Milli a new composite mill kit

Hi Spumco - My original thought was a lifting, moving gantry. But it was less stiff and more complex then the other configurations. This is intended as a benchtop machine but its growing to a stand machine. I may shrink it a little after this next Round. We are now at version No8 so there has been plenty of variations looked at. I spent a lot of time developing a moving column design prior to going into this mill. Seems its a sweet spot. Its the Y travel that drives some of these decisions. I have made it a little narrower then first spec to remove the issues with double columns, gantries and saddles etc. Being my first mill I want to keep it simple KISS is important. And since I'll be making moulds I need to b certain its the best solution. if I made a steel or aluminium one and needed to change it that would be easy... But a production cast machine is bit of a risk that needs to be minimised.. Keep Making Peter

If you look at the Vulcan early videos you may pick up that they originally designed it as a double column or bridge machine ie fixed gantry and the Y passed under the bridge. Somewhere along the path they changed it to a long X model vs a long Y model and since they had committed to castings they had to rejig a few things. So I think their design falls short of their original intent. I prefer to do nearly exhaustive CAD/FE work to prevent such mid track changes and to understand the design envelope and opportunities/downfalls as best as possible.

Hi All - Been rebuilding Milli with the FE cars and rails. Results are good. Can now do some more optimisation and maybe more generative work. This model uses fiberglass infused laminate with E=30GPa. Can get to 40GPa but will start low. So now I've decided to aim at 20N/um this means the bonded FE model target is 40N/um which may not be possible with glass (or epoxy granite). Or I aim at 10N/um so need 20 in the model and I'm there. Will look at its envelope now and adjust then refine. Spumco may have some stiffness figures for us soon?

I also ran a couple of models with the machine sitting on a floor with friction and applying two opposing internal loads as it would be in reality. The tool deflection was very similar so I haven't bothered to add that complexity to the development model yet. The base deflected in a slightly different way so it will be good to do it again once the details firm up... Peter

by the way the current model weighs 670kg so its a heavy bench mill... so a diet maybe required I was thinking about 300-400kg Hmmmm

Measuring head deflection has been a little challenging; a fishing scale seems to be impossible to obtain this past week, go figure...

Kitchen scale didn't range high enough (5000 grams) to move it more than 0.00015 or so, so I had to rig up the bathroom scale with a jacking screw and a sliding wedge to load it up. The stupid digital scale does not have direct reading; you have to tap it to turn it on, then wait for 0.0, then apply a load. One it settles it gives a reading, so no way to creep up/down on a particular weight. Just shove the wedge in and then read the scale and indicator.

Hence my weight readings below are a bit scattered. And only in X - nothing to brace the jack screw against in Y just yet, but I figure X is usually the least stiff direction.

It looks like the force is being applied way off-center, but it's not. The feet on the scale are the actual transducers and I'm pressing directly against one of the feet.

Without futher ado:

(pounds-inches)
10.8 - 0.00025
30.2 - 0.0005
62.8 - 0.0011
67.0 - 0.0012
120.2 - 0.0024
123.0 - 0.0026
157.0 - 0.003
159.0 - 0.0031
230.2 - 0.0045 (got a good shove on that one)

Hope this is remotely useful. Note - head was about 75% at the top of the travel during testing.

-Ralph

https://www.cnczone.com/forums/attac...d=451534&stc=1https://www.cnczone.com/forums/attac...d=451536&stc=1

That's fantastic Ralph 10/10 - doing astrology chart now, what is the name of the machine?

see attachment - your test method is very good because your correlation is =1 There is a slight preload in the machine which is good as it passes above zero . The machine stiffness is 8.8N/um which is very good. this is the slope of the load curve. Does it cut steel easily? Peter

It's a Mikini 1610L.

Mikini is a now-defunct US importer. Long, LONG story about it, but it's a very early version of the SkyfireCNC SVM-2 mill. Mikini imported the machines, and the designer was (I believe) an ex-Syil gent who struck out on his own. When Mikini went belly-up, the designer started his own company and started selling his own design - he was already supplying the castings and all the hardware to Mikini.

Skyfire has upgraded quite a bit over the original Mikini offering, but the fundamental casting, rails, and so forth are essentially identical as far as I can tell. Mikini's were plagued with spindle motor drive issues; I think that's what did them in. There's nothing seriously wrong with the machine in general - ground ballscrews, very nice rails, decent castings. The open-back column is the only major flaw I've found so far. Plenty of poor QC, of course, and a few questionable design choices, but overall it's not bad.

