Catahoula's router/mill build - welded steel with EG

[catahoula]

The beginnings of my fixed gantry router/mill. 40 hours of fabrication in the books since last Friday. Most tubes and hollows to be filled with epoxy granite after stress relief and machining. Still need to do the columns, but pretty close to done with the welding. Proud of my baby 110V welder for surviving with this super heavy wall steel-- greatly aided by a 300-400 degree F preheat, though. Low power makes it hard to weld over the tacks and keep it looking nice, but overall I think good for minimizing heat, and penetration is enough for this stiffness-critical machine. I started out doing a lot of stitch welding to minimize heat, but after working on the frame and feeling vibrations move through the structure, I decided to fully weld everything except the long, heavy rail mount bars, but using more narrow stringer beads than weaves where possible. Preheating thick steel plus proper weld sequence makes this take about 10x longer than it would otherwise!

Tip: granite countertop scraps are readily available at the used building materials stores. $5, and flatter than about anything short of a blanchard ground top or actual surface plate.

Thanks for all the help so far; great forum.

Similar Threads:

• Welded steel frame router build
• New Machine Build- Welded steel mill hybrid
• CNC ROUTER MADE FOR AND BY AMERICANS , WELDED STEEL
• CNC ROUTER MADE FOR AND BY AMERICANS , WELDED STEEL
• My First All Welded Steel Router Design

[catahoula]

Question for the forum- I will have the long steel flats milled to accept linear rails at a local machine shop. The machine in question is an absolutely massive old Hydrotel milling machine, so fitting this on the table is no problem. Apparently the machine is quite accurate, but I don't have a way to verify this. Should I have shoulders machined for both rails of each pair? Or just have one side of each pair done with a shoulder and let the other rail be adjusted to match? Obviously I'd prefer to have the rails a completely set-and-forget affair with two shoulders, but if the two shoulders aren't very close to parallel it won't be much good. Thoughts?

Edit to add: I've also worked with the machinist to figure out a workholding situation in advance, so I'm pretty confident that my router frame will fit up on the machine table without much distortion or flex.

[jackjr-123]

You should first stress-relieve the whole thing if you seek accuracy.
If you think you can enhance the flatness of the mounting surfaces after the milling (by scraping, grinding, lapping, ...), you should not mill shoulders for the rails as they will be annoying to work around.
Note that these shoulders in small hobby mills have absolutely no benefits regarding lateral forces. They are only useful for installing the rails quickly IF they are milled accurately to about 0.01mm/m. If the mill can't achieve that, you can do a whole lot better by aligning the rails manually with a precision straight edge.

[pippin88]

An alternative to shoulders is to get the side of the rail mounting milled so you can indicate along it.

This means a shoulder is not in the way if you need to scrape the mounting surface.

[catahoula]

Thanks guys.

jackjr- definitely going to have this commercially stress relieved in a large vacuum oven. Walls are heavy so am optimistic about the resulting stability of the structure. Agreed about lateral forces- shoulders would just be to provide a hard datum edge. I did find a 36" straight edge at Shars tool for $450 USD, which is less expensive than I'd thought (starrett is $1k usd), Thanks for the guideline on straightness, that's very helpful. I will check with the machinist to see if 5 tenths over the whole thing is a reasonable ask.

pippin- interesting idea. But maybe if the machine is good enough to mill a datum edge it will also be able to mill accurate shoulder? I may have the edge milled as well for good measure though.

[pippin88]

Datum edge doesn't lock you in.

You can double check with a straight edge / master reference.

Hard to adjust a shoulder.

Also errors in different planes may not be the same.

Known good machine - shoulder should be fine.

[catahoula]

Had to take a break from fabrication for a bit, but have now built the gantry supports. Now it's just weld on a few lifting points and other small features, and the parts are ready to go to the heat treater for thermal stress relief. Exciting as it's starting to look like a machine!

I've been struggling a bit with weld distortion with the large plates being attached as flanges to the gantry support tubes. Even with what I thought was an ironclad weld sequence, there's some distortion. It's not a problem for the end result as the important interfaces will be machined, but it will add to the machining time and will make the resulting plates a bit thinner than expected (though still plenty thick enough). Moral of the story is go as heavy as possible. 19mm plate was massively better than 12.7 mm plate for pieces of this size.

Linear motion components are ordered and controller will be next.

