Added steel channel to X-axis c-beam

[wiremonkey]

Great! I'm going to put together a part 3 summation video and I want to think about the next step/upgrade. I do want the machine stiff enough to mill aluminum, but maybe I keep this machine for wood an other materials and start from scratch, save my pennies and build a stout little milling machine.
In turn:
1. There was no before data, unfortunately. If I had to guess, I'd say that my modifications both to the Z axis: steel V-track, bearings and stainless channel, and to the X axis: steel channel, added very little rigidity where it was needed, in the middle of the X beam.

2. OK, so v-track is probably fine to stick with if I try an X beam upgrade to steel.

3. Should I use: http://www.mycncuk.com/threads/2214-...ess-calculator URL for the calculator and the general mycncuk.com URL in general elsewhere? I'm happy to drive traffic here if I can.

4. I entirely understand that your time is limited and I appreciate that you've already helped me so much! I can't believe you ran my numbers, that's so kind.

5. What you're describing is actually what I did, I think. I applied say 20lbs to the scale by pushing the router collet down, which amount to upward force. Then I did the same 20lbs for 7 points all over the machine. That is what the data represents. Maybe I'm misunderstanding.

I do have more questions about the next step, steel. But, I'll start a new thread so all can benefit.
Thank you again, routercnc!

[routercnc]

Hi Tyler,

1) OK, but read 5) below before concluding (although you might still be right!)

2) No, what I meant was just upgrading to profile rail would not help unless the underlying structure was improved. So I agree if going with the steel beam upgrade (which is your plan) then I would recommend doing both the steel beam and the profile rail.

3) I'm OK with you including the URL in the text box below the video on YouTube for the calculator and the forum site. Lee Roberts is the site owner and I'm sure he will chip in if there are any issues with that.

4) All I did was divide the force you applied (converted to N) by the displacement you measured. The only with any real meaning is the force and displacement applied at the collet/cutting tool so make sure you include that.

5) To get the bigger picture when the force is applied to the collet - what I understood you had done is apply force at one location (say location A) , and measure the displacement at that same location (location A). Then applied the force to ANOTHER location (say location B), and measure the displacement at location B, and then the force at location C and the displacement at location C.

I'm saying don't do it like that. It is much more helpful to apply the force to location A, then measure at A (A being the collet/tool). Then apply at A again, and measure at B. Then apply at A again and measure at C, etc. Do not move the force location, only the measurement location for the deflection. Then by looking at the numbers, or plotting them on a graph, you can see where the stiffness falls away and where the improvement is required in the structure. You can also apply at A, and measure the displacement at A,B,C,D etc if you have enough DTIs. If this is what you did then carry on.

The short answer to all the above is:
Used closed sections - e.g. rectangular hollow sections [], not [, U, or I

Maximise their outer dimensions (this gives stiffness improvements to the ^3 for a linear size increase)

Then up the wall thickness until the weight is getting too high (this gives stiffness improvements directly proportional to increase so is not as effective as outer dimensions)

Minimise offsets, these create moments which you will not fully recover through beefing up elsewhere

Be practical - consider the principles, but you have to be able to build the machine !

[wiremonkey]

Ah yes, I did take measurements the preferred way. For each measurement, I simply plunged the router down onto the scale, the same way for all measurements taken. Sorry I wasn't clear about that. I will also take some deflection at the collet while applying lateral force in both the X and Y directions.
Yes, from what I'm learning, closed structure for beam is best. Presently I'm thinking about using a 2" x 5", 3/16" thick steel beam (8.15lbs per foot) with 1/2" steel end caps welded on, for mounting to the gantries. The way I see it, since the worst of the deflection is clearly from twisting in the middle of the X beam, I'll start there. I can upgrade the rest in stages, as money allows. The idea is that I can reuse the steel beam in whatever configuration I end up with. If I go thicker on the steel tube walls I think it gets too heavy for my design.
I may go with linear bearings, but I'm also considering v-track and v bearings. The the prices are $300 and $200 respectively, in the long run, not much of a difference.
Again, thank you for all of the feedback! I'll post a part 3 video soon in which I discuss some of things we're talking about here and consider options. On another note, I'm seriously considering leaving this CNC machine as is and saving my pennies for a mini mill for metal work and leaving the larger machine for wood and softer materials. The CNC, as is, is quite adequate for such purposes. Taig, I found them today, makes some incredible tiny mills capable of milling mild steel for short money, and you can order them CNC ready, just hook up your steppers and off you go. Have a look: http://www.taigtools.com/mmill.html
Pip, pip!

