Nomad with deep DOC, shallow WOC

[Moded1952 (Lucas)]

@LiamN

If you go servo

Don’t tempt me… The main thing keeping me away from them is the size. Even the smaller ones are nearly 100mm long, which I don’t think I can accommodate.

@LiamN

Like spargeltarzan says, a consistent force is quite important here.

Generally I agree but in this case I’m not really looking for an accurate quantitative measurement (how wobbly the parts are) but more a qualitative one (are the parts wobbly).

I’d try the luggage scale idea again (I did it earlier) but it’s quite difficult to measure most of these parts using it in my environment.

@LiamN

Also, being very consistent and clear about the deflection between which two parts is key. Measuring between the Y travelling build plate and the spindle tells you about the achieved cutter to workpiece deflection but not necessarily where the components that make it up come from.

I’d love to mount the DTI to various other surfaces but I’m not sure how to mount it to anything other than the bed (where I can use a vise) and a rod sticking out of the spindle (which has an unfortunate habit of rotating, making it not very useful).

Given that constraint, I’ve been trying to isolate different parts of the machine by applying force to different parts of the machine. For example if I have the DTI mounted to the bed with the tip on the spindle and push on the spindle tip, that will include the deflection from the spindle, the Z-axis bearings, the X-axis bearings, the steel rod, the vertical supports and the bed itself. However if I push on the steel rod, that should eliminate the effects from everything preceding it.

I know it’s not perfect, especially for a good quantitative measurement but I think it’s enough for me right now. Do you disagree?

[LiamN (Liam Newcombe)]

@Moded1952

@LiamN

If you go servo

Don’t tempt me… The main thing keeping me away from them is the size. Even the smaller ones are nearly 100mm long, which I don’t think I can accommodate.

I’m not sure they’ll be any better on this machine even if they do fit.

@Moded1952

Given that constraint, I’ve been trying to isolate different parts of the machine by applying force to different parts of the machine. For example if I have the DTI mounted to the bed with the tip on the spindle and push on the spindle tip, that will include the deflection from the spindle, the Z-axis bearings, the X-axis bearings, the steel rod, the vertical supports and the bed itself. However if I push on the steel rod, that should eliminate the effects from everything preceding it.

I know it’s not perfect, especially for a good quantitative measurement but I think it’s enough for me right now. Do you disagree?

Ah,

I found places to clamp the base of my dial indicator to such as the X beam or Y plate and then measure deflection between parts such as X beam to Z carriage, Y plates to baseboard etc.

If you remove some of the covers from the Nomad you should be able to mount to those big base rails, the Y plates etc. to start isolating what flexes where.

All those measurements were done with the force applied in a consistent place, the cutter in the spindle, as that was the most representative of cutting forces applied to the machine.

[spargeltarzan (caveat calculator)]

@Moded1952

It’s definitely not a repeatable amount of force but the thing I’ve noticed is that the deflection in the Nomad isn’t anywhere close to linear. If you were to plot force vs. deflection you’d get a curve with a very very steep section where very little force results in a lot of deflection but then you’d hit a plateau where it’s hard to move the machine at all.

Aha! That’s important.

@LiamN

The backlash and slack are the slop in the leadscrews, bearings for the ends of the leadscrews, any slack in the linear bearings etc. These will all move initially for not much force and are probably where a lot of that rattling vibration comes from when the machine chatters in a cut.

This sounds very plausible to me. Because the actuators look like integrated stepper+leadscrews, there is little chance getting at the axial bearing. So then, the key modifications (low hanging fruit…) would not so much concern increasing the stiffness of the contraption but rather to eliminate backlash - decent ballscrews can do that. I wonder whether the leadscrews are the only source, though.

(Now, how to account for backlash in millalyzer… )

[Moded1952 (Lucas)]

@LiamN

I’m not sure they’ll be any better on this machine even if they do fit.

In terms of power it’s overkill for sure but in terms of accuracy, servos + ballscrews should be able to get me close to µm precision, shouldn’t they?

And it looks like I actually could fit some of the smaller servos in my enclosure…

@LiamN

I found places to clamp the base of my dial indicator

Ah, that’s a good idea. I’ll need to take the machine out of its enclosure though.

@spargeltarzan

Because the actuators look like integrated stepper+leadscrews

They are. They’re made by JUGETEK and look similar to these, though the company told me that Carbide 3D’s are a custom model.

@spargeltarzan

there is little chance getting at the axial bearing

I think I remember seeing @Vince.Fab tearing apart a stepper… Vince, was it a Nomad stepper?

[Vince.Fab (Chippy McChipload)]

Stepper was Nomad. Pretty easy to disassemble the thing.My ballscrews are actually solid coupled to leadscrew nubs.

AB nuts are only good until a certain limit, then they just bounce around.

