if u want to calculate the amount of load the linear bearings can handle, refer to the datasheet of the bearing. What you want to lookup is max permissible static roll, pitch and yaw moment. In that, you just need to plug in the values of maximum moment(force x distance) each bearing will experience in a given direction.
1. What kind of acceleration do you need? (Minor weight savings can increase manufacturing costs). What is your expected cutting force to force needed for acceleration ratio? About one clamping setup - it's not always more accurate, if you remove a lot of material the part can bend from internal stress after unclamping. Also if you specify that front and back faces of the plate have to be parallel but use loose tolerance for thickness it might reduce cost (less machining to do) 2. The plate behind Z axis rails looks relatively thin, especially the corners. 3. You can think about other materials, like steel or granite (especially for the base) as they are cheaper and dampen vibrations better. The bed doesn't have to be machined, only the mounting points are important, you can plan or lap the work area yourself later! 4. Instead of making everything in one plane focus more on leverage effect. (use 4 carriages for z axis, it can extend above gantry)
With the big hole for the nut of y the axis you have weaken the structure. For the best stiffness put the z and y axis carriages back to each other on same plate and put the z linear slides on the back of the spindle plate
You get the minimum deflection off the machine when you get closer the ball screw to the tip that cuts the material, when the ball screw is collinear to the point that cuts then you have zero tongue from the cutting forces and ballscrew. When you put the ball screw between the rails you minimize nothing, it is the symmetry that make you thing that is the best placement.
deflection is handled not with ball-screw, but with rails. You want to have rails as far apart as possible, carriages as far apart as possible, and yes, you want to have smallest distance between cutting surface and rails. And this one is a tricky one to have. I just calculated that i will have 26cm stick-out of spindle below rail support I am also trying to make it as compact as possible, since this will minimize deflection due to small leaver effect. well,where are so many decisions regarding CNC constructions that i can't wrap my head around.
Indeed the stiffness is greater the more distance the carriers have, but the placement of the ball screw is what I am talking about. For example you can see that the designers of Bridgeport have put the screw for the knee collinear to the center of gravity off the moving part and not between the slides. If they have put it between the slides like in the old mahos, then they had to use more rigid square slides like on the mahos or excess wear of the slide will be the result do the greater torques on the slides.
The placement of the rails and the shape of the machine is determine the stiffness the machine will wave. but the forces on the machine is gravity and the forces from ballscrew and cutting forces, the combination off thaws forces is torques that trying to twist the parts of the machine. Put the forces in line and the torques goes to zero.
Another example, the ballscrew is not at center distance between the linear guides of Y axis, but at close to the bottom as possible. www.yamaseiki.com/images/milling/5axis/fmv/construction/fmv-construction-fmv-99.jpg
7:35 looks like it is impossible do that in spec so longer axis move smooth. and linear rails should be longer than skrew to get full movement and put 4 carts. that i would do lol
It would be great to see an update on the driver! You are using GaN transistors? 200kHz is pretty high switching frequency! Any reason why the typical 20kHz is not enough for this application? That filter seems very small, but working at those frequencies have that advantage!
I was pushing 2.2kW from that TI module at 500Khz with no problem. Also GaN can take heat. Normal working temperature can be 120C. Using 200KHz switching will yeld better results, yes, for 400Hz tone, you can do that with 20Khz, you just have high distortion , and it can be audible noise. I have GaN mosfets, and a will to use it. It just very nice transistor, ultralow Rdson, sub ns switching and so on. Also, this makes filtering simpler
But with the output filter I think is not possible to have sensorless vector control, right? I suppose you will need an encoder if you want FOC... I'm not 100% sure about that but if you can't read the unfiltered current I don't see a way to do it EDIT: I was curious about the output filter+sensorless FOC and actually there are publications on how to do it, interesting project!!
ballanux I see you know what you are talking. That's good. I am planing to put 16k encoder inside spindle with home index, so I don't need to read voltages/currents :)
Insomnious1000 Its from local hiwin reseller. Links will not help you. I ordered all parts based on hiwin website. Rails and screws are cut from rods, you have to machine ends too
Faidul Islam I am planning to develop VFD with encoder feedback. So I will have better control over power and cutting speed. Also will be easier to make Estop on fail conditions, like overload :)
Nice work if you want us to quote for machining then feel free to provide STEP files and tolerance required. The z axis looks a little week i think the longest part looks to be the thinnest?
