i also fixed the z leadscrew backlash problem with the mpcnc but i approached it differently. i bought a second brass nut , flipped it upside down from the original, spun it backwards till the holes lined up and used longer screws to take out the backlash. same as your method but like a spring pushing the nuts together instead of apart.
That digital dial indicator was an impulse buy like 6 months ago which has now been fully, retroactively justified 😅 . It was genuinely helpful in troubleshooting that stiffness issue. Thanks for watching!
This is a great video, thanks heaps! FYI you can chuck a little masking tape over the corresponding pin in the USB cable to save yourself having to slice the cable open
@@ObsessiveEngineering Checked out your build time lapse, really nice. I'm half way through my MPCNC build and will likely come back, really like your upgrades, would love to get mine cutting aluminium as cleanly as yours!
top class video, much appreciated. Building a CNC myself right now but I am using the approach which u see in onefinity. I use 4 pcs 1100 mm 40 mm solid rod linear rods for Y axis and 1700 mm of the same for the gantry. Not sure about the Z axis at the moment but it's pretty straightforward. I'll add a laser as well and possibly some other attachments but not sure. It will run on rootCNC, a grbl based solution and the motors for all all axes with encoders, to make it more reliable I hope. The movement is done with 2 x 16/10 ballscrews on Y and 1 x 16/10 on the X axis. The base and gantry will prob be printed for the beginnig or made of some plywood but when I'm confident it works it will be replaced with solid steel and/or some epoxygranite cast parts. We'll see. Have some fun time on that beast you built man!
Because you asked, I just posted it to printables.com! I parameterized the design in Fusion360, but it's a very elaborate design, and the parameterization doesn't fully work. You are stuck with my ~17"x13" work area if you use it (or quite a bit of tinkering with the design). That said, it can still be used as a reference to design your own--I am quite happy with it, not sure I would change anything about it! www.printables.com/model/1062115-mpcnc-enclosure-design Note: you also asked about "build process", you can see my build timelapse here, which should cover most of that: th-cam.com/video/GPkXe8oY7z4/w-d-xo.html
For cutting aluminium you'd get a lot more gain by replacing that router with a proper spindle that can achieve much lower RPMs. It'll also probably be a fair bit quieter too.
I'm not sure about that, but check my logic: X and Y axes are moved only by two NEMA 17 motors each, which aren't very strong. The frame is also not super rigid by most CNC machine design standards (though I obviously made it a bit stronger through these upgrades). The entire thing runs on a single 48W power supply. Through experimentation, I can confirm that I can't push the feeds and speeds with much further without occasionally suffering skipped steps. OTOH, the router I have on there is 600W. By running high RPMs and surface speed, I'm essentially shifting much of the cutting stresses from the XY motors and frame rigidity, into the spindle motor. The faster it spins, the less force I need from the motors and the less frame deflection I get. Even though the chips aren't huge, it's not dust either, which means I'm not being (totally) irresponsible with feeds and speeds, I just have to stick with single-flute carbide cutters. Slower RPMs and larger chips would probably be ideal, but I'm not sure the machine can take it. Also, this is a budget machine, and a functional router for $100 is pretty good value. If I ever decide to build a beefier/sturdier/expensiver machine (PrintNC? RatRig mill?) I'll invest in a more professional spindle.
Nice upgrades! What Kobalt router model are you using? Reason why I ask is because I have a Kobalt router and was hoping to commandeer it for my build, but always thought it was too heavy. Would be cool to know if it’s the same as yours.
I bought this one directly from Lowes. Read the linked article, apparently designed in coordination with the MPCNC guy. Only $100, and I'm m extremely happy with it so far. I've probably done 50 hrs with it. www.v1e.com/blogs/news/kobalt-router-featuring-the-lr3
The work area is about 17"x13" (and 3" Z-travel). My modifications with the linear rails would've reduced that slightly, but I made the base new base a bit bigger to compensate.
uhh... do some research first. Aluminum extrusion vrail is cheap and easy to find and builds are much more rigid. the MPCNC is a fairly old project that I simply do not see recommended anymore.
