Small Mill very early design stage

At the moment my new machine is in a very early design stage: it was quite far advanced until I started asking questions on this forum and found that many of the things I was going to do wouldn't work, so at the moment even the basic architecture is in the air.
The eventual aim is to make a milling machine to make very small parts in anything from machinable wax to steel. The main requirements are accuracy, accuracy and more accuracy. Something slightly bigger than my existing Proxxon MF70 would be nice - a Y axis limited to 40mm is a bit of a drag, but 200mm by 200mm with enough space in the Z axis to allow 4th and even 5th axis work (eventually) would cover everything I want to make.
I am impatiently waiting for some linear guides that I bough on ebay to arrive so I can start serious design work - the manufacturer's site does not have datasheets for the ones I bought. Meanwhile I have attached some JPEGs, A formation of components done in Turbocad - mostly not applicable, I am (probably) not using round rail, the toothed pulley data was taken from an RS datasheet which was just plain wrong. The second photo is of a 218W brushless DC motor from Zapp, along with a controller that I built - PCB made on the Proxxon. Since that photo I have had the motor drive a spindle at 55,000RPM.

Quote:

---

Originally Posted by leadinglights

The main requirements are accuracy, accuracy and more accuracy.

---

IMHO the path to accuracy is twofold, lots of metal and springs.

Everything will bend, more metal bends less.

Everything will wear and bed in. Springs take up the slack.

Robin

Quote:

---

Originally Posted by leadinglights

The main requirements are accuracy, accuracy and more accuracy.

---

In which case you need to look larger in terms of design even though you want a smaller machine.

Start off with a large chequebook, a large bed plate about 3 tonnes to mount it on and build a temperature controlled workshop.

Forget steppers with their inefficient micro stepping and go to servos to start with.

.

I think the point that both Robin and John are making in their own inimitable ways is that accuracy = rigidity and rigidity = mass, and damped mass at that. You haven't said what accuracy you want/expect... a few spec's would be a good start. It is quite feasible to make a reasonably accurate router/gantry style machine that can do 0.05mm or better accuracy (NB accuracy is NOT the same as resolution) in engineering plastics and aluminium... especially if the depth of cut (DOC) is small (.1- .2mm) It is another thing entirely (by a factor of 3 or more) to do the same thing in free-cutting mild steel, and another jump up in stainless steel...

O.K., I will rephrase. Accuracy, accuracy, accuracy within limits, within budget and within practicability. My target is 0.001" throughout the range, but I know that this is hard. Hopefully the same will be achievable on a small part. As far as rigidity goes, I am the fortunate owner of an old (1960s)sensitive drill that weighs about 300lbs and teaches you that a big lump will always beat stylish when it comes to accuracy.
For rigidity I am hoping to make the frame and non-moving gantry in 15mm steel plate but may drop back to 20mm aluminium with some form of temperature control - possibly peltier heater/cooler elements in the vertical and cross members.

Be careful as high accuracy will come with a high price tag, what are you making that needs to be manufactured to that degree of accuracy???

Not so sure about your cooling/heating idea for the aluminium frame, aluminium will expand at twice the amount of steel but for a machine as small as you are describing then not so sure it would be necessary.

For instance, a 100mm length of steel will increase or decrease in length by 0.00126mm for every degree C of temperature change so a temperature change of say 20 degrees C will change the length by 0.0252mm, aluminium will be approx twice this with a change of 0.048mm.

Buy yourself a good heater for the winter and open the doors in the summer then you should not see much difference!

Why do I need accuracy? I am afraid the best answer I can give is, in the words of Sir Edmund Hillary, "because it's there". In constructional hobbies, and this is my hobby, most people strive for some principle in what they make: elegance, simplicity, beauty etc.. For me, the principle is accuracy.
As far as expence goes, what is a reasonable price to pay for a CNC milling machine? If you regard Arc Euro Trade's KX1 at £2350 as reasonable, then I think that I can build something quite special for less than that - especially as I have controllers, stepper motors etc.
What will I make with it? Clocks, small but accurate tools, a co-axial brushless DC motor for a model aircraft, wax models for investment casting, you name it. How long will it take? A couple of months if I work on it all of my spare time, a year ot two if I only spend the time "She Who Must Be Obeyed" allows me, but probably somewhere in between. Will it be worth it? If you are in business the answer must be "Hell no!" but if you are a hobbiest then the answer is "Damn yes!"
Regarding the idea of using Peltier temperature control. After the normal suspects, backlash, vibration etc., thermal differences in the workpiece and machine contribute a lot to errors. Controlling the temperature of bits that get hot such as the spindle and table, and bits that are long such as the upright bits of the gantry with Peltier devices is cheap - about £5 for a 90W device and somewhat less than that for the bits to control it, a few pounds for heatsinks and tubing and perhaps £20 for one of the water cooled radiators sold to computer modders. Including fans and concertina hose to dump the stepper and spindle motor heat outside the case I think there should be some change from a couple of hundred pounds.

