[dlyork]

Hello,

I have been designing my first cnc router over the past couple of months while surfing around this website picking up bits and pieces of advise on how to build one of these. I don't have any way of machining metal so I decided to use 8020 and cnc router parts linear motion system for ease of construction. Currently I am going through the analysis of my design to get a better feel of how much it will deflect under use. I'm most worried about the cutting platform sagging when the gantry is in the middle. My design right now has two extra pieces of 1530 that are a couple of inches set in from the rails. Would it be better if I just had 3030 along the edges? It will primarily be used to shape foam and cutout skateboards (7-11 plys of hard rock maple). Any advise or insight would be greatly appreciated.

Doug

[Outlander]

I agree that unless you some how support the middle you're going to have table sag. Also at that length and width I think you're going to have problems using the single under the table screw. Not only with whip (I cant tell from the pictures but it doesn't look like your screw is supported ) but also with racking of the gantry. I see you're using CNCrouterparts components have you considered using their rack and pinion setup? With the R&P setup you could get rid of the under the table part of the gantry and make room from some more supports.

[dlyork]

I did a cost analysis of the R&P system vs ACME screw and the R&P system costs about $500 more than the ACME screw ($3300 vs $2800 whole machine cost). Because of the cost difference I would like to find a way around having to use R&P. Each axis has a bearing block supporting each end of the screws. I might put two bearing blocks at the non drive end of the x axis screw in order to get more of a fixed-fixed support and thus higher critical speed. With the gantry at the ends of table where there will be the greatest length of unsupported screw it will be slowing down to change direction, so the angular velocity will be much less than the critical speed which makes me think that whipping will not be that big of an issue. Also I only need about 50 ipm working speed which is only 100 rpm on the 5 start screw. I don't exactly understand what racking of the gantry is. What kind of deformation is it?

Thanks for your help,
Doug

[Outlander]

As long as you aren't looking for lightning speed I agree that you probably wont run into whip. What diameter screw are you going to use? Although I don't think you'll have a too much if any issue with racking as Ive been looking at design some more and your y axis is only 24" its something to consider. Racking is, if you're looking down on the top of the machine, when the gantry is subjected to rotation due to the cutting force at the edges of the y axis travel. It can cause the gantry to bind on the screw. Honestly I think if you can sort out a little more support for the center of the table you'll have a machine that will do what you're asking.

Adam

[jsheerin]

Racking is when the gantry yaws in the x-y plane (rotates around the z axis). Think about having the router at one side of the gantry and the screw driving the gantry in the middle. The offset force of the screw will try to pivot the gantry around the router bit. If you have a drive mechanism (rack and pinion or screw) on both sides of the gantry, then the racking will be prevented (as long as the force available from the drive mechanism exceeds the reaction force of cutting and your motors don't stall).

50 ipm is pretty slow... How big is your design? I can cut above 100ipm using 1/2-10 5 start screws on a 48x48" machine.

[Drools]

Using Ahren's carriages, you might want to think about protecting that X-Axis rail from debris (saw dust, shavings etc). My router is very similar to your design but I upgraded to Ahren's carriages. It just worked out that I had a ~1/8 gap between the back of the carriage and the upright. Like in the right side of the picture. I used a long piece of click-floor underlay on both sides to protect the X-axis CRS which is working perfectly. Now I have to figure out how to protect the Y-Axis.
On the racking issue, you could always build the whipperstopper using 8020 bearings.

[dlyork]

Thanks for the definition of racking. The working volume is 5' x 2' x 5". I know that 50 ipm is slow but in all honesty it will make a big improvement in the production speed of decks being able to cut and drill one every 3 minuets vs the 30-45 min it currently takes with a jigsaw and hand drill. If I can get it to run stable faster then great. I have attached a pic of just the x axis and how i was thinking of stiffening it. Another thing I was thinking of but have yet to model is using 3030 as the outside edges.

[ger21]

You're going to want to cut a lot faster than 50ipm.

I have 60" long 1/2-8 2 start and get 150ipm (600rpm). There's a little whipping at that speed, but partially due to slightly bent screws.

I only have 45" of cutting area, but even at that size, 150ipm is painfully slow for larger parts.

[dlyork]

The first thing I am going to make with it is going to be a vacuum mount to suck up most of the debris. I will most likely also find something to shield the rails maybe by making the mdf cutting platform a hair smaller than the gantry uprights. At 600rpm I would be getting 300 ipm plenty fast for anything I could think I would need to do.

Thanks for everybody's input.

[dlyork]

I have been working on finalizing the the design so I can find the reaction forces at the bearings, but I keep running into a problem that I have 6 unknowns but 5 equations. This makes the system statically indeterminate and unsolvable. I have attached a picture of the built up Z-plate assembly with the bearings I have selected (green) as having reaction forces and simple cutting forces as shown. Does anyone have suggestions on how to solve this system? Thanks.

[jsheerin]

It should be pretty simple as you have it drawn, but I don't think it's correct. I think you'd need to also include the reaction at the nut. Solve in each direction separately. Combine your 4 bearings in the same plane into two (since it's symmetric). Iirc, the way you do these kinds of problems is calculate the direct reaction force on each support and then the reaction force from the moment created by the applied force. Then combine them into total reaction forces (but it has been quite a while since I've done these).

