Is there anyway to turn up the logging in the console? My machine randomly stops but the console thinks everything is fine when it has in fact destroyed yet another product.
Short of re-wiring my house I would like some kind of solution for this problem… SD Card, Bluetooth, more console logging, really anything.
I followed the guide for connection issues two computers ago and it still randomly fails. So to me the issue is
This machine is made of cotton candy and its raining
USB sucks
Lightburn doesnt say when communication stops
If I could fix 2 or 3 that would be most cost effective.
Yes starting to realize these GRBL machines have many quirks and issues.
Thank you for your reply. I ended up moving the PC right next to the machine and it appears to have solved the issue for now. Every once and a while the job shows running but at least it isnt messing up my job during.
Equally unfortunately, LightBurn must detect those failures at the far end of the USB connection, where it doesn’t get much information. I have only a general idea of the USB driver interface in Windows / Mac / Linux, but I’ve made the mistake of assuming a synchronous communication link would report a failure, rather than hanging up without notice and pinning my code to the mat.
LightBurn gets all the blame because it’s the visible user interface, but the real problems lie at the far end of the cable.
Sculpfun S30 Pro Max 20w diode laser
Sculpfun grbl board
1.1h
Stops in different places each time, also job stays running after completion
Laser is off but makes the sound like it’s on
I have already changed usb ports so they do not sleep, baud is 115200 as recommended, dtr signal in light burn makes no difference
No gcode was created exclusively in light burn 1.4
I was able to get the job to complete by “moving the pc closer” but still jobs don’t autocomplete so I click stop when I see it’s done.
I assume this is because the machine mother board is built of pure chineseium. Maybe in the future they build these better. Can’t believe it livestreams the whole job to the machine in real-time. No memory or sd card. Having a single point of failure like that is such a horribly bad idea. Soon I will try Bluetooth and see how well it operates as it couldn’t get worse could it?
I read the article about grounding the machine and static build up but TBH I dont have a tin foil hat. Can’t believe they are real suggestions. This laser is a house of cards. The software seems to be fine though, can’t point a finger that way.
It’s a very common issue that you describe. Here are the top causes for this problem:
USB cable. Most lasers are shipped with very cheap cables, using a high-quality cable solved the issues for 80% of the users
electrical disturbances: many connections issues arise from electrical interference and power supply issues. Having the laser on a power line with high load devices is a bad idea. Long cables and high-power devices close to the laser as well.
Driver issues: the CH340 drivers are somewhat picky. There are different versions out there and there are a multitude of driver-OS-combinations that are unstable / fail.
Firmware issues: very rare. In some cases, it helped to re-flash the firmware, but I’d say, that’s below 2%
I don’t agree here. The mainboards are kind of cheap, but the same as in nearly all 3D printers and all diode lasers. Not much they could have done wrong here.
Streaming the data as a “live stream” is state of the art in this area. It’s even like every usual printer does. You can add offline capabilities, but this will make more people use the laser unattended, which is VERY dangerous. Otherwise, you also wouldn’t be able to change parameters during runtime.
If you use an SD card, you have the same single point of failure, just at a different location. SD card control is as crappy as USB communication at those microcontrollers.
Bluetooth is no solution here, wireless communication is more unstable than wired connections per definition. Always prefer wired connections.
Basically, you get what you pay for. If you would buy a “real” laser cutter, you won’t find anything below $5000. Even more than $10000. So if you pay $300-900 for a complete laser package, you have to be aware that this is no professional equipment, consider it as a TOY. That’s what they are. They are neither reliable, nor safe (absolutely not, they are dangerous as weapons). You need to tinker a lot to make them run fine. You already agreed to this by selecting a toy as tool.
I don’t know if it’s helpful, but I had a similar problem a while ago with my Sculpfun S9. It turned out that on my setup turning off the compressor for the air assist generated some stray signals that were often (but not always) disrupting the USB connection. Thus, that often happened after the last layer, i.e. just as the actual job was finished, but still before LightBurn reported it as finished. But if I had layers with and without air assist, it would also happen in the middle of the job, which of course was worse. I resorted to switching on the air assist manually before starting the job, and switching it off manually again once the job was completed. When I later replaced the compressor by another model, the problem disappeared completely and I can now use LightBurn to switch the air assist as needed.