Hmmm

Perhaps confidence is that feeling you get just before you understand the problem? :whistling:

I don't think you have to worry about heat quite yet. I suspect you are losing the plot and zooming off on a tangent.

You need a lot of iron and the cheapest way to get it is an older machine that you tart up. Scrape the slides, add ball screws etc.

It may seem a good idea to buy a shiny new wonder mill aimed at the CNC hobby market, but an older industrial or (better) ex-school machine is probably a much better bet, not so pretty but considerably cheaper and with more possibility for improvement.

I suggest you aim for 5 micron accuracy, the resolution of your everyday digi-caliper. You will then get 25 microns fairly easily. You will then realise 25um it isn't enough.

If you shopped with 5um in mind you don't wind up replacing the whole works like I had to.

Robin

Well good luck, since its for hobby use then you will probably find that you are continually tweaking and changing things anyway

If you are worried about thermal effects then you will definitely need the consider dynamics as well, beam/column frequency and moving masses etc. just adding more mass can lead to resonance:eek: not to mention problems with accel/decel.

As with the thermal effects I suspect that the small size wont be to much of an issue but might be some thing to look at on your pursuit of perfection.....

Have you ordered the C1 precision ground ball screws yet? :tongue:

Quote:

---

Originally Posted by Ross77

If you are worried about thermal effects then you will definitely need the consider dynamics as well, beam/column frequency and moving masses etc.

---

I see the big problem is that the spindle connects to a slide that connects to a slide that connects to a slide that holds the workpiece. All this with a total slop of < 0.001" so there is something left to allow for flex in the system.

I just don't think it's possible using commercial bearings, there simply isn't enough contact area to take the load.

Robin

Quote:

---

Originally Posted by Robin Hewitt

Hmmm
...
I suggest you aim for 5 micron accuracy, the resolution of your everyday digi-caliper. You will then get 25 microns fairly easily. You will then realise 25um it isn't enough.
...
Robin

---

Hi Robin,

1 thou , 0.001 inch, 25 microns for the youngsters, is quite easy over a small distance with a well adjusted machine, but gets progressively harder over long distances. I have aimed at this over the 200mm by 200mm by 100 mm range, quite tight enough - and even if I get close it should be good.

I have rejected converting an old milling machine on a simple basis, space available is of the order of 1000mm by 600mm including access so say 600mm by 600mm.

Quote:

---

Originally Posted by Ross77

...
If you are worried about thermal effects then you will definitely need the consider dynamics as well, beam/column frequency and moving masses etc. just adding more mass can lead to resonance:eek: not to mention problems with accel/decel....

---

Hi Ross,
It was the mass/rigidity that convinced me that thermal probles should be addressed. The same mass of aluminium should be more rigid than steel, but has about twice the thermal expansion and also has much better thermal conductivity - hence easier to control the temp of the whole device.
Hidden agendas are, aluminium is also much easier to work with with the equipment that I have and my workshop temperature varies from freezing (4C) to bloody'ot (33C)

on the C1 ballscrews, I keep looking on ebay.
Mike

Quote:

---

Originally Posted by leadinglights

.... The same mass of aluminium should be more rigid than steel...

---

about 3 times more rigid... but also 1.7x linear length, i.e a beam 5cm x 5cm steel will need to be 8.5cm x 8.5cm in aluminium for the same mass, or 6.5cm x 6.5cm for the same rigidity, but because it is less dense it will more easily transmit vibrations from motor to workpiece partly destroying the benefits...

Quote:

---

Originally Posted by irving2008

about 3 times more rigid... but also 1.7x linear length, i.e a beam 5cm x 5cm steel will need to be 8.5cm x 8.5cm in aluminium for the same mass, or 6.5cm x 6.5cm for the same rigidity, but because it is less dense it will more easily transmit vibrations from motor to workpiece partly destroying the benefits...

---

Hopefully easier to address than many of the other potential problems. Modern motors, whether stepper, servo or BLDC tend to be smooth and powerful and compliance for damping in the mountings of less consequence than in ballscrews, guides etc.

