Opinion about the structure of the router ...

Hello,

I'm designing a router to make molds, it is I need a lot of precision. (For this kind of work an error of 0.1 mm (0.0039 inches) is an huge error.)

I did some research in terms of design and have something very much like the photo you posted yesterday ... JAZZ

The router will have 1200 x 1500 (useful area 900x1200x350)

I'm using tube 80x40x3, 2

The CAD software I'm using is Solidworks.

Would you, please, give me your opinion about the structure?

Thanks.

Sorry for my English ...

-

__________________
Sincerely

CarloSilva

Attachment 5659Attachment 5660Attachment 5661Attachment 5662Attachment 5663Attachment 5664Attachment 5658

For a CNC router it's not difficult to achieve +-0.05mm.

Your Y and Z axis designs look good.

I would increase the size of the triangular plates and add some to the other two sides. You could also use diagonal pieces of box section to reinforce the X-axis. If you can add a piece of box section across the back between the X-axis rails that will make it a lot stronger.

What thickness are the gantry side plates? I'd reccomend 15-20mm aluminium. The X-axis ballnut mounts will be much stronger if you extend them down and mount the ballnut directly to the gantry side. The ballnut mounts are particularly important since any deflection in the ballnut mount directly leads to tool deflection and inaccuracy. You could also add a triangular support between the X-axis bearing block mounts and the gantry sides to help stop that joint bending.

If you don't mind posting the solidworks model I can open it to have a better look.

What ballscrews are you intending to use? For that size I would reccomend RM1610 on X and Y and RM1605 for Z with timing belts and pulleys on all of them.

You will be hard pressed to achieve that accuracy on 10mm pitch screws and gearing down say 2:1 to get resolution will only be the same as running a 5mm pitch direct drive.

Commercial routers use high lead screws because they are servo driven with at least a 2500 count encoder and running in quadrature that's a resolution of 10,000 pulses per rev

Yes I mentioned pulleys but forgot to say you should gear down 2:1.

My machine easily obtains better than 0.1mm, often measure with the caliper and find it's within 0.01-0.02mm and that's on 1:1. Assuming the molds are out of wood you're surely not going to be able to measure much less than 0.1mm difference?

I think using 10mm pitch screws is more versatile since you always have the option for higher feedrates than you could get with 5mm pitch should you need it, at the expense of worse resolution of course.

Quote:

---

Originally Posted by Jonathan

I think using 10mm pitch screws is more versatile since you always have the option for higher feedrates than you could get with 5mm pitch should you need it, at the expense of worse resolution of course.

---

Why does everyone assume you have to travel at warp speed, is it a dick slapping contest ?
Anyway when you are cutting unless you are doing simple carcase routing where all moves are straight and long the home stepper driven machine can never accelerate to these insane speeds everyone is bandying about.

Program a part to run at 2000 mm/min then watch the screen and read the actual feed-rates. No way does it get anywhere near this figure because it never get to speed before it has to perform the next line of code.

Now I could understand 10mm pitch screws if you were on servo's and production work but personally I'd much prefer working more slowly and accurately and get the job done whist the machine is under control, not be boreline running away, especially for a beginner, after all this IS a hobby forum isn't it ?

I tend to mention 10mm pitch on this size machine as the machines I've seen in person with 5mm pitch screws don't seem to run as smoothly as mine or get as high acceleration.

Quote:

---

Originally Posted by John S

Anyway when you are cutting unless you are doing simple carcase routing where all moves are straight and long the home stepper driven machine can never accelerate to these insane speeds everyone is bandying about.

Program a part to run at 2000 mm/min then watch the screen and read the actual feed-rates. No way does it get anywhere near this figure because it never get to speed before it has to perform the next line of code.

---

With the 3nm steppers on 70V and 10mm pitch my machine on the vast majority of jobs would get to 2000mm/min. I have the acceleration set to 1000mm/s^2 on X and more on Y. So to get from zero to 2000mm/min from zero takes:
v^2=u^2+2as
s=(v^2-u^2)/2a
s=((2/60)^2)-0)/(2*1)
s=0.000555...

