[Vinnito1]

Howdy folks,

I've been working the last few months on converting my G0704 to cnc. I've got most of the design complete, ordered a majority of the parts, and now starting to machine everything. My first goal is to get the machine working and them make additional improvements to it. My list of major parts for the build includes Acorn controller, 60VDC power supply, DMM servo drives, and roton ballscrews (should of looked harder).

I took the time to model the machine in CAD and boy there is hardly anything that is square or symmetric on this thing. I had to get creative to measure some of the parts that were longer than my calipers.


I made the mistake of trying to measure and model the dovetails. Did i mention there isn't anything square on this thing



Below is the final design. I initially wanted to use Clearpaths but moved away from them due my limited budget and found a great deal on some used double stack NEMA 23 Servo motors. I haven't model the head yet as its going to be a future project.




I will mention that all the motor mounts and ball screw mounts are being made out of steel as I'm going to try to have both ends of the ball screws fixed. The choice to use metal is try to match the CTE to the cast iron as i'm hoping for very little lateral movement in the ball screws and prevent it from binding up during the large temperature swings ( that's what I have in my head at least). From what I have read, this will have the most stiffness vs a floating end or unsupported end. T


The biggest issue I have at the moment is how all the mounts are attached to the saddle. I think the two screws jamming the nut against the saddle is a very poor design choice but a very go choice to assembly it easily. I don't think I have seen this before but I'm going to bolt the ball nut mounts to the saddle. This will take additional machining of the saddle, but it is well worth it in my opinion not having the jam screws loosen up or cause the ball nut mount to deform. Below is the design for the X Axis mount


Below is the Y axis mount. Again this requires more machining on the saddle and base.


I'll share more pictures of the build as it progresses. I also need to give a shout out to my new friend Lucas for the machining he has done and is doing for me!! With out his help, this project would be much more difficult to do

[CS900]

There is no real good reason to have fixed ends on both sides of the ballscrews. You're assuming that the screw and cast iron will be the same temperature if you're trying to match CTEs, and this simply won't be the case. The ballnut will have a higher temp than everything else, followed by the ballscrew. And even then it'll be more localized in the area you're performing the work in. The cast iron itself should have very little heat if it's lubricated properly.

[Vinnito1]

There is no real good reason to have fixed ends on both sides of the ballscrews. You're assuming that the screw and cast iron will be the same temperature if you're trying to match CTEs, and this simply won't be the case. The ballnut will have a higher temp than everything else, followed by the ballscrew. And even then it'll be more localized in the area you're performing the work in. The cast iron itself should have very little heat if it's lubricated properly.

The reason why I would like to used fixed is ends is that the ball screw is able to carry more load and have a higher max speed, according the NSK and THK guides I have looked at


I know that I will not be able to push the ball screw to its limits because of the machine design limiting factors, but I was trying to increase the overall stiffness of the ballscrew to limit the amount of deflection it had under load.

Last year I was able to measure the saddle on some fancy measuring equipment using a Reinshaw touch probe system. This is much easier to taking measurements with calipers!


My friend got recently got a Precision Matthews PM-949, so it is going to be much easier making parts

[Vinnito1]

Here is the first modification to the saddle using my friends PM 949 pictures above. We created clearance for the ball screw, ball nut and the new attachment method for the ball nut mount.

[CS900]

The stiffens comes in the form of pre-tensioning the ballscrew. I think you'll find that putting that load on your casting will probably cause things to warp more than anything (especially on the table). I think you hit it on the head when you said the limiting factor will be your machine. That column is going to flex before anything else.

that said, looks like your making good progress. looking forward to more.

[Vinnito1]

These are the first parts that have been made. My friend Lucas has been kind enough to lend me his equipment and time to make parts. I have the X and Y ball nut mounts complete and the machining done on the saddle for the X axis

Below are the X and Y ball nut mounts that attach to the saddle




Next is the clearance machined out on the saddle for the X ball nut mount. I also cleaned up the face of the saddle to get rid of the jagged edge left by the casting process.


This is the test fit of the X ball mount on the saddle



Luckily, I guess all the measuring paid off as I don't need to add clearance between the X-axis ball nut mount and table!


Once I get everything together and working, I'll post drawing up of these parts if anyone want them.

