Motors for small production machine.

[Jim Dawson]

The Galil is a motion controller, and could operate as a stand alone in some applications, but not in a CNC machine tool. It has no idea what G code is, the only thing it understands is encoder pulses or steps. It is furnished from the Galil factory as a blank slate, ready for the user to program the unit for their application. The good news is that you can easily program it with any text editor, and they can be made to operate just about any system you can imagine, very flexible. Used in everything from simple table top machines, to military targeting systems.

In the case of a CNC machine tool, a UI is required to operate the machine. The UI consists of the normal file handling stuff, a G code translator, operator interface display, and other stuff as needed. The G code translator converts the G code into move commands and into a language that the Galil can understand. It's a little more complicated than this, but those are the highlights.

[Al_The_Man]

When I started using Galil, it was the old DMC1000 cards.using an ISA slot.
At that time Galil came out with a neat little interface program called Opint, you could build your own screens in minutes, and embed G code commands
It ceased at the DMC1000.
Unfortunately they never did come up with their own after that and left it to after market users such as Camsoft etc.
Max.

[joeavaerage]

Hi,
I don't know a huge amount about Galil, they've always beyond my means.

Galil does have an Mach3 plugin. That would allow you to run Mach3 on your regular PC and that would in turn generate motion codes
which would be consumed and enacted by the Galil controller. As such the Galil recedes form the operators view, despite it being so critical to motion.

I'm not sure whom wrote the Mach3 plugin, it may have been Galil, or maybe someone else.

What certainly is true is that Galil have never written a Mach4 plugin. Steve M^cMurphry, one of the lead developers of Mach4 and a longtime
Galil afficiado has written a Mach4 plugin.

My understanding and reading about Galil (and Jim Dawson confirms) is that there is no particular requirement that you run a plugin, the controller will digest Galils proprietary
motion codes from a C program, VBasic, text file or whatever, and generate the pulses or analog voltages necessary for the machine. What the plugin
does however is present a familiar CNC software solution to the user that consumes regular Gcode without the user being required to understand
the inner workings of the controller.

The Dynamotion Kflop is rather similar, you can supply it with strings of Dynamotions motion codes from a text file or whatever and those codes
will be translated in the required pulse streams for the machine. There is a Mach3 plugin for the Kflop as well, thus you can run a familiar CNC software
solution and the plugin translates to the Kflops motion codes. To my knowledge Dynamotion has not written a Mach4 plugin.

Craig

[hanermo]

Lemme put my oar in.
I actually have a 12x24 semi industrial lathe, with high end cnc industrial kit on it.
450 kg mass, 32 mm screw/z, 25 mm x, scratch built saddle on linear rails, c axis of 2.5 kW ac brushless servo, 750W ac servos on x and z.

2 toolchanger, 4-revolver for front and 7 rotary (servo in back).
Toolchangers are not quite in service yet - 3 -phase and mass issue (the servo turret is 70 kg, need the linear guide support at back).
300 mm gang tool capacity on x.

I´ve spent several thousand hours at this, and a lot of money.
Over about 17 years.
I was recruited by Haas, and was their country sales manager for Spain, 2011-2012, growing sales 11-fold.
And lots of training, even from the guys who actually designed their lathes.

ANY of the modern servos will work fine, and most support s/d just like steppers, as well as analog input.
Imo. Ime.
You can plug in the servo just like a stepper, and then gradually increase the p/d values until they are close to what you want.
No, it might not be "optimal" but it will track extremely close.
You can see the error real-time on the led on the drive.
The error is gonna be much less than any stepper system, worst case, and with minimal tuning (trivial) will perform extremely well.

I use mine with very loose "soft" tuning.
So i get some following error under heavy cuts.
I could not care less.
Likewise, I use about 1/10 of the max speed for rapids.
I could not care less.

And in case of overloads like drilling into tool steel with 32 mm drill, the servo just faults (under 2 ms) and nothing breaks.
My controller is 4 Mhz with 10 kHz servo resolution.
Everything is built industrial level - the z axis motor mount is 40 kg of steel.

I recommend You buy motors a bit better than You think You need.
But everything else in Your motion-control train is probably going to need 1-2 levels of upgrade, for an actual production system.

Saying as someone who did this upgrade circus, and then got trained on this, and then sold lots of these, and visited 200+ shops in 5 countries.
I know perfectly well how painful the money is.
I would never have afforded this if I knew at start how expensive it would be.

