Another homebrew CNC miller

[Roger HartParticipant @rogerhart88496]

[Roger HartParticipant @rogerhart88496]

Looking at feasibility of building a small miller for watch parts – so about 30mm X and Y and say 20mm Z. I have a microscope XY table as a possible start but the question of feedscrews arises. I reckon I can make a reasonable 0.5mm x 5mm screw and lap it and make an adjustable nut but is this likely to have small enough backlash to allow dial engraving/rose engine work?

Alternatively I have some optical grating strip – so would it be better to go the servo route? Or get a big lump of wood, some CI and brass and make a rose engine and do the numbers by hand.

[Ady1Participant @ady1]

I have an old precision pultra with anti backlash nuts etc, it was used to make bits for the Thunderbird missile program in the 1960s

An effective anti backlash system has 2 feedscrew nuts, one for each direction of travel, a near perfect leadscrew along its entire length or you will get binding, and tapered gibs for the slide

It's a lot of work for relatively little gain unless you plan on doing a lot of that kind of work IMO

[Gary WoodingParticipant @garywooding25363]

A friend tried utilising the conventional leadscrews on a new mill that he was converting to CNC. No matter what he did he was unable to eliminate, or compensate for, the backlash. I suggested the method *here* which I thought would probably work, but he never tried it.

In the end he used ballscrews, which, or course, did the trick, but I've no idea if they are available in the size you would require.

[Bob Brown 1Participant @bobbrown1]

There are miniature ball screws available 4mm diameter x 1mm lead **LINK** just another option.

Bob

[Michael GilliganParticipant @michaelgilligan61133]

Roger,

I would look out for an old Toolmaker's Microscope as the basis.

Societé Genevoise; Zeiss Jena; Gaertner … that sort of thing

These typically have ball slides and high precision lead screws.

The screws on my Gaertner are even adjustable for effective pitch.

… They do come up on ebay.

MichaelG.

[Roger HartParticipant @rogerhart88496]

Thank you everyone. Will have to consider whether the effort is worth it or to stick to old fashioned methods. I like the miniature ball screws – have not yet enquired re price…..

What is frustrating is that most of cnc seems doable fairly cheaply right up to the mechanics – but whichever way I twist the problem there seems no cheap or easy way past that problem. So to forget it or JFDI is the question.

[Michael GilliganParticipant @michaelgilligan61133]

Roger,

You may find it worth looking at the Airpax Digital Linear Actuators.

I bought some [ebay] a while back, and they seem to work well

… On a small XY stage they should be about right for watch work.

… Not sure about their realistic life expectancy though.

MichaelG.

[BazyleParticipant @bazyle]

A pantograph is sometimes used for watch work. How about CNCing the pattern side of that to reduce the leadscrew problem if it doesn't introduce new ones.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Bazyle on 15/06/2014 22:04:16:

A pantograph is sometimes used for watch work. How about CNCing the pattern side of that to reduce the leadscrew problem if it doesn't introduce new ones.

.

Something I have been pondering for a while

Go for it Roger !!

… Then I can learn from your experience

MichaelG.

[John HaineParticipant @johnhaine32865]

The Airpax unit specs say nothing about backlash and don't appear to use ballscrews.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by John Haine on 16/06/2014 21:35:37:

The Airpax unit specs say nothing about backlash and don't appear to use ballscrews.

.

John,

True … That's why I said:

"… Not sure about their realistic life expectancy though."

Given the small working envelope and low forces that Roger needs:

• Does it matter if they don't use ballscrews?
• Might it be cheaper to just swap the unit if it develops too much backlash?

I don't know the answers … just wondering.

MichaelG.

[Marcus BowmanParticipant @marcusbowman28936]

Ballscrews alone, especially small diameter ballscrews, don't have zero backlash. You would need two nuts per screw, adjusted to eliminate the backlash. What ballscrews do offer is a relatively constant backlash along the length of the screw. Ground screws have less backlash than rolled screws.

Yes; you are right: good mechanics don't come cheap, but they are essential if you are to realise the promise of CNC. As usual, you won't get what you don't pay for…

I am also of the opinion that just because the workpieces will be small that doesn't necessarily mean the CNC machine will be small. Modern Swiss watchmaking machines tend to be large and heavy, if pictures are anything to go by. Size and weight are associated with rigidity, and that's often a problem when making DIY CNC machines. In fact it is a problem with many of the commercially available small mills etc (Hobbyhorse…).

One way to be able to use large diameter screws is to use differential screws which effectively provide a fine pitch despite the screws themselves having a coarser pitch. Expensive, though.

A problem with a fine X or Y feed is that when machining small parts you will need a high spindle speed and fast feeds to allow the cutter to cut efficiently.

Marcus

[John McNamaraParticipant @johnmcnamara74883]

Hi Roger

Have you considered using a steel cable drive for your XYZ axis?
Steel bands are also feasible but difficult to source in small sizes

You will need guide pulleys at each end of the travel They can be ball bearings with pressed on flanges leaving a slot for the cable that runs directly on the bearing. a V pulley is not required for the kinds of load this machine will impose on the cable and bearings are very accurate. A 2mm 7×19 Stainless steel cable has a breaking strain of over 330 KG (2.5mm over 500 KG) It will be a very stiff drive with very little backlash between the carriage and the drum. Cable drives are pre tensioned.