Nema17s will provide enough force for this size machine. I understand going all the way…but just like feeds and speeds, there’s a certain sweetspot where this machine will accel.

[gmack (Gerald Mackelburg)]

My forum addiction started with attempts to get answers to the types of questions you guys are now asking and addressing. I found it surprising that very little information was available on machine rigidity and cutting force ability. Stepcraft seemed unique in that they formerly provided “Torsional stiffness” specs. ShopBot also apparently uncharacteristically, provided cutting force specs for one of their products (to help sell its closed loop stepper option?)

IMO you guys are doing “Gods work” with your endeavors - thank you!

[Moded1952 (Lucas)]

That’s an odd way to specify it, isn’t it? A quick Google says that what stiffness is to force, so torsion is to torque (edit: s/torsion/torsional stiffness/).

So why are they listing a force in Newtons and a displacement in mm?

[gmack (Gerald Mackelburg)]

IDK, maybe a language translation issue?

[Moded1952 (Lucas)]

Hmm, it wouldn’t surprise me but I don’t think so. The German phrase they use is “Verwindungssteifigkeit”, where “steifigkeit” = “stiffness” and “Verwindung” = “torsion”.

Also, I wonder why they include that number for the hobbyist D-series but not the higher M and Q series? (Note: you said “formerly” but for the D-series, the numbers are still advertised on the main product page)

I wish I knew what I know now when I bought the Nomad though. For the $3k of the Nomad, I could have had a larger machine, with a real spindle, linear rails, ballscrews, more powerful motors, more detailed specifications and a better ecosystem. Or, if I’m reading their specs right, I could just have had a machine about as wobbly as the Nomad for about half the price.

Anyhow, I have the Nomad now so that’s what I’ll stick with.

In other news, I emailed Mechatron to ask about a spindle. I was thinking to start on the Z-axis but then realized that I really need to know which spindle I’ll be using so I know what the spindle mount will look like. Now I’m pretty sure I’ll be going with the HFP-6508-30. It’s more ~1k€ instead of ~0.5k€ like the cheaper spindle but the extra low runout and extra 6k RPM seems worth it, especially if I’ll be using my now huge collection of 3mm endmills.

So that leaves me with a 65mm spindle for which I’ll be using a spindle holder with a 120mm x 80mm mounting flange. That doesn’t leave a lot of space for bearing blocks so I’m wondering whether I should increase the height of the mounting plate (e.g. make it 120mm x 120mm) or just use two longer bearing blocks (TBI Motion TRS15VN) . Using single blocks does subject them to moment but they’re rated for 15000 kg-mm on all axes. Assuming they’re ~50mm from the spindle and my maths isn’t crazy, that’s fine though, as it leaves me with 300kg of force. So I’m inclined to go for a simple 120x90mm adapter plate for the spindle, that just bolts the spindle holder to the bearing blocks.

One question I’m yet to answer there: should I just tap holes in the adapter plate or should I use nuts or inserts? I’ve heard that Aluminium threads aren’t the most reliable things in the world.

I also looked at ballscrews and couplings.

If I go for 12mm ballscrews, I get 26kgf/µm of rigidity and a load rating of 900kgf, which seems plenty and I like that they’re compact, so I think I’ll go with the SFU01204-4 nuts and screws.

Speaking of ballscrews, @Vince.Fab, what size did you end up going with? Did you have any issues fitting the ballscrew nuts onto the X and Y axis carriages? I haven’t been able to get a good look at the hole that the AB nuts fit into yet. Will a 24mm diameter nut fit in there?

And finally, I looked at steppers. I’m thinking about whether I need to upgrade to NEMA23 for Z since it has to lift a 2.4kg spindle now and keep it still while it’s under the axial load from cutting.

Anywho, I think that’s enough of my day’s thoughts written down for now. Hopefully tomorrow Mechatron will reply and I’ll finally buy the spindle and move forward with the Z-axis design.

[gmack (Gerald Mackelburg)]

@Moded1952

Now I’m pretty sure I’ll be going with the HFP-6508-30. It’s more ~1k€

Wow! That’s a lot less than the quote that I got for the 60kRPM version last December. "The price for our spindle HFP-6508-60-ER11 is 2358€ " (albeit for 1/2 the cutting force at the same MRR).

@Moded1952

I just tap holes in the adapter plate or should I use nuts or inserts? I’ve heard that Aluminium threads aren’t the most reliable things in the world

I believe that a steel adapter plate would be much stiffer (and harder to tap).

@Moded1952

I’m thinking about whether I need to upgrade to NEMA23 for Z since it has to lift a 2.4kg spindle now and keep it still while it’s under the axial load from cutting.

I’m not a ME, but I’ve heard from MEs that leadscrews can’t be back-driven. Wouldn’t that apply to ball screws too?