Machine tools are typically max-loaded at 2% of their breaking strength.
They are stressed-skin structures with thin skins carrying the load, and most of the volume is air.
My VMC could support 100-200 metric tons, on the columns. Imo.
Practical working rating is == 1% == 1000 kgf, just like the Haas VF4 (both using 32 mm ballscrews).

A Haas ST10 lathe (11 kW) 3500 kg iirc for mass has about 102 Nm peak torque @ 1200 rpm, less than half at start.
Mine has 90 Nm 0-1000 rpm.

I recommend You build Your lathe really stiff n heavy. For actual production, as You said.

Like 3x the biggest expected.
If everyone laughs at your section sizes, then they are just right.

For example, my z axis servo motor mount plate in tool steel is 50 mm thick, 140x200 mm iirc.
About 20 kg in mass.

Every single great machine tool in the world is exceptionally heavy.
Hardinge HLVH, Monarch10EE, Sharp, SIP, Weiler, etc.

[strantor]

I recommend You build Your lathe really stiff n heavy. For actual production, as You said.

Like 3x the biggest expected.
If everyone laughs at your section sizes, then they are just right.

For example, my z axis servo motor mount plate in tool steel is 50 mm thick, 140x200 mm iirc.
About 20 kg in mass.

Every single great machine tool in the world is exceptionally heavy.
Hardinge HLVH, Monarch10EE, Sharp, SIP, Weiler, etc.

Thanks for the input. Yes, it will be unreasonably heavy. I would like to make it out of cast iron billets for vibration dampening but those are hard to find and expensive. I have a bunch of 3/4" (19mm) steel plate that I will make it out of. Anything tubular will be full of sand.

[joeavaerage]

Hi,

. Anything tubular will be full of sand.

Nice thought but does not work.

For a material to dampen vibration first the vibratory movement must be coupled to the material then internal friction of the material must dissipate
the vibration energy.

If you have a steel tube and it is exhibiting vibration, because steel is so stiff, it will have considerable energy despite the vibration movements being very small, of the order of um.

If you have sand in the tube the wall of the tube may move back and forth by 1um, but will the sand? Some people have filled tubes with cement or epoxy/sand mix so that
the movement of the tube is coupled to the damping material. It relies on the adhesion of the damping material to the inside of the tube, which is problematic at best.

Lets assume say that you fill the cavity with epoxy/sand and that you manage to get perfect adhesion to the tube. Steel is about ten times stiffer than epoxy/sand, so if the 1um
vibration of the steel represents a vibrational energy of 10 Joule say, the epoxy/sand material with the same 1um vibration will have only 1 Joule of energy. You've gone
to a lot of effort and expense only to at best reduce the vibrational energy by 10%.

You would be better to spend the extra on increasing the wall thickness thereby improving stiffness and forget damping. Stiffness is way WAY more important than damping.
Otherwise use cast iron and pay the cost.

The other alternative is put the damping material on the outside of the tube seeking to increase the coupling of the movement of the steel into the damping material,
or better yet do away with the steel altogether and the damping material is the structure. That is the evolution of epoxy/granite machines.

Ordinary grey cast iron has good damping properties but is about 1/2 as stiff as mild steel. Of course you can get much stronger cast iron, SG for instance, but then
while the stiffness is up the vibration damping is down. So if you are going to use cast iron then you need to make it approximately double the weight of steel
to achieve the same stiffness.

I had axis beds cast for my new build mill, and they were expensive, $3500NZD, for three beds of 115kg each. Even worse was to come when I had them machined
on a large late model Okuma mill, another $5000NZD. I've got what I wanted......but it cost a fortune. I am going to presume that I will use this machine for many years to come,
and so the cost may not seem so bad in that context, but it sure hurts now!

As I've already posted I've decided, and paid for, profile cut steel for the machine frame as my budget would not permit me casting an SG frame and paying for time on a big
horizontal mill. If the frame vibrates I hope its an 'A' or 'C' or some other pleasing note!!

Craig.

[Jim Dawson]

About half the mass of my CNC lathe is concrete, some special blend developed by Hardinge, ''Harcrete''. They built the frame, presumably cast iron, then filled everything in with concrete. You could do the same thing with a steel tube frame.

[strantor]

Hi,


Nice thought but does not work.

Tell all that to a dead blow hammer!

[strantor]

About half the mass of my CNC lathe is concrete, some special blend developed by Hardinge, ''Harcrete''. They built the frame, presumably cast iron, then filled everything in with concrete. You could do the same thing with a steel tube frame.