@@electronicsandewastescrapp7384 I would argue that it's a relevant project, but if your primary goal is DIY + cutting aluminum, I would lean towards PrintNC, though I don't have experience with it. I would promote MPCNC if you primarily want to cut wood and would like to be able to cut aluminum occasionally. The extra money I put into this upgrade was a great experience but I might've been better off starting with PrintNC now that I am focused on aluminum cutting. Obviously I made it work, though, and the whole system works way better than I otherwise would've expected
@@electronicsandewastescrapp7384 Additionally, the max cutting dimensions on the MPCNC are about 24" x 24" IIRC. They recommend the "Low rider" from the same group if you want to cut bigger areas in wood. I built an MPCNC and was rebuilding it as a newer model, but decided not to continue assembling it. I'd rather cut bigger things, and it just can't.
Did you use the standard PI image that has octoprint and cnc.js (I have that set up currently) from V1, there also seemed to be a question if the CNC.js was still an active project. I am just about to make chips with my mpcnc so I am at the suck end of the learning curve.
I did not use a canned RPi image. I used the basic raspberry pi OS and manually installed CNC.js on it. I'm very experienced with Linux and RPi, I didn't even look if there was a preconfigured one. Given my Linux experience, I was actually going to do LinuxCNC, but it requires a computer with a parallel port, and that just wasn't with the effort for me.
There are screws that drop through the lead scream nut into the top of the core so that the lead screw isnt just free spinning. Did you use longer ones that drop through the entire double lead screw setup? Please advise
Almost! You do need longer screws to get through the bottom printed fixture half and lead screw nut, instead of just through the nut. And you separately need screws to fix the top nut to the top fixture half. Then 4 more to hold them together after tightening. So yeah, it's a lot of M3 screws for this thing...
It's a great job, but let's not forget how we got here - the community behind MPCNC is amazing, and going off on a few imperfections isn't the look. I don't know if I would lean into CNC.js that might, as they have essentially abandoned the project.
They were printed flat. I believe the thinnest part was 3x 0.3mm layers, and I made sure that it printed with an orientation so the first and third layer were using lines directly perpendicular to the hinge.
Since my slicer is set to do infill at 45deg, I rotated the part 45 deg in the XY plane to match. Then I previewed the gcode to confirm the first layer was across the hinge. That said, having the first layer lines 45deg relative to the hinge may be totally fine with multiple layers. I did it this way because I've made living hinges for fun with a single layer and that is best if it's perpendicular. May not matter here. Let me know how it goes
Hi Alan, Did you need custom firmware for the tool centering fixture there, or is that a standard setup for CNC.js? Did you make the probe or did you make it with the cnc? EDIT: Saw you machined the block. Can you describe the math behind finding your position and then how you're implementing that through CNC.JS?
Yup, I machined the XYZ plate from a 10mm aluminum plate, to roughly match the specs of the $50-100 ones I see online. I got the G-code for the macro from this video that shows it in action, and has a link to the source code in the description: th-cam.com/video/YRHsWt38QY8/w-d-xo.html The gist of the math is this: every circle is fully defined by any three points on it. So if you start the probe anywhere in the circle, and probe in 3 different directions, you will stop at 3 different points on the circle, which gives you enough information to compute the center and radius. The neat thing about this is that you don't even need to know the bit diameter. If you move the bit to the edge of the circle and trace it all the way around with any bit, you're just drawing a smaller circle that has the same center as the XYZ plate circle. So the larger the bit, the smaller the circle you're probing, but you'll still compute the same correct answer from the 3 points. The G-code has the equations in it. That's one reason I like CNC.js, because you can use variables and equations in macros, which makes it very versatile. For instance I made a macro for simply cutting out a rectangle. At the top of the macro I have the variables for the bit size, rectangle length, width, total depth, depth-per-pass, and max feedrate. I just have to tweak those few numbers at the top, put in a wood sheet, set the XYZ offsets where I want the bottom left corner, and run the macro.