However I'm not sure how useful this is... I guess if you know the load ratings of your bearings you can check that for a given cutting force you're not exceeding the rating. However I'd be more interested in calculating the deflection of the tool tip for an applied load. This is pretty tough to do by hand for a realistic design such as you have sketched, but you can use FEA to do it if you make some assumptions (or spend a lot of time building a really complex model). From this you can get an idea of what portion of the machine needs reinforcement to reduce deflection. I've been doing this in a thread for a mill I'm working on: http://www.cnczone.com/forums/showthread.php?p=815261

[ahren]

Doug,

I'm excited to see your machine design coming together, and obviously excited to see you using our parts. Let me know what I can do to help!

I'm not too sure where the $500 premium calculation came from for the R&P -- the way I look at it, you'd need a couple of pieces of rack from Moore ($80), some rack clamps ($84), and a couple of R&P units for your long axis ($89*2), for a total of $342 (assuming you still use ACME for the cross axis) . For the screw, you need an acme screw ($57.50), acme nut ($18.50), bearing blocks, motor mount, shaft coupler, shaft collars, and thrust bearings ($67.50) for a total of $143.50 (or $156 if you you two bearing blocks on one end), so the premium looks more like $200, for a dual drive machine that can easily rapid at 600 IPM (and cut foam at this speed as well). You will need another axis of electronics, which may be part of your calculation, but from a performance standpoint there's a huge improvement. There's a reason I've helped over two dozen Joe's builders convert from multi-start ACME to rack and pinion.

Don't get me wrong, I think your machine will be great either way. Whatever you decide, keep us posted on your progress and I'll watch your thread if there are any questions you have.

Best regards,

Ahren
www.cncrouterparts.com

[dlyork]

jsheerin,

I have combined each group of four bearing into two which is how I end up with 6 reaction forces. The force from the Z axis motor only offsets the weight of the assembly and therefore is a known force. It dose contribute to a moment about the y axis, but the problem is that none of the 6 bearing forces aren't coupled to the z axis equilibrium. Is there a way to nullify one of the forces at a bearing? Maybe if I calculate the moments about one of the bearings instead of the center of mass. This will be my goal for the weekend.

Thanks ahren for your support. I did include the cost of the electronics in my calculation to get the $500 extra. How much better is the accuracy with R&P vs. Lead screw? This will eventually be used for production so repeatability accuracy and precision are all important. I can have .05" inaccuracy on cutting the outline of decks probably even .1" and nobody would notice, but the holes for mounting the trucks need to be pretty on point and aligned or the whole deck is scrap. If the accuracy is there it'll be worth waiting for more funding to build a higher quality machine.

If I do go with R&P, I will redesign the X axis (long axis) to be like Arbo's, for dust control and stiffness. I'll keep posting models as the design comes together.

[ahren]

Doug,

The accuracy of both systems is about the same. I have a friend who makes custom snowboarding bindings on a machine with ACME 5 start, and has no problems keeping the mounting holes aligned (similar application to yours, I think). From the R&P side, I have several customers making cabinets with their machines, which requires boring hinge recesses and shelf pin holes. I think either one will work well for your application. It's more about speed (both positioning and cutting) that makes the R&P a higher-end solution.

Best regards,

Ahren
www.cncrouterparts.com

[jsheerin]

Hi dlyork,
Here's an example of a simple way to do the x axis. I also included a sketch of a more realistic way to do the x axis with the acme screw as point 3. Hopefully you can see this in the pic... Then you'd do the y axis and z axis the same way (z axis would have a force from plunging the router into the cut, for example) and then sum the forces at all bearings. Also, you'd want to divide the resultant forces at the bearings by 2 since there are two of each bearing. Hope that helps.

[dlyork]

I just cant seem to get it right. I can solve it for the X-Z and the Y-Z axis but the third one doesn't match up with the first two.

[jsheerin]

Solve for reaction forces on all bearings with a force in x, then y, then z. Different forces will involve different bearings. Then sum the forces at each bearing for whatever combination of forces you're interested in.

However I still don't think this is very useful unless you know the load ratings of the bearings (and we already know that these bearings work through actual usage, so why bother?). What would be more useful is looking at deflection of the frame, but I think you'll have a tough time taking this information and calculating deflection of the frame.

[dlyork]

By doing this I am trying to find what forces will be acting on the frame to get a more accurate approximation of what the deflections will be. Deflection is proportionate to the force acting on it.

[jsheerin]

You already know what forces are acting on the frame. All the bearing reactions will be transferred to the frame through the steel rail and the bolts holding it to the 80/20. Sure, there will be local deformation of the steel rail due to the localized loading of the bearings, but I doubt you will calculate that by hand. Additionally, are you going to calculate the effect of the multiple loads on various points of the frame? Typically you would use a beam model and apply moments and distributed or point loads. Going beyond that is when finite element analysis makes more sense. So why not just assume the z axis is a rigid body mounted to the frame than then calculate what the frame will do with your loads applied at the appropriate distance from the frame? This is what I did for my first machine design to size my 80/20 beams.