The only way to address losing the wifi connection is to reduce the distance
to between the maslow and the device you are connecting to or put a
directional antenna on the thing you are connecting to.
If @bar ever does a run of controllers with a different EXP32 module that
accepts an external antenna, that could be a fix on the maslow side, but
unless you are going to unsolder the module and solder on a replacement,
there is no fix that we can do.
Sorry, I should have added that this is after you make sure that your computer
is not disconnecting your for it’s reasons.
These reasons can be:
there is no Internet connection when the maslow is in AP mode, some OSs
really want the Internet and will disconnect from the maslow to try other
networks.
aggressive power savings, windows in particular will try to put the wifi to
sleep to save power, disconnecting you
AP refers to access point mode. When your computer is directly connected to the maslow WiFi. You would see in you Wi-Fi settings in your computer that you were attached to Maslow as a network. This generally is used to talk to the robot in order to set it up and give the Maslow the WiFi password to your main home network. Once the Maslow can connect to your local home network then people usually connect to the maslow by passing through their local network router by going to an address like maslow.local or 192.1……. Which you can find by logging in to your home router.
Your laptop may drop this connection because current computer security doesn’t like to connect to networks that aren’t connected to the Internet partly for convenience and partly because people set up dummy networks to trap people and grab credentials
I am having problems with Wifi disconnecting as well. Most notably whenever I go to start a job.
I am new to the Maslow 4 and just got the anchors located and it calibrated. It keeps dropping connection and then will not reconnect unless I cycle the power on the Maslow.
I am running Windows 10, version 1.16 firmware. I have set the computer to connect automatically to the Maslow network and disabled automatic connection to all other networks. The computer is only 5 feet away from the Maslow, so I don’t think signal strength is a problem.
I am in a public school, so Wifi is always an issue. I would prefer to just run this off the USBC to minimize the potential of interference. Is this a possibility?
I tried the ping technique you described. It would give me a ping time of 3 - 4 ms even as the Maslow 4 reported the state as unknown. While in that state it would not allow me to slacken the lines, apply tension, or retract them. It would report that the 4 motors passed the tests.
After cycling the power I was able to slacken the cables, then retract them, and then fully extend them. After that I could jog the machine to start a job. But as soon as I did start the job it disconnected. The ping was still active in the cmd prompt while it was disconnected. Attached is the serial log.
So I was trying this again, from the State of Unknown and tried to cut another job. He is my workflow:
First, cut power to the Maslow 4. Waited for ground control to tell me it was disconnected. Closed that window, then turned on the power and waited for the wifi to come back on and it prompt me to open the browser and connect.
Then I started the ping in the cmd prompt.
After that I saw I was again in unknown state, so again I clicked alarm to release it, went into the settings and slackened the cables. Then I disconnected them, retracted them, and fully extended them. Once hooked back up, i clicked apply tension and it picked up nicely.
Then I added a gcode to test. It still recognized the lower left as home since I wanted to cut their earlier today. It showed the gcode on Ground Control, so I clicked play.
The Maslow then plunged the bit all the way down. It bottomed out the Z axis. then began moving across the plywood to the home position.
If anyone has any information on what I am describing, I would greatly appreciate your insight.
Thank you,
This sounds like you might have a G28 in your code see Supported Gcodes | Wiki.js which has not been defined for your setup (random numbers in memory). If you open the job file in a text editor and search for G28 and delete the line it is on, alternatively you can move to a safe location and Z setting and send a G28.1 to set it up.
G28 Predefined Position
G28 is a machine coordinate location stored in non volatile memory. It will go to the same location regardless of the coordinate system or G92 offsets. You can view the current setting via the $# command.
G28 This does a rapid move to the location.
G28 axes If you specify axis values it will first do a rapid move to that location in your current work coordinate system. Then it will do a rapid to the stored G28 location. It will only move on axes specified.
G28 Z10 will move to work Z10 then move to the Z location stored in G28. No motion on X or Y.
G28 X10 Y20 will move to work X10 Y20 then to the G28 location for X and Y. There will be no motion on Z or any other axes you have defined.
G28 G91 X0 This is a way to move only to the X stored in G28. G91 X0 is a relative move of 0 so there will be no move before the G28 move. Again, only the X axis will move. Note: You will remain in G91 after the move. If you were in G90 before the move, you may want to send a G90 to return to that mode.