Grbl state of the art? I like your analogy to a printer but it seems that even a printer from 1990 has better operation; they usually have a buffer, some memory local to the machine and also bidirectional communication where as this grbl controller seemingly lacks everything? Even udp checks to ensure packets arrive before sending more.
How is this house of cards stacked?? If a single bit 1 or 0 gets flipped due to emi is the whole laser job now lost? How delicate is this process?
To me avoiding usb is both the problem and the fix. There are no other issues with this machine but the USB. Normally I would agree a physical cable connection is best but if that’s failing then what’s the recourse but to try another communication method. Since all I can do is throw spaghetti at the wall since event log is not showing anything and both lightburn the laser lack ability or transparency to show what’s going on.
Sd card would be in fact be a good solution as having the job completely stored on the machine would avoid the whole USB connection. At that point the manufacturer would be 100% in control of the issue. Right now is like purgatory; lightburn points the finger one way while the manufacturer points the other way.
I will give Bluetooth a try later today and report back. I just need the machine to work. I don’t care if I need to do a handstand while drinking a glass of milk. Whatever floats the boat as they say…
It would be nice to confirm the issue but it seems nothing here is going to point the way. Lightburn thinks it’s communicating to the laser and the laser thinks it’s doing laser things… Normally I would search for a log file but is that possible with this hardware? Where can the disconnect be found?
Yes I’m aware that I should have bought a $10,000 laser machine, but to be honest you don’t get good at driving by getting a Lamborghini. And while the marketing teams that make all this would like you to believe it’s all 100% they all have their very own issues. Just like the light burn software seems to be perfect but contains silly bugs like rotating an image then the adjust preview window is garbage, I don’t expect things to be perfect but I would like a methodical approach to solving the issue.
I just had another job fail after 50 minutes on the last pass.
The job was 4 passes on a single image with air assist on, the only other layer is a tool layer for framing. Do you think this would cause the issue you have mentioned?
How do you suggest to turn the air assist off manually? I have all stock components from Sculpfun S30 Pro Max.
It turns out that GRBL is based on a single-chip microcontroller designed in the mid-1990s:
The typical GRBL controller runs on the cheapest possible knockoff Arduino PCB with an ATmega 328 microcontroller:
32 8-bit registers for computation
2 KB RAM for all variables
2 KB EEPROM for GRBL config parameters
Note that those are all “kilobytes”: that’s all there is. Also, it’s an 8 bit microcontroller, so all that fancy floating point math happens in software, one byte at a time.
The chip does have a serial port, but when crappy hardware glitches the power supply, the microcontroller can do nothing but reset and start over again. The lack of diagnostic logging comes from a complete lack of hardware to store anything past the crash, which is internally indistinguishable from the usual power-off and power-on.
GRBL simply stops running when the microcontroller shuts off, so LightBurn has no way to know anything more than whatever was on the far end of the wire has stopped sending data.
Now, fancier 32 bit microcontrollers may be somewhat more capable, but that firmware consists of a branch ripped from the GRBL tree and patched to get it barely working with whatever hardware might be out there. Given that everybody ignores GRBL’s GPL license and keeps their source secret, there’s no opportunity for anyone other than the OEM to fix the code and they ain’t got time for that.
What makes it worse is the USB interface depends heavily on good hardware design, which absolutely does not happen on those knockoff PCBs out there. They may look pretty, but robust design intended to survive the normal rigors of being out in the open, rather than in an RF-shielded / glitch-protected case, while switching multi-ampere motor currents, simply isn’t there.
So, yeah, it’s awful, but getting away from that requires spending nontrivial money for industrial hardware.
FWIW, CO₂ lasers in the few-kilobuck range don’t have much better designs, but the additional money gets you replaceable parts hung on a better mechanical frame with less-marginal power supplies.
How do you suggest to turn the air assist off manually?
As my machine is DIY, I simple removed the relay and just used the switch to turn air assist on and off. Using Lightburn, you could try to disable air assist on all your layers and use the console window to send manually an M8 (or M7, depending on your device settings) before you start the job, and after the job has finished send the same way an M9 to stop the air assist. However, this assumes that LightBurn only sends M9 commands when air assist is actually used by the layers (which should be so, but I am not really sure). However, you say that the job failed “on the last pass”, so if this was not at the end but in the middle of a layer, I doubt that in your case the switching of the air pump is the problem…
I think Ed already explained it well. I truly agree with you that it’s a pity that the hardware does not have advanced functionality, but that’s how they are designed. You could build your own mainboard using just a few Arduino and parts. So all these lasers are not far away from a kid’s school project.