The monsterous "how to build a CNC from beginning to end" thread from 2011 is, I believe, the beginnings of SkyfireCNC. Some of the photos even show green-painted mills in the background - presumably unsold to Mikini when they folded.

If you've got more time than sense, have a browse here:

EDIT - the SVM-2 photos start appearing on page 9 of the thread. But he'd been making those castings since at least 2008-2009 when Mikini started up.

https://www.cnczone.com/forums/vertical-mill-lathe-project-log/133172-cnc-software-forum.html

Quote:

---

Originally Posted by peteeng

Does it cut steel easily? Peter

---

Thanks for the encouragement. I'll try to get better/more numbers soon.

How does it cut steel?...hmm. Good question.

I can only use 3/8" end mills in steel for side-milling. 1/2" will chatter unless I back off the depth/width of cut to the point where a 3/8" tool would be more efficient. Facing steel with an insert mill or flycutter is hopeless - I use a 5/8" bull-nose with lots (7) of flutes when facing steel, and even then it's rather shallow DOC.

My 'go-to' end mill for steel is a carbide 3/8", 5-flute, stub length, roughing tooth. That one can plow through steel with no chatter and it holds dimensions well. If I use the same cutter but with normal flute length (7/8") it will chatter at full DOC.

So the max it can do in steel (smooth cutting without chatter) is about 5/8" DOC, 15% WOC, 0.0025" IPT at around 450SFM. Call it 2 cubic inches MRR.

In 6061 aluminum, I can get 5.5 MRR with a 1/2" 4-flute roughing end mill no problem.

I'm convinced that there are two issues limiting this mill:

1. Column flex (or column to base, or both)
2. R8 spindle with TTS tool holders

I intend to mitigate both by installing a BT30 spindle and by filling the column (and perhaps the base) with epoxy granite (of some flavor) and adding a steel back plate to the column.

-Ralph

Hi Ralph - What is the power of the spindle and its speed range? Gearbox or pulleys? Thanks for your efforts. Peter

Again - good question. I've replaced the BLDC spindle motor due to the bad drive, but will be reinstalling it during the overhaul as I've discovered that when driven by a proper servo drive it's a rather good motor. Tons of torque and very light weight.

Current spindle motor
2hp induction w/ closed-loop vector drive VFD. (Marathon BlackMax, Hitachi WJ200, 200-line (800 count) encoder). Motor is an 1800rpm 4-pole, and it can go to 5400RPM with the VFD.

Transmission: single ratio 1:1.56 using 20mm HTD5M belt. I could theoretically get >8kRPM at the spindle, but I keep it at 7k or less to save the spindle bearings for now.

1/2" drill in steel is no problem. Rigid tapping 1/2-13 in steel is possible with a very sharp tap and not too deep, but that's about the torque limit at low speed.

The only downsides to the spindle motor are that it's way too heavy for the mill (>70lbs), and it's also way too large (diameter). I can't fit a normal PDB cylinder directly over the spindle.

With a 2-speed belt drive 1.5hp would be fine for this machine. The ideal setup would be an automatic 2-speed belt drive like on the Fadals. Smaller, lighter motor, plus plenty of low speed torque, and a 10k+ upper RPM limit for small cutters.

-Ralph

Hi Ralph - Thanks for the info. My research at the moment points to a 750W (1hp) servo or a 1kw (1.25HP) servo. The other solutions are very heavy and maybe the DMM servo is not much more $$$. Final figures aren't in on that. If we start with the 130mm frame it allows 1kW, 1.3kW and 1.8kW choices. Sounds like that maybe the go even if it feels a bit big to me. If I use the 750W its a 86mm frame so upgrading to 1kW maybe a mission due to lack of real estate... What's the slowest speed the Mikini is comfortable with? I'll think about that... Peter

Also my model is at 20N/um so maybe 10N/um in reality which is not much stiffer then the Mikini so maybe aim at 20N/um real so 40N/um model... Got to aim high...

