I have cnc router that uses a Spark Concept xPro V5 with a spindle. I wanted to add a laser to as well so I can expand my projects. The xPro has FluidNc on it with functioning config for a spindle with a RS485 connection. I was hoping to the laser using the PWM and switch between them using the switch on the xPro. The RS458 uses gpio.4 as the txd pin and the PWM uses gpio.4 as the enable pin. When I add the PWM to my configuration, I get an error that gpio.4 is already used and the device seams to error out. Is it not possible to have two setups and switch back and forth? I am really hopping someone can help me set this up correctly. I will include my config.yaml file. Thanks
You can have multiple 'spindles', but you must use separate i/o pins for each one: Spindles It's a while since I've delved into FluidNC, and don't actively used it, so can't really offer any help, sorry. I have the following YAML file that switches between a LASER and a relay controlled spindle, but I'm still using grbl_ESP32 to do the same thing because it's working, and I'm not going to fix it You could try Bart's FluidNC support forum on Discord: Discord - Group Chat That’s All Fun & Games Edit to add: The OpenBuilds stuff runs GRBL HAL - I don't know whether it supports multiple spindles.
Perhaps I need to find a work around for my gpio4 and use a relay for my pin for my laser. I will need to think on that. I tried the dicord but I keep getting logged out and account cancelled. I will try again.
FluidNC supports multiple configuration files. You can put the configuration with the pin setup for the spindle in the main yaml file, and then a separate yaml file with just the configuration for the laser (therefore allowing gpio4 to be used just once in each case). You can write a macro to set the config file to be the alternative config, and then force a reset which will load in the new configuration. If you use say 'macro 0' for doing the configuration switching, then each configuration file can be set to switch to the other. If you have say config.yaml defines the spindle, and config_laser.yaml has the laser option, then in config.yaml the macro section would be something like: macros: macro0: $Config/Filename=config_laser.yaml&$System/Control=RESTART then in the file config_laser.yaml you use macros: macro0: $Config/Filename=config.yaml&$System/Control=RESTART The macro can be called from the gcode window with $RM=0 or, if you have a spare line, could wire it to a switch. When the machine changes over to the new configuration, it will do a fluidnc reboot, and will squirt out into the command window of whatever interface software you are using a bunch of messages: the spindle option is usually listed near the bottom so a quick visual check is always worth while to see if it has changed over to the one you want! If you can find a different spare pin other than gpio.4 that could be used for the laser PWM, then fluidnc supports multiple spindles and you can swap between them in the same GCODE file, which means CNC and then laser is possible in one Gcode file. EDIT, I did not notice before, but after re-reading your config file, gpio.13 seems to be wired to a switch, which is configured to trigger macro0 already
EvanH, I really like this idea. The FluidNc dicord made it sound like this was not possible. On a very high level I understand what you are saying, I am not sure I have the knowledge base to fully set this up. Is there any way you would be willing to help me?
Yes, no problem. I have attached two files; upload both (as .yaml versions) and fingers crossed, sending the command $rm=0 should switch between them.... For the laser option, I have made a couple of tweaks to the suggested configuration. The first is to use the 'laser' spindle type rather than just pwm as fluidnc may make other changes to the operation (such as M3/M4 behaviour etc.) that may not necessarily occur if just a pwm type is used. I am not sure if there are many differences yet, but there could be in the future. The second is to add a line at the end to turn off the laser if there is a fault condition. Good luck! Evan EDIT: I have just noticed the laser in your original file was set for S100 to give 100% output. I cut and pasted my configuration over, which is set for a more common S1000 to give 100% output, so you may want to change that back to 100=100% in the mapping. I also forgot to mention, on the macro definition line, any spaces around the '&' between the commands, and stray spaces at the end of the line can cause odd effects! I also forgot to add that when you switch between the configurations, you will also have to flick the switch to swap what gpio.4 is connected too. The latest version of fluidnc (3.8.0) has much more scope for running files from macros, and it may be possible to trigger a file on switch-over that puts out more 'text' as a reminder to flip the switch for gpio.4... I am exploring what may be possible... I have also just noticed that the PWM frequency for the laser is at 10kHz. It may be worth checking that as although some lasers can run with PWM that fast, many cannot (mine is happiest at 2kHz and has stopped working by 5kHz).