Linas K you can not count the linear rail for strength, the rails actually bend from one mounting hole to the next this is why its recommended you machine a wall to press it up against and thats also why you run indicator all the way down rail when fixing. Im not sure on grade of aluminum you will use but after the spindle , spindle holder , z axis plate and servo are mounted you will have to consider that wight wanting to push down then add the force experienced when plunging into metal then add force of x y movements then finish this of with the vibration traveling through it. Then you can decide if 2cm is good. We have an old haas here and under all the rails are a min of 5 inches of cast iron (very small machine) Do you know the type of cuts you want to preform? Ie heavy removal low rpm or small passes high rpm?
I can't tell you right now. Since firm that makes it is a bit strange. They forget to fill half of my order, motor was damaged during shipping , and they are extremely slow to react. I can't recommend them right now. Also, just after i bought this spindle, i try to read some comments about this firm, and yes, many experienced same problems as I, so not good. Hope it performs , sine 3k$ is not cheap
Hi I can see that it is very long time ago that you have made this project. I dont know how long you have made it. But I cross my fingers for you. YOU have good idea's. I just wish that you had a better dialog with your subscribers. I can see that many people have asked you different things but they never got any answers from you. Or is it me that is making a mistake...!!? Anyways GOOD LUCK
@@Kilohercas First of all I think you really deserve getting all the bets and pices that you need, and I thank you very much for your time that you put into these projects and share it with us. I just got my cnc machine that I have to upgrade if I really want to cut anything serious with it, As soon as I get it going and I have the Money I think my first REAL Project will be a DESKTOP CNC MACHINE. We shal see :). Thank's again and I do hope that you start making lots of video's about cnc ...:)
You definitely want 2 more Z carriages there for stiffness. How does the remainder of the design look like? I found your video via your other one, where you were trying to design an outrunner-BT30-ATC-spindle. Is the spindle you got now capable of low rpm/high torque - for something like 50 mm face mills in aluminium? As you seem to be looking for input, you might want to look at this build - czcams.com/video/bX2Eg18AzTA/video.html Details are over here - www.cnczone.com/forums/uncategorised-metalworking-machines/327914-gantry-mill-precise-milling-aluminium.html The specs of what you want and he has built match pretty closely and his results are impressive - take note of the reasons behind some design decisions (gantry stiffness, etc.). Be aware of the other video of his, where he mills some aluminium part for a bigger machine and where the swarf goes everywhere - something I'd want to avoid. Another example (fixed gantry) of a seemingly sturdy design (note the vertical z-plate stiffening), but also swarf problems - czcams.com/video/o__Uk_ek8Ec/video.html Personally I'm planning a portal milling machine with a similar bed size as you (1200mm x 800mm) but a Z height of 350mm and the y-rails/drives up on 'walls', similar to this - czcams.com/video/fIVL83AwXrE/video.html Some early musings can be found here - www.cnczone.com/forums/uncategorised-metalworking-machines/347540-arent-portal-mills-popular.html
Definately over-engineered but not as optimized as you think. It looks great but wastes a lot of money for material that is not needed or in the wrong place. 30mm rails for a small 2.2Kw router? That's about a Kilogramm per carriage that has to be accelerated. 25mm for Y and X axis is more than enough and 20mm for Z. If you increase the hight of your Z assembly and use four 20mm carriages instead of the two 30s you'll get more rigidity with less weight. And that huge piece of aluminium...huge waste of money. If you want stiffness then you want a large mostly hollow cross-section, not a thin solid piece. Two heavy 90x90 Bosch extrusions sandwiched between two 10mm cast aluminum plates ... simple, clean looking and stiffer than this design - at a fraction of the cost. The ballscrew - as nice as it looks, hidden in the middle, behind a super flat Z assembly - should be placed between the X guide rails and the Z assembly, as close as possible to the mass it's supposed to move. Think about inertia, leverage and resulting torque in the rails. Your current design does the opposite of what you're trying to achieve. I know that's a lot of critique ... but ...
