plugins and super easy to macro. Very excited to see this!!
Do the jogging controls work, including the jog rate?
1 Like
Carson Barry wrote:
These macros look like a great solution to an idea I had for one-click jogging to common non-0,0 home locations.
Way better than my original idea of creating gcode to do it for me and having those files clog the list.
that sounds like something where the G54 family of commands should be used.
David Lang
1 Like
Well, after installing 0.75, adding a UPS to the mix, and waiting until after 5pm due to a noise complaint, here is the late night update. I donāt think serial control is likely to address most of these issues, other than possibly being able to stop itā¦
===
I am officially no longer of the opinion that ādirty powerā has been the cause of any of our problems. At this point, demonic possession is pretty much the most rational explanation for the behavior of both machines ā after running a half dozen or so sheets without trouble, both machines are now refusing to do my bidding, choosing instead to go off on long, meandering strolls, plunging twice as deep as they are told, and reversing the direction of the belt drums so that they are spitting out the belts, instead of pulling themselves along with them. And when they start going nuts, they wonāt respond to any of my commands (including the stop button), even though they report that they are still connected, so I have to race over to yank the power cord out before they damage themselves, because we no longer have a kill switch.
We will see what the day brings.
1 Like
Sorry to hear
Regarding power, have you tried running mainboard from a 24v bench supply?
Our stock power supply output cord and connector is not shielded and the power supply is not grounded. Using a power source with DC output cord that is shielded to earth ground is something to test if possible.
1 Like
I am hoping our firmware is updated so we have a persistent boot counter so we can know when panic resets occur by the increment of the boot counter on some UI/log.
1 Like
This also sounds like ESD buildup. A grounded 24v power supply might help, as would grounded vacuum hoses.
1 Like
Also, does it heal itself if you let it sit overnight or at lease for a while? That would support the idea of an ESD problem.
1 Like
I can at least posit that, if it were wholly down to the router producing interference or ungrounded dust collection, I wouldnāt have been able to run a 5 hour job and he wouldnāt have been able to get through as many sheets as he did. That is, unless itās static build up only mattering when the device is booting because itās flipping the encoders and doesnāt have the opportunity to do that unless itās present during the boot process, which could easily be the case if youāre doing a bunch of jobs in a row.
@WFD All of the things that I have heard have resolved the issue in every thread Iāve seen have included time passing or some degree of disassembly and reassembly which could have resolved static build up on their own. There have been some other things tested, but no really controlled tests to make it so we can be sure in any direction.
@boxomatic This most recent account of letting belts out when it should be pulling them in while meandering definitely sounds like those encoders running backwards.
Some questions:
Did this start occurring after having them powered off and start immediately when you powered them on and tried to cut?
How did you power them on?
Are they both on the same power strip?
1 Like
I hope to get some more answers today as I wasnāt there.
My assumption is that they are not powered down between jobs. We have been successfully releasing tension, detaching, moving the Maslow out of the way, unloading and loading another sheet of 5/8 plywood, re-attaching, applying tension and starting the cutā¦ but when there are connectivity/control issues might involve a power cycle.
The Maslowās were wired into a hand-built box with outlets and separate switches for the routers, mallows and dust collector so we could manage them independently. The box is fed from two different circuits, but not sure how far away the panel box is in this ware-house style space.
The power supplys are mounted on swing arms overhead and have an extra 20ā of 12 gauge DC power extension attached. And extra wire coiled up, etc. So lots of antenna space!
So the belt direction / encoder issue Iām not sure yet if that happened after power cycles or not. My schedule is tight today with some personal commitments, but I am hoping to solder up some encoder boards to have one set with the direction input connected. And maybe shielded cat6 arriving today.
That might address the belt direction issue.
Iām not sure about the z-axis though. Could be values getting stomped if there has been a power cycle or software failure I guess and the z-value for zero is lost or corrupted. not sure on that one.
The command/control/connection loss issue Iām not sure if the USB connection will help or notā¦ but hope to do some experiments. I suppose there is always the chance that even if directly wired that the wifi/webserver/etc portion might still fault out and impact the lower level command processing. I saw a mention somewhere else of building FluidNC without the wifi capability so thatās a long shot. Iām hopeful there is something concrete and addressable in the interaction between FluidNC/server/webui/maslow specific-code that will be found and fixed.
There are definitely times when it seems connected, but doesnāt respond. Now Iām wondering if it seems connected because we get some status showing up in the console log, but it might be in a stop state but the button still says idle, or similar internal state so it is not responsiveā¦ lots of conditions to isolate. Iād say the webui feels loosely coupled with the machine stateā¦
2 Likes
For those following along:
Using OpenBuilds Control software on MacOS:
- Connected to Maslow using 20ā USB-C cable.
- User defined macros to Release Tension, Retract All, Extend All and Apply Tension all worked.
- Was able to move Z axis up and down and set Z Zero
- After a false start, some GRBL alarm issue, another cycle of release/retract/extend/apply
- Was able to use jog controls to move X/Y in positive and negative directions.
- Established āapply tensionā as X and Y zero
- Freshly loaded ārepairā g-code (skipping the stuff that finished cutting last night)
- Router on and dust collector on
- Pressed Play, and it started to cut just fine.
-
Here is short line segmented started near the edge and connecting to existing line - nice repeatability after 3-4 retract/extend cycles, etc! Canāt tell it was cut 12 hours later with different control software.
It completed one or two more of the short cuts in the file, then this happened:
You can see it didnāt pull all the way up during the move, and then when it went to cut again, it was that much deeper than it should have been.
This is due to the z-axis getting racked somehow. Not sure what is going on. Last week we were having similar problems that we thought were due to tipping at the upper left, the file and a bunch of other gremlins. Finally realized that we have the Maslow cut all over the board and it finishes in the upper left quadrant, then moves to the upper left edge to do some of these short finishing cuts. And for whatever reason, it was failing to lift the router on one side.
I thought it was the stepper motor, so replaced it (thanks @bar) and that seemed to have helped.
Now getting it on the other Maslow. This time I figured out that I could unplug one of them at a time and run the other one to get them at the same level. When they were at the same level they would go up and down.
This is where the other one was last week, and after another cut or two it was failing consistently so I replaced it. Now Iām going to hook it up to my controller and see if the lead screw got bent or somethingā¦ because it looked fine visually.
Anyway, interesting that something is going on with the z-axis - either one side is missing steps, or maybe when its under tension it doesnāt move as much as it should and gets further and further out of alignment, etc.
1 Like