[rherman]

I am embarking on building a high-speed cnc router table. I only need a work envelope of 24" x 24" x 6" in the xyz. Actually the z only needs to be 1" (.5" cut depth and .5" clearance). I have two machines, since they are general use, they are not designed for high-speed. I am cutting wood only to a max. z of .5". I am looking to cut at 600ipm and up with 1/8" and 1/4" shanked bits. Right now, 180ipm is as fast as my 4' x 8' x 6" router table with steppers can go. Since my loads will not be great, I am thinking a rigid, but light gantry and high speed spindle with servos will get me where I need to go. Anyone have info. on cheap servos and drives? Also, accuracy requirements are not as tight as my metal cnc machine. I can probably get away with +-.010" (.020" total delta). Great forum.

Thanks,

Rob

[rherman]

I dislike answerning myself, but has anyone built anything that can cut in light passes at 600 ipm? Thanks.

Rob

[ger21]

I doub't you can get to 600ipm on a 24"x24" machine. That's 10 inches/sec, barely enough room to accel and decel.On a small machine like that, even 400ipm might be more than you need. But you can probably get 400ipm with rack and pinion and steppers. Shopbots use closed loop steppers and can cut at 600ipm, rapid at 1800ipm. You shouldn't really need closed loop for light wood cutting. A good source for servos is http://www.homecnc.info. And Gecko drives to run them.

[wisp]

You also really need to think about a spindle that can run high enough rpm for that feedrate. At 600ipm, with a 2 flute cutting bit, to get a .006" chipload per tooth, you need 50,000 RPM.

Something to think about.

[rherman]

Thanks for the info. Wouldn't a 3 flute bit at 24000 rpm (hitachi router with speed feedback) give me approx. .008 at 600 ipm? Maybe 300 ipm is more realistic. My bought machine, a Practical CNC machine, runs open loop steppers on rack and pinion. I can rapid at 200 to 250 ipm, but when I am cutting complex 3d surfaces it actually cuts at around 90 ipm as opposed to the 135 ipm I command it to run. I am looking to significantly cut down on my machining times. How about servos and 36" x 36" table? Thanks.

Rob

[ger21]

You also really need to think about a spindle that can run high enough rpm for that feedrate. At 600ipm, with a 2 flute cutting bit, to get a .006" chipload per tooth, you need 50,000 RPM.

Something to think about.

You're not using the right tooling then. I looked in an Onsrud catalog, and the recommended chipload for a 3 flute spiral chipbreaker (1/2"D) is .02, which is 10,000 rpm at 600ipm. THis is for hardwoods, plywoods and mdf. Our router at work only goes 400ipm, and I use similar tooling @ 14,000. http://www.vortextool.com

[Chris D]

The slowdown you see could be the result of the computer, not the machine. When machining complex 3D shapes, you are often cutting lot's of little X,Y,Z moves, the computer has to process a lot more data for that than if cutting in a single axis as the same speed. This problem exists even on industrial machine tools.

Thanks for the info. Wouldn't a 3 flute bit at 24000 rpm (hitachi router with speed feedback) give me approx. .008 at 600 ipm? Maybe 300 ipm is more realistic. My bought machine, a Practical CNC machine, runs open loop steppers on rack and pinion. I can rapid at 200 to 250 ipm, but when I am cutting complex 3d surfaces it actually cuts at around 90 ipm as opposed to the 135 ipm I command it to run. I am looking to significantly cut down on my machining times. How about servos and 36" x 36" table? Thanks.

Rob

[rherman]

ger21: That's because you're using a 1/2" D tool. I am using 1/8" and 1/32" Dia. tooling. Angular velocity at that diameter is a lot slower than your 1/2" bit at the same spindle speed.

Chris D: I don't think it's the computer. I have a WinCNC controller with high-speed card. I think the software may be limiting it due to the accelerations. I generated the same toolpath in different Cam package, and it spaced out the line segments at .020 on average as opposed to .009 or .0012 on average. This is not my stepover, but the CAM packages interpolation to fit the surface data. I am cutting with a stepover of .015 max.

Maybe I'm just asking too much given the work envelope is so small and tight; however, I was motivated by seeing those high-speed machining centers cutting metal almost as fast as I cut wood. Thanks.

Rob

[Halfnutz]

Please excuse my ignorance, but I am new to this and dont understand what the need or desire to go that fast is based on. It is obvious that this is a major concern for many people, but I dont understand why. Why are people so concerned with the speed of thier machines, especially while cutting? Is this production based, ie to produce more parts faster, or performance, ie do the finishes turn out better, or both, please clue me in. Thank you,

ps also to note the quick change R8 collet set shown is a nice tool, (I have a set) but TIR is much greater with these than with a standard collet. I get MUCH better finishes with standard collet's than with the QC set.

[rherman]

In the manufacturing business speed is key, if finish is within spec. But, I am machining 3d surfaces over an area from 4"x5" to 24"x24". It takes me 4 hours to machine a 5"x6" surface. I am using v-bits or 1/32" ballnose bits with a < .015" stepover. Take a look at my user gallery here. If I could cut twice as fast with the same quality, that's twice the money made that day.

Rob

[Halfnutz]

Where is your user gallery? I would like to see it. Thanks,

[ger21]

http://www.cnczone.com/gallery/showg...0&ppuser=12847

[rherman]

In the Member's Galleries, under rherman. Here's a link: http://www.cnczone.com/gallery/showm...p?sl=r&cat=500

This was my fifth generation test of my software. I am at gen 7 now. I will post more pics as I can. Thanks.