I've looked into all manner of concrete casting for improvised machine tools. All in the past. When I had little money (and now I have little time). I don't know what kind of concrete they used but I imagine it's some pretty specialized stuff. My research lead me to the conclusion that concrete (at least every commercially available type) has too much dimensional drift to make a reliable machine tool. If you cast a tool in concrete you would have to wait a couple of months to use it, and then you would have to take measurements and make adjustments every week or maybe every day that it's in use (and everything would have to be adjustable). Fluctuations in Ambient temp, humidity, etc. Would throw it off. I know they did it in WWII but I don't know what tolerances they were working with or how often they were making tweaks. Nobody has done it since, so I assume a lesson was learned.


Is it a Portland cement based concrete or an epoxy concrete or something else maybe? Bondo? I'm curious. Never heard of this.

[Jim Dawson]

I've looked into all manner of concrete casting for improvised machine tools. All in the past. When I had little money (and now I have little time). I don't know what kind of concrete they used but I imagine it's some pretty specialized stuff. My research lead me to the conclusion that concrete (at least every commercially available type) has too much dimensional drift to make a reliable machine tool. If you cast a tool in concrete you would have to wait a couple of months to use it, and then you would have to take measurements and make adjustments every week or maybe every day that it's in use (and everything would have to be adjustable). Fluctuations in Ambient temp, humidity, etc. Would throw it off. I know they did it in WWII but I don't know what tolerances they were working with or how often they were making tweaks. Nobody has done it since, so I assume a lesson was learned.


Is it a Portland cement based concrete or an epoxy concrete or something else maybe? Bondo? I'm curious. Never heard of this.

Not exactly sure what the base of the stuff is. It's very fine grain, I accidentally chipped a small piece off when I was moving the machine, so could look at the structure. It looks like morter. I'm assuming that it is not affected much by changing ambient conditions, at least I have not noted any change in the machine as conditions change. I'm guessing in the case of my machine there is at least 4500 lbs or so of concrete in it. The total machine weight is 10,200 lbs. EMCO makes their lathes the same way.

There are some very low shrinkage, high strength morters available, normally used for grouting in machinery. There are some chemicals that you can add to the concrete mix that will change the viscosity, without changing the water content, to make the pour easier and don't seem to affect the strength and stability.

[spumco]

I've looked into all manner of concrete casting for improvised machine tools. All in the past. When I had little money (and now I have little time). I don't know what kind of concrete they used but I imagine it's some pretty specialized stuff. My research lead me to the conclusion that concrete (at least every commercially available type) has too much dimensional drift to make a reliable machine tool. If you cast a tool in concrete you would have to wait a couple of months to use it, and then you would have to take measurements and make adjustments every week or maybe every day that it's in use (and everything would have to be adjustable). Fluctuations in Ambient temp, humidity, etc. Would throw it off. I know they did it in WWII but I don't know what tolerances they were working with or how often they were making tweaks. Nobody has done it since, so I assume a lesson was learned.


Is it a Portland cement based concrete or an epoxy concrete or something else maybe? Bondo? I'm curious. Never heard of this.

There's a LONG thread underway in the DIY CNC sub-forum discussing alternatives to cast iron and/or steel in the context of a DIY milling machine. Have a look, especially at the later posts, where CSA concrete is 'discovered'. The OP in that thread is about to start doing some test casts to check for stiffness and workability.

Non-shrink, extremely fast strength gain (think hours, not weeks). If the literature is to be believed, it appears to expand very slightly if the water ratio is within spec.

I just finished filling my mill castings with a consumer variety of this stuff (Rapid-Set) mixed with basalt fibers. And yes, 'various chemicals' can be added. I believe Jim is referring to superplasticizers, which really help the viscosity and workability without degrading the final product.

Frankly, I couldn't have worked the CSA without it. Once I added the fibers it approximated a cat's hairball mixed with peanut butter - not good for casting. Add the superplasticizer and it didn't turn self-leveling, but I could 'pour' it and get it where I wanted it.

I'm a while away from post-upgrade stiffness tests, but I'll be reporting back in the other thread once I finish getting it back together. I have no data to confirm or dispute damping or vibration improvement clams, but I do have baseline stiffness measurements.

-Ralph

[joeavaerage]

Hi,
This is a study done on concrete filled steel tube structures. for CNC machines.

https://inter-eng.umfst.ro/2012/file...ers/paper4.pdf

In section 4: Results and Conclusions look at graph fig 8 and look at the damped natural frequencies. The concrete filled structure shows better damping as you would
expect, but rather less dramatic than you might have thought.

Craig

[strantor]

Ok I've made some decisions and some purchases, now narrowed down the options to just two.

