[MichaelGolden]

Hey guys. I'm in the process of getting parts together for a diy 5 axis cnc.

Lets not talk about the g-code or cam software. I'm only interested in talking about the electronics. (which im not an expert with) and the trunnion table rotary bed build.

I'm not concerned about head clearances either. This can be changed. The rotary bed on the trunion table will be 6" and will hold a 5" chuck.

I'm going to build it around a 'little machine shop hi-torque mini-mill model 3900'.

I spent some time looking at stepper motor sizes. I got some calculators. It was a decision for me to go with overkill using g203v drivers to allow for upgrading the machine later. (stiffer and better spindle.)
I decided to go with 381oz/in steppers all around. With g504 4 axis and g251 driver. Smooth stepper. And a C11G breakoutboard for the 5th axis breakout and pump/spindle control. 48VDC 20A power supply

Heres the current order list for the electronics from AutomationTechnologies.

Gecko Diver G540 4 Axis Driver, Current Version
NEMA23 381oz/in 3.5A Single Shaft Stepper Motor KL23H2100-35-4A
Ethernet SmoothStepper Motion Control Board for Mach3
Geckodrive G251 50V/3.5A Stepper Motor Driver
E-stop Switch
Coupling 1/4" and 1/4 "
Micro Limit switch
Ethernet SmoothStepper Cable Combo
4 conductor 22 GA Shielded Motor wires
Motor cable connector
C11G - Multifunction CNC Breakout Board
Unregulated Linear 960W/ 48 VDC/20A Toroidal PSU (KL-4820)


I also plan to build an enclosure for it.
I could use some suggestions on the electronics. I wasn't sure about the C11G breakout board. But it seems like it has a lot of room for expansion.
Lets hear your thoughts.

Cheers

[LongRat]

I can't see any problems with the electronics you have listed. Going with the SS is a good idea, I have built a 5 axis mill and before the SS my rapids were extremely slow. The PC couldn't produce a clean enough pulse stream at the sort of rate needed to move all 5 axes simultaneously.
You might not get too many replies because 5 axis builds are not common here. There are a couple of reasons for that. Most people doing hobby stuff just don't need it, and ignoring it or not, the software is expensive and can be tricky to learn. Most times it is just more than you would need. I would be interested to hear your intended application for this mill.

[MichaelGolden]

Thanks for the reply,
I've been working on designs of model jets at home. Am interested in producing custom impellers and blisks. Mainly from aluminum. If possible I would chew away at some one off parts in inconel. I believe its possible at much slower speeds. We will find out.

I will most likely have to put a spacer block under the gantry support to gain some headway for the trunnion table. and add steel tubing support to the top of the collumn, which will also be apart of the enclosure. Which costs me nothing but my time.

I appreciate the positive feedback!

Attachment 216060Attachment 216062Attachment 216064Attachment 216066

[zamazz]

Very neat! That compressor blade looks nice! I'd like to do the same thing at some point, although it will be years before I'm at that point. Once I get my mill running a Wren 54 is on the top of my list of projects. I have too little time to even think about engineering one at this point. As for the Inconel, it may be difficult with a machine that small, especially holding the tolerances necessary for a jet engine. Do you already have the mill?

[neilw20]

Without zero backlash, or a way to compensate it (and that will leave marks) what you are trying to achieve will be difficult.
How are you going to get zero backlash on the tilt table and the rotating parts?
Harmonic drives, come to mind.
Servo drives, running against high friction drag brakes, and positional feed back is also an option on any/all axis.
Double loop servo controls, which are rather difficult to make work with Mach3, if you can make them work at all.
One loop for the servo control, and an external one to account for position errors in drive (reduction) system.
Backlash compensation will always make all axis pause for the one changing direction (removing it's backlash).
Could never get 5 axis compensation on Mach3 working, and then if it did the referencing would not work properly on all axis.
referencing, or compensation, but not both!! Does anybody know a Mach3 version with NONE of these features broken?
I have worked on a 5 axis machine, and initial planning to achieve repeatability/reliable zeroing/referencing on the 4th and 5th axis (and other axis) will make or break the project.