Rob

[ger21]

Are you still using Blender, or did you write a stand alone app?

[Chris D]

ger21: That's because you're using a 1/2" D tool. I am using 1/8" and 1/32" Dia. tooling. Angular velocity at that diameter is a lot slower than your 1/2" bit at the same spindle speed.

Chris D: I don't think it's the computer. I have a WinCNC controller with high-speed card. I think the software may be limiting it due to the accelerations. I generated the same toolpath in different Cam package, and it spaced out the line segments at .020 on average as opposed to .009 or .0012 on average. This is not my stepover, but the CAM packages interpolation to fit the surface data. I am cutting with a stepover of .015 max.

Maybe I'm just asking too much given the work envelope is so small and tight; however, I was motivated by seeing those high-speed machining centers cutting metal almost as fast as I cut wood. Thanks.

Rob

Yeah, I know what you mean. Keep in mind too though that industrial machines have everything going for them, not everything going against them. BIG servo motors, lot's of power, course pitch ball screws, dedicated controller serving a single purpose - making the machine move. Absolute precision with regard to alignment of the screws to the axes. The list goes on and on.

Some machines use linear motors as opposed to ball screws - this really makes the machines move fast - however, at a price too!

It has also taken the industry as a whole, many years of development to get up to the speeds they are running at now. In the late 70s, 160 IPM rapids were pretty fast, today, that is child's play :-(

[rherman]

I've since wrote a stand alone program in java and scheme (Scheme, since The Gimp, an open source program has a scheme scripting interface). I own BobCad and their BoBArt program almost seemed a fit, but my program takes the photo as is with no pre-processing and outputs a pretty good mesh. I am also designing a special cnc machine just for what I am trying to do. Ultra-light materials and sections to keep inertia low for accel and decel. A small, high speed spindle for cutting up to 1/4" thick wood with 1/8" collet. Ball screws geared for speed not resolution or torque. My goal is to be able to cut x-y rasters in z with a max. depth of .25" at speeds around 350 ipm to 600 ipm. I realize some machines already cut that fast, but they can't seem to handle the fine detail work of photo carving. My current machine can cut at 160 ipm, but quality suffers, so I usually cut around 100 ipm.

Rob

[wizard]

Another option I see is to remove more material per pass. That would seem to be doable if you can switch tooling quickly and get your cad system to ouput two files. That would be one file to rough out the image and a second to do the finish work.

This of course is a different way to address the issue. Instead of going fast you attempt to do more on each go around. This would certainly result in variable performance depending on ho much you can rough out.

Actually I think what you are doing could be handled really well via linear motors. If that is you can keep the mass down. An Option here is to make Y axis light and mak sure your software keeps movements on the X down. Think of a raster scan, this limits your need to make improvements to one axis. Well maybe two as you will have to move Z pretty fast also. Either way though you will need hi performance motors.

Others have already mentioned software as a potential issue. Don't discount this evne though todays computers are vary fast. In the old days I workd on some PC based CNC lathes that we upgraded from 386 to 486 computers. Doing nothing else this sliced
two seconds off a 13 second cycle time. Your applicaitons sounds like it is a lot more involved than that.

Dave

In the manufacturing business speed is key, if finish is within spec. But, I am machining 3d surfaces over an area from 4"x5" to 24"x24". It takes me 4 hours to machine a 5"x6" surface. I am using v-bits or 1/32" ballnose bits with a < .015" stepover. Take a look at my user gallery here. If I could cut twice as fast with the same quality, that's twice the money made that day.

Rob

[rherman]

Yes, concentrating on one axis, the y, has been my plan since the x is just a step over move at the top and bottom of the scan. I am removing all material in one pass since my doc is .25" or less usually .1". A three flute cutter is currently being used. I may opt for a 4 flute for faster feeds with the same chip per tooth removal. Does anyone know of any spindle ideas? Light-weight, fast, low horsepower reqs?

Rob

[wizard]

Have you considered driving the Z-Axis up and down with a voice coil. Syncing it to the Y-axis will be a bit of electronic engineering but should be doable on a FPGA. You may be able to do this also on the PC if you have a high speed D/A card and can hack the code doing pulse generation for the Y. You would need a table of values to walk through for each scan that would be output after each step signal is generated.

This really wouldn't be CNC but your should be able to run that Y very fast. You would need a wide bandwidth audio Amp to drive the voice coil. In effect instead of moving a cone to move air you would be driving the spindle up and down with the voice coil.

Yes this is a non trivial exercise and your position would be some what open loop in a basic implementation but boy would it be fast. I forgot who it was, possibly a German manufacture of lathes, but they had such an arrangement on a lathe to turn pistons just ever so slightly out of round. That worked well there apparently, I'm not sure how it would translate into a back and forth device like a router.

Dave

Yes, concentrating on one axis, the y, has been my plan since the x is just a step over move at the top and bottom of the scan. I am removing all material in one pass since my doc is .25" or less usually .1". A three flute cutter is currently being used. I may opt for a 4 flute for faster feeds with the same chip per tooth removal. Does anyone know of any spindle ideas? Light-weight, fast, low horsepower reqs?

Rob

[turmite]

rherman is this complete package you intend to sell to the public? Machine, software controller ...... everything?

Mike