I've purchased:
(2) 300mm glass scales
Raspberry Pi 4 w/10" touchscreen
Mesa 7i97 ethernet 6 axis analog card with 6 encoder inputs.

I've chosen the analog card with encoder feedback because:
- I understand it well enough to explain to a child.
- I have experience with this setup and I think the project will go faster.
- I still opine that dual closed loop; encoder (velocity) feedback to the drive and glass scale (position) feedback to the controller is the most bulletproof setup there is. I'm not 100% confident in that, some compelling arguments for distributed motion control were delivered and I heard them and respect them but I confess I didn't invest the required time to understand and be ready to adopt them. There are too many aspects requiring research and I had to start pulling some triggers, getting out of theory and into action, so I defaulted to what I know.
- This enables me to use cheaper drives.

I chose the RPi because it's cheap and born for Linux. And because I found others using it, so I think any questions/problems that come up will have already be addressed in a forum somewhere on the internet.

I chose the 7i97 ethernet card over the 7c8x RPi card because I'm not 100% confident that the RPi will perform, and I want to reserve the option to use something else. And there's no RPi card for analog w/encoders.




So now I need to order servo motors and drives. Going with 400W servos. Now that I don't need glass scale feedback to drives, Delta is not necessary. Now that I don't need encoder output from the drives, Dyn4 is not necessary. I could go with Dyn2 or Cheap China, but strongly leaning away from Cheap China, except maybe for the spindle. The only thing stopping me from pulling the trigger on a Dyn2 setup is the need for a high power DC supply. When I add that to the cart, the total comes up almost to the cost of a Dyn4 setup. I could save money by getting a cheaper power supply from eBay or rolling my own power supply, but in investigating these options I generate more questions.

1. How much power do these Dyn2 drives actually need? They say 60V, 16A. But DMM sells the same power supply with all their multi-axis kits; 48V, 21A. I guess they assume no two, three, or even four drives will ever be pulling max amps at the same time, but on what would they base that assumption? If 21A is really all that's needed for 4 axes given that's what they sell with a 4 axis kit, then is 12A sufficient for two axes? It's hard for me to confidently pull the trigger on a 12A supply for 2 drives when the label says 16A just for one of them. If 16A is just a brief spike then I think the caps in any cheap supply could handle it, but I don't know that.

2. Why are they selling a 48V supply for 60V drives? Is 48V really enough to hit the high speeds? It seems like it could be on the ragged edge, and if I wanted to employ the "field weakening" (or whatever that's called when dealing in permanent magnet motors) to get the advertised maximum of 5kRPM, I wouldn't be able to do so, and I might not even be able to get the 3K nameplate value.

3. How much ripple can these drives handle? I don't see a spec anywhere in DMM documentation for DC ripple tolerance. I could easily make an unregulated transformer/bridge rectifier supply with a beefy electrolytic capacitor tank but I don't know if the drives would like it.

4. I guess a consolidation of 1 - 3. What power supply would you fine gentlemen recommend for a pair of Dyn2 drives? Or would you recommend just going for a pair of Dyn4s and put all these questions to bed? I'm seriously considering that anyway.

[Al_The_Man]

AC, DC & BLDC Servo's are very P.S. tolerant, one thing to avoid, however is a SMPS, linear is far more rugged.
The only feedback to the analogue BLDC motors/drives I use is the commutation (hall equivalent) markers on the encoder.
P.S. design attached.
Al.

[rodw]

Ok I've made some decisions and some purchases, now narrowed down the options to just two.

I chose the RPi because it's cheap and born for Linux. And because I found others using it, so I think any questions/problems that come up will have already be addressed in a forum somewhere on the internet.

I chose the 7i97 ethernet card over the 7c8x RPi card because I'm not 100% confident that the RPi will perform, and I want to reserve the option to use something else. And there's no RPi card for analog w/encoders.

If you do decide to turf out the RPi, the Odroid H2+ is a good small form factor substitute (and a real PC). Its only problem is the twin 2.5gb Realtek 8125 network interfaces do not yet have a Driver in release versions of Linux. It is easy to install if you grab the driver form the realtek site.

I've attached a pic of my H2+ attached to the back of a 15.6" touch screen with a Mesa 7i76e, a Mesa 7i73 HMI interface and a wireless receiver for a 433 Mhz Industrial Pendant. Hopefully it will work with the 7i73.

[pippin88]

I vaguely remember something about DYN4 being better in terms of input / control options and preferred?