@@ObsessiveEngineering Thanks for the explanation and for the source! I wonder if one adds additional circles At say (0,50) and (50,0), they could account for the stock's skew, like so: th-cam.com/video/bhZdbgM6S70/w-d-xo.html
@@ObsessiveEngineering Thanks for the explanation and for the source! I wonder if one adds additional circles At say (0,50) and (50,0), they could account for the stock's skew, like so: th-cam.com/video/bhZdbgM6S70/w-d-xo.html
At first I thought I would lose some Z-height but because of the geometry of the machine, the core never has a reason to travel low enough to interfere with it. So it turned out to be a compromise-free upgrade. It's just two TR8 leadscrew nuts instead of one. So nothing changes about the lead screw, though I did switch to a new one (of the same size), simply because i assumed the old one was worn down based on that original backlash. Thanks for watching!
thanks for the details. I’ll give it a try a soon as i have some abs that matches my mpcnc color. did you ever get any feedback from the mpcnc team on this mod? this whole backlash thing seems intentional or an oversight?
Honestly, having access to the big CNC was a luxury, but it would've been doable without it. The grooves for the T-track weren't really necessary -- they're only 2mm deep, primarily for positioning. And all the holes for everything else are normally measured out by hand anyway (per the original MPCNC design/instructions). I only had trouble the first time because I didn't have a large carpentry square, and wasn't as rigorous as I should've been.
CAREFULL what you call this. The project owners threatened to take legal action against Thomas S. for an enhanced part he did and offered via Open Source.
Thanks for the warning. I heard about that saga and have mixed feelings about it. But worst case I'll change the title and take down the model for the linear rail mod. That said, i did post about the video in the V1E forums and he hasn't said anything so I might be safe 🤞
Just want to chime in here, that is not true at all. I use a share-alike license, I asked Tom to change his license to comply. He said no, we gave each other our point of view. He still said no, that is how it ended between us. Neither person threatened anything. No hard feelings even.
@@allted I don't think you are being as open as you would like us to think. After your conversation, Thomas did an entire video on DISASSEMBLING the MPCNC and shared his displeasure with the chats with you. We can tag that video here, so everyone can watch him take it apart and get their opinions. I think you may be being disingenuous about this, again that is my take on it. I would invite everyone to watch it and formulate their own positions.
(EDIT: I'm not aiming directly at any particular post in this thread, but simply addressing the high-intensity opinions I see about this everywhere else) I didn't really follow the saga very closely. As always, the internet is full of people exaggerating or reading exaggerated viewpoints on things and feeling strongly about them based on personal factors. I try to give each side benefit of the doubt, and I'm not going to take any purist views. @allted put his time and effort into the MPCNC, and makes money from it, as is his right to do. He never claimed it was open-source, just a freely-available set of files and instructions, which helps the community (even if not fully open-source), and also funnels people to his shop. As a source of income he wants to try to protect it. It's tough to do that in an consumer industry where open-source is kind of the baseline, especially for a DIY project. But that doesn't make it wrong. If you release things (semi-)publicly without a strong grip/claim/license on them, and refuse to publicly defend against people overstepping, the courts won't help you recover your intellectual property (IP) in the future. You kinda have to stand your ground or you just bleed out over time (in a legal sense). It's not fun to end up doing that publicly in a community like this, but it's kinda necessary to simply maintain the baseline stated from the beginning. And it's a mode of operation that many people don't understand, and don't want to simply because it's not "fully open". I'm sympathetic to the whole thing on both sides. It's not fully open-source, so some people will never like it, and others will simply think less of it. I think @allted has accepted that's the tradeoff for protecting his design and his income source around it. I would just hope that people could understand that in perspective, rather than trashing the whole thing because it doesn't conform to purist open-source ideals.
You can also kapton tape the USB connector to disable the power. This was popular a few years ago on certain 3D printers
i also fixed the z leadscrew backlash problem with the mpcnc but i approached it differently. i bought a second brass nut , flipped it upside down from the original, spun it backwards till the holes lined up and used longer screws to take out the backlash. same as your method but like a spring pushing the nuts together instead of apart.
The deflection measurement is so cool
That digital dial indicator was an impulse buy like 6 months ago which has now been fully, retroactively justified 😅 . It was genuinely helpful in troubleshooting that stiffness issue. Thanks for watching!