[Vinnito1]

As previously commented, the column is the weakest part of the mill or I think it is. Since I'm waiting on parts and time to machine parts, I'll share with everyone my wonder how much the design of the column could be improved using FEA. I want to make it clear that this is a qualitative attempt as I don't have the right tools and fixture setup to measure actual stiffness. This was done a few months ago over X-mas break as I had the time to sit down

I started with a nominal 100lb static load applied parallel to the X direction about 3/4 of the way up on the column and used generic cast iron material properties in the mode. I know that the cutting force applied to the column is never "static" but since I have never measured or seen anyone post any measured loads, 100lbs is as good as a starting point as any. I replicated how the column attaches to the base with 4ea M12 fasteners torqued to a value around 55 ft-lbs. I'm not aware of any torque specificied by girzzly, so I made some material assumptions and used an online calculator found here, FUTEK Bolt Torque Calculator | Bolt Torque Calculators


I was very surprised to see the column chad a max deflection of ~9mm with a 100 lb static load. Again I don't have any way to actually measure. I then thought if the column was shaped like an I beam design that could be made out of a solid piece of cast iron. I choose an I shape to provide room for the ball screw in the front and have a place to mount the linear rails that the head could mount too.


The solid piece cut down max deflection to ~2.5mm. This was a drastic improvement over the 9mm. Thinking about how to improve the design, I added gussets to the side and added additional bolts to attache the gussets feature to the bottom of the base

Adding the gussets cut down deflection to ~.25mm, IMO, that was pretty amazing but not practical as of how large a piece of material was getting. I started to think about bolting multiple pieces together to form the column. At this point I took the base out of the model because it was taking a long time to complete the simulations. I started with the column supplied by grizzly.

I was initial surprised by how much stiffer the column was by itself when compare to it being attached to the base. But thinking about it, the results make some sense as I cannot replicate the bolted joint interaction with a single part. So I loose alot of localized deflection at the bolt joint interface and make the part seem stiffer that it may actually be. That's my working theory at least and I'm sure someone will comment on this. At this point, I created a new column made out several pieces that were all bolted together. Throughout the simulation, I have continued to use 100lb static load and generic cast iron as the material properties.


I saw that this column is much stiffer than what is supplied with our mills and could realistically be made if a person wanted to. So I began to looking into material costs, linear rail costs, grinding costs, etc and thought how about I get what I have running first.. I also included a stress plot for everyone to see the localized stress created at the bolted joints. I know I learned more about my G0704 mill , hopefully someone else will too.

[ewick]

PM sent to you Vinnito1!

Thanks
Eric W.

[Vinnito1]

Figure I would update this thread with a few pictures

[vectorsc]

So a fun note - what happens when you add gussets to the existing column? Heck, you could almost capture the existing column within the i beam pocket and bolt it down.

[Vinnito1]

I never looked at adding gussets to the original column. I think there would be challenges in attaching them since the mating surface has a chamfer and some draft, but it is possible. If you wanted to capture the original column in an I beam, i think you could do it. I've seen some smaller benchtop conversions where something similar that was done.

The other item I want to do is space the column off the back of the base like so many other people have done. I'm doing this to be able to get more clearance between the table saddle and column to put protective bellows in places This is hopefully what it will look like when its all done

[vectorsc]

So odd note. I would finish the 0704 mostly stock and see if you can sell it. Why? If you are making your own column and using rails just go all the way and go full custom. Maybe order just the table from grizzly

the base cost of the 0704 could be used to get a 10k spindle with atc built in.

[hanermo]

Nah.
Does not work like that.
Building scratch is a huge project, 30x more work.
I did it, so...

So odd note. I would finish the 0704 mostly stock and see if you can sell it. Why? If you are making your own column and using rails just go all the way and go full custom. Maybe order just the table from grizzly

the base cost of the 0704 could be used to get a 10k spindle with atc built in
.

[vectorsc]

What was the biggest challenge? I mean compared to him milling the ways off to use linear slides and such. And he has the new column already it looks like.

Not doubting you - you have done it lol. I’m just going to do a scratch build next time and am curious. If I had to guess I would say it’s modtly parallelism problems.

[hanermo]

Just work.
Endless, endless issues with everything in scratch builds.

None of the issues are particularly difficult - in machinist terms.

For example I have now had the new saddle plate off the lathe 4 times, due to fitting issues.
My 25 mm linear guides barely fit in the volume, and the 32/5 ballnut yoke has less than .5 mm clearance or 0.25 mm per side on x axis.

It has taken about 60 hours of fiddling, and the current x axis is almost done.
And the lathe x axis is a minor work.
Then add the 2 toolchangers, waycovers, wipers, limit switches, chain quide for wires.

Aligning the x axis linear guides was about 3-6 hours with dtis and temporary saddles and forward dti to indicate errors.
It was quite fast and quite easy.
The final alignment was less than 2 hours, to less than 0.01 mm avg indicated error in travel along the linear guides.

Any error above 0.01 mm is very apparent by feel, with hand, moving the guide blocks.

What was the biggest challenge?

I mean compared to him milling the ways off to use linear slides and such. And he has the new column already it looks like.

Not doubting you - you have done it lol. I’m just going to do a scratch build next time and am curious. If I had to guess I would say it’s modtly parallelism problems.