[mitip]

I had two high speed CNCs, one 50,000rpm the other 40,000, air driven heads with refirgeration units for the head bearings. Both used Creative Evolution controllers, one was a Defiance VTX the other a Strathclyde HT500. Had to sell them when I closed the business and am intruiged to find out how good these smaller machinesb are by comparison and if I could use them for similar work. We used Delcam Powermill to programme offline, post processed and dumped files that were sometimes 40Mb onto the controllers with masses of points in order to make 30mm x 20mm 3D object very precisely. Seemed vast overkill, the machine had a travel of 1000mm x 400mm. I recall we were needing lookup(?) times less than 4ms and acceleration/deceleration and such to be set up as we sometimes used 0.5mm or 1mm (occcasionally 0.25mm) cutters.

I dont mind if cutting takes longer but I am trying to understand if I could still use Powermill (I know it reasonably well) to programme, I gather the typical controllers drip feed via a parallel port, is it G code and post processed? Does Mach3 send data the same way, not clear how it works. Anyone got their head around these issues that I could chat with?

Mike

[SORCHEROR]

mach 3 accepts G code,u might have to play around to find the correct post,i use a fanc
post and its been fine so far,u can use a high rpm spindle with a cooler,but unless u build a highly precision benchtop u wont be able to cut at a high feedrate

[ihavenofish]

mach 3 accepts G code,u might have to play around to find the correct post,i use a fanc
post and its been fine so far,u can use a high rpm spindle with a cooler,but unless u build a highly precision benchtop u wont be able to cut at a high feedrate

being a step and direction controller, you also end up balancing precision and speed - even with using servos.

keeping better than .001 precision with steppers usually limits mach to under 500ipm, and not much better with servos. its not bad at all really, but not typically thought of as "high speed" these days. it simply cant spit out the pulses fast enough. this also limits acceleration somewhat i believe, but you might still get .5g with a good servo setup (and capable machine).

it does have look ahead of up to 1000 lines, constant velocity and other high speed friendly features though.

i think emc2 might be better suited to high speed controlling servos with torque instead of steps. ive not had a chance to use it yet though (will be playing with it on a servo setup in the near future though).

[mitip]

Thanks guys, I am trying to understand the differences between what we used to do and what these benchtop machines can do. We used to post as Fanuc, so that is OK. Also I read that steppers are nearly as accurate as servos, but you are suggesting a servo set up would be better?
You make a point about the accuracy of the machine needed, I am not clear how rigid these smaller machines are or whether it is worth considering an older machine and doing a refit. I came across a machine that is half way between a Bridgeport knee mill and a benchtop, very solid cast construction, I could scrape the slides and get it working well but would there be any benefit? All the parts are heavy, especially in z, would the torque needed be too much for steppers?
Last point, I gather with Mach3 the data drip fed to the controller via parallel, is the point you are making that this is not a limiting factor? I recall early Interact controllers that would simply stop and dwell when they ran out of data. You mention 1000 lines look ahead, which I recall was a figure from the past, is this OK when doing high feed rates. I have a brand new NSK 65,000rpm air head worth about $4000 I was planning to use but it does not have much torque but should be OK with 1-1.5mm cutters although someone mentioned electronic heads that are OK for 36,000rpm. One of the guys who did our programming never used more than 20,000rpm but for plastics I found the faster the better as long as you can get the heat and swarf away and stop burning. What is the emc2?
Mike

[ihavenofish]

Thanks guys, I am trying to understand the differences between what we used to do and what these benchtop machines can do. We used to post as Fanuc, so that is OK. Also I read that steppers are nearly as accurate as servos, but you are suggesting a servo set up would be better?
You make a point about the accuracy of the machine needed, I am not clear how rigid these smaller machines are or whether it is worth considering an older machine and doing a refit. I came across a machine that is half way between a Bridgeport knee mill and a benchtop, very solid cast construction, I could scrape the slides and get it working well but would there be any benefit? All the parts are heavy, especially in z, would the torque needed be too much for steppers?
Last point, I gather with Mach3 the data drip fed to the controller via parallel, is the point you are making that this is not a limiting factor? I recall early Interact controllers that would simply stop and dwell when they ran out of data. You mention 1000 lines look ahead, which I recall was a figure from the past, is this OK when doing high feed rates. I have a brand new NSK 65,000rpm air head worth about $4000 I was planning to use but it does not have much torque but should be OK with 1-1.5mm cutters although someone mentioned electronic heads that are OK for 36,000rpm. One of the guys who did our programming never used more than 20,000rpm but for plastics I found the faster the better as long as you can get the heat and swarf away and stop burning. What is the emc2?
Mike

mach doesnt drip feed in the normal sense. mach IS the controller, and simply outputs step and direction pulses. 2 pulses = one command to move one step.