Mike

So has there been some confusion over the level of accuracy then??? :whistling:

0.025mm doesn't sound that unachievable. when I was looking for the xy table for my mill I'm sure I saw some precision tables in that travel range for around the £300-£400. bolt it to tee slot table you've all ready got, add a solid fixed gantry for the z and go from there on a trial and error basis.

I'm sure most of the accuracy you require will be achievable by finding the machines best running conditions for each material type and maybe establish a warm up routine so its not used when really cold......then it wouldnt matter if everthing was tight and binding when cold.

Quote:

---

Originally Posted by Ross77

0.025mm doesn't sound that unachievable.

---

Set your micrometer to .001", look at it and decide how you're going to divide this tolerance up between the slides, the screws, the spindle runout, the surface finish and flex in the system.

On a good day with an obliging material I can cut better than .001" in the Y, but the X cuts around +0.0015" or worse. All bearings and ballnuts are preloaded, the Gibbs have fine adjustments and pukka slide oil, I even have a pneumatic quill lock to de-slop the Z, BUT, it's a round column mill and even wound right down that 6" diameter, cast iron tube at the back is more willing to twist than it is to bend. It's a kind of torsion bar suspension.

You might think that I could just go around again and whisk off an errant .0015" but it doesn't work like that. Once you get inside the flex parameter it prefers to rub rather than cut. Wind it in until you pass the flex point to get a cut and whammo, instant undersize.

You might think that I could cut the X undersize on the finishing path, but it's erratic. I think I've reached the limit of the machines' capability.

As soon as you skimp on the iron you also get to worry about sympathetic vibration. There's a reason why the troops break step when they cross an iron bridge even if it can carry a much heavier load than them.

Perhaps milling to fine tolerance with less than 2 tons of cast iron to back up the tool is a black art. I could try hanging around midnight crossroads in case Old Nick turns up so I can do a deal for some fern seed but I may just try filling the column with ferro concrete first :whistling:

Robin

Robin,
Is your round column hollow, if it is you could put a screwed rod down the middle and put the column under tension as I'm sure that would help without adding extra weight.

Peter

Quote:

---

Originally Posted by Robin Hewitt

On a good day with an obliging material I can cut better than .001" in the Y, but the X cuts around +0.0015" or worse. All bearings and ballnuts are preloaded, the Gibbs have fine adjustments and pukka slide oil, I even have a pneumatic quill lock to de-slop the Z, BUT, it's a round column mill and even wound right down that 6" diameter, cast iron tube at the back is more willing to twist than it is to bend. It's a kind of torsion bar suspension.

You might think that I could just go around again and whisk off an errant .0015" but it doesn't work like that. Once you get inside the flex parameter it prefers to rub rather than cut. Wind it in until you pass the flex point to get a cut and whammo, instant undersize.

You might think that I could cut the X undersize on the finishing path, but it's erratic. I think I've reached the limit of the machines' capability.

As soon as you skimp on the iron you also get to worry about sympathetic vibration. There's a reason why the troops break step when they cross an iron bridge even if it can carry a much heavier load than them.

Perhaps milling to fine tolerance with less than 2 tons of cast iron to back up the tool is a black art. I could try hanging around midnight crossroads in case Old Nick turns up so I can do a deal for some fern seed but I may just try filling the column with ferro concrete first :whistling:

Robin

---

Robin has nicely quoted an example of what happens in the real world regardless of endless calculations.

Quote:

---

Originally Posted by ptjw7uk

Robin, Is your round column hollow, if it is you could put a screwed rod down the middle and put the column under tension as I'm sure that would help without adding extra weight.

---

Hi Peter

The twist is too small, like .005 degrees. Adding extra tension wouldn't help, neither would concrete for that matter but it might make me feel better :heehee:

Robin

Status of the project.

Having been disabused of my idea that I could get the kind of accuracy that I want in a desktop sized milling machine, I have settled back to trying to get the best I can with my limited resources. Previously I had the new milling machine all planned out down to the level of suppliers and part numbers, but all of my plans now lying in ruins. So far I have come up with three options which are in varying stages of taking shape in the immagination of my Turbocad program.
Attachment 2167
The first option has reached the stage where I can show you a graphic of the X and Y axes. Construction is intended to be in one inch thick aluminium tooling plate for frame as well as the table. Frame parts held together with 10mm socket screws with 6mm dowels fitted after it has been trued up. Initial trials with a wooden mock up shows that the geometry is stiffer than you would expect. Cost of tooling plate is quite expensive, but not breathtakingly so - the bed at 520mm by 240mm of 1" plate is $108 from an American company (Onlinemetalstore) - I assume not absurdly different in Blighty. I am awaiting with trepidation some quotes for milling the edges - including four accurate datum faces.