So for any linear move over 0.55mm it will get to 2m/min speed, or similarly it will maintain 2m/min for any arc above 1.1mm radius. When I cut woods I generally go at around 6m/min, been known to use 8m/min when roughing. High feedrates are especially good with MDF as you can maintain a big enough chipload to make chips, not dust which is bad for you, without lowering the spindle speed to the point where there is not enough power.

I made the right size pulleys to run my machine on 2:1 for better resolution, but I've never needed to use them.

Here we go again?????

Quote:

---

Originally Posted by Jonathan

For a CNC router it's not difficult to achieve +-0.05mm.

Your Y and Z axis designs look good.

I would increase the size of the triangular plates and add some to the other two sides. You could also use diagonal pieces of box section to reinforce the X-axis. If you can add a piece of box section across the back between the X-axis rails that will make it a lot stronger.

What thickness are the gantry side plates? I'd reccomend 15-20mm aluminium. The X-axis ballnut mounts will be much stronger if you extend them down and mount the ballnut directly to the gantry side. The ballnut mounts are particularly important since any deflection in the ballnut mount directly leads to tool deflection and inaccuracy. You could also add a triangular support between the X-axis bearing block mounts and the gantry sides to help stop that joint bending.

If you don't mind posting the solidworks model I can open it to have a better look.

What ballscrews are you intending to use? For that size I would reccomend RM1610 on X and Y and RM1605 for Z with timing belts and pulleys on all of them.

---

where can I post the solidworks files???

Quote:

---

Originally Posted by C.AlveSilva

where can I post the solidworks files???

---

You should be able to attach them to a post using the advanced view. Same method you used for attaching the original pictures.

Quote:

---

Originally Posted by Jonathan

You should be able to attach them to a post using the advanced view. Same method you used for attaching the original pictures.

---

My Zip file. It's too big "5.76 MB" ...

I sent an email with all the zip files, but in the meantime I made changes. Can I send again to your e-mail?

Hi again.
I made ​​a few changes in my router project. after taking account of the costs, I came across the need to make some adjustments. My budject is smalest. But what I ask is that comment on the change in the yy axis (the botom linear guide).

My YY axis is all in alumineo, 20mm thickness, is better than steel tube? probably more expensive... but less heavy.

I also change the ball-screw diameter and pitch instead 2005 I use 1610. I could control the resolution with pulleys.

Please comment.


Thanks.

Sorry my English ...

-

__________________
Sincerely

CarloSilva

Attachment 5896Attachment 5897Attachment 5898Attachment 5899Attachment 5900Attachment 5901Attachment 5902Attachment 5903Attachment 5904Attachment 5896

Quote:

---

Originally Posted by C.AlveSilva

Hello,

I'm designing a router to make molds, it is I need a lot of precision. (For this kind of work an error of 0.1 mm (0.0039 inches) is an huge error.)

I did some research in terms of design and have something very much like the photo you posted yesterday ... JAZZ

The router will have 1200 x 1500 (useful area 900x1200x350)

I'm using tube 80x40x3, 2

The CAD software I'm using is Solidworks.

Would you, please, give me your opinion about the structure?

Thanks.

Sorry for my English ...

-

__________________
Sincerely

CarloSilva

Attachment 5659Attachment 5660Attachment 5661Attachment 5662Attachment 5663Attachment 5664Attachment 5658

---

Hi, I don’t want to burst your bubble but also do not want you to waste your time. I have built injection moulds for Gillette and the motor industry. Assume you are not thinking of moulding plastics or any material that flows through an injection mould sprue bush when hot as there is no chance that a gantry mill (Router) is accurate enough for a precision mould, particularly an injection mould using materials such as nylon or plastics where the flashing gap causes many issues. These moulds are constructed using highly polished surfaces for the part and extremely accurate Attachment 5906part ejection pins that are cylinder ground with the mould parts hardened then fitted after surface grinding to within 0.005mm tolerances way beyond gantry or any other mill capability. The tolerances needed for mould making far exceed that of most other precision engineering needs
It is possible to make a simple mould with a gantry mill providing the moulding material is quite thick. Clay moulds are an obvious option, however if you need to build the two sliding parts (Back and Front) of a mould accurately, you will have to apply all your engineering skills in precision part manufacturing for the mould part construction to achieve a working result from a gantry mill (Router).