the limiting factor for mach speedwise is the number of pulses you can send out per second. in the current incarnation of mach3, using my 2.8ghz pentium 4 pc i can safely output the maximum of 100,000 pulses per seconds. safely, you could average 40,000 pulses per axis per second in typical 3 axis 3d contouring. thats 2.4 million pulses per minute.

in microstep mode, you have 2000 pulses per revolution on a stepper motor. using my sieg kx1 (desktop mill) which has 4mm pitch screws i need 12700 pulses to move 1 inch. this means i have a practical speed limit of 188 inches per minute for 3d contouring and an absolute maximum of 472 ipm one axis at a time. reliable positioning accuracy barring mechanical considerations is .0004" which is good. the acceleration on my machine is tuned to about .05g if i want to get to 188ipm. this is decent enough for my use, but not great. this is a limitation of the stepper motor power (190 oz-in).

so, properly configured, my machine is accurate and more than fast enough for the 5000rpm spindle, but nowhere near high speed and under stepper control wont get there without going to a larger lead screw and sacrificing accuracy and even more acceleration.

using mach with a servo system allows alot more fun. you can use the servo drives internal "digital gearing" to make better use of the steps from mach - instead of microstepping each step could then corespond to a line of the encoder. you could easily get up to 10 times the speed yet maintain the same accuracy.

from EMC2 you could use the servos via torque control as well, which as my understanding goes can allow another 2x speed up with the same amount of pulses. you would also get the most torque in this mode making the best of acceleration... my math says my 300oz-in (peak 6 seconds) nema 23 motors could push my particular mill at .5g happily.

BUT. and this is the big one with these little machines... the mecanical limitations of my kx1 (which is a nice inexpensive but high quality machine) wont allow this. at best, 250 ipm would be great on this little machine given the bare iron dovetail ways, tiny ball screws, play and flex in they system. top speed is also inhibited by how much it can ramp up over the mere 5" of travel on the y axis. even at .5g, it take a few inches to get to the theoretical 1880 ipm, so in most contouring with small sharp details, its barely going to get past 2-300ipm in the best case scenario (this is true also for many super expensive machines).

so basically what im getting at is "high speed" is alot more complicated than slapping on a 60k spindle and hoping for the best

to get the best of it, you want a linear way machine with peppy servos and that starts costing 5 figures real quick.

[mitip]

Again thanks for a full reply. I hoped someone would comment on the stiffness of these smaller machines, from what you say I suspected they may be too flimsy for both fast and accurate work.
I think the smaller of our two mills a Defiance VTX was only slightly larger than these new benchtop machines in terms of travel but had a granite structure and massive reinforcement. Even then when it was moving fast the whole massive steel bench it was mounted on would waft about as the drives changed direction. It would leave small dwell marks and have evidence of slight over run into walls on small parts compared to our much larger machining centre.
I guess I ought to concentrate on converting an older, bigger and probably far more rigid mill with servo motors. I would appreciate any links to possible suppliers. Is it possible to retrofit knee mills where the motor has to move the whole mass of x and y for z movement, or will that be too much mass to drive directly?

[ihavenofish]

Again thanks for a full reply. I hoped someone would comment on the stiffness of these smaller machines, from what you say I suspected they may be too flimsy for both fast and accurate work.
I think the smaller of our two mills a Defiance VTX was only slightly larger than these new benchtop machines in terms of travel but had a granite structure and massive reinforcement. Even then when it was moving fast the whole massive steel bench it was mounted on would waft about as the drives changed direction. It would leave small dwell marks and have evidence of slight over run into walls on small parts compared to our much larger machining centre.
I guess I ought to concentrate on converting an older, bigger and probably far more rigid mill with servo motors. I would appreciate any links to possible suppliers. Is it possible to retrofit knee mills where the motor has to move the whole mass of x and y for z movement, or will that be too much mass to drive directly?