This is awsome! I am away from my machine for a about a week, but will look into it when I get back. I know in the past I havd struggled to load anything not named config.yaml through fluidterm. It also seems to over right any file that exists so I am not sure how to place 2 files at once.
In Fluidterm, do CTRL-u to upload the file, or it can be done via a web interface if you have that set up. On windows, the fluidterm option pops up a file browser where you can select a file (just one at a time), and then it asks to confirm the filename (just hit return) and it should upload. On Linux, I find I have to type in the filenames, so that can be a bit more tricky (for me anyway!). If you do have the web interface running, then there is a tab where you can manage the local files, and they can be uploaded there. Once the files are up, reboot and it should pull in the new versions. Do you know what version of fluidnc is running on the machine? Evan
I am not sure of the version of fluidnc off the top of my head. I am using widows and have loaded new config files with the ctrl+u. It seems to give me an error when I load anything but config.yaml I should set up the web interface as bringing my controller on the house to update is getting old
Ok, I finally got this today. I am running V3.7.6 I was able to run the upload on both files without getting an error. I am able to confirm in fluidterm that the file config.yaml is there with all the correct info after restarting the device. In cncjs I run the $RM=0 the device does not restart. I manually restart it and it is still running the config.yaml and not the configlaser.yaml. What should be my next steps?
V3.7.6 is older than I tested on. I have a fluidnc test-rig that I checked out the file operation on, and will install 3.7.6 and check to see what happens. I seem to remember there may have been updates on the macro operation around the time that 3.7.6 was released. You should be able to run $Config/Filename=configlaser.yaml from fluidterm, then reboot to force a switch over manually on 3.7.6, even if the macro operation does not work on that version. Evan
I have just run up 3.7.6 on my test rig, and the macros work ok. One thing to check is that when you do: $LocalFS/List to list what is on the local flash, it should have something like (mine may have different file sizes as I have had to change some of the limit pin definitions to prevent my rig from crashing). [FILE: configlaser.yaml|SIZE:5353] [FILE: config.yaml|SIZE:5369] I have been caught out in the past by an errant space being added either before or after a filename, which can prevent files from loading. Hopefully it is something that simple! If switching using $Config/Filename=configlaser.yaml (plus a CTRL-R reboot) and then switch back with $Config/Filename=config.yaml work, then it must be the macro not running properly. The macro also runs $System/Control=restart which should cause a reboot. Evan
Ok, my error was inputting the command for the macro, in cncjs I tried to paste it and it did not work. I typed it in by hand in both cncjs and in fluidterm and got it to switch file. Would you suggest I update fluidnc? Where do I find the current version? Do I need to erase everything or can the current version be installed over the old one? This has been a ton of help, thank you. Update, cncjs does not like the $RM=0. I did create a macro within it that looks like this and seems to work fine, so thats great. $Config/Filename=config.yaml $System/Control=restart I am waiting on a few parts to arrive so that I can wire this all up and test it. You mentioned that you had to run a lower frequency, what are the indicators that I will need to watch for to know I need to change mine?