As fabricated, FEA analysis indicates about 50 N/um of deflection after the epoxy granite fill, assuming the rails and carriages are rigid, which of course is not the case. If it comes in at half of that in reality I'll be happy. I haven't shown the z axis yet but it's fully boxed from heavy aluminum plate. I'm using a 25 mm Hiwin rail on the x axis with medium ZA preload and long "H" length carriages, as that one sees the most moment, and the Y and Z will be 20mm with long carriages and ZA preload too.

[peteeng]

Excellent work Catahoula - Really interested in your machine stiffness when its done. What modulus are you using for the EG? I'd be interested in knowing the FE static stiffness with and without EG....Peter

[routalot]

Its quite refreshing to see a really rugged machine coming together.Make sure it doesn't slip off the bench and land on your foot!Seriously,it looks like the foundation of a very useful machine.

[jackjr-123]

Looks good!

Just curious, what's the size of the gantry tube?

[catahoula]

Thanks all; I started this project trying to make a "minimum necessary" machine, but quickly realized that it's too much time and money to spend and not just do it right. Of course, it's taking much longer than I would like but I'm optimistic about the final result. The machining of the linear guide mounts is the critical step, as long as those come out within 0.0005" of straight and parallel, everything else can be adjusted to a good end result. (gantry will be bolted and epoxied to the support towers, and the support towers will be epoxy-bedded and bolted to the frame (aka "Conformal Shimming"), leaving room for adjustment in all 3 axes). Ball screws will be TBI brand C5 ground and motors will be DMM Dyn2 400w servos unless the hive mind thinks Dyn4 is much better. After speaking with DMM and looking at the charts, it seems that Dyn4 is only necessary for good performance at 750W and above. Servos will be geared down 2:1 with HTD5M belt drive

jackjr- gantry tube is 200x150x12.7wall, with 38mm W x 19mm H flat bar welded on for the linear rail mounting. It's 860mm long. Last night I was looking at the x axis rail spacing on the DMG Mori gantry machines and having a panic that I did not get enough vertical spacing on the gantry X axis rails. I'm wishing I just fabricated the gantry from welded plates- the 12.7mm wall thickness means the the tube corner radius is about 32mm, more than I expected from DXF's of standard AISC tube sizes. I don't know why I got so hung up on using a stock steel tube, could've easily used a 10"/250mm wide plate for the front face and that would have made a lot of design elements a lot cleaner and easier. But at this point I don't see a good way to boost the rail spacing without a lot of messy welding so I will probably just leave it as is, about 175mm rail spacing outside-to-outside. It's not terrible, and hopefully the extra heavy load carriages with preload will compensate to some extent.

Edit: gantry tube is 200Hx150W (8"x6" here in USA) not 200x200

Edit to also add: 2:1 reduction with 5mm lead screws give 7.5 m/min cutting speeds (300 IPM) with full motor torque and 10 m/min (400 IPM) peak rapid speed (max servo rpm). This isn't crazy fast but I think plenty fast enough for appropriate chip loads on any tooling I'm likely to use in a 24k rpm 2.2kw ER20 spindle. The ballscrews at this length and size (2005 X&Y, 1605 Z) would go a bit faster (up to 12-13 m/min) before risking a wobble.

Edit again to add: "conformal shimming"

[peteeng]

Hi Cata - Seems there's another machine in you after this one!! Automotive call your thinking "minimal functional prototype" MFP. Std tubing with radii does limit a few things. I'm sure this machine will be a cracker and allow you to make the next and greatest. Looking fwd to swarf shots. Peter

[jackjr-123]

jackjr- gantry tube is 200x150x12.7wall, with 38mm W x 19mm H flat bar welded on for the linear rail mounting. It's 860mm long. Last night I was looking at the x axis rail spacing on the DMG Mori gantry machines and having a panic that I did not get enough vertical spacing on the gantry X axis rails. I'm wishing I just fabricated the gantry from welded plates- the 12.7mm wall thickness means the the tube corner radius is about 32mm, more than I expected from DXF's of standard AISC tube sizes. I don't know why I got so hung up on using a stock steel tube, could've easily used a 10"/250mm wide plate for the front face and that would have made a lot of design elements a lot cleaner and easier. But at this point I don't see a good way to boost the rail spacing without a lot of messy welding so I will probably just leave it as is, about 175mm rail spacing outside-to-outside. It's not terrible, and hopefully the extra heavy load carriages with preload will compensate to some extent.

Edit: gantry tube is 200Hx150W (8"x6" here in USA) not 200x200

Ah yes I had the same feeling about the rail spacing when I built mine (with a 260x180x12.5mm tube). The tube always ends up looking too small... I placed a rail on top to increase the spacing.