bigger machines are more rigid, but get you diminishing returns on speed when you convert them if they arent linear. you need massive motors, gear downs etc to get enough torque to move them at all.. resulting usually in no top speed and super slow accelerations.

its going to be hard to get something premade and inexpensive. ive been trying to design a small (8"x4" envelope) high performance, high speed mill for years. the instructions are "must cost under $10,000 complete". thus far, its been impossible. even scouring ebay for cheap surplus industrial gear hasnt helped. i can easily make a good mill for that cost. a really good mill, but it just wont be super fast.

i do have an idea on how to take my machine (sieg kx1 - $2900 base price new 5"x10"x8" travel) and push it much further to an entry level high speed territory. using surplus servos i have ($3x$750 new, $3x$90 ebay) instead of the stock steppers and AMC drives (3x$260 new). psu and other accessory boards add about $350. mach3 is $160, but one could use emc2 which is free. thats stage 1 and will move my machine safely at 250 ipm at a startup cost of $4460.

stage 2 is to one by one change the axes to tkh linear rails using the machine itself to make any accessory bits and modifications. the rails run new about $600 per axis (hsr20 model) and can be often found much less on ebay. cnczones own widgetmaster sells a nice cast iron table perfectly sized for the machine already flat to accept rails for another $220. a new base for modification is $1oo at littlemachineshop.com. just need a steel/iron plate to make a y saddle and a few accessory plates and the x/y should be good to go at much much higher accelerations (nearly 1g) and a top speed of over 600ipm (limited by ball screw specs and 80v servo drives). total price up to $6k or there abouts.

the z axis it trickier to modify on the machine itself, as its not big enough to mill the lands for the linear ways. we also need a riser because the xy table assembly will be taller now. even with a bit of outsourcing for a new column, it should come in under $1000 bringing the total now to 7k. in your case, youd make a sleeve to fit the nsk spindle into the 50mm spindle head clamp and be about done. in my case, im looking for cheap spindle options still (i have a few ideas which makes auto tool changers possible).

you could do this same idea of mod on a bigger machine like the kx3($3600 base 12"x7"x11" travel) but all the motors, drives, rails, hardware and scale is bigger an more expensive. probably cost well over 10k all told unless you found some great deals and had facility to machine the modified parts yourself.


i should note that even with high speed spindles and specialized cutters, you usually wont be cutting metal at more than 2-400ipm with small bits. the extra speed comes in very handy though when the machine isnt cutting.

[mitip]

Hi ihavenofish your posts are really useful. I somehow stumbled on high speed work a few years ago and as I said in my first post managed to pick up some ex demo machines that were really fast. Once you have worked with small cutters at high speeds about 20,000 rpm and seen the quality and accuracy it produces you never want to go back. I see Datron are selling some special cutters aimed at use on their high speed machines that I would like to try. We sometimes ground our own cutters from TC stock and I even tried using diamond cutters after fruitlessly searching for the single point diamond cutters they use for optical machining. I agree it is as much the time the cutter spends in the air between plunges not cutting that eats up time, especially if you need to go back to safe Z between plunges which cheaper CAM software often seems to.
You suggest Z conversion is the hardest but I was thinking that z speed and acceleration is often less a problem if the machining strategy is doing a series of z level plunges and area clearance. It is only when profiling over 3D surfaces that the head needs to change direction a couple of times each pass. So possibly a slower Z would only be a restriction on certain types of machining. Having said that we were often using 0.1mm 0.004" step overs with 20,000 rpm and 1.0mm dia ball ended cutters to get finishes that were nearly perfect and needed no further work.
I would be keen to work in parallel with you here in the UK if you plan to go ahead and build the machine you describe, I am in no massive hurry and would enjoy the 'journey'. I have 7 seats of ProEngineer sitting doing nothing right now so CAD work would be simple and I could model up custom parts if that helps. Getting 2 sets of bits made may help with cost also.
I have searched through endless auctions and sites to see if I could find a suitable donor machine that might save time or money but going the THK route I cant see that is not going to help much. I found a few Denford and Boxford benchtop machine for about $1000 but I am not sure if many bits are worth salvaging. From what you are suggesting as your stage 2 machine is there much point in starting with any donor machine? Widgitmaster site is down at the moment but I was thinking that I have got a large granite surface table that could use to make the base structure and it would need a cast column bolted to it. Plenty to think about, I will research the components you describe and post again in a few days. Mike