If cost is similar I would go for DYN4 and avoid DC

[pippin88]

Ordinary grey cast iron has good damping properties but is about 1/2 as stiff as mild steel. Of course you can get much stronger cast iron, SG for instance, but then
while the stiffness is
I had axis beds cast for my new build mill, and they were expensive, $3500NZD, for three beds of 115kg each. Even worse was to come when I had them machined
on a large late model Okuma mill, another $5000NZD. I've got what I wanted......but it cost a fortune. I am going to presume that I will use this machine for many years to come,
and so the cost may not seem so bad in that context, but it sure hurts now!


Craig.

5 thousand for machining?

Surely these are actually simple parts to machine? Face rail mounting, drill and tap a few holes?

How many hours / hourly rate?

[strantor]

I vaguely remember something about DYN4 being better in terms of input / control options and preferred?

If cost is similar I would go for DYN4 and avoid DC

Yes I've approved a quote from DMM for a pair Dyn4 packages, just waiting on an invoice.

Want even better prices on DMM gear? Check out their eBay store. Prices are 25% less than on their website and shipping is free. If they don't have what you want listed on eBay, contact them and talk about it. They worked out a deal with me that saved me a substantial amount of money.

[Iron-Man]

I'm starting a scratch-build. It will be a dedicated purpose tiny desktop CNC lathe. It will be probably smaller than a HF mini lathe but weigh twice as much. This will not be a hobby machine. It will be putting in serious hours. I want it to be fast, accurate, rigid, robust.

I am debating about what motors to put on it. I would like to save money where possible. AC servos would be ideal but they're expensive. There are some ridiculously cheap ones on eBay, $150 for 500W motor + drive. I assume they're crap and should be avoided, but thought I would check here and see if anyone has used them and had anything to say.

Also considering the hybrid stepper w/ encoder feedback option. Never used those, are they any good? Can they hold their own against an AC servo?

Also considering DC servos. I would have no qualms using DC servos but I can't seem to find any. Where can one buy decent DC servo motors?

BLDC? I know very little about them other than they seem to have much lower torque specs than the other motor technologies for a given size. I have that notion only after a cursory glance, I could be way off.

This will be a LinuxCNC machine, and I'm considering using one of the Mesa boards to drive the servos. Does anyone have anything to say about their motor driver boards? 8i20, 7i39, 7i32, 7i54, etc. etc.?

This will 100% absolutely have closed loop control no matter what motors are chosen.

Hello Strantor

I refitted a Denford NovaTurn with IBOB2, AMD GizmoSphere I, Yaskawa SGDA 01AS Servos, Mesa 6i25 / 7i77. Pathpilot is the control. Analog +- 10V fully closed loop.

See my work here; https://www.cnczone.com/forums/bench...754-cnc-2.html

More is posted on the Denford site; NovaTurn Servo retrofit AMD Gizmosphere iBOB2 PCIe x1 - Denford Software & Machines

As far as servos go, ebay can be an excellent resource for industrial drives (Yaskawa, Allen Bradley, Mitsubishi, Omron, etc. The only downside is that you may have to wait, to find good deals.

Hope this gives you some ideas.

Iron-Man

[strantor]

Hello Strantor

I refitted a Denford NovaTurn with IBOB2, AMD GizmoSphere I, Yaskawa SGDA 01AS Servos, Mesa 6i25 / 7i77. Pathpilot is the control. Analog +- 10V fully closed loop.

See my work here; https://www.cnczone.com/forums/bench...754-cnc-2.html

More is posted on the Denford site; NovaTurn Servo retrofit AMD Gizmosphere iBOB2 PCIe x1 - Denford Software & Machines

As far as servos go, ebay can be an excellent resource for industrial drives (Yaskawa, Allen Bradley, Mitsubishi, Omron, etc. The only downside is that you may have to wait, to find good deals.

Hope this gives you some ideas.

Iron-Man

Nice! Similar story...

I'm a Controls Engineer; I design control systems for industrial process equipment (conveyors, palletizers, extruders, etc.) and CNC is more of a hobby. But people usually assume I'm some sort of "whiz" or "hacker" and throw random stuff at me. Just so happens I had a customer with a Denford NovaTurn he got at auction. He couldn't run it because it didn't come with the license floppies. He sent me home with the PC and wanted me to "hack" it. I told him I would evaluate it and see if what could be done. I spent a couple of hours and came to the conclusion that it needed a retrofit. I offered to quote him, told him it would probably cost more than the machine was worth, and of course he declined. I think he sold it on eBay. Maybe you bought it?