In my case I get parts as support, so i had to chose with no regard for cost. It is overkill, and now its more a problem not a feature. I am making 3d model for XZ carage in completely different way (testing idea). Spindle and all bearings will be mounted inside a square box, it like exoskeleton, In that case, i will have 4 rails for z, and X rails will be centered between spindle, so rails will be placed along the axis in between mass center of the spindle. This should make more sturdy design as well as put X rails closer to the bottom, and it will be super easy to protect from dust, since everything will be inside hollow tube. But this will place lead-screws off axis of center of the mass, i don't know should i be worried :?
the way you are putting the longest linear rails is complex and you going to have a problem when assembling the machine physically this is due to the tolerance of the machine that you will use to mill the 300 euro component better put the linear rails on top than on the side avoid reducing the thickness of the aluminum if you want maximum stiffness use gearbox for your servos instead of belts to minimise backlash
Awesome, keep the vids comming!
I can't wait to see it Finished. Good luck.
Damn... you're ballin' - This is really industry-grade engineering. But with all the costs.
if u want to calculate the amount of load the linear bearings can handle, refer to the datasheet of the bearing. What you want to lookup is max permissible static roll, pitch and yaw moment. In that, you just need to plug in the values of maximum moment(force x distance) each bearing will experience in a given direction.
1. What kind of acceleration do you need? (Minor weight savings can increase manufacturing costs). What is your expected cutting force to force needed for acceleration ratio?
About one clamping setup - it's not always more accurate, if you remove a lot of material the part can bend from internal stress after unclamping.
Also if you specify that front and back faces of the plate have to be parallel but use loose tolerance for thickness it might reduce cost (less machining to do)
2. The plate behind Z axis rails looks relatively thin, especially the corners.
3. You can think about other materials, like steel or granite (especially for the base) as they are cheaper and dampen vibrations better.
The bed doesn't have to be machined, only the mounting points are important, you can plan or lap the work area yourself later!
4. Instead of making everything in one plane focus more on leverage effect. (use 4 carriages for z axis, it can extend above gantry)
With the big hole for the nut of y the axis you have weaken the structure. For the best stiffness put the z and y axis carriages back to each other on same plate and put the z linear slides on the back of the spindle plate
Any chance you would be willing to passing along the Solidworks files? I'm working on my own design too.
Do you have a link for that 3 phase filter?
Good effort dude by the way spindle is from which brand..?
You get the minimum deflection off the machine when you get closer the ball screw to the tip that cuts the material, when the ball screw is collinear to the point that cuts then you have zero tongue from the cutting forces and ballscrew. When you put the ball screw between the rails you minimize nothing, it is the symmetry that make you thing that is the best placement.
deflection is handled not with ball-screw, but with rails. You want to have rails as far apart as possible, carriages as far apart as possible, and yes, you want to have smallest distance between cutting surface and rails. And this one is a tricky one to have.
I just calculated that i will have 26cm stick-out of spindle below rail support
I am also trying to make it as compact as possible, since this will minimize deflection due to small leaver effect. well,where are so many decisions regarding CNC constructions that i can't wrap my head around.
Indeed the stiffness is greater the more distance the carriers have, but the placement of the ball screw is what I am talking about. For example you can see that the designers of Bridgeport have put the screw for the knee collinear to the center of gravity off the moving part and not between the slides. If they have put it between the slides like in the old mahos, then they had to use more rigid square slides like on the mahos or excess wear of the slide will be the result do the greater torques on the slides.
The placement of the rails and the shape of the machine is determine the stiffness the machine will wave. but the forces on the machine is gravity and the forces from ballscrew and cutting forces, the combination off thaws forces is torques that trying to twist the parts of the machine. Put the forces in line and the torques goes to zero.
