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

[luke peter]

Thanks Bongodrummer.

The structure is pretty small all things considered, TSR will thus be reasonable. I did briefly consider diy VSR, but I dont have the time or energy for the R&D

Its always tempting to keep adding to the machine in terms of stiffness and damping etc, trying to build a perfect machine on iteration 1, for me, it turns into analysis paralysis - and I never get anything done.

[luke peter]

Love the boxy design, although all that grid on the spindle box is actually detrimental unless it's gonna involve some damping material later on. Only make cutouts where its necessary for things like visibility or cables etc.

The edge cutouts are for weight, and draw bar spanner access, the non-triangulated cutouts are for bolt and belt drive inclusion for the servo spindle. the weakest part of the Z frame is the back plate - the 10mm steel bows more accross its width than the frame bends. It will probably be replaced at some point.

[Bongodrummer]

Thanks Bongodrummer.

The structure is pretty small all things considered, TSR will thus be reasonable. I did briefly consider diy VSR, but I dont have the time or energy for the R&D

Its always tempting to keep adding to the machine in terms of stiffness and damping etc, trying to build a perfect machine on iteration 1, for me, it turns into analysis paralysis - and I never get anything done.

If you see my concrete incorporated build you realise I have the opposite problem

[luke peter]

If you see my concrete incorporated build you realise I have the opposite problem

Ah, that's your machine

I have watched a bunch of your videos already, impressively resourceful gantry build. I am seriously considering surface replication for my rail support beam to upright joints - how has that played out for you?

I am also interested in how your machine is performing w.r.t. stiffness - I cant remember if you did any final compliance measurements with your spindle extended? what was the ultimate stiffness at your collet? Your machine performs really nicely, and it would be a good benchmark for what I can expect out of my mechanical assembly.

Genuinely enjoyed your content.

[Bongodrummer]

Hey thanks
The surface replication worked perfectly for me. I didn't really know what to expect going into it, but yeah - no complaints on that front.

w.r.t. stiffness: I don't think it's in the same league as yours, being a big router really. I hadn't actually measured this, despite testing it on steel etc. So I had a look today...
By far the most flex is coming from the spindle plate, as it has a lot of z travel (400mm from the bed to the gantry, and it has a bit extra to reach things in the gap). I always thought that if I was going to be machining metals I would do them towards the top of the z's range....
So fully extended out it's getting 0.14mm deflection at the tool when I put 30kg of sideways force on the tool holder... What surprised me is that deflection is roughly the same when I pull in both the x and y direction. I expected to see much more parallel to the long rails when it's trying to twist the gantry and bend the z spindle plate in the direction I assumed it would be weakest. Am I right in saying therefore a rather disappointing 2.1um/N
Here was my setup for measuring in the other direction - the woodworking clamp made it easy to regulate the force:


When near the top of z travel where I would have it (if I possibly could for metal), it's deflecting 0.04mm with 30kg force. Again both the x and y directions were the same. When I measure parallel to the gantry travel (what I call x), so not like the photo, I put the spindle in the middle of the 1.2m gantry, assuming that would give the most deflection... a bit better 7.5um/N if my calls are right?

One last thing to note is that the ballscrews on this are 20mm pitch and with the servos unpowered I can push it by hand, so I did the tests with servos on, and I could hear them working as I cranked the force on.
Ok hope that helps.

[luke peter]

Your test methodology is good as far as I can tell, and the numbers I managed to find put 3000N/mm (3N/um) at a very comfortable place for ali milling. I don't think my machine as built will be anywhere near as stiff as the sims, since it assumes bonded joints that are 100% efficient, looking at how your machine is cutting, I will be more than happy to be in the same ball park.

Your machine has a lot of really well executed stiffness fundamentals, its still really stiff for a router, and it has big levers. Its cool that your x and y stiffness have come out similar, it will translate into a more uniform surface finish (at least this makes sense to me - I have almost 0 actual CNC machining experience . Your bed frame is much more robust than mine, I am relying on the 20mm steel bed base plate to add a lot of stiffness for me.

Do you have a video planned testing your machine in some steel (DOC and speeds/feeds ramping until you reach a stiffness roadblock with chatter)? So many people write off DIY routers for metal work, and I have a feeling your machine, and hopefully mine, will stand out as counterpoints to that narrative.

