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

[peteeng]

Hi Cat - What are the restraints in the model? If you are restraining the base feet the model will be much stiffer then in practice. Restrain the table or better still say a block "vice" on the table so the load loop goes from the tool to the vice which is then correct. CSA or UHPC start at 26GPa and through to 80GPa. Epoxy granite starts at 20 and can get to 40 as epoxy has poor stiffness. cheers Peter

[catahoula]

Yep, it's not clear in that photo but the restraint is a 200x200x12 square plate in the middle of the moving table- that way the table and frame are free to flex. The typical use for this machine is milling aluminum molds about 300 x 180 x 75mm, so I think it's a reasonably accurate setup.

[catahoula]

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.

jack, thanks for the tip on the yaskawas. Are you referring to used/refurbished prices? I looked into those and it turns out that yaskawa drives are able to do fully closed loop control within the drive using a linear scale. That's the holy grail for me, although I don't think I want to tackle it right away. I might start with DMM so i can buy a pre-built control unit, and then sell those later and upgrade if I find the need. At some point I need to actually start using this machine

spumco, thanks for the info about the drives. DMM assured me that 400W was plenty strong given the loads. The fixed gantry layout is nice for that, relatively light moving components. I initially was looking at 3:1 reduction but they said 2:1 is fine, which is good because finding room for 60t HTD5M pulleys on the ballscrews turned out to be really difficult, and I am set on not going smaller than 20t for the motor pulley out of concern about backlash and wear.

Sterob- yes heat treaters are fairly common in any medium sized city. I'll have to travel about 3 hours to neighboring town. They charge by weight, this frame at about 275 kg will cost $200 USD.

[footpetaljones]

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.

Do you have a link to where Yaskawa servos can be bought?

[spumco]

Some - not all - of the Delta drives can do dual closed loop as well. If you're willing to go used, Allen Bradley Ultra 3000 and Copley Xenus drives can also to dual closed loop, and they're both pretty flexible on encoder inputs.

Ultra 3k's can be found for $100-$200 pretty frequently on ebay and are pretty bullet-proof drives. There's a good reason A-B discontinued them and then immediately brought them back - customers loved them and howled.

If you eventually want dual closed loop at the drives (and not at the controller like with LinuxCNC), you can get the drives and not fiddle with mounting linear scales right away. I've fiddled with the A-B and Copley software and it doesn't look that challenging to set up.

[jackjr-123]

jack, thanks for the tip on the yaskawas. Are you referring to used/refurbished prices? I looked into those and it turns out that yaskawa drives are able to do fully closed loop control within the drive using a linear scale. That's the holy grail for me, although I don't think I want to tackle it right away. I might start with DMM so i can buy a pre-built control unit, and then sell those later and upgrade if I find the need. At some point I need to actually start using this machine

No I'm talking new. However I think the fully-closed loop feature is an option?

Do you have a link to where Yaskawa servos can be bought?

China of course
You can find plenty on eBay but you'll get better prices by asking for quotes on Alibaba. About $400 for a 400W motor+drive+cables set (SGM7J series). If that's still too much Delta latest ASD-A3 series are nearly as good and less expensive (I was quoted $370 for a 750W set).



Here's some pictures (sorry for the poor quality) of my gantry being assembled, lying on its back. Interior was epoxy-granite filled with a tube in the center for cables. Top rail mounting surface was epoxy molded with a granite straight edge.
Bottom rail is a piece of steel 50x15 glued (epoxy injection after rough alignment with grub screws) and bolted to the tube, then scraped for flatness. The small pockets at both ends for mounting the ballscrew supports were milled in-place with a temporary Y/Z axis setup...

[catahoula]

Here's some pictures

Nice work, it's so much cleaner with epoxy bonding than welding. Gives me some ideas for how to increase the rail spacing on mine...



Dug into the Allen Bradley ultra 3000 manual, one of the easiest to digest I've encountered, but still a bit daunting. Pretty tempting, thanks, and the auxiliary encoder feedback seems to be standard. Generally speaking how bad/good of an idea is it buying used or refurbished servo drives? Do they burn out? I'd given up on used stuff after struggling to find the right ballscrews and linear rails, but the servo drives actually seem possible as there's fewer variables

[spumco]

Generally speaking how bad/good of an idea is it buying used or refurbished servo drives?