The protocol and communication could be enhanced, but it’s open-source software. If nobody does such in their spare time, it will never be implemented.
I went ahead and flashed FLUIDNC to my Sculpfun S30 Pro Max board, can you please tell me the difference in the config files on your website?
Specifically these two mentioned below, just trying to figure out which is best for use.
FluidNC configuration for MKS-DLC32 board using Sculpfun lasers: Download (new! not extensively tested)
FluidNC configuration for S30 board: Download (new)
The reason I ask the question is I have opened each file and it appears that coolant is showing as gpio21 in your file “XY-DLC32-MC V1.1” and the other is not showing a configuration for this gpio “MKS-DLC32 V2.1”, in this regard it appears your file is better.
Although looking at the “x max_travel_mm: 370.000” in your file shows what I believe to be an incorrect value as this should be 410mm as seen in the other config file “MKS-DLC32 V2.1”, in this regard the other file offered is better.
Just trying to figure out which file is best for use and it seems maybe they both have good settings and need to combine into one file?? Maybe I can help dial this down to 1 file.
-Moved the PC closer to the laser (1 foot away)
-Changed USB Cables (tried 3 different cables)
-Disabled seletive suspend USB port in Windows Power Options
-Changed Windows to High Performance, set computer & screen to NEVER SLEEP
-Ensured BAUD rate was set for 115200
-Enabled DTR signal in Lightburn (no change whether this was enabled or disabled)
-Reinstalled ch340 driver
-Uninstalled ch340 driver, rebooted machine and reinstalled
-Uninstalled ch340 driver, rebooted machine and reinstalled the BAUD Compensation version
-Tried the old version of ch340 driver
-Flashed latest firmware to Sculpfun S30 Pro Max (actually caused issue with current position, and jobs still fail)
Finally I flashed FLUIDNC 3.7.0 - wifi, seems all is well now.
Just completed a 50 minute job over Wifi with Lightburn, the job DID NOT FAIL and it was able to detect that it finish on it’s own without me having to click stop at the end.
It appears everything is working better than stock as of now. A big plus is I don’t have to sit 1 foot from the noisy laser with my green glasses on. Oh and I can stop wasting good materials becuase of failing jobs.
Lots of pro’s to mention here.
Great verbose logging too, you can see exactly what is going on the whole time. Not just by layer…
Those files are for slightly different board layout. Though those board origin from the same schematics, Sculpfun did some changes and therefore you need to use the MKS-DLC32 configuration file for the MKS board and the S30-configuration file for the S30 board. Both files are optimized for each variant.
I don’t remember the coolant pin, but it might have changed on the S30 board.
No, the 370.000 is the correct value. Be aware that the Pro Max (that you have, right?) has a working area of 370 x 363 mm. The frame is the same as the older models that could make use of the full space of roughly 410x400, but the module of the Pro Max is such big that you lose parts of the workspace. Limit switches subtract another few millimeters.
You can test yourself by home the laser and moving it manually until you hit the frame. Then set your dimensions accordingly.
The S30 file is already optimized for a Pro Max, no need to change anything, just tune the parameters if you like to.
I just checked again. The MKS DLC32 has no coolant pin in that sense. You can connect a relay, using any available pin. Most people use the I2C-pins for this, but in my configuration file, those are used for the display. Check here on how to connect a relay for air assist control to the MKS board: Relay Control - Diode Laser Wiki The S30 board has a dedicated coolant pin for the automatic air assist.
Additionally, the MKS board uses a different control scheme for the stepper drivers, so those configurations also differ significantly.
Awesome! Thank you so much for your contributions, you are a wealth of knowlege in this area!!!
It appears that everything is 100% working. Air assist, homing, limit switches, stepper motors, communication. I couldn’t be happier now!
I guess I dont need a relay or anything since its 100% working.
I have noticed little bugs here and there related to FluidNC but nothing that was stopping and destorying my work like the original S30 Pro Max firmware.