This is a great video, thanks heaps! FYI you can chuck a little masking tape over the corresponding pin in the USB cable to save yourself having to slice the cable open
Oh yeah, that certainly would've been easier. Thanks for the tip!
@@ObsessiveEngineering Checked out your build time lapse, really nice. I'm half way through my MPCNC build and will likely come back, really like your upgrades, would love to get mine cutting aluminium as cleanly as yours!
top class video, much appreciated. Building a CNC myself right now but I am using the approach which u see in onefinity. I use 4 pcs 1100 mm 40 mm solid rod linear rods for Y axis and 1700 mm of the same for the gantry. Not sure about the Z axis at the moment but it's pretty straightforward. I'll add a laser as well and possibly some other attachments but not sure. It will run on rootCNC, a grbl based solution and the motors for all all axes with encoders, to make it more reliable I hope. The movement is done with 2 x 16/10 ballscrews on Y and 1 x 16/10 on the X axis. The base and gantry will prob be printed for the beginnig or made of some plywood but when I'm confident it works it will be replaced with solid steel and/or some epoxygranite cast parts. We'll see.
Have some fun time on that beast you built man!
Would you be willing to share you drawing (and build process) for the enclosure?
Because you asked, I just posted it to printables.com! I parameterized the design in Fusion360, but it's a very elaborate design, and the parameterization doesn't fully work. You are stuck with my ~17"x13" work area if you use it (or quite a bit of tinkering with the design). That said, it can still be used as a reference to design your own--I am quite happy with it, not sure I would change anything about it!
www.printables.com/model/1062115-mpcnc-enclosure-design
Note: you also asked about "build process", you can see my build timelapse here, which should cover most of that: th-cam.com/video/GPkXe8oY7z4/w-d-xo.html
For cutting aluminium you'd get a lot more gain by replacing that router with a proper spindle that can achieve much lower RPMs. It'll also probably be a fair bit quieter too.
I'm not sure about that, but check my logic: X and Y axes are moved only by two NEMA 17 motors each, which aren't very strong. The frame is also not super rigid by most CNC machine design standards (though I obviously made it a bit stronger through these upgrades). The entire thing runs on a single 48W power supply. Through experimentation, I can confirm that I can't push the feeds and speeds with much further without occasionally suffering skipped steps.
OTOH, the router I have on there is 600W. By running high RPMs and surface speed, I'm essentially shifting much of the cutting stresses from the XY motors and frame rigidity, into the spindle motor. The faster it spins, the less force I need from the motors and the less frame deflection I get. Even though the chips aren't huge, it's not dust either, which means I'm not being (totally) irresponsible with feeds and speeds, I just have to stick with single-flute carbide cutters. Slower RPMs and larger chips would probably be ideal, but I'm not sure the machine can take it.
Also, this is a budget machine, and a functional router for $100 is pretty good value. If I ever decide to build a beefier/sturdier/expensiver machine (PrintNC? RatRig mill?) I'll invest in a more professional spindle.
definitely not enough credits for the science you're doing there. I'm looking forward to build my mpcnc and will include your upgrades for sure!
Freaking Boss level 💯
Nice upgrades! What Kobalt router model are you using? Reason why I ask is because I have a Kobalt router and was hoping to commandeer it for my build, but always thought it was too heavy. Would be cool to know if it’s the same as yours.
I bought this one directly from Lowes. Read the linked article, apparently designed in coordination with the MPCNC guy. Only $100, and I'm m extremely happy with it so far. I've probably done 50 hrs with it.
www.v1e.com/blogs/news/kobalt-router-featuring-the-lr3
Because of this video I am going to build an MPCNC! What is the cutting area of yours?
The work area is about 17"x13" (and 3" Z-travel). My modifications with the linear rails would've reduced that slightly, but I made the base new base a bit bigger to compensate.
uhh... do some research first. Aluminum extrusion vrail is cheap and easy to find and builds are much more rigid. the MPCNC is a fairly old project that I simply do not see recommended anymore.