[ihavenofish]

Hi ihavenofish your posts are really useful. I somehow stumbled on high speed work a few years ago and as I said in my first post managed to pick up some ex demo machines that were really fast. Once you have worked with small cutters at high speeds about 20,000 rpm and seen the quality and accuracy it produces you never want to go back. I see Datron are selling some special cutters aimed at use on their high speed machines that I would like to try. We sometimes ground our own cutters from TC stock and I even tried using diamond cutters after fruitlessly searching for the single point diamond cutters they use for optical machining. I agree it is as much the time the cutter spends in the air between plunges not cutting that eats up time, especially if you need to go back to safe Z between plunges which cheaper CAM software often seems to.
You suggest Z conversion is the hardest but I was thinking that z speed and acceleration is often less a problem if the machining strategy is doing a series of z level plunges and area clearance. It is only when profiling over 3D surfaces that the head needs to change direction a couple of times each pass. So possibly a slower Z would only be a restriction on certain types of machining. Having said that we were often using 0.1mm 0.004" step overs with 20,000 rpm and 1.0mm dia ball ended cutters to get finishes that were nearly perfect and needed no further work.
I would be keen to work in parallel with you here in the UK if you plan to go ahead and build the machine you describe, I am in no massive hurry and would enjoy the 'journey'. I have 7 seats of ProEngineer sitting doing nothing right now so CAD work would be simple and I could model up custom parts if that helps. Getting 2 sets of bits made may help with cost also.
I have searched through endless auctions and sites to see if I could find a suitable donor machine that might save time or money but going the THK route I cant see that is not going to help much. I found a few Denford and Boxford benchtop machine for about $1000 but I am not sure if many bits are worth salvaging. From what you are suggesting as your stage 2 machine is there much point in starting with any donor machine? Widgitmaster site is down at the moment but I was thinking that I have got a large granite surface table that could use to make the base structure and it would need a cast column bolted to it. Plenty to think about, I will research the components you describe and post again in a few days. Mike

you seem on the right track. im too busy to go back to working on this at the moment, but one day soon i will return to my mini cheap hsm centre idea. for the moment, im just getting my little kx1 fully functional and makin useful parts

[Jason3]

Hi all, thought I might check in here in case I can offer something.

From my perspective, I believe it's quite possible to build your own desktop machine and have it produce parts with excellent accuracy and a very nice surface finish. The little machine I'm using lately runs a 50 Krpm spindle with pneumatic collet release, and feed is up to 200 ipm. Accuracy is better than .01 mm, the limit of my measuring equipment. The resolution is 0.00025 mm.

It's rigid enough to hold accuracy with this 0.5 Kw spindle with it's max. 1/4" cutters. I wouldn't count on that with a more powerful spindle, or a slower one with more torque though.

Here's a recent clip of this machine in action from my lathe build thread:

[ame="http://www.youtube.com/watch?v=OOesDXSxzqI"]YouTube - ER32 spanner set[/ame]

Best regards,

Jason

[ihavenofish]

Hi all, thought I might check in here in case I can offer something.

From my perspective, I believe it's quite possible to build your own desktop machine and have it produce parts with excellent accuracy and a very nice surface finish. The little machine I'm using lately runs a 50 Krpm spindle with pneumatic collet release, and feed is up to 200 ipm. Accuracy is better than .01 mm, the limit of my measuring equipment. The resolution is 0.00025 mm.

It's rigid enough to hold accuracy with this 0.5 Kw spindle with it's max. 1/4" cutters. I wouldn't count on that with a more powerful spindle, or a slower one with more torque though.

Here's a recent clip of this machine in action from my lathe build thread:

YouTube - ER32 spanner set

Best regards,

Jason

i like youre machine alot. if you wait around ebay and are doing a one off, i know you can definitely built an amazing machine for a small price. people on this forum prove that every day. my original criteria for a mill though was "built like a production vmc, but smaller". it was meant to handle alot of torque, not just light cuts so alot of the hobby build techniques dont apply.

my ideas for the kx1 stem from it being cheap and small (and already owning it), and being able to use it to self replicate its own new components. scratch building is fun and all, but for myself, it takes too long to get up and running especially if you dont have any other equipment as is my case.