Reasonably Rigid Router (RRR) - a cautionary tale (likely) in the making.

Hi all.

I have lurked here since varsity when I was building my first little machine, and have slowly been accumulating parts and designing components over the last 8-ish years - as the budget of a middle class family man (with diverse hobbies) allows...

This will be a thread chronicling the design (briefly) and construction (likely protracted) of the machine in question, which shall lovingly be referred to as the Reasonably Rigid Router (RRR) for the duration of this build thread.

I began with a goal in mind, cut all the things, and do it to 0.01mm repeatably - alas, this will likely not be realised...

The design spec evolved to where I am currently:

Machine footprint - 1538x1366mm
Usable cutting space - roughly 1250x950x170mm
I have 3Nm Nema 23 steppers - they will be driven by dm860a drivers at 65v.
2 spindles, one standard 2.2kw Chinese 24krpm air cooled flange mounted router, and a bt30 6kRPM rate spindle cartridge which will be belt driven by a 1.8kw servo.
Ball screw drive on all axes (2x 2005 on X (regret not getting 2010), 1605 on Y, 1204 on Z), with 20mm HIWIN linear guides.

There are many conventional design choices, and some unusual ones (which I am sure will be cause for comment). I suspect I will need to put little AC servos on my axes to get decent performance out of what will be a heavy machine, but the idea was a solid mechanical build primarily, which could have electrical and electronic upgrades applied as funds allow.


Where I began:
https://i.imgur.com/7d6t4QD.png

Intermediate design using square tube frame elements:
https://i.imgur.com/ExkJWOe.png

Where I have ended up (showing the servo spindle mounted)
https://i.imgur.com/WV3tT2d.png
Change to heavy C Channel because I got a bunch of scrap channel for small money.

The bed will be slightly different, with 20mm steel plate drilled and tapped with a bunch of m12 holes for clamping, and 16mm steel plate on the sections that cant be reached by the cutter.

I have started fabrication of the frame elements, and will share some pics of the process shortly.

Very simple FEA indicates an at the tool stiffness of between 4000 and 6000N/mm(10 component spring factors added in series), which according to the late D. Malicky (of this forum), should be ok for Ali.

Thanks for looking.

Luke

Hi Luke - Looking good. The steppers will move that around easily, especially with 5mm and 4mm pitch screws. Keep t it. Peter

Thanks Peter. I am concerned about gantry weight (121kg), but the maths suggests the steppers will be able to accelerate it without too much strain (its just so much heavier than most gantries, so it will be a case of seeing is believing).

Hi Luke - The gantry "weight" is actually a misconception. The motors are moving the inertia plus friction not the gantry weight itself. You will be able to push the gantry with your little finger when you assemble it. Initially the seal friction will be the dominant load. Once these settle in its the inertia thats the deal. ie how fast do you want to accelerate the gantry+z axis is the real issue. Take it slowly a 5mm pitch screw with 3Nm will "lift" 173kg, you have two screws on the gantry so at near slow you have 340kgf available. If you look at the hiwin design manual they spec the friction at 0.004 of the weight so lets say 121x0.004 = 0.48kg plus the seal friction (say 16x1=20N/10=2kg) 2.5kgf is much less then 340kgf....... Same as moving the Z axis along the gantry , its weight is supported by the structure so its friction and seal friction that you are pushing (same as you can push a 1 tonne car on a flat surface). The Z is being lifted but that's using a 4mm pitch and it will lift it easily. So to do the math you will find the 20mm and 16mm screws rotational inertia will be the biggest load if you want to accelerate fast.