I was looking at BLDC's as well what was yours? The DMM is 1500-3000rpm so intended a 1:1.5 or 1:2. The 750W weighs 3.1kg while the 1.3kW weighs 6.7kg.... thanks again Peter

Hi All & Sundry - Been looking at the model and the column is the critical bit to get to 20N/um. Its currently 300mm square which is BIG for this size machine. So I looked at how I could improve this. The obvious is to use CF but that's a lot of CF and lots of $$$. So I looked at an aluminium enhancement. A 250x250x12t aluminium tube filled with Tetrium-S is the same rigidity as a solid 300mm square FG infused column (The al by itself is about half the rigidity). I calculated this first then modelled it to check torsion. Torsion is difficult to hand calc. So the simulation shows they are equivalent. I have also decided to go with the 1kW (130mm flange) as the base option for the spindle which gives other power options.... The machine will be bigger then I imagined but I think it will be a great mill. So now I will optimise around the 300mm column as this is easier to do then the 250 al/Tet hybrid.

torsion -1000Nmm applied to a 2m long column, both twist 0.03mm
bending -1000N applied to the top edge. Both 1.33mm RHS is hybrid column
calc - shows calcs for sections

I looked back thru some recent laminate tests and found the compression E=35GPa and the tension E is around 32MPa. I will change the spec a little to get up to 35GPa all round maybe 40GPa this will help. Keep making Peter

Hi All - Started detailing a few parts. bolt holes subplates etc. I want to get the table and bearings back into the model and the drives.... I lightened the column to 95% the solid rigidity. So the 300x300 has a 162mm hole down the middle. cheers Peter

Peter,

1. DMM is rumored to have 1kw 86mm frame servo available - but it's not on their web site. I don't have any figures on torque, but if you contact them I'm sure they will provide what you need. That would permit a 5kRPM upper limit rather than 3k on the 130mm servo (assuming a 220/240VAC DYN4 drive)

2. The slowest speed my mill is comfortable with is about 200RPM. That's about 150RPM motor speed - but remember I've got an induction motor with encoder feedback and a good VFD. I can use a 1/2" drill in steel at 500RPM (spindle) without stalling, but I wouldn't want to do a million holes like that. The VFD and motor are quite capable of full torque at slower speeds, but my rather coarse encoder count limits the accuracy at really low RPM. (The Hitachi can't use anything over 2kHz, and I'd saturate it at 5kRPM with a higher encoder count). My 200-line encoder is fine for synchronized tapping at 'normal' tapping speeds.

3. The BLDC that came with the mill is a 130mm frame 2.2kw Chinese thing. It's essentially a 3-stack BLDC, and is virtually identical to many 'big' BLDC's available through the usual Chinese sources. Automation Technologies imports a version to the US, and variations are found on quite a few smallish Alibaba-type mills.

https://www.automationtechnologiesin...tor-and-driver

The driver supplied with the AT (and I assume most others) is not the same proprietary drive that came with mine - and which was utter rubbish. I have no idea how these perform nowdays but I suspect they're better than they were.

Mine, on the other hand has been converted to a true servo. I replaced the Halls with a commutation encoder, and am driving it with a Copley Controls Xenus servo drive. The difference is night and day. I've not reinstalled it on my mill yet, but bench testing indicates it's going to be a monster - at less than half the weight of an induction motor. And with double the top speed (6kRPM) of most 130mm frame commercially-available, low-cost servos (DMM, Delta, Leadshine, etc).

The nice thing is that if it isn't as awesome as I predict other 130mm frame servos are a bolt-in replacement. And the Copley drive can just (barely) handle a 7kRPM, 3.5kw(!) Baldor BSM90-375AA.

This is what I was hinting at earlier in the thread - there are better solutions available, but not really suitable for a production mill. The Copley drive is >$2k new, and having to buy a BLDC, encoder, and bodge up some cables is not what you need to be doing when trying to sell a product.

4. Suggestion... Get the specs for the DMM 1kw Nema34 servo and set up a spreadsheet. Compare the 1kw to the 1.8kw servo. Pick a target top speed - perhaps 6kRPM - and plot the torque available over the RPM range given the reduction ratios required to achieve the target top end.

Then, if the 130mm (at 1:2) is favorable compared ot the 1kw (at 1:1.2), is it still favorable given the increased weight and costs?

Or do you want a 8kRPM top end to match most of the BT30 spindles available from China? If so the torque/speed/drive ratios will probably change your results.

5. Finally - another option is to just buy a servo spindle motor & drive combo suitable for the top speed you choose. These are starting to become more prevalent via internet commerce, and are really the right thing for a bolt-in solution. I don't have torque curves available, but a 2.2kw 0-8kRPM servo (with motor cooling fan) and drive is about $1.5kUS on ebay now. If you're importing from China they may be less, especially in quantity for a production machine. This is essentially the same setup as a SkyfireCNC, Syil, and virtually all other 'off-brand' small VMC's are using. Fan-cooled high-speed servo driving a BT30 spindle cartridge.

So... $600(ish) for a DMM 1k motor & drive, $700(ish) for the 1.8k & drive, or $1000-$1500 for a much higher spec (but maybe spotty QC) servo spindle motor & drive?