[Moded1952 (Lucas)]

@gmack

Wow! That’s a lot less than the quote that I got for the 60kRPM version last December. "The price for our spindle HFP-6508-60-ER11 is 2358€ " (albeit for 1/2 the cutting force at the same MRR).

It could be that the 60kRPM model is just more expensive, or it could be that the price was for a full quote, including all the accessories.

@gmack

I believe that a steel adapter plate would be much stiffer (and harder to tap).

Unfortunately, it’d cost me a few hundred euro to have it machined. For now, I’d like to make as much as I can on the Nomad itself, which means Aluminium. I’ll keep nicer materials in mind for later though. If I put rails on the other axes, I’m going to have to have plates machined elsewhere anyway.

@gmack

I’m not a ME, but I’ve heard from MEs that leadscrews can’t be back-driven. Wouldn’t that apply to ball screws too?

“Can’t” is a strong word. I back-drive my lead screws all the time (moving the axes by hand), you just need enough force.

[Moded1952 (Lucas)]

I’ve been working on the CAD and roughed out most of the important bits of the Z-axis at this point.

There’s a lot I could do to improve on this but I’m aiming to make the upgrade work with no mechanical modifications to the Nomad (it’ll just bolt on to the X-axis carriage), with parts that are machineable on the Nomad and without changing the travel of the Nomad.

But I think I’ve finally at least picked all the parts. The remaining pieces now are mainly basic stuff like adding some more screw holes and countersinks.

[LiamN (Liam Newcombe)]

So…

One of the interesting things I’ve seen looking at designs for bigger CNC homebrew machines evolve is that frequently the advice to get travel + stiffness on the Z axis is to flip around the rail blocks and rails.

If you put the rails on the moving Z axis with the spindle and the rail blocks on the Z part fixed to the X axis then you can space the blocks further apart for better stiffness whilst not losing clearance, your spindle is way bigger than the XZ fixed part there, might be worth flipping those round in CAD and seeing how it looks?

[spargeltarzan (caveat calculator)]

@Moded1952

One question I’m yet to answer there: should I just tap holes in the adapter plate or should I use nuts or inserts? I’ve heard that Aluminium threads aren’t the most reliable things in the world.

My experience: Threads in 7075 and 2017 work fine, unless you want to unscrew them every day. In 6082, OK-ish, 5083 not so much. So - I’d recommend helicoils or other inserts in 5083, or if your want to frequently loosen the screws. Pre-milled plates of 7075 are not much more expensive and it actually cuts pretty nicely in the sense that for some reason it doesn’t appear to stick to the endmills as easily.

[Moded1952 (Lucas)]

@LiamN

your spindle is way bigger than the XZ fixed part there

I really like your idea but I think the spindle placeholder is misleading. The actual spindle is 217.4mm end to end, so I modelled the placeholder as a 217.4mm long, 65mm wide cylinder. However the 217.4mm length includes the stickout of the ER-11 collet holder and its nut, which is ~30mm. If you take that away, the length of the spindle is ~187.4mm, which isn’t much better than the 185mm height of the X-axis carriage.

I could get that extra 30mm if I let the rail supports stick out past the spindle body but I’m afraid that might give me issues with clearing workholding fixtures at some point.

I have been thinking about whether it’s possible to extend those rails a bit more and get bigger carriages or more separation but I think to do that I really need to increase the height of the rail support plate (which can be mounted to the X-axis carriage with simple 90° brackets). I have plenty ~120mm of vertical space left.

The main reason I haven’t done that is that I’m worried about balance with the X-axis rods. Right now, even though I’m adding a bunch of weight and force to the carriage, I think the center of mass for the assembly will remain between the two rods. If I extend the rail support plate further upwards, there’s going to be a point where I create a big lever using one of the rods as a pivot point to exert force on the other.

@spargeltarzan

My experience: Threads in 7075 and 2017 work fine, unless you want to unscrew them every day. In 6082, OK-ish, 5083 not so much. So - I’d recommend helicoils or other inserts in 5083, or if your want to frequently loosen the screws.

Ah, that’s very good to know. I can only get 5083 pre-milled from my local supplier so that’s what I probably would have gone with if you hadn’t told me otherwise.

@spargeltarzan

Pre-milled plates of 7075 are not much more expensive and it actually cuts pretty nicely in the sense that for some reason it doesn’t appear to stick to the endmills as easily.

And once again you point me to a European supplier I was previously unaware of (previously Spanflug and Weerg). Shipping cost is a tad high but I can live with that.

[Moded1952 (Lucas)]

@Moded1952

I really like your idea but I think the spindle placeholder is misleading.

To clear this up, here’s what that picture looks like with the a real spindle model mounted for 80mm of clearance:

[Moded1952 (Lucas)]

Minor update: I made the various vertical parts parameterized based on height, so I can change the height of the spindle adapter plate with one parameter and the height of the rails, ballscrew and support plate with another.