I have not used cncjs, but yes the macro you have created (and I assume a 2nd to switch back) in cncjs looks just as useful for switching the files. For most diode lasers, they switch on/off very fast so the PWM usually defines a duty cycle where the laser is on at 100% for a period, then at 0%, rather than creating an analogue change in the power output. Running the PWM rate too slowly may make the on/off modulation visible if you have super fast travel speeds like some of the bespoke laser rigs. For most dual-purpose cnc systems, the gantry cannot move fast enough for the dot-space modulation to be seen: at a 500Hz PWM modulation and 10 metres per minute, the laser would modulate on/off in 0.33mm, which can be visible, but in practice for 'burning' applications would not be seen. I only saw it as I was trying to expose a UV coating on a PCB and in that case, 0.33mm in relation to fine circuit track spacing was a problem. Running from around 1kHz and up for my system is ideal (where 10m/minute is my top speed), and if I go for a faster PWM on my laser, then it starts to have problems switching to be full-on within the short duty periods. That is, if you took the PWM as the S command in GCODE running from 0 to 1000 to give 0% to 100%, at 1kHz PWM, setting the power to 'S1' would be 1us on and 999us off. If you ran the PWM at 10kHz, but the minimum time the laser can switch in was 1us, then settings from S1 to S9 are not likely to turn the laser on fully, and you may see at the low power levels a very rapid fall-off in effective output power. I have a couple of lasers and the minimum switching time of one looks to be about 2us, but I have a cheap one that takes around 50us to first turn on, but then the on-time can be changed in small increments. When I run the cheap laser at 1kHz PWM, nothing happens for S1 to S50, and then it suddenly starts to come on, and can be controlled quite well thereafter, with the power output appearing to be more linear with changes in S for the higher powers. If I run the cheap laser any faster PWM, then the low-end lack of power becomes more problematic. For exposing circuit boards, the low-duty performance is important for me, but for burning applications, I never go that low down so it matters less. Really you have to do some test burns at different PWM rates to work out what is best, and it can be as simple as setting it to 1kHz and if it works for what you make, use that I ran tests on UV pcb coating so I could see the laser dots, but if you want to 'tune' the PWM, a simpler way to check would be to write some GCode to just burn a line at say 10% power for 10mm, then 20% for 10mm and so on to say 50% with the laser running at 1kHz, and a travel speed slow enough that the 10% section just makes a mark. Then tweak the configuration so the laser runs at 2kHz, and burn another line next to the first, with then faster or slower PWM runs if needed. For low PWM speeds that are all slow enough for the laser to respond fully, the burned tracks should look the same. When the PWM speed starts to get too fast, then the 10% line will suddenly start to fade, but the higher percentage regions will be the same. If you keep pushing the PWM higher, then more of the lower percentage regions will start to dim as the burn profile becomes very non-linear. The best PWM setting can then be adjusted to find the fastest rate that preserves useful low-power performance, but tuning like this does take a while and the simpler option of 'try a few PWM rates and find one that works' is much easier! For the firmware update, there is a list of all the builds (including 3.7.6) and what has changed between builds on GitHub at Releases · bdring/FluidNC and the installers can also be downloaded from there for each version. I would suggest looking through what has been changed before considering updating as if 3.7.6 is working, then at least you have a cnc machine that runs Evan
This is amazing info and I am thankfull you shared it. I have been busy getting thingthing ready to run my 1st sample. I would like to verify that I am understanding your post correctly. The limitation to the speed of the laser is the speed of the pwm, so I will want to fine tune my config for optimal use? The S comand determis the laser power and is created by turning the laser on and off, not by actually limiting the output power? So the laser will actually make lots of little dots similar tonold comics? I plan to use laserburn, which has a test file already created. My plan is to bring the frequency down to 3khz in my config and then run the test, which makes lines at different speeds and powers. I am wondering if this seems line a good plan to start? Thanks again Evan, this info has been great.