I generally shy away from used electronics, unless...

The Ultra3k's are all over ebay. Tons of them for sale, many with 'working' warranties of various periods. If you get your ducks in a row (get a servo, linear scale, cables, and bench breaker/power supply, and PC with appropriate software before you buy a drive) you can bench test a drive pretty quickly.

Just pay the small premium for a drive listed as 'working pull' or similar with a DOA return policy. Those two features will up the price 20%-50%, but I think it's worth it.

You can also get deals on multiple drives if the seller has more than one. If they've got multiples, send an offer for 80% of list for a handful of drives. They may bite, or at least drop shipping a bit.

At the price of the AB drives it's not worth repairing them if they're dead (unless you can do it yourself). They tend to be pretty bullet proof, and as you've read the manuals are in English and comprehensive. And the tuning/programming software will run on a modern computer (unlike the still-good but obsolete Parker Gemini series).

Same goes for the Copley Xenus stuff. Used Copleys are usually a bit more expensive than Ultra 3k's for the same amerage rating, but the Copleys can be had in 20A continuous (40A peak!) on 240 single phase. The biggest single-phase U3k (20A) tends to be rare and expensive.

Not many single phase drives can handle that load (4kw servo spindle, for instance). The really cool thing about the Copley Xenus Plus (newer) models is that they are extremely agnostic - they can interpret just about any encoder on the planet, including proprietary Yaskawa/BISS/etc.

[catahoula]

Cool, thanks for the tips. Always blown away at the depth of knowledge on this forum. Cheers

[ardenum]

Bottom rail is a piece of steel 50x15 glued (epoxy injection after rough alignment with grub screws) and bolted to the tube, then scraped for flatness.

The left side looks like its a regular tube but the right side, is this how you made it?

[jackjr-123]

This is just a plate glued and bolted to the tube.

[catahoula]

Got a bit nervous on the stability of the front part of the frame where the angled pieces come together with the rails, so decided to add some additional tubes last night (seen in second photo). I also finished out the underside with more ribs welded into the hollows and the three-point mounting pads. The three point mounting means the machinist can lay the frame upside-down, surface the three feet flat to the machine, flip it over and do the rest of the work. Part of the reason i went so heavy on the tubes in this build is to make sure that the machining is accurate- that the frame has enough mass and rigidity to not distort in the machining process. Time will tell if I did enough. The frame certainly could be a bit thicker in the z-axis, but it's 5"/125mm deep as is, which is pretty solid. I'm thinking I will pour at least some of the EG fill before machining to beef it up even more.

Tomorrow, I'll fix the gantry tube so as to increase rail spacing from ~165mm center-to-center to ~210mm center-to-center. Honestly this is probably unnecessary and I'm fairly certain I'm being paranoid, but at this point it's worth another day and another $100 of steel to make sure it's right. This way, c-c rail spacing will exceed the cantilever (or z-travel) of 200mm. It'll also result in more surface area for bonding to the columns and a lower position for the lower rail of the two, so all good things for real-world stiffness.

(please forgive the dirtiness of the shop floor, the angle grinder + cut-off wheel makes a royal mess and I'm almost done using it)

[Sterob]

I have forgiven you Bro.
Nice work.

[mactec54]

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.

The Yaskawa is more than twice the price of any Dmm product, I use Yaskawa and Dmm and for the money DMM does the same job or if you need / want a higher Encoder PPR then Yaskawa or Delta DMM 20Bit Encoder is in testing is close to being released

[mactec54]

The left side looks like its a regular tube but the right side, is this how you made it?

That is Cast Iron a lot of commercial Routers use custom cast beams like this

[mactec54]

This is just a plate glued and bolted to the tube.

Dreams are free too

[catahoula]

I have forgiven you Bro.
Nice work.

Ha I think I forgot the laughing emoji there

Finally got the frame off to the heat treater for stress relief. It'll be done next week and then I can take it to the machinist.

Getting kicked out of my shop space at the end of the month, and unfortunately I live in one of those "desirable" places to live that saw spikes in property values around 25% or more in the last year, along with record low inventory. It's making finding a place rather difficult, so this frame may sit on the pallet in storage for a bit.