Looked at the manual each seal has 1N friction and the car friction co-efficient is 0.004 for ball cars cheers Peter

Main frame material - Salvaged mezzanine support channel (65x160mm)
https://i.imgur.com/uhlc574.jpg

I cut all the channel frame elements with a 115mm angle grinder, most came out pretty square (not critical, since I will be welding end plates onto each one in a jig for perfect final lengths etc.).
https://i.imgur.com/XMB0tng.jpg

Two 40x16mm flat bar lengths will be the y axis rail mounting surfaces, these rails and the 90deg mounting surfaces (bottom plates for the gantry) will be machined post welding.
https://i.imgur.com/1UMeeD6.jpg

Finished the seam welds for the gantry box, as well as the two bulkhead plates (115mm inside the gantry ends) yesterday. I do not consider myself a good welder, but this will be good enough considering the small actual forces. Grinder and paint...
https://i.imgur.com/LI21Xr4.jpg

Note: Welding, at least for a reasonable time period, puts out a lot of UV radiation. I have sunburned arms and an odd patch of sunburn on my abdomen (welding in a somewhat tatty garage work shirt with holes :) )

I save a lot of money on material (vs getting laser cut tube/channel) but it was, and will still be, massively labour intensive to accurately cut, strip paint and rust, and prep for welding(tip, strip and clean discs work amazingly for this, far better than flap discs and wire wheels). Its time that most people would use for media consumption and rotting though, so while its just for me, I consider it free.
https://i.imgur.com/IgLJYaU.jpg

I appreciate the effort in that response Peter.

I did do similar maths, and determined my kit should be well within spec, but feeds for ali milling (especially with the high RPM spindle) are going to be a bit of a stretch for the little 3Nm motors to get to (considering their torque drop off).
The Z will have 100% pneumatic relief, so the ball screw will also only see inertial load. I must admit, I did not do calculations for screw inertia, how very un-thorough of me... I did put my machine info into the oriental motor screw sizing calculator (https://www.orientalmotor.com/motor-...ew-sizing.html) and it gave me the proverbial green light for my intended accelerations.

I remember reading (about 10 years ago, when writing my undergrad lit review) that steppers handle inertial loads poorly. I think this is the major source of my undercurrents of concern :) Thanks again for your input - its nice to get an informed "it will be OK"

Hi Luke and RRR - please attach your images using the advanced button. The imgur links will be disconnected at some point whereas attached files are here forever for us to look back on. "handling" inertial loads is a relative statement, all motors don't cope with inertia at some point. Modern digital stepper drives do very well and hybrid drives are a bit better... Your 65V will solve many speed/torque issues, it will surprise you... and what's your target acceleration? Cheers Peter

Quote:

---

Originally Posted by peteeng

Hi Luke and RRR - please attach your images using the advanced button. The imgur links will be disconnected at some point whereas attached files are here forever for us to look back on. "handling" inertial loads is a relative statement, all motors don't cope with inertia at some point. Modern digital stepper drives do very well and hybrid drives are a bit better... Your 65V will solve many speed/torque issues, it will surprise you... and what's your target acceleration? Cheers Peter

---

Noted, I will add images on CNC zones storage in future.

I want the machine to get to 2500mm/min for rapids, within 1-3 second (so call it 850-2500mm/s^2 ?) - should be a motor RPM of 500 give or take, and torque down to about 15% of holding IIRC. These are just thumb suck figures though, as I don't actually have any useful experience cutting stuff with a cnc... Would you say that is reasonable?

Edit: I read in your build log, that you move the z axis motor with the Z assembly, have you had any issues with cable longevity etc? RRR has the stepper mounted inside the Z assembly frame, and I have had the components cut already - I don't relish the concept of a redesign...
Edit 2: motor selection based on 1s acceleration to 2500mm/min of 125kg gantry, with a 120N cutting load, single screw, 80% efficiency...
https://www.cnczone.com/forums/attac...4&d=1645565474
I have done no manual calculations to corroborate this, but I would expect it to be theoretically sound... seems like there is headroom.

Hi Luke - I use high flex multicore cable for my wiring. No issues yet. One machine over 1000hrs .. Here's my numbers on your machine. Due to the 5mm pitch screw you have heaps of force. So much available that the rotational torque is the dominant requirement for accel. I expect you'll get from 0-2500mm/min in less then 0.5secs. Check my math I have been wrong. The interesting outcome is that due to the small pitch this system is highly rotational dominant. It will have mega grunt. When you start this machine up set the driver amps to the lowest settings. If set high and something goes wrong it will tear itself apart. I had a signal issue on last build that did this. I then set the amps to min while I was debugging it and haven't bumped it up. Does everything it should at 1A I think. Peter

Thanks for this Peter.