Consider the following:

Travel needed 100mm

Rule: less than one turn of the drum allowed to avoid non linearity issues with a laid multi turn drum (Quite feasible but more complicated)

The steel drum that cable drive runs on is 40mm using PI 3.141… the circumference is 125.66mm.
Therefore rule one is met. 100mm of travel is less than one turn.

Now how to drive the drum

Step motor using half step = 800 counts per rev

11 tooth driver pinion on motor
Driving a 110 tooth Driven gear attached to the drum
This pair gives a ratio of 10 to 1
Of course this ratio can be increased to reduce the step length and increase the resolution. at the expense of a larger driven gear.

Translated to movement per step this equals .015708 mm per step

Or .00062 inches per step

I have avoided micro stepping beyond 800 (standard step motors step 400 per rev) more is possible at the expense of linearity.

With 10 to 0ne reduction there will be plenty of torque available. smallish step motors can be used

The timing belt drive will have a little backlash but the 10 to run ratio will help. most of the error will be at the small pinion and that is divided by 10 before it affects the driven gear.

A while back I posted a thread on building a Timing belt driven router using a similar setup with a friend, but in that case a timing belt itself is used for the drive. not as good as the steel cable drive suggested above but the travels for the router are a lot longer, over a metre.

The router is in constant use and has worked perfectly, the pre loaded steel cable core T2.5 drive belts have not stretched and there is very little backlash, they have not yet needed to be adjusted. All the rotating parts run in ball bearings. I think essential for this type of drive but easy to do!. Here is a link. **LINK** Page down a bit to see the illustrations. There is also a short video.

For this tiny mill with 100mm travels the drives might be mounted under the machine out of the way? That would make it a good conversation piece no motors! I would be happy to collaborate with the design… It would be a fun to do project.

Regards
John

Edited By John McNamara on 19/06/2014 16:35:49

[Russell EberhardtParticipant @russelleberhardt48058]

I've seen high precision semiconductor manufacturing kit that uses a nut split lengthwise so that it can be compressed onto the screw as it wears and it achieves sub micron accuracy. Just an idea.

Russell

[jason udallParticipant @jasonudall57142]

For 25 mm throws…and low force….mmm where have I seen that……MICROMETER……just a thought…
But seriously steppers are used in this way are an open loop system…

The half nut solution suggestion seems good….ever wonder how the next most accurate screw thread was made.. ( think long screw thread squeased between two flats of wood. .holding say 50 leads and travelling 10)…worked for the romans.

[frank brownParticipant @frankbrown22225]

Anti-backlash nut1, make long nut, cut off 3/4 threads worth at one end, refix to long nut with screws which are parallel to the leadscrew, with a space between the two nuts of one pitch of the lead screw. So when the nuts wear, tighten the screws, so pressing the nuts together on to the leadscrew thread flanks.

Anti-backleash nut2. Make a nut with a taper at one end, cut external thread on the taper. make a nut to suit the external thread. Split the tapered end into four fingers. Tightening the external nut onto the taper will press the fingers in and so tighten on to the leadscrew thread.

Splitting the nut axially and tightening it up, will cause the centre of the thread to drift from where it should be.

Frank

[Michael GilliganParticipant @michaelgilligan61133]

Posted by jason udall on 19/06/2014 17:05:59:
For 25 mm throws…and low force….mmm where have I seen that……MICROMETER……just a thought…

.

Jason,

That is one of the several reasons why I first suggested an old "Toolmaker's Microscope" as a basis for Roger's machine. … all it would need is a cutter head and some steppers.

It's probably just my paranoia showing; but …

MichaelG.

[Russell EberhardtParticipant @russelleberhardt48058]

Posted by Michael Gilligan on 19/06/2014 21:40:59:

That is one of the several reasons why I first suggested an old "Toolmaker's Microscope" as a basis for Roger's machine. … all it would need is a cutter head and some steppers.

Wouldn't it be sacrilege to destroy such a fine instrument – if you could find one? Rather an expensive option I think.

It might be possible to find a decent microscope X-Y stage though.

Russell.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Russell Eberhardt on 22/06/2014 10:32:14:

Posted by Michael Gilligan on 19/06/2014 21:40:59:

That is one of the several reasons why I first suggested an old "Toolmaker's Microscope" as a basis for Roger's machine. … all it would need is a cutter head and some steppers.

Wouldn't it be sacrilege to destroy such a fine instrument – if you could find one? Rather an expensive option I think.

It might be possible to find a decent microscope X-Y stage though.

Russell.

.

Russell,

Such things turn up in various states of dereliction … some are worth using, some are worth restoring, some are beyond use as a microscope but mechanically sound.

As for prices … they vary greatly.

MichaelG.

[Author]

Posts