NC-Joey
09-08-2007, 01:19 PM
Hi Guys.
Sorry I`m not able to post any pics yet as my daughter stood on and broke the Belkin adapter so I cant
get pics onto the pc.
Yeah I had this crazy idea to use ice as a mould for producing 3D prototype models. I know there are loads of
different materials being used on other machines but hey, water is the cheapest and cleanest so why not give it a try.
I originally went for the Router type design for my machine thinking it could double as a small very `light` milling machine as well.
Bear in mind that my whole machine will eventually be dumped into and run inside a household chest freezer.
Ok so the principal design is :-
Water that is filtered down to 10 micron or less will be pumped thru a micro-dispensing valve which has a nozzle outlet
of either .005", .007" or .010" dia (not sure which will work best yet in practise). This produces a very fine spray, the nozzle will be set around 3mm from a frozen metal surface plate to start the ice building. As with most prototyping machines the 3D shapes will be made of 2D layers probably between 0.1 and 0.2mm in Z height.
I`m hoping for build speeds of around 35 mm per second which would be awesome in terms of build speed.
There will be a small electric heater circuit built into the nozzle housing which will keep the nozzle from freezing (being so close in contant with the build layers) Many many thanks to Neil at Reuk.co.uk for such a great circuit design.
He has also built me a 24V to 3V Regulator circuit to control the Micro-dispensing valve Spike and Hold Driver, which basically opens and closes the valve from TTL signals given from the machine controller.
I have done a full loop using the Fab@home setup with my router build. In fact I had it all installed and working (in a fashion anyway) but have now opted for Mach3 as I`m comfortable using Gcode and I think I`ll have better control of the finished output.
In order to get the 2D slice info into Gcode I`m looking at some German software called VisCam which is used in all manner of 3D rapid prototype systems. Gonna cost around £1000 for the software tho. But it is sweet. I can import a 3D STL file, select my build step over and Z height increments and output directly into Gcode. This also has a customizable post processor so hopefully I can fine tune it with Mach3.
I`ll use either spindle or coolant on/off commands within Mach3 to run the valve.
Sorry I never created build logs of my machine, not enough hours in the day !
But I could expand on any of the issues I`ve had to get the machine built.
I`ve had to be pretty lean with how much cash I spent and as a result can give any new CNC builder some good advice.
Namely - It is more Expensive to go Cheap !!
Belt Drive
Thinking my icemould machine wouldn`t need micron accuracy, I first tried with a simple toothed belt drive on the X and Y axes. The machine ran nice and quick but lines drawn (using a pen attached to the Z) on paper soon revealed a staggered effect to them, any angular moves were particularily bad. After I fitted Readout Scales to all Axes did I find
that the machine was not repeating, by as much as .3 to .5 mm in places.
Well (I thought ) the stepper motors were ony doing about two complete revolutions for the full axis travel, what I need is geared steppers for more accuracy. So I bought some off ebay and after making new mounting brackets I found that the Advertised `Zero Backlash` gear motors were crap. as there was at least .8 mm backlash on a 50mm dia toothed wheel mounted to the motor spindle. So they were no good.
Feeling very frustrated now, I tossed the whole lot out and did what I should have done in the beginning. Use Ballscrews. My X axis has a nice 16 mm Schneeberger Ballscrew, my Y axis is a THK KR26 linear actuator, 8 mm ballscrew. and My Z axis uses a NSK 8mm Ballscrew.
Linear System
haha this has been an interesting ride.
Initially I bought linear bearings and ground rod from Marchant Dice. SC16UU for X axis and SC12UU for Y+Z axes.
Well what a pile of poo in terms of accuracy, very sloppy fit and no good if you require any sort of precision.
Granted I only used 2 of SC12UU bearings, which are 30mm long each, on two rods spaced 75mm apart for my Y axis.
If you are thinking of going down this route I recommend to use 4 bearings, 2 on each shaft, and spaced as wide apart as possible.
Part of the reason the bearings are`nt so good is due to the fact that the ground rod is not very round and parallel.
And here`s how I know that. I decided to bin the bearings and use aluminium bronze bushes which I could make at work. They were bored as close to size as possible and lapped the rest of the way, I had a slit in the bushes and could adjust tightness to the shaft with grub screws. It took ages to get them reasonably true ( lapping with a sacrificial bush ) Had to remove high spots in the region of .03 on diameter. This doesn`t sound like a lot but if you take that movement multiplied by the pivoting and levering action of your machine slide setup once it`s assembled, it turns into loads of `play`. In other words your mate who comes to look at your machine, rocks the column from side to side and says " humph that`s not very good".
I learnt the hard way, the most important thing is the rigidity of the frame that holds your slide system in place.
I had the silly situation of `as you tighten the thing into place you can feel the slides binding up`. Causing the stepper motors to stall.
Ideally what you want is to make your slide frame independant of the main machine frame. in other words get the slide system running well on it`s own frame/brackets and bolt this onto machined flat surfaces of your main frame.
Anyway once I had (lapped) reasonably good shaft rods I thought I`d try something with the linear bearings, I ground a thin slit 20mm long in the sidewall of the bearings outer case, in between the bearing rows and parallel with the bore axis, using the mentioned grub screws I could then nip up the bearing dia to reduce the play. It does help and still have these in my X Axis. There is a slight gritty noise to the bearings at high speeds.
After all the struggling I ditched the Y + Z axis bearings and bought the linear actuator for the Y axis. My Z axis is homemade using the NSK Ballscrew. I`ll post some pics of that when I can as it`s a bit unusual in design, but works great for what I want.
At the moment I`m doing the electrics on my machine so still a little way off doing some full blown trials.