@@electronicsandewastescrapp7384 I would argue that it's a relevant project, but if your primary goal is DIY + cutting aluminum, I would lean towards PrintNC, though I don't have experience with it. I would promote MPCNC if you primarily want to cut wood and would like to be able to cut aluminum occasionally. The extra money I put into this upgrade was a great experience but I might've been better off starting with PrintNC now that I am focused on aluminum cutting. Obviously I made it work, though, and the whole system works way better than I otherwise would've expected
@@electronicsandewastescrapp7384 Additionally, the max cutting dimensions on the MPCNC are about 24" x 24" IIRC. They recommend the "Low rider" from the same group if you want to cut bigger areas in wood.
I built an MPCNC and was rebuilding it as a newer model, but decided not to continue assembling it. I'd rather cut bigger things, and it just can't.
Did you use the standard PI image that has octoprint and cnc.js (I have that set up currently) from V1, there also seemed to be a question if the CNC.js was still an active project. I am just about to make chips with my mpcnc so I am at the suck end of the learning curve.
I did not use a canned RPi image. I used the basic raspberry pi OS and manually installed CNC.js on it. I'm very experienced with Linux and RPi, I didn't even look if there was a preconfigured one.
Given my Linux experience, I was actually going to do LinuxCNC, but it requires a computer with a parallel port, and that just wasn't with the effort for me.
Nice. 🙂
There are screws that drop through the lead scream nut into the top of the core so that the lead screw isnt just free spinning. Did you use longer ones that drop through the entire double lead screw setup? Please advise
Almost! You do need longer screws to get through the bottom printed fixture half and lead screw nut, instead of just through the nut. And you separately need screws to fix the top nut to the top fixture half. Then 4 more to hold them together after tightening.
So yeah, it's a lot of M3 screws for this thing...
@@ObsessiveEngineering thanks!
It's a great job, but let's not forget how we got here - the community behind MPCNC is amazing, and going off on a few imperfections isn't the look. I don't know if I would lean into CNC.js that might, as they have essentially abandoned the project.
The TPU higens, what orientation you printed them in?
They were printed flat. I believe the thinnest part was 3x 0.3mm layers, and I made sure that it printed with an orientation so the first and third layer were using lines directly perpendicular to the hinge.
@@ObsessiveEngineering Thanks! will be interesting to test live hinges durability, material and print patterns
Since my slicer is set to do infill at 45deg, I rotated the part 45 deg in the XY plane to match. Then I previewed the gcode to confirm the first layer was across the hinge.
That said, having the first layer lines 45deg relative to the hinge may be totally fine with multiple layers. I did it this way because I've made living hinges for fun with a single layer and that is best if it's perpendicular. May not matter here. Let me know how it goes
Hi Alan, Did you need custom firmware for the tool centering fixture there, or is that a standard setup for CNC.js? Did you make the probe or did you make it with the cnc?
EDIT: Saw you machined the block. Can you describe the math behind finding your position and then how you're implementing that through CNC.JS?
Yup, I machined the XYZ plate from a 10mm aluminum plate, to roughly match the specs of the $50-100 ones I see online. I got the G-code for the macro from this video that shows it in action, and has a link to the source code in the description:
th-cam.com/video/YRHsWt38QY8/w-d-xo.html
The gist of the math is this: every circle is fully defined by any three points on it. So if you start the probe anywhere in the circle, and probe in 3 different directions, you will stop at 3 different points on the circle, which gives you enough information to compute the center and radius.
The neat thing about this is that you don't even need to know the bit diameter. If you move the bit to the edge of the circle and trace it all the way around with any bit, you're just drawing a smaller circle that has the same center as the XYZ plate circle. So the larger the bit, the smaller the circle you're probing, but you'll still compute the same correct answer from the 3 points.
The G-code has the equations in it. That's one reason I like CNC.js, because you can use variables and equations in macros, which makes it very versatile. For instance I made a macro for simply cutting out a rectangle. At the top of the macro I have the variables for the bit size, rectangle length, width, total depth, depth-per-pass, and max feedrate. I just have to tweak those few numbers at the top, put in a wood sheet, set the XYZ offsets where I want the bottom left corner, and run the macro.