[catahoula]

Happy to see some of the linear motion components arrive from China today! I ordered with Fred at the BST Motion Aliexpress store after reading some positive reviews here. Super helpful and positive experience. His english is pretty good, certainly good enough for reasonable communication with only a few corrections. Linear guides haven't arrived yet but he had to go straight to Hiwin to get H accuracy class rails, so I'm not surprised. These packages arrived after only a few weeks, including custom end machining of the ballscrews and air freight. Ballscrews were packed well and tightly with foam blocks in a doubled cardboard box.

I went with ground C5 TBI OFUR (offset style double nut) ballscrews, 2005 size, with P1 preload (zero clearance, no preload). Reading the TBI manual I thought P2 would best ensure zero backlash over time, but he said when another customer had demanded P2 against advice, the ballnuts turned out too tight and had to be changed for P1. So I went with P1. In the hand there's certainly no perceptible backlash. They turn pretty easily, to make a very subjective statement, with a little bit of roughness here and there. This was one of the toughest choices- I was initially set on traditional double nut ballnuts so the preload could be adjusted as they wear. But I couldn't find a supplier for those from TBI even though they're shown in the catalog, and new THK or the like would be out of my budget. There's a good used supplier for THK/NSK etc on Ebay, dyglobal I believe, but without being able to be certain of the preload, I couldn't quite go through with it. Ultimately it seems like to get higher preload you probably need to go up to C3, or maybe C5 from THK/NSK etc. I almost went with the BST rolled double nut screws and then use a DRO for screw-mapping to gain back the accuracy, but thinking about trying to shim out backlash to a screw that has quite a bit of variance just seemed a little optimistic. TBI C5 were around $700 for 2x 2005 800mm-ish and 1x 1605 450mmish. The other reason I moved away from true double nut screws was out of concern that most wear will happen in the middle of the screw, so shimming for preload down the road will have varying results across the screw anyways. If I get a couple years from these and then need to spring for THK, that'll be OK. Mounting blocks are the BST C5-grade house brand. These are angular contact and if necessary I'll shim the fixed end bearings to remove any axial play.

Also here are some basic stamped steel servo motor mounts (will likely machine rigid aluminum mounts eventually) and HTD 5M 15mm belts with 20t x 40t pulleys for a 2:1 drive ratio. CAD is one thing but these look very stout in real life. This stuff was all the BST house brand and seems decent on cursory inspection.

Cost for linear guides, ballscrews with end machining, mounts and belts/pulleys was about $1450 + $400 shipping. I think that compares pretty favorably, as long as the linear guides show up some time

[catahoula]

And here's the modified gantry. I flipped it around so the old rail pads will now just be hanging out on the back of the gantry. I welded 25x63mm flat bar on the new front face for the rail pads, and then added a 12mm plate top and bottom to support the overhang. Pretty similar in effect to what jackjr showed, but I wanted to stick with fully welded construction (for a one-time thermal stress relief and just general stability). This will allow me to move the rails 50mm farther apart for a center to center spacing of about 210mm. It also made a larger, flatter surface for the support columns. Previously the support column top plate would have been hanging over the tube corner radius a bit. It was quite a bit of fabrication, as to combine two flat pieces of steel you need holes or cutouts to weld. Killed another 5 cut-offs wheels or so, and the better part of an oxyacetylene tank fill for preheat

The former gantry was maybe 50kg or so and I could manage it myself, but this modification nearly doubled the weight. Yikes. I'm now reconsidering the necessity of filling the tube with EG. I'll need to have it on the engine hoist to position it anyways, so maybe i should just fill it anyway. Might do a partial fill with a PVC tube down the middle. The Datron M8Cube has a steel gantry and given the acceleration of the machine it seems unlikely that it's filled with anything, but that's just speculation....

Altogether this modification probably falls in the superfluous category, but I only want to do this once!

Edit to add: my super sketchy file cabinet work table has officially survived this project and will now be discarded

[catahoula]

Update: linear guide package was separated from the rest of the shipment at the warehouse. Seller arranged expedited shipping of it over the weekend. Solid customer service so far.