Geez i`m tired of typing now
regards to all
Sorry I`m not able to post any pics yet as my daughter stood on and broke the Belkin adapter so I cant
get pics onto the pc.
Yeah I had this crazy idea to use ice as a mould for producing 3D prototype models. I know there are loads of
different materials being used on other machines but hey, water is the cheapest and cleanest so why not give it a try.
I originally went for the Router type design for my machine thinking it could double as a small very `light` milling machine as well.
Bear in mind that my whole machine will eventually be dumped into and run inside a household chest freezer.
Ok so the principal design is :-
Water that is filtered down to 10 micron or less will be pumped thru a micro-dispensing valve which has a nozzle outlet
of either .005", .007" or .010" dia (not sure which will work best yet in practise). This produces a very fine spray, the nozzle will be set around 3mm from a frozen metal surface plate to start the ice building. As with most prototyping machines the 3D shapes will be made of 2D layers probably between 0.1 and 0.2mm in Z height.
I`m hoping for build speeds of around 35 mm per second which would be awesome in terms of build speed.
There will be a small electric heater circuit built into the nozzle housing which will keep the nozzle from freezing (being so close in contant with the build layers) Many many thanks to Neil at Reuk.co.uk for such a great circuit design.
He has also built me a 24V to 3V Regulator circuit to control the Micro-dispensing valve Spike and Hold Driver, which basically opens and closes the valve from TTL signals given from the machine controller.
I have done a full loop using the Fab@home setup with my router build. In fact I had it all installed and working (in a fashion anyway) but have now opted for Mach3 as I`m comfortable using Gcode and I think I`ll have better control of the finished output.
In order to get the 2D slice info into Gcode I`m looking at some German software called VisCam which is used in all manner of 3D rapid prototype systems. Gonna cost around £1000 for the software tho. But it is sweet. I can import a 3D STL file, select my build step over and Z height increments and output directly into Gcode. This also has a customizable post processor so hopefully I can fine tune it with Mach3.
I`ll use either spindle or coolant on/off commands within Mach3 to run the valve.
Sorry I never created build logs of my machine, not enough hours in the day !
But I could expand on any of the issues I`ve had to get the machine built.
I`ve had to be pretty lean with how much cash I spent and as a result can give any new CNC builder some good advice.
Namely - It is more Expensive to go Cheap !!
Belt Drive
Thinking my icemould machine wouldn`t need micron accuracy, I first tried with a simple toothed belt drive on the X and Y axes. The machine ran nice and quick but lines drawn (using a pen attached to the Z) on paper soon revealed a staggered effect to them, any angular moves were particularily bad. After I fitted Readout Scales to all Axes did I find
that the machine was not repeating, by as much as .3 to .5 mm in places.
Well (I thought ) the stepper motors were ony doing about two complete revolutions for the full axis travel, what I need is geared steppers for more accuracy. So I bought some off ebay and after making new mounting brackets I found that the Advertised `Zero Backlash` gear motors were crap. as there was at least .8 mm backlash on a 50mm dia toothed wheel mounted to the motor spindle. So they were no good.
Feeling very frustrated now, I tossed the whole lot out and did what I should have done in the beginning. Use Ballscrews. My X axis has a nice 16 mm Schneeberger Ballscrew, my Y axis is a THK KR26 linear actuator, 8 mm ballscrew. and My Z axis uses a NSK 8mm Ballscrew.
Linear System
haha this has been an interesting ride.
Initially I bought linear bearings and ground rod from Marchant Dice. SC16UU for X axis and SC12UU for Y+Z axes.
Well what a pile of poo in terms of accuracy, very sloppy fit and no good if you require any sort of precision.
Granted I only used 2 of SC12UU bearings, which are 30mm long each, on two rods spaced 75mm apart for my Y axis.
If you are thinking of going down this route I recommend to use 4 bearings, 2 on each shaft, and spaced as wide apart as possible.
Part of the reason the bearings are`nt so good is due to the fact that the ground rod is not very round and parallel.
And here`s how I know that. I decided to bin the bearings and use aluminium bronze bushes which I could make at work. They were bored as close to size as possible and lapped the rest of the way, I had a slit in the bushes and could adjust tightness to the shaft with grub screws. It took ages to get them reasonably true ( lapping with a sacrificial bush ) Had to remove high spots in the region of .03 on diameter. This doesn`t sound like a lot but if you take that movement multiplied by the pivoting and levering action of your machine slide setup once it`s assembled, it turns into loads of `play`. In other words your mate who comes to look at your machine, rocks the column from side to side and says " humph that`s not very good".
I learnt the hard way, the most important thing is the rigidity of the frame that holds your slide system in place.
I had the silly situation of `as you tighten the thing into place you can feel the slides binding up`. Causing the stepper motors to stall.
Ideally what you want is to make your slide frame independant of the main machine frame. in other words get the slide system running well on it`s own frame/brackets and bolt this onto machined flat surfaces of your main frame.
Anyway once I had (lapped) reasonably good shaft rods I thought I`d try something with the linear bearings, I ground a thin slit 20mm long in the sidewall of the bearings outer case, in between the bearing rows and parallel with the bore axis, using the mentioned grub screws I could then nip up the bearing dia to reduce the play. It does help and still have these in my X Axis. There is a slight gritty noise to the bearings at high speeds.
After all the struggling I ditched the Y + Z axis bearings and bought the linear actuator for the Y axis. My Z axis is homemade using the NSK Ballscrew. I`ll post some pics of that when I can as it`s a bit unusual in design, but works great for what I want.
At the moment I`m doing the electrics on my machine so still a little way off doing some full blown trials.
Geez i`m tired of typing now
regards to all