I too have a Gantry Mill and have increased its accuracy simply because it reduces the hand work needed to make the parts fit properly. Just finished my Kitchen where machining the new gas hob and new sink aperture in a worktop was a doddle on the CNC machine. If you want to increase the accuracy of a gantry mill you need to apply the following:

• Use high precision linier bearings on each axis
• Reduce the spindle head moment distance as much as possible
• Keep the work piece inside the bearing boundary. (I noticed that your design allows the Z axis to drop below the X axis which will work ok but will increase the moment and reduce the machining accuracy.)
• The spindle head needs to be high power variable speed AC motor using a VFD frequency controller and at least 3kw able to draw 8-10 amps under load. High load side pressure bearings with a collet run off less than 0.005mm
• When machining more accurately “slow down” the faster your machine runs the more overrun is used.
• Become familiar with tool speeds and feeds for different material machining.

This gantry type design is easy to construct but has an inherent problem as each axis compounds the axis tolerance accuracy error, so the closer the spindle is to the Y axis the better. When the tool head (6) is under Z+ pressure any head movement depends on the Z tolerance which depends on the Y tolerance which depends on the X tolerance so the head movement is compounded by the bearing tolerances’ and overall design. Spindle wobble Attachment 5905can be as much as 0.2mm, I use the type of bearing in the photo above which has much more strength and accuracy than the round type. If each axis baring has a tolerance of 0.02 movement this compounded by the distance from each baring to the next X axis has 0.02 by the time the moment reaches the top of the Y axis (4) the moment is increased by the width and height of the side, 40% is estimated this is then increased again by the Y axis to Z (5) where another 20% is applied and finally this is increased by the Z to the tool bit (6) there again half the distance from the side giving another 20% not forgetting that these are compound slackness errors. Starting at the X axis with 0.02mm movement at the top of the Y site this will be 0.03mm movement, then the Y moment to Z 0.02 added to this is 0.05 and then back down the moment for the Z axis to the tool bit could end up with an X axis movement of 0.1mm (0.005”) and that is without any pressure on the tool (6). This is the sort of tolerance expected from a router for machining wood, which is of course what they are designed for. Another area of concern with this type of design is the accuracy in terms of X and Y axis squareness, this design lends itself to almost no adjustment and of course this is the most important aspect of any CNC machine. If you make a part that is not precisely square its fitting ability is seriously affected. It is important to use the right tolerances and material hardness, in my experience building press tool and injection moulds even the standard milling machines used in a tool room like a Bridgeport mill, even though design to overcome temperature changes and are very rigid, are still not accurate enough for mould making as they can not machine hardened surfaces or produce a sliding surface for a mould part to work without gaps and problems. I can see in your design, the Z axis is inside the Y bearings between (1) and (2). However, the design will still compound axis bearing tolerance but at least when under pressure the load will be between the X axis bearings (1 and 2)
Attachment 5907
Other things to consider are:

• The affect temperature has on the frame design; the coefficient of linier expansion will increase and decrease the frame members potentially putting your job position in a different place on a hot day or if the machine increases in temperature from its use.

• The distance between bearing 1 and 2 in the diagram make a big difference on the machine accuracy, however the longer this is the less relative X axis movement you have.

• Fit linier scales and a closed loop system. I have fitted a DRO and linier scales and am working on a MACH3 positioning monitor to dynamically adjust to the scales position.

• Square X and Y; the only way to check this is to machine along the Y axis and back down the X axis after setting the test piece parallel to the X axis with a DTI and flip it over and check the run out with a DTI along the Y axis. The reading difference will be double the machines error.

Attachment 5908

I have managed to reduce the “moment” on my router to half of it original design and I now get an accuracy on +- 0.0254mm (0.001”) with much effort put into adjusting the X and Y squareness I can now produce a successful circuit board and flip the board over only compounding a squareness error of 0.04mm run out over 500mm. The machine is now also rigid enough to machine aluminium, so I have changed the spindle to reduce the RPM needed for light engineering work, now making new parts for my CNC machine.

Hope this all helps and is beneficial.