The 'depth of burn' is determined by the energy delivered to the surface, which is power times time. For the cnc, that translates into both the power of the laser, and how long it spends in one place. So controlling the PWM value will essentially change the power of the laser, and then the feed rate of the axis alters how much time the laser is shining on one place. If you ran the laser at 100% and say 500mm/minute, and then ran the laser at 50% but 250mm/minute, 'in theory' the lines burned should be to the same depth/shade of grey as all points on the line have had the same amount of energy applied. In reality, there may be small differences as materials that have a partially reflective surface may need quite a lot of energy to start to burn the surface, that then becomes black and is much less reflective, and so needs less energy to burn deeper, and also when moving slowly with a low power laser, some surfaces may cool quite quickly, especially if you have an air assist. The Gcode that the laser software produces will have M3 or M4 commands in it to enable the laser operation, and then the PWM output percentage is changed with the S code. Generating Gcode from the test file is ideal. Yes most lasers turn on/off at the PWM frequency and if you could see the pattern, it would look like the dots in comics. For engraving, the size of the 'dot' that is burned usually ends up larger than the gap between dots, so they all blend together and you cannot see the individual dots, but if you are engraving pictures and select say 5 lines per millimetre, then you would probably see individual lines engraved. Yes trying a run at 3kHz sounds like a good place to start. If you get a test pattern out, try doubling the rate to 6kHz and see if the test patterns look the same. If they do, then try going faster, but if the patterns are different, try a run at 1kHz etc. It will be where the output is at a low percentage that the differences due to PWM frequency are most noticeable. Depending on the maximum power of your laser, you may want to adjust the axis feed rate too. As a starting point, a 1Watt (optical output) laser diode running at 100% and a feed rate of 500mm/minute should burn a nice dark patch on MDF, but not burn it too excessively!
Evan, I finally got things put together and have been learning a lot, there is more than I thought I would have needed to know considering my CNC was operating well. The macro and two configs seem to be working well. I have not been able to test the laser yet as I am still setting a few things up. I do have a new question. Here is the situation. I am using CNCjs as the software for my CNC and form what I have gathered the CNC uses a negative coordinate system where 0,0 is in the top right corner. I am going to use Lightburn which does not use a negative coordinate system. They have documentation for a command (G10 L2 P1 X-310 Y-180) that creates an offset to allow things to work. The documentation states that the command needs to be reversed (G10 L2 P1 X0 Y0) when going back to the CNC. In my mind, if I am changing confing files, wouldn't the G10 command stick with the config loaded when it was given? so if I run my laser macro, run the G10 to make an offset, then run my laser macro the offset would not be there as I did not run the G10 for config? I am really hoping I only need to run this once for each config and not have to run it each time I load a new config.
Yes the convention is for a negative coordinate system with the Machine Coordinates of X0 Y0 being the top right-hand (and Z0 being the Z axis as high up above the workpiece as possible). Not all machines are configured this way, but I am assuming yours is. When Gcode is run, it operates in the Work Coordinate frame, with the G10 command being what is needed to generate an offset between the work and machine coordinates. The G10 command is the normal route for telling the machine where the work zero origin is, irrespective of whether CNC or laser is being used and it is a non-volatile offset, so when you power cycle the machine, the offset is remembered. Some gcode senders use the G92 command that also generates an offset, but is volatile and lost between resets (I know Candle uses G92, not sure about CNCjs). Using G10 to do the machine->work offset is the much better route as jobs can be restarted from where they left off more easily if an issue occurs (I do know UGS uses the G10 command when you hit the 'set zero position' buttons, and it will work with your fluidnc control system ). If you used the following sequence: 1. design your laser job with 0,0 being the lower-left corner and X and Y being positive values going right and up (i.e. the normal Lightburn layout) 2. home the machine to the top-right corner of the workspace 3. Jog to where you want 0,0 to be in the laser system, including in Z to get Z at 'Z0' where your laser is nicely focussed 4. run the Gcode G10 L20 X0 Y0 Z0 The gcode will create the offsets necessary for where you jogged to, which may not be X-310 Y-180... 5. run the laser code (or CNC for that matter, it is the same process for setting Workspace zero) The G10 command uses 'L2' to take the specified X,Y values and puts them as the offset to use. The 'L20' option takes where the machine is at the moment, and calculates what the X and Y (and Z) offset values need to be to make the current point in work coordinates the value specified in the command (i.e. the X0 Y0 Z0 as above). The 'P1' part of the command you have seen is used to set the offset in the G54 workspace; that is the default so for most jobs the 'P1' can be left out. I do not use lightburn, so I am not sure if it puts unusual Gcodes in the header section of the milling file, so it may be worth saving off the output from a simple Lightburn design and then have a look at what Gcode header it is producing.