Another example, the ballscrew is not at center distance between the linear guides of Y axis, but at close to the bottom as possible.
www.yamaseiki.com/images/milling/5axis/fmv/construction/fmv-construction-fmv-99.jpg
Gotta link to that spindle?
Does the belt/pulley off of the motor introduce extra backlash?
Yes, it should. But also, this is the best pulley/belt you can get. This is gates 25mm GT2 profile
7:35 looks like it is impossible do that in spec so longer axis move smooth. and linear rails should be longer than skrew to get full movement and put 4 carts. that i would do lol
It would be great to see an update on the driver!
You are using GaN transistors? 200kHz is pretty high switching frequency! Any reason why the typical 20kHz is not enough for this application?
That filter seems very small, but working at those frequencies have that advantage!
I was pushing 2.2kW from that TI module at 500Khz with no problem. Also GaN can take heat. Normal working temperature can be 120C. Using 200KHz switching will yeld better results, yes, for 400Hz tone, you can do that with 20Khz, you just have high distortion , and it can be audible noise. I have GaN mosfets, and a will to use it. It just very nice transistor, ultralow Rdson, sub ns switching and so on. Also, this makes filtering simpler
But with the output filter I think is not possible to have sensorless vector control, right? I suppose you will need an encoder if you want FOC... I'm not 100% sure about that but if you can't read the unfiltered current I don't see a way to do it
EDIT: I was curious about the output filter+sensorless FOC and actually there are publications on how to do it, interesting project!!
ballanux
I see you know what you are talking. That's good. I am planing to put 16k encoder inside spindle with home index, so I don't need to read voltages/currents :)
Is this cnc depot s30 spindle? I just ordered one and it looks identical without the blue
Yes
Very interesting project! Can you post some links to parts that you use in your build?
Insomnious1000
Its from local hiwin reseller. Links will not help you. I ordered all parts based on hiwin website. Rails and screws are cut from rods, you have to machine ends too
Awesome
Thank you for sharing this
It's 2019, any updates on this?
Looks like the hope wasn't achieved.
Yet? )))
Hi. Have no time and space. Need to buy apartment first, but it is hard to have CNC inside, but i will make it, one way or another :)
Do you devlop that's Vfd?
Faidul Islam
I am planning to develop VFD with encoder feedback. So I will have better control over power and cutting speed. Also will be easier to make Estop on fail conditions, like overload :)
Nice work if you want us to quote for machining then feel free to provide STEP files and tolerance required. The z axis looks a little week i think the longest part looks to be the thinnest?
monit111
2cm plate. And also massive 30mm rails. They are just as stiff as some one Z axis on their own :D
Linas K you can not count the linear rail for strength, the rails actually bend from one mounting hole to the next this is why its recommended you machine a wall to press it up against and thats also why you run indicator all the way down rail when fixing.
Im not sure on grade of aluminum you will use but after the spindle , spindle holder , z axis plate and servo are mounted you will have to consider that wight wanting to push down then add the force experienced when plunging into metal then add force of x y movements then finish this of with the vibration traveling through it. Then you can decide if 2cm is good. We have an old haas here and under all the rails are a min of 5 inches of cast iron (very small machine)
Do you know the type of cuts you want to preform? Ie heavy removal low rpm or small passes high rpm?
What spindle is that?
I can't tell you right now. Since firm that makes it is a bit strange. They forget to fill half of my order, motor was damaged during shipping , and they are extremely slow to react. I can't recommend them right now. Also, just after i bought this spindle, i try to read some comments about this firm, and yes, many experienced same problems as I, so not good. Hope it performs , sine 3k$ is not cheap
Linas K so, which company is it?