G28.1 Use this to set the G28 offset to the current location in machine coordinates.
Thank you for your comment. I appreciate your insight.
You are absolutely right: There were 3 G28 commands in the code. This was never an issue with the original Maslow, but now is a problem.
I use Autodesk Inventor in my classes to design parts, and then Autodesk Inventor CAM Ultimate to generate the g-code. In the past I have always told my students to not click rapid retract, but looking back I see it in all my code students gcode. In the past we used Garble, which I believe still works for Maslow 4. I found a way in the post processing steps to set the Property of the Safe Retract from G28 to the Clearance Height.
I tested this by generating my gcode, and running it through .ncviewer to see if it still made the proper shape, as well as doing a search in Word to find any instances of G28. Both came back correct, so I’m going to retry this code tomorrow. Assuming I can fix this WiFi disconnecting issue. I see that 1.17 was just posted this afternoon. I’m really hopeful to make a lot of sawdust in the next few weeks.
If you move the Maslow to a known position, set the Z to a safe height and Enter a G28.1 it will save that position in non volatile memory. If you go to the pull down to the left of FluidNC and go down you can enter the G28.1 command there, which will at least save you if you miss a G28 command in a future cut.
With the wifi problem I am having a problem starting this afternoon after a windows update. I can only reach the Maslow with an IP address. not maslow.local
I will look for the that in FluidNC in the future.
With the 1.17, in addition to using cmd to ping the sled I am able to maintain connection. So I would call that a win.
My problem now is the sled lifting up while cutting and all the controls are locked up. The bottom lifts and it kind of rolls around on the curve along the top edge of the sled. Then when I try to jog the Maslow up so I can try to sand the plywood, all the controls are locked. I can’t jog, I can’t even release tension. The state says ready to cut, but nothing works.
I am in a vertical orientation, around 20 degrees from vertical.
Kind of a rough morning for the Maslow. So far it can’t stay connected to the computer, even with the cmd ping.
The Maslow hasn’t locked up on my again yet. But I did have to retract & extend the cables again already this morning. Even so I never go the “ready to cut” as the status. Once I loaded a sample gcode, it disconnected as soon as I hit play and turned the router on.
The issue is for sure being caused by this error message:
[MSG:WARN: Previous error was 54.872mm]
[MSG:WARN: Position error on Top Left axis exceeded 15mm while running. Error is -53.112mm Counter: 3]
Basically what that means is that the machine is trying to move faster than it is physically able to and is shutting down to prevent it from hurting itself.
There should be a speed limiting feature which should quietly slow things down without any error message so that message is still a sign that there is something weird going on.
Thank you for the reply. My feed rate is set to 30 in / min in Inventor CAM, but I see my plunge was set to 30 in/min, in the past I used 13. Could that be the cause of the error?
I updated my gcode to limit to a 13 in / min plunge speed.
Since it would not exit the state of Extended, I detached the cables, retracted and extended them. Then I was able to jog and move the sled to the home position (where I wanted my cut to be made).
However when I started the new gcode it disconnected again.
So are you saying that it is disconnecting from the computer to stop the operation when it detects a problem in the gcode?
I just tested running the file on my machine and everything worked normally so I don’t think that it is an issue with the gcode itself.
Are you seeing the error instantly as soon as you press the play button or does it happen a few seconds into running the file?
[MSG:WARN: Position error on Bottom Right axis exceeded 15mm while running. Error is 15.056mm Counter: 1]
[MSG:WARN: Previous error was 15.056mm]
This error message basically means that the bottom right belt is far enough away from where it should be that that the machine is stopping because it is detecting that something is really wrong.
How does the tension in your belts feel? Does it seem like they are pulling really really hard?
[MSG:INFO: Center point deviation: TL: 0.000 TR: 0.000 BL: 7.102 BR: 11.669]
This line tells me that the bottom right belt is starting out quite far from where we want it to be after “Apply Tension” is run. The threshold for passing that check is 12mm so 11.69 is barely good enough to unlock the machine.
That tells us that the machine doesn’t have a great understanding about it’s surroundings. That could either be the Z-axis height or the anchor point locations.
I would try taking the router bit out, lowering the z-axis all the way down and running “Find Anchor Points” again. That should give us a clean slate with the machine understanding the world around it better.