I was able to find some more info, Lightburn has a CNC Machine option, when selecting that I was able to Set G10 L2 X0 Y0 and have the machine move as expected. I then went back into CNCjs and validated that the machine still moves as intended. One thing the instructions from the Lightburns webpage was to add $10=0 after the G10 command. Now when I type $$ in the command line it only reports back $10=0 instead of a list of $*. Is this something that happened when I made $10=0 or is this something in FluidNC? After further looking, FluidNc's support page says it only supports $10?
Fluidnc does odd things with the $10 and other classic GRBL commands, especially the earlier versions of fluidnc as all the settings are handled through the .yaml files, rather than $ options. Only a small subset of $ options are reported; the later versions of fluidnc report more of them to help with compatibility with CAM software, but there are issues still. If $$ is now reporting the value expected for $10, then hopefully it should run ok. There should be a fall-back option in Lightburn where instead of using Lightburn to drive the machine directly (which is were access to $10 may be needed), a Gcode file of the laser path could be saved off, which can then be sent to the machine via CNCjs or other Gcode sender that can talk to fluidnc more natively.
Some really exciting stuff is taking place. The laser moves as anticipated, I am able to turn it on manually with the light burn software. It has came with a pre programed material test file to help determine speed and power. I am running it now but the laser does not seem to be turning on. Could this be the frequency setting in the config? Seems odd as it did turn on with the manual button. **** Ok, I spoke a little to soon, the test worked. However the only visible markings are at 10mm/s across the full power range and and some of the 42mm/s at higher power ranges. None of the text showed up in the test from the built in material test. Are the changes I can make or this a limitation of my laser? Does my Frequency setting have anything to do with this?
It may possibly be that the PWM frequency is too high, but it may also be that the pre-programmed file is doing something that fluidnc or the laser is not quite expecting. A direct 'manual' check is to try from the gcode command prompt (or fluidterm) M3S0 G1 G91 X0.01 F1000 which should fire up any laser enable signals (gpio.4 if I remember from your configuration), but have the pwm set as zero (so laser is not shining). In laser mode, often the laser is off unless the machine moves, hence the G1 command, with the G91 putting the machine into incremental movement mode. Then trying S100 X0.01 should be enough to turn the laser on to 10%, with then higher values for S turning it on brighter each time, up to S1000 X0.01 causing the laser to fire at 100%. Sending M5 should shut the laser off. Before you run the above commands though, check that the laser is firing onto something that is not going to go up in flames if the laser is on for a short period while you send the commands! (Been there, done that...) The other 'gotcha' that can occur is if the tool number is changed to something that is outside of the laser tool number range, but that should not happen with your configuration file as when in laser mode, the laser is the only spindle defined.
If the laser only comes on when S1000 is sent to drive it to 100%, then it could well be the PWM, as at 100%, the pwm line is on all the time and the actual pwm frequency makes no difference.
What is the PWM currently set at? If it is running too high, then 100% on should still be full power, but other settings would appear to lose power very quickly. What wattage is your laser?
My PWM is set at 3k, per the vender (ebay) it is a 10k hz 80W I made a quick smiley face, ran it at 100% and 50 mm/s, got a very deep burn with almost solid color and a lot of excess side burn, I turned it up to 110 mm/s and still got a deep burn with less uniform color and still some excess burn on the sides. I will try crate a file for a single line burn increasing power but not speed and see what happens. I will post pictures then as well
Ok, I ran a test of lines and this is what it looks like. The top line is at 100% speed 100mm/s, I stepped down in 10% increments. Then started at 10% at 50 mm/s and moved up to 100%. This is at 3k. I was able to notice a difference in light brightness at higher % but lower seemed to be random as it changed and not really able to see a brightness difference. Material is pine. Any tips? On a side note, I would like to set up fluidNC so I can update the config via wifi, but I do not understand the wiki instructions, any advice would be greatly appreciated.