[catahoula]

Ah yes I had the same feeling about the rail spacing when I built mine (with a 260x180x12.5mm tube). The tube always ends up looking too small... I placed a rail on top to increase the spacing.

That's a serious gantry! Do you have a link to photos or a write-up of your machine? I'd be interested to check it out.

Ive shied away from the top mounted rail due to the extra machining difficulty, though it's definitely an optimal layout when done right. The limiting factor for me in adding height to the rail is actually the z-axis motor position, with the servo brake it hangs down very far, but I'd I extend the z-plate I think it'll work. Also has the bonus of pulling the z ballscrew up out of the way of chips

[spumco]

unless the hive mind thinks Dyn4 is much better...

Hive mind checking in...

I won't speculate on whether the DYN2 & 400W combo is better or worse than a DYN4 & 750W combo for your application - you and the folks at DMM should be able to work out what you need for speed/resolution/torque.

However, the DYN2 will require a separate power supply. That alone turns me off from the DYN2 as building the electrical enclosure is much easier when everything is the same voltage and more or less the same package shape/size.

Depending on your controller and accessories, you could wind up with just 240Vac and 24Vdc in your entire build - much simpler and makes for tidy wiring and easier troubleshooting down the road.

The downside with the DYN4 is that they are very deep compared to other drives and generally require a 10" deep enclosure (for a standard NEMA 3/4/12 thing). Even with 90deg DB25 signal cable ends they don't really fit in an 8" box.

I would consider a DYN2 if I were converting an already-working machine from steppers to servos. Keep the power supply (assuming it's an appropriate voltage) and everything else, and just toss in some DYN2's.

But otherwise the DYN4, plus some 750W servos (direct or belt-driven) is a pretty nice setup.

Nice build, by the way. Looking forward to your progress.

[jackjr-123]

That's a serious gantry! Do you have a link to photos or a write-up of your machine? I'd be interested to check it out.

No build log but I'll try to post some pictures.


For the servos these DYN4 400W with brake are overpriced. You can get Yaskawa Sigma7 for less, and they are miles ahead in every metric.

[Sterob]

Had to take a break from fabrication for a bit, but have now built the gantry supports. Now it's just weld on a few lifting points and other small features, and the parts are ready to go to the heat treater for thermal stress relief. Exciting as it's starting to look like a machine!
.

What sort of business can do the heat treating for you? Someone with a furnace?
I think I will struggle to find someone to be able to do this where I live.

Steve

[peteeng]

Hi Steve - There are heat treaters in all of the major centres in Oz. Do a search for heat treaters. Peter

[catahoula]

What modulus are you using for the EG?

I went to double check this and turns out I had mixed up the material properties from another model and was actually using the value for granite, so 55 GPa, which is definitely too much. I just re-ran it at 20 GPa and found 6.9 um deflection at 300N with the EG fill in the gantry, and 7.8 um /300N without. So 43 N/um with and 38 N/um without, a difference but not massive. That's just for the gantry tube, I was still using EG fill in the base frame and gantry supports. So maybe another 1 um deflection with those unfilled too.

The reason these values are lower than what I quoted above is that I've messed about with the model and didn't have the Z-axis box in place, this was just with the spindle mounted on a 35mm thick aluminum plate which has a bit of cantilever to it. So with the Z boxed up it'll be back up to 50-60 N/um in FEA. It'll be a few months before I can attempt to compare this to real measurements.

All in all it shows how difficult it is to get a gantry machine much beyond 50-60 N/um. The best results I've been able to get utilized a custom machined spindle mount that also served as the spindle mounting plate, so it was basically a solid block of aluminum 125mm x 125mm x 350mm bored down the middle to accept the spindle. Inspired by the Datron Neo spindle housing- they must be doing something similar to that. That made a big difference in managing the cantilever and stiffness issues with shimming tramming plates, but that's quite the machining project so I'll only tackle that it if I really need to. Of course, this difficultly is mostly because I'm set on having close to 200mm Z work envelope. If you drop the Z height to 125mm things get massively easier, because most of the cantilever can be taken up by the spindle itself. For my application I could actually be OK with that, but I still want the workholding flexibility of being able to mount a standard 75-100mm vise when necessary, and still have a bit of travel to work with.

Edit to add: I found this example of modulus (https://www.basetek.com/zanite-plus-polymer-composite), so de-rated a bit from 32 to 20 GPa for good measure. Just an approximate guess tho

[catahoula]

A screenshot for reference. this is without the EG fill