-Ralph

Thanks for the summary Ralph - Milli has grown somewhat and I'd prefer to cut her back a bit. If the 1kW N34 or 86mm frame is available then I'll go there. The 130mm frame results in a bigger machine then I think I want for the market. So 1) I'll send a note to Stephen at DMM and figure it out 2) I'll continue detailing out some of the bits that are not affected by this issue. 3) I may bite the bullet and just go with the 750W, we shall see. But when you look at the DMM range it seems there is a gap in the N34 1kW space so it would seem prudent for DMM to fill that and a 500W is lacking as well.... Peter

Is there some reason why you favour BLDC spindle motor over an AC servo motor ? Google suggests BLDC motors suffer from torque ripple.

Hi Zorbit - If the question is directed to me; I have no preference for any motor type. I have no direct experience in this area so I'm looking at all solutions to pick the best one. I want a small light motor 500W-1kW that has good torque and speed suitable for a small mill. It may have to use a pulley drive to get the required speed range. I want a basic version (cheap no rigid tapping) and a rigid tapping version. What I have looked at so far is :
1) steppers 2) induction with VSD 3) DC motors with speed control 4) servos 5) axial flux motors 6) briefly BLDC is there anything else that is suitable? Seems the issue is a) cost b) getting a speed range for the material. Some will want 100-8000rpm some will want 100-5000rpm. Still investigating... Regards Peter

The goalposts are still moving around, perhaps a poll of hobbyists would be a good idea - find out what they really want and what is missing from their current workshops.

I'd want a higher speed spindle, the Bridgeport VMC I used for years has a 15k spindle, I regularly used 10k and that wasn't using really small tools that a hobbyist might want.

Hi Zorbit - Speed depends on the "hobby" so a pulley system is needed for speed ranging. Or a high speed spindle option like a router plus the power head. Its quite a puzzle... Peter

update, email just in: DMM do make a 1kW 86mm frame servo so I think that's on the option list. See attached spec. They use the same driver so that's easy. Now to figure a non rigid taping version or decide "thats it for you lot" so it would be:

750W, 1000W servo then a 1.5kW high speed spindle. Would be nice if there was a 1.5kW low speed spindle.... I don't know if there is name for it but making a design commensurate or balanced $$$ vs performance is always a juggle and every machine seems to have a crux component. Milli's crux seems to be the motor...

Today is D day!! The two final doors get fitted to the workshop. They delivered the roller doors with no tracks!! so been waiting on tracks. Tomorrow the 30m trench gets dug for the power supply and Friday the workshop has lights, GPOs and juice. WooHoo!!

Morning all - Stephen at DMM confirmed that a 1kW 86mm servo is available so I put the 86mm motor back on Milli. She looks a bit big now!! So I'm going to cut the column back from its current 300mmSQ to 300 deep by 250mm wide and balance out some of the bits. I'll add a subplate to the front of the arm so spindles & power heads can be worked on without pulling the arm off and then we can deal with some more of the details. I'll also look at servos for motion control. So Milli will have stepper and servo options.... Thanks for participating people, you guys (maybe gals? out there) have filled in gaps that would be impossible otherwise... cheers Peter

Morning All again - I updated Milli's column and arm. She's become cranky and won't move anymore so some of the constraints have become rusty. Instead of figuring it out or using WD-40 I think it's time to archive and rebuild with the new ideas and info. Onto Milli No9... will try to get rid of the sticky out motors this time as well... Peter

If you go with the DMM servo, rigid tapping should be pretty easy to implement if the controller supports it. The DYN4 has encoder outputs that can be scaled so as not to overwhelm the controller (if it has low-speed optocoupler inputs). And for a non 1:1 belt ratio a simple proxy sensor, optical or magnetic switch on the spindle pully can be the index pulse.

As for hobbyists and what they want... we want it all.

Some people want high speed (10k+ RPM) to use little tools on softer materials at reasonable feed rates - i.e. they want a Datron.
Others want to be able to put a 1/2" end mill in and hog steel.

This is why vendors offer different spindle configurations. Without a seriously expensive setup, or a multi-speed spindle drive, you can't please everyone.

My personal wishlist for a hobby-grade mill that won't drive me crazy would be a 10k top end, and the ability to drill 1/2" holes or use a 2" index mill in steel without chatter. Face mill doesn't have to be ripping through the material, but a moderate DOC to face off stock would be nice.

Before you make a decision on the spindle motor, have you thought about the control system? Both hardware and software?

-Ralph