Second option is begining to take shape and is more like a normal milling machine with an XY table. The basic idea is to have two large "L" shaped pieces of 7075 aluminium bolted to separators to form a stout "I" beam structure.

Final option is the same as option 2, but is based on a rusty hulk of a milling machine(?) of unknown make. The vertical beds would be milled and hand-scraped flat to fit the guide rails - simlarly the horizontal beds. A new cross table would be built from scratch. So far this is only a picture in my mind.

Mike

Looking good and glad you haven't been put off by the other miserable B*****ds:heehee:

I'd go with the first option but if it was mine I would put the y-axis lead screw between the blocks, like you have done on the x-axis. The gantry could also do with a bit of stiffening to prevent racking, either another plate on the other side or a length of 8020 bolted to the ends and the y-axis plate would really improve it.

the lower gantry fixing may also be a weak spot and again I would use some 8020 (or angle plate) to strengthen the base plate and prove a larger fixing surface to the gantry.

How did you get on with thk rails?

Quote:

---

Originally Posted by Ross77

Looking good and glad you haven't been put off by the other miserable B*****ds :heehee:

---

Mea culpa, you go that way after a few years in this game :whistling:

It looks perfectly serviceable now that Mike has cut his tolerances to 'whatever is achievable'.

Mike, I have bags of Belleville washers and springs left over from preloading my 16mm ball screw nuts and bearings if you want some (assuming no one has pinched them). True zero backlash is a delight after you have experienced the alternatives, I often find myself idly rocking the X axis handle a thou either way and feeling it move the table :beer:

regards

Miserable git

Other things have got in the way of my replying earlier, but in the interim I have been doing a lot of reading - mostly on the web, but a few books. Nothing I have read has made up my mind what kind of machine to build - or even what is needed to get the sort of accuracy I want. As an example of this, I was just looking at a Renishaw (dental milling) site where they said that too rigid a frame worked against accuracy. (something to do with compliance missmatch)

The upshot of this is that the only certainty is the need to machine parts to try different things out - I need a (CNC) milling machine to make one. As I enjoyed converting my Proxxon, hopefully I can find something that will both be useful, and become the basis of a conversion.

By the way, Ross77, the THK rails are fine. They have been swaddled up in protective paper against use in the, hopefully, not too distant future.

Mike

Have you got a link to that article? I know that just adding more mass can lead to vibration problems but not heard that a machine can be to stiff !!!!!

Is that the type of machine you are after? I know what you mean about needing one to make one. I'm in the same position (although nearly there now)

Link is http://resources.renishaw.com/downlo...en&inline=true

Probably a little lighter than I want, but not much. The sort of thing I want to make may be (examples to show range)

Frame side in aluminium machined from 2mm alloy plate, ribs "T section" 2mm by 2mm with up to 60mm unsupported.
Compressed air turbine in titanium (a bugger to machine with anything) 26mm dia 6 grams weight - must be inherently balanced (hence need accuracy)
machinable castable wax model to try casting iconel model of same if titanium doesn't work.

The most common needs are accuracy and a really high spindle speed.

Mike

What a fascinating machine. An upside down mill using a die sinker's rib cutter to machine ceramic. Presume "non-Cartesian" means they have done away with the need for slides by using two, offset rotation centres. I've never been a particular fan of "skimp on the design then try to fix it with software", but that looks like it might just work :beer:

I can't find the reference at the moment, but I think that it is a sort of stabilised tripod with five legs - two of them are cross braces. Each leg can be individually jacked up or down giving X, Y, Z and two limited axes of rotation. There was something vaguely similar in the form of the "hexapod robot milling machine" on Youtube.

Of course, I could be wrong and it may be something else entirely.

Mike

Edit:- Robin, I have just been thinking about your two offset rotation centers - and fix it with software - what a great idea, have you patented it yet?

Simplest would be a single, horizontal, swinging arm with a bearing at one end for the workpiece and a bearing rigid to the machine at the other.

The available work area would be an odd shape, the maths would be tricky, but replacing linear bearings with good old taper rollers could solve a lot of rigidity issues especially if you dumped their C shaped frame in favour of something tripedal.

Quote:

---

Originally Posted by leadinglights

Edit:- Robin, I have just been thinking about your two offset rotation centers - and fix it with software - what a great idea, have you patented it yet?

---

With a bit of clever design you might be able to use bearings that were physically larger than the offset. A bit of luck on ebay could make it quite solid at a reasonable price, large bearings with no particular purpose don't attract a lot of bids :beer:

How big do you need? Ive got quite a few under the bench . if theres one the right size then it yours for the cost of the postage.