@@ObsessiveEngineering Thanks for the explanation and for the source! I wonder if one adds additional circles At say (0,50) and (50,0), they could account for the stock's skew, like so: th-cam.com/video/bhZdbgM6S70/w-d-xo.html
@@ObsessiveEngineering Thanks for the explanation and for the source! I wonder if one adds additional circles At say (0,50) and (50,0), they could account for the stock's skew, like so: th-cam.com/video/bhZdbgM6S70/w-d-xo.html
What is that alternative lead screw coupler you switched to?
I believe it's called a "flexible jaw coupling". Here's one listing on Amazon
amzn.to/3UZYM9H
do you lose any z-height with the 3d printed backlash thing? do you need a different size leadscrew?
At first I thought I would lose some Z-height but because of the geometry of the machine, the core never has a reason to travel low enough to interfere with it. So it turned out to be a compromise-free upgrade.
It's just two TR8 leadscrew nuts instead of one. So nothing changes about the lead screw, though I did switch to a new one (of the same size), simply because i assumed the old one was worn down based on that original backlash.
Thanks for watching!
thanks for the details. I’ll give it a try a soon as i have some abs that matches my mpcnc color. did you ever get any feedback from the mpcnc team on this mod? this whole backlash thing seems intentional or an oversight?
Where is all my GRBL 5x users at 😂
Wonder how much more annoying it'd be to make the upgrade without access to any other CNC.
Honestly, having access to the big CNC was a luxury, but it would've been doable without it. The grooves for the T-track weren't really necessary -- they're only 2mm deep, primarily for positioning. And all the holes for everything else are normally measured out by hand anyway (per the original MPCNC design/instructions). I only had trouble the first time because I didn't have a large carpentry square, and wasn't as rigorous as I should've been.
CAREFULL what you call this. The project owners threatened to take legal action against Thomas S. for an enhanced part he did and offered via Open Source.
Thanks for the warning. I heard about that saga and have mixed feelings about it. But worst case I'll change the title and take down the model for the linear rail mod. That said, i did post about the video in the V1E forums and he hasn't said anything so I might be safe 🤞
Just want to chime in here, that is not true at all. I use a share-alike license, I asked Tom to change his license to comply. He said no, we gave each other our point of view. He still said no, that is how it ended between us. Neither person threatened anything. No hard feelings even.
@@ObsessiveEngineering Your parts your choice, I have no issue with mods. You have been in the forums for a long time, I am sure you know that.
@@allted I don't think you are being as open as you would like us to think. After your conversation, Thomas did an entire video on DISASSEMBLING the MPCNC and shared his displeasure with the chats with you. We can tag that video here, so everyone can watch him take it apart and get their opinions. I think you may be being disingenuous about this, again that is my take on it. I would invite everyone to watch it and formulate their own positions.
(EDIT: I'm not aiming directly at any particular post in this thread, but simply addressing the high-intensity opinions I see about this everywhere else)
I didn't really follow the saga very closely. As always, the internet is full of people exaggerating or reading exaggerated viewpoints on things and feeling strongly about them based on personal factors. I try to give each side benefit of the doubt, and I'm not going to take any purist views.
@allted put his time and effort into the MPCNC, and makes money from it, as is his right to do. He never claimed it was open-source, just a freely-available set of files and instructions, which helps the community (even if not fully open-source), and also funnels people to his shop. As a source of income he wants to try to protect it. It's tough to do that in an consumer industry where open-source is kind of the baseline, especially for a DIY project. But that doesn't make it wrong.
If you release things (semi-)publicly without a strong grip/claim/license on them, and refuse to publicly defend against people overstepping, the courts won't help you recover your intellectual property (IP) in the future. You kinda have to stand your ground or you just bleed out over time (in a legal sense). It's not fun to end up doing that publicly in a community like this, but it's kinda necessary to simply maintain the baseline stated from the beginning. And it's a mode of operation that many people don't understand, and don't want to simply because it's not "fully open".
I'm sympathetic to the whole thing on both sides. It's not fully open-source, so some people will never like it, and others will simply think less of it. I think @allted has accepted that's the tradeoff for protecting his design and his income source around it. I would just hope that people could understand that in perspective, rather than trashing the whole thing because it doesn't conform to purist open-source ideals.
$600 for THAT? You have to be kidding!
Careful. The designer is an asshole and may harass you for doing mods to his design