OK, update, I ran a successful test sample and it looks good. I had the program set to 100mm/sec at 65%. But I found out that my machine was set to run at a max of 41mm/s. I did try to bump it up in the config but the machine does not like 100mm/s. As it moves at that speed it racks out of square and locks up. Not sure if I need to adjust something or if the machine is just capable of those speeds. It is larger at 1500x1500mm Here is a my trial sample, its on pine so the there seems to be variation along the grain. Let me know what you think, what I can do to make it better. Thanks
Awesome, looking good! Yes woods with distinct hard/soft parts of the grain can give variation in burn depth/darkness, so that is not unusual. I try to minimise how much I watch the laser, as through my glasses it still looks brighter than I am comfortable with, so on mine too I cannot really see much difference between low and high powers: I often go on the quantity of smoke as to whether I have got the setting right or not! The 2500mm/minute set in your original config should be handled by most drive systems at 1500mm length (e.g. belts or screws), but going faster will depend on how the drive system is configured. That is, belt or rack & pinion systems can usually go a fair bit faster, but 1500mm long lead-screws or ball-screws may need to be tensioned, be large diameter, or have a steep-pitch to move the axis faster without whipping. The lockup of an axis is often due to accelerating too fast to reach the wanted top speed, but your config had an acceleration of 50mm/s^2, which is not usually excessive. When you said it racked, I am assuming it was the dual-motor Y axis that had issues. What drive method does your machine use (i.e. belts or screws etc.) and if you manually move one side of the axis, is one side noticeably stiffer than the other? As stepper motors lose torque at high rotation rates, if one side gets a bit too stiff, it can potentially stall a stepper when turning at speed. On my machine, I have ball-screws (one of which is 1500mm long) and have a 1:2 ratio belt-drive system so the stepper spins at half the speed of the screw to better exploit the higher-torque at lower step speeds, allowing me to run that little bit faster. For the fluidnc web updates, I have only ever done it via a cable unfortunately.
The machine is a lead screw system with a dual Y. I have had very little problems running the cnc with the 2500mm/min. Learning about the laser and the setup is what actually got me to realize why my cnc projects were taking a bit longer than calculated. The burned logo was projected to be done in 5 is minuets with the laser but took almost 20, that's when I realized that the speeds I was giving was faster than the config. So I made some adjustments and tried to move the machine. I up the x and y to 6000 and tried to move it from the back right to the front left. It got about half way, stalled in the x, the left y stalled, the right y moved a little further and stalled. I was able to move it in small increments to the front so I could square it back up. I Dropped the speed to what I do not remember, and tried again. The y moved fine but it still stalled in the x. I set it back to the 2500 and am running a new test and all seems to be fine. I am going to slowly try and work my way up to 4000mm/min and see if I can find a happy place. I may try and retention my lead screws also.
I have ball-screws and linear rails on my current machine, but have had lead screws and rollers in the past. Even with ball-screws, debris finds its way into bearings that should only have grease and things can get that little bit stiffer for the steppers to turn. I have a couple of different configurations I use where one has a lower top-speed but high acceleration that I use for 3D carving (and also sometimes on the laser for cutting), and another that accelerates more slowly, but can reach a much higher top speed for when I have long-straight paths and few turns. One of my ball-screws does have a slight bend in now though as one time when I was tuning the speeds, an errant extra zero on the maximum travel speed made the machine move far faster than I intended. I don't know how fast it got, but the screw whipped like crazy and luckily the motors stalled before it reached the 100m/minute I had put in!