Hi I can see that it is very long time ago that you have made this project. I dont know how long you have made it. But I cross my fingers for you. YOU have good idea's. I just wish that you had a better dialog with your subscribers. I can see that many people have asked you different things but they never got any answers from you. Or is it me that is making a mistake...!!? Anyways GOOD LUCK
Hi. project still collecting dust, waiting for more space and better time, other projects takes too much of my time :(
@@Kilohercas First of all I think you really deserve getting all the bets and pices that you need, and I thank you very much for your time that you put into these projects and share it with us. I just got my cnc machine that I have to upgrade if I really want to cut anything serious with it, As soon as I get it going and I have the Money I think my first REAL Project will be a DESKTOP CNC MACHINE. We shal see :). Thank's again and I do hope that you start making lots of video's about cnc ...:)
You definitely want 2 more Z carriages there for stiffness.
How does the remainder of the design look like?
I found your video via your other one, where you were trying to design an outrunner-BT30-ATC-spindle.
Is the spindle you got now capable of low rpm/high torque - for something like 50 mm face mills in aluminium?
As you seem to be looking for input, you might want to look at this build - czcams.com/video/bX2Eg18AzTA/video.html
Details are over here - www.cnczone.com/forums/uncategorised-metalworking-machines/327914-gantry-mill-precise-milling-aluminium.html
The specs of what you want and he has built match pretty closely and his results are impressive - take note of the reasons behind some design decisions (gantry stiffness, etc.).
Be aware of the other video of his, where he mills some aluminium part for a bigger machine and where the swarf goes everywhere - something I'd want to avoid.
Another example (fixed gantry) of a seemingly sturdy design (note the vertical z-plate stiffening), but also swarf problems - czcams.com/video/o__Uk_ek8Ec/video.html
Personally I'm planning a portal milling machine with a similar bed size as you (1200mm x 800mm) but a Z height of 350mm and the y-rails/drives up on 'walls', similar to this - czcams.com/video/fIVL83AwXrE/video.html
Some early musings can be found here - www.cnczone.com/forums/uncategorised-metalworking-machines/347540-arent-portal-mills-popular.html
Definately over-engineered but not as optimized as you think. It looks great but wastes a lot of money for material that is not needed or in the wrong place.
30mm rails for a small 2.2Kw router? That's about a Kilogramm per carriage that has to be accelerated.
25mm for Y and X axis is more than enough and 20mm for Z. If you increase the hight of your Z assembly and use four 20mm carriages instead of the two 30s you'll get more rigidity with less weight.
And that huge piece of aluminium...huge waste of money. If you want stiffness then you want a large mostly hollow cross-section, not a thin solid piece.
Two heavy 90x90 Bosch extrusions sandwiched between two 10mm cast aluminum plates ... simple, clean looking and stiffer than this design - at a fraction of the cost.
The ballscrew - as nice as it looks, hidden in the middle, behind a super flat Z assembly - should be placed between the X guide rails and the Z assembly, as close as possible to the mass it's supposed to move. Think about inertia, leverage and resulting torque in the rails. Your current design does the opposite of what you're trying to achieve.
I know that's a lot of critique ... but ...
In my case I get parts as support, so i had to chose with no regard for cost. It is overkill, and now its more a problem not a feature.
I am making 3d model for XZ carage in completely different way (testing idea). Spindle and all bearings will be mounted inside a square box, it like exoskeleton, In that case, i will have 4 rails for z, and X rails will be centered between spindle, so rails will be placed along the axis in between mass center of the spindle.
This should make more sturdy design as well as put X rails closer to the bottom, and it will be super easy to protect from dust, since everything will be inside hollow tube.
But this will place lead-screws off axis of center of the mass, i don't know should i be worried :?
You can place two ballscrews on z-axis.
the way you are putting the longest linear rails is complex and you going to have a problem when assembling the machine physically this is due to the tolerance of the machine that you will use to mill the 300 euro component better put the linear rails on top than on the side
avoid reducing the thickness of the aluminum if you want maximum stiffness
use gearbox for your servos instead of belts to minimise backlash
if you melt all your iphone you bought last 10 years you safe from aluminium LOL
Lost me at "Mach"
why ?
You prefer LinuxCNC or other program ?