[ardenum]

Hey thanks
The surface replication worked perfectly for me. I didn't really know what to expect going into it, but yeah - no complaints on that front.

w.r.t. stiffness: I don't think it's in the same league as yours, being a big router really. I hadn't actually measured this, despite testing it on steel etc. So I had a look today...
By far the most flex is coming from the spindle plate, as it has a lot of z travel (400mm from the bed to the gantry, and it has a bit extra to reach things in the gap). I always thought that if I was going to be machining metals I would do them towards the top of the z's range....
So fully extended out it's getting 0.14mm deflection at the tool when I put 30kg of sideways force on the tool holder... What surprised me is that deflection is roughly the same when I pull in both the x and y direction. I expected to see much more parallel to the long rails when it's trying to twist the gantry and bend the z spindle plate in the direction I assumed it would be weakest. Am I right in saying therefore a rather disappointing 2.1um/N

When near the top of z travel where I would have it (if I possibly could for metal), it's deflecting 0.04mm with 30kg force. Again both the x and y directions were the same. When I measure parallel to the gantry travel (what I call x), so not like the photo, I put the spindle in the middle of the 1.2m gantry, assuming that would give the most deflection... a bit better 7.5um/N if my calls are right?

One last thing to note is that the ballscrews on this are 20mm pitch and with the servos unpowered I can push it by hand, so I did the tests with servos on, and I could hear them working as I cranked the force on.
Ok hope that helps.

how's the rotating nut coming along? I remember you had some speed issues/ bend ballscrew?...

[Bongodrummer]

Well, thanks for the encouragement! I must say I felt a bit down about the reading earlier today... But hay, I do know it can cut steel, it might not be amazing at it, but it works. I did always plan to box in the spindle with another aluminum casting, wich should add significantly to the rigidity, but since it was working, that dropped to the bottom of the jobs list.

I actually think the bed of my machine could do with some kind of metal plate material (in a section at least) if I end up doing much metal with it. The 3/4" ply 3/4"mdf sandwich I have at the moment was just what I had to hand, but between the steel box the mdf can be deflected quite a bit. As well as the obvious stability issues - but that's not such a big thing as I'm just learning and cutting into it and doing different jobs on it day by day ATM - it's a spoil board after all.

Yep, I am planning such a video - to talk a bit more about the spindle (which is ace but a bit fast for steel)... Am just making some tool changer rack things with it ATM and then will be trying it cutting stone, and then some surface grinding... Basically all the abuse... Anyway, this is about your build, and I'm looking forward to more...

[Bongodrummer]

how's the rotating nut coming along? I remember you had some speed issues/ bend ballscrew?...

Working great now. Changed up the ballscrews for something much lass bent, better quality and 20mm pitch. Which of course if I was to do over would be more like 15 or 10!

[peteeng]

Hi Bongo - Thanks for doing the test. To do this test better you need to measure at least 3 load points. You start with a low preload say 2kgf, this takes out any mechanical play, then reset the dial guage. Then go 10,20 & 30 kgf . This will allow you to do a excel chart and regress the line. You may find the line stiffness to be quite a bit different to the single test figure as the single test includes the mechanical play. Then if you find play you need to get rid of it! The regressed line will not go thru "zero" and if you have a good correlation for the 3 load points then the "missed" zero value will be your play.... Peter

[luke peter]

Hi Bongo - Thanks for doing the test. To do this test better you need to measure at least 3 load points. You start with a low preload say 2kgf, this takes out any mechanical play, then reset the dial guage. Then go 10,20 & 30 kgf . This will allow you to do a excel chart and regress the line. You may find the line stiffness to be quite a bit different to the single test figure as the single test includes the mechanical play. Then if you find play you need to get rid of it! The regressed line will not go thru "zero" and if you have a good correlation for the 3 load points then the "missed" zero value will be your play.... Peter

Thanks for this Peter. It makes sense and will help differentiate looseness vs flex.

Well, thanks for the encouragement! I must say I felt a bit down about the reading earlier today... But hay, I do know it can cut steel, it might not be amazing at it, but it works. I did always plan to box in the spindle with another aluminum casting, wich should add significantly to the rigidity, but since it was working, that dropped to the bottom of the jobs list.

I actually think the bed of my machine could do with some kind of metal plate material (in a section at least) if I end up doing much metal with it. The 3/4" ply 3/4"mdf sandwich I have at the moment was just what I had to hand, but between the steel box the mdf can be deflected quite a bit. As well as the obvious stability issues - but that's not such a big thing as I'm just learning and cutting into it and doing different jobs on it day by day ATM - it's a spoil board after all.

Yep, I am planning such a video - to talk a bit more about the spindle (which is ace but a bit fast for steel)... Am just making some tool changer rack things with it ATM and then will be trying it cutting stone, and then some surface grinding... Basically all the abuse... Anyway, this is about your build, and I'm looking forward to more...

A indicator reading cant take away from what your machine has done, or will do. This kind of thing is iterative, 80% of the fun for me is in the build and subsequent improvements. Looking forward to the next video

I welcome comparisons and shared learning of other machines in this thread - it's the strong point of forum interaction, the shared experience of a bunch of people with a similar passion - and a history of dialog for everyone to see and process later.

[luke peter]

Its been a while since I have been CNC useful. I have had a bit of a sideline project going installing klipper and a small rpi on my 3d printer to decouple it from the workshop laptop - Thats mostly wrapped up now, and I have made a plan to have my parts machined by a friend next weekend - he has a chunky Doosan DNM6700 and he has very graciously offered to do the rail cuts and drilling and facing for me.