[wiremonkey]

So, after much deliberation, I've decided to go with a 2" x 6" x 3/16" x ~750mm long piece of mild steel for the X beam. I found a local machine shop willing to flatten the face where the linear rails will go and make the bottom perpendicular along with plates that I'll weld onto the ends for mounting. I found a good source on eBay for the steel. The reason I opted for 2" x 6" x 3/16" was both budgetary and design based. I couldn't source the size tube I wanted locally so shipping was prohibitively expensive after a certain wight/size point, massive jump in price. Literally from around $36 shipping to well over $100! And, since I'm doing this upgrade in increments, I'll eventually switch over to a fixed gantry with Y table movement, I want to keep the weight somewhat reasonable. This beam is 9.42lbs per foot making the total weight around 23.5lbs. That's a goodly chunk of steel to be throwing around along with the whole X/Z assemblies and router! Ooftah. Let's see what these poly carb wheels can take (on the gantry plates). The total upgrade is going to be close to $500, not including all the swearing I will do. I'm teaching myself to weld with a pitiful little DC arc welder I picked up, and off to the races! Wish me luck! Video to come.

[wiremonkey]

So, here is my current plan for the X beam, 2" x 6" x 3/16" steel beam with linear rails. I'll weld at least .25" thick plate onto the ends for mounting to the gantries. Maybe thicker? Not sure if it's necessary, will just add more weight.
Over on the Openbuilds forum, M90Ranger suggested I split up the rails and use a design like this: Certainly two .25" wall 2" or 3" square tubing would be cheaper than the single 2" x 6" I purchased. Somehow I like the overall feel of the single rectangular tube. Which do you think would be more rigid?

[magicniner]

Two tubes could be almost but not quite as good if the end plates were of infinite stiffness, the guy who suggested that obviously isn't aware of the concept of Triangulation in structural design or the properties of different cross sections in torsional load.
I don't find that surprising any more though ;-)

[wiremonkey]

Well, to be fair, most of us building our own CNC mills are hobbyists, not engineers. So, the closed rectangular tube is acting as a triangle, in a sense? What if the walls on the smaller set of two square tubes were thicker, say .25" in stead of my 2" x 6", which has 3/16" walls? The weight would be about the same, within a pound or two. I'm leaning toward the singular rectangular tube for sure. Although my hunch is that they would be similar in rigidity due to the fact that the linear rails would ensure that torsion would be applied to both rails evenly, no?
Thanks for the input magicniner!

[routercnc]

The split tubes have the advantage that the Y axis can be very close to the gantry as it does not need to be offset to accommodate the ballscrew behind it which usually takes up more space than the rail and carriage.

But it is nowhere near as stiff a shape as a single rectangular gantry. In vertical bending the 2 tubes are only twice as stiff as one of the tubes on its own. But when joined as a single large tube they are 2^3 times stiffer (8 times) for example.
As mentioned by magicniner there are similar benefits for torsion for similar reasons.

That is why when you see 2 tube designs with the screw in the middle the builder often plates them together at the back to try and gain some of this advantage back.

If you go with the single large rectangle then the rails can be spaced out on thick solid bars welded to the section to give the ballscrew clearance. Sure, this puts a moment on the section but you can afford a bit of this loss as you have so much to start with.

Another variation is to have a tall more slender section vertically and have another horizontal section behind in a reverse L shape. This is popular when used with aluminium extrusions and the ballscrew can sit behind the section driven off a good bracket that often houses the Z stepper too then the rails can be tight against the section again.

I’d favour the large rectangle as per your blue drawing but the above gives you some options and reasons.

[wiremonkey]

Here is the Taig in action. I mean, those are some seriously deep cuts! I'm not affiliated with them, I just like the design. https://www.youtube.com/watch?v=_9y-_9SKJ8s