Maths looks good, barring your value for rotor inertia (I realize I didn't provide motor specs...). Manufacturer quoted rotor inertial as 800g.cm^2 (8x10^-5 kg.m^2) - that would put Jsys at 2.49exp-4 per screw assembly, and a rotational torque requirement of 0.3Nm.

That is close enough to the calculated value on the online calculator for your imposed short acceleration. I dont have high hopes for the steppers (1000rpm is, according to my research, well past stall for them), but I do think they will give me 0.3Nm at 500rpm, which is more than enough. I will likely run the drivers at 8 microsteps, as the screws are C7 at best. Cheap chinese usb breakout board tells me it can to 100kHz, even if the step rate is a third of that there will be no controller bottleneck up to 600ish rpm.

Thanks again for your input, and the time you have freely given - it is very much appreciated.

I re-ran some FEA for the Z and Y assemblies (simplified). FEA is not my bread and butter, and Inventor is not a purpose designed FEA package, but its an OK indication.

The z axis was modelled in its fully extended position with the load applied inside the spindle taper (1000N for easy maths)
https://www.cnczone.com/forums/attac...6&d=1645602264
28000N/mm of deflection - a large contributor is actual concave flex in the 10mm back plate, I did think that would be a concern, but the displacement is low enough that I will leave it be. I will possibly increase back plate thickness if I make a longer reach Z assembly as required.

I added a "place holder" part in the y axis assembly to put the load where it should be with an extended spindle, I did not convert the load conditions to a torque and a direct load as I should have, and the place holder will contribute slightly to deflection as a result (not too much I think). I havn't added weld beads to the assembly, so the reality should be stiffer by a bit.
https://www.cnczone.com/forums/attac...4&d=1645602264
20600N/mm of deflection

With HIWIN indicating a stiffness of 460N/um for each of the HG20C ZA bearings I have purchased(1840000N/mm per axis connection, 4 blocks), the 6 component assembly (z, z interface, y interface, y, x interface, x) comes to 11643N/mm of stiffness. This does not include bed and vise, and assumes bonded connections. so the reality will be far less. It is, however, reasonably rigid

I am not going to model the base and X assembly, mostly because its going to be an order of magnitude or more stiffer than the rest of the assembly...

I am preparing to weld the gantry riser/base plates onto the gantry box now - The laser cutters cut 4 instead of 2, so I am going to sandwich them first (so 20mm thick, instead of 10), then weld them to the box with the correct spacing. I will then use the lasercut holes in the base plates as drill guides to drill the holes through the gantry box.

I have beveled the edges of the plates to get some additional penetration for the welding, but there is still a large "void" as a result of the sandwiched attachment...

I am considering the best way to fill this cavity, and perhaps improve the bond. My initial idea is to apply a layer of epoxy a little short of the heat effected zone in-between the layers (one when sandwiching 10mm plates into a 20mm plate, and then another when attaching the 20mm plate onto the gantry box). Does anyone know of any reason why this is a poor idea?

For illustration, it is in the below section (model shows one plate, this will be a sandwich of two):
https://www.cnczone.com/forums/attac...6&d=1645770157

Hi Luke - do you intend bonding then welding? That's not a good idea. If you are bolting right thru the three layers Then you need to check that each "ply" of the 10mm is flat by lapping and the faying surface to the gantry is flat by lapping. Otherwise the three layers will spring and you won't get the proper grip in the joint. If the surfaces are flat and pull up correctly no need to bond them. Check fit with engineers blue... Or paper tissue or bond paper. Are you welding the spacer to the gantry? or setting it to the bearing carrier? Peter

I intend to weld post bonding (I figured if I left a reasonable border so the heat doesn't damage the epoxy), and I intend to weld the risers to the gantry and machine them orthogonal to the rail bars after welding, the x carriage plate will be tapped. I am considering doweling the gantry to the carriage plates after squaring the gantry also, but that something I can do down the road.

The welding is intended to hold the plates together(but only on the edges, no plugs), and to the gantry, the epoxy is really there only as a filler, and possibly to provide a modicum of damping. This is not an area I have any useful experience in.