This has added an element of urgency to the last bit of fabrication for the parts that need to be milled... Thus we have the following progress:

Closure plates arrived last week - I made them each a single part to replace the bearing and motor mount plates


Painted what will be the insides leaving some space for HAZ. Not perfect, but will help a little with the mild internal corrosion potential.


Because I didn't intend to close the outsides, the spacing of the bulkheads was not perfect. Each closure plate is therefore slightly different w.r.t. slot position. Test fit:


The slots will be plug welded onto the bulkheads:


I got the stuck tap out by welding over it, then centre punching, and using a 4mm universal drill bit (basically a masonry bit sharpened to acceptable geometry for steel) to drill out the tap. there will be one through bolt for the high speed spindle mount.


In preparation for assembling the frame, I printed a hole alignment jig to drill the allen key clearance holes for but jointing screws. The frame will also have 90 deg brackets bolted on, but I thought the but joint screws will add a little joint effectiveness.



The kids are away at their grandparents this week, so assuming our defunct power utility manage to keep the lights on while I am at home, I will be sorted with the important welding shortly.

[luke peter]

I have run into a snag...

When Welding the closure plates on, the x rail support beam has developed a bow towards the closure plate (5mm off straight in the middle of its span). I suspect it is from the plug slots contracting. I tried welding lines a cross the other side in the same position as the plugs, but it hasn't made much difference.

Any suggestions? I have a 100ton press at work, I can try and bend it back... I can also make the bowed rail support work by cutting a straight slot into it and shimming the ball screw mounts... Sigh. Everything was going so nicely.

[luke peter]

Workshop 200t press did the trick - I was assisted by a very experienced boilermaker, took him 5 mins to get the bow out.

[peteeng]

Hi LP - Ahhh a 200T press will fix most things Peter

[Texasbowhunter]

It seems I have read this thread and noticed that seeming through the entire thread the were pics along the way...Wasn't able to view them until the last couple pics
What's the reason for this? My end?
Paul

[peteeng]

Hi Paul - seems the entire cnczone has an issue at the moment with attachments. Peter

[luke peter]

So I requested input on the facebook group to see what more experienced people would do to avoid the warp during welding - I got many suggestions, many of which were beyond my tooling and capabilities.

What did make sense, was to clamp the beam before putting all the heat in - In a perfect world, I would have a 20mm thick steel welding table, and I would tack the beam down before getting started. I don't, so I didnt...

The next best thing was to find something stiff, and clamp to that - stiffest assembly in my garage is the gantry:



It looked like it was going to work, up until the last 400mm of welding on each seam. It has warped less than the last one, but I think the clamps themselves were too compliant, and opened up fractionally. About 2mm total bow this time around, I will be taking it in for nudging like the last one.



The plug welds went better this time, and this is the last bit of fabrication I need to do before the machining next week.
After discussion with my friend, we decided to face the rail surfaces, and drill and ream holes for alignment dowels for the master rail in each pair, rather than cut a close tolerance slot. I believe it will be OK in terms of straightness, and much easier and quicker for the machine.

[peteeng]

Hi LR - Sorry but no amount of constraint can fix the issue. Steel shrinks 6% linearly when it transforms from liquidus to solid. That's a huge strain that has to be taken up somewhere. You can get wire that is more "plastic" and has a long liquidus time. This allows the weld to stretch as it cools helping the situation. Or you can use bronze wire so the parent does not melt. This is the main issue the parent melting and the combined parent/weld volume is large. If you braze your connections then no parent shrinkage and bronze is plastic. Brazing also has a local stress relief affect as the surrounding metal is brought to red heat at which temp the metal is quite weak. Oxy welding has a similar effect. But don't weld is the answer. I'm not sure if you can get bronze wire for a mig welder? Just had a quick look yes bronze mig is out there... It melts at 1000C so is less then the steel.... this would help lots. Peter

[joeavaerage]

Hi,

I'm not sure if you can get bronze wire for a mig welder? Just had a quick look yes bronze mig is out there... It melts at 1000C so is less then the steel.... this would help lots. Peter

Yes you can get bronze MIG wire, its worth a fortune. We used to sell a 15kg roll of 1mm wire for $685NZD+GST.....and that was several years ago. The only people who used it were panel beaters
where they were trying to avoid distortion in car panels, and a architectural fence and gate manufacturer whom used electro-galvanised steel and could not tolerate ordinary steel welds rusting.

I've used it and it works OK but is not easy to get a good job, mind you, while I've done a lot of welding I am no expert either.

It sounds like an appealing technique but I think its practicality is marginal. I'd rather 'bite the bullet' and pay for regular full solution heat treatment, do it once and do it right.

Craig