I came up with this:

I’m going to try and fit longer carriages in there and though there’s not enough space for a 16mm ballscrew, I think there is space for a 20mm rail. I realized I can get 10mm more space on X without reducing working area due to the linear bushings protruding from the carriage.

And I think even if I can’t add in bigger carriages, just moving them further apart should have a huge effect, since I think it gives them mechanical advantage?

The question I’m trying to deal with now is tramming. I’m hoping this will be accurate enough that I can use what little slop remains to handle it, e.g. by putting some shims under the spindle holder.

People who’ve actually trammed a machine before, how does that sound?

Also, Mechatron is trying to sell me their gigantic KG5000 chiller unit for cooling the spindle. Can anyone with a watercooled spindle chime in on the practicality of that? Isn’t a repurposed PC watercooling loop plenty?

[LiamN (Liam Newcombe)]

@Moded1952

I’m going to try and fit longer carriages in there and though there’s not enough space for a 16mm ballscrew, I think there is space for a 20mm rail. I realized I can get 10mm more space on X without reducing working area due to the linear bushings protruding from the carriage.

And I think even if I can’t add in bigger carriages, just moving them further apart should have a huge effect, since I think it gives them mechanical advantage?

The HGR15 sized rails are rated for some pretty heavy loads already.

Spacing the rails and bearing blocks further apart reduces the leverage that the spindle loads have over them when loaded in X and Y. At what spacing it becomes the mounting materials which are deflecting more than the rails becomes the key question there. HiWin do publish deflection loadings for their rail blocks.

@Moded1952

The question I’m trying to deal with now is tramming. I’m hoping this will be accurate enough that I can use what little slop remains to handle it, e.g. by putting some shims under the spindle holder.

Ideally you’d want flat face contact for the spindle holder to the Z carriage, or any tramming adjustment. You can rotate this, ideally with eccentric or set-screws pushing the spindle mount to rotate into X alignment and then lock it in place with the main bolts.

The Y alignment is harder, either you rotate the X rail mountings at the Y plates or have to do things with spacers & shims. Like on the Shapeoko there doesn’t seem to be any obvious bolted joint that allows for Y alignment of the spindle bracket.

@Moded1952

Also, Mechatron is trying to sell me their gigantic KG5000 chiller unit for cooling the spindle. Can anyone with a watercooled spindle chime in on the practicality of that? Isn’t a repurposed PC watercooling loop plenty?

You may have trouble using all the Watts of the spindle even in a super-Nomad. Unless there’s a big saving buying it all in one go I might grab a CW3000 or clone first.

I believe Mechatron are decent enough to specify the required flow rate and supply water temp for their spindles.

[spargeltarzan (caveat calculator)]

@Moded1952

I’m going to try and fit longer carriages in there and though there’s not enough space for a 16mm ballscrew, I think there is space for a 20mm rail. I realized I can get 10mm more space on X without reducing working area due to the linear bushings protruding from the carriage.

And I think even if I can’t add in bigger carriages, just moving them further apart should have a huge effect, since I think it gives them mechanical advantage?

HIWIN stiffness specification:

So,

1. HG15 with preload class ZA is considerably stiffer than HG20 with Z0.
2. those stiffness values are huge, compared to everything else on that machine…

It’s probably not worth spending extra effort to try and make bigger rails fit. It may be worthwhile to consider buying well-defined quality however. I’m not sure all of the aliexpress vendors know which preload class they are selling. (Tuli, that you linked to earlier, looks pretty reliable).

[Moded1952 (Lucas)]

@LiamN

Ideally you’d want flat face contact for the spindle holder to the Z carriage, or any tramming adjustment. You can rotate this, ideally with eccentric or set-screws pushing the spindle mount to rotate into X alignment and then lock it in place with the main bolts.

Thanks, that sounds like a good idea, much simpler than the things I was looking at.

@LiamN

The Y alignment is harder, either you rotate the X rail mountings at the Y plates or have to do things with spacers & shims.

I think I can deal with Y alignment where the plate attaches to the X-axis carriage. I think I’ll have to use right-angle brackets, so I can just leave a little bit of wiggle room in the holes for them.

@LiamN

Unless there’s a big saving buying it all in one go I might grab a CW3000 or clone first.

How about a PC water cooling loop? Just a couple of radiators, fans and a pump? I’m looking for quiet and I’m afraid these larger chillers will be too loud. Have you used one? Do you maybe know how loud they are?

@spargeltarzan

It may be worthwhile to consider buying well-defined quality however.

Yeah, that I’m definitely doing. I’m not buying random stuff from AliExpress for sure.

The problem for me at the moment si that Tuli charges a bunch for machining the ends of the ballscrews. Dold Mechatronic sells ballscrews pre-machined for a much better price but they have literally no specifications for their ballscrews…