I can get the plates surface ground pretty easily as a batch, but the assembly will move while I perform the rest of the welds, and the plan is to stress relieve and machine post welding to get suitable mating surfaces for the bolted connections.

Hi Luke - your welding at 1400C you'll need a long distance to not affect the epoxy. The welding will be adequate. If you want to bond then do it after you weld. Stitch weld or leave a gap and use wicking loctite. Look up loctite 290. But welding will be fine. Reread - stress relief is at 650C plus the epoxy will burn out. Don't have anything in there for SR only steel...Peter

Thanks Peter - appreciate the sensible input.

Hi Luke - Since you are thermally stress relieving you don't do things like grinding or fine machining as these will be negated or changed in the TSR. Often things are rough machined, then TSR then finish machined. If you TSR then remove a lot of metal it can still move a little. Talk to your machinist and the SR people to find out suitable material allowances for your processing... I say TSR because you can also mechanically SR using vibration which is called vibration SR or VSR. I do work for a company that makes very large mining equipment and most of their fabrications use VSR not heat. Ovens aren't big enough for the things they make... Peter

https://en.wikipedia.org/wiki/Vibratory_stress_relief

I am halfway convinced to try VSR myself, I have some very competent vibration analysts sitting in the office next door to me, and I am sure we could find the structures resonant frequencies without much effort, Then I imagine some sort of oscillatory vibrator or frequency controlled eccentric weighted shaft would need to be made... Sigh, perhaps a project for another year.

I will only be taking 1-2mm off the critical surfaces (assuming I don't absolutely botch the fabrication), is that considered "ä lot of metal" in terms of residual stress depth? There is a company about half an hour away from me that will be able to TSR treat my kit.

In your opinion, is it a wasted endeavor? considering the ZA grade linear bearings and rails, and the C7 screws? I know its good practice, and I would insist on it if it was for work, but the TSR cost will likely buy a bunch of cutters etc. that will make chips for me:) I will get a quote from the heat treatment place anyway and see what I would be in for.

I should have put it in the OP, but please excuse the picture heavy nature of this thread, It is as much to share my blunders and successes with you all, as it is for me to have somewhere to keep progress photos and a build log.

Following Peters advice, I have gone ahead and welded the gantry risers sans epoxy. They feel solid enough.

I cleaned up the bottom of the gantry, and ground the welds flush where the risers will be attached:
https://www.cnczone.com/forums/attac...8&d=1645798237

Flux peels are satisfying for me...
https://www.cnczone.com/forums/attac...2&d=1645798237

Gantry risers post seaming:
https://www.cnczone.com/forums/attac...6&d=1645798237

I built this little 50x2000 belt grinder as a knife making tool, but it is really valuable when doing general fabrication also. 1.1kw motor on a VFD, does around 33m/s belt surface speed at 55Hz:
https://www.cnczone.com/forums/attac...0&d=1645798237

The 36 grit ceramic belt made very short work of cleaning up the risers:
https://www.cnczone.com/forums/attac...4&d=1645798237

Time is still relatively scarce for me, as I have a busy family and probably too many hobbies, but I am trying to be disciplined enough to do a little CNC building every day.

Ta for looking.

Luke

Hi Luke - I do modal analysis occasionally pre VSR for the company I do stress analysis for. Attached is a typical trailer that I do. This is a small 360T payload one. They do up to 600T ones. Everything is VSR'ed on these in house by the company. The tyres are 3m high. 1-2mm is a good machining allowance, 2mm good. SR is definitely worthwhile. My sheet metal parts just work with the square rails and it has taken time and effort to get the metal guys to make things in spec. & sometimes they have to be redone. But my machines are intended for production so a reliable part shape is important.

Everytime I have something welded the cars jam. Its impossible to keep things in spec for square cars if welded. I have to have parts corrected. I have given up on welding as a production process for my machines. Hobbyists / DIY do this by mucking around with it until they get it working or use epoxy/shims/luck. I do braze things and the assembly stays in spec. But if you braze you cant TSR as it melts. If you have TSR handy it pays to get it done prior to any machining. If you take a punt on it being OK you will spend more $$$ fixing it then the TSR would cost so go to TSR at least have it quoted... Peter