[LongRat]

Are those some of your parts? Very nice work.
If you build your machine stiff enough and use the right tooling, you will be able to machine inconel. I think you will need to make a lot of the much-documented improvements for that machine, and the size of the rotaries you wish to fit will be tight. You need to know what you are doing but if you are building engines like these already, you do.

[MichaelGolden]

Hey all, thanks for the reply's. Happy holidays/new year!

I'm planning on reinforcing as much as is needed with the machine. I'm not going to spend a fortune on it. I'll do what is necessary. I'll have to find and read up on the improvements that have already been done! The rotating table is 6" but I will be using a 5" chuck making mostly 4" OD diameter parts. The LMS model has 5.1" movement in the Y axis. I've seen a few extended bed mods. We'll see what happens..

I'm not looking for spaceX quality parts. But getting zero backlash is definitely a concern. I think I'll be using worm gearing on the trunnion/rotary table. I've seen belt drives but im curious what the backlash with a belt would be like. The current engine plans available, such as a KJ-66, suggests accuracy within +/- 0.05 mm on some parts.

Then again I've seen running inconel turbine's been made from hand. So even if I was able to get an accurate rough cut. I could finish the blades by hand.

The parts in the photo's I've posted are done by hand. Based upon a lot of math! Its all experimental stuff still. I have a manual 7 x 12 lathe at home. It takes patients. But I have no time limit. And one of the best tips/sayings I've read on the dedicated machining forums is 'you cant rush art'.

[Khalid]

You are a real craftman. Are you from USA ? USA workmanship is well reknowned in the world. Love the parts so neat and fine that i wonder how long it would have taken to complete them by hand. Sorry for my fractured english

[MichaelGolden]

Khalid, I hadn't noticed. Your English is great.

All the parts are made in Canada!

[Maglin]

I think that .005mm accuracy might not be achievable with a worm gear drive. I would think it could on a belt drive with a lot of reduction like 5:1 to give you maximum holding torque. I also like the idea of duel closed loop servo system to ensure positional accuracy. LinuxCNC will definitely be better suited for control until Mach4 comes out.

Sent from my KFTHWI using Tapatalk

[MichaelGolden]

Hey guys, got the electronics last week. Just a little update getting everything together in my spare time.

If theres any problem with the backlash I can always go another route. I'm just not to keen on spending too much on this little toy right now. One step at a time. Lets try it out and see what happens.

I decided to get into hobby machining a year ago. I should give props to all you guys with build logs and youtube videos before me. Its helped a lot. Going to be an awesome year.

Mike

Attachment 221810

[lsfoils]

Greetings,
Worms and wheels are problematic... running them in works, to a degree, for those that are already very close. I would recommend against using any lapping compound as the wheel or worm can be permanently loaded with the compound and things will wear out quickly.

Bought a small rotab and found the wheel was out considerably. Took it apart and realized the worm was very close to 8 tpi. + or - just a tiny bit. The wheel was cast iron. So I made 2 worms on my lathe (no modular function on that machine for worm gears) and cut 2 duplicates for the worm gear/shaft out of SS 304; 8tpi, 29 degree included angle. Then I gashed one new worm gear at an angle offset from the shaft's center axis to make cutting teeth using a 60 degree dovetail cutter. I loaded up this new hob into the rotary table and set the backlash to near zero. Instead of using a stepper, I hooked onto it with my cordless drill. After maybe 30 seconds, I could hear the drill's motor smoothly running for the full rotation of the table. I readjusted the lash ring on the table and did it again. I think it took 3 cycles to get the worm to run clean. Wish I'd taken pictures. Anyway, if you use this method, be sure you take the table completely apart and clean it! I found quite a pile of shavings inside (surprising amount!) and you'll want to make sure all the bearings and bearing surfaces are clean. The teeth on the worm wheel looked perfect. There is still a tiny bit of backlash in the unit. Has to be to turn. But it works for what I do. Hope this helps. -Doug

[TJones79]

Doug - this method sounds nice I might have to use it myself as I've had some problems with the worm wearing out too quickly.
Mike - your parts look nice. Glad you're making progress

[dmauch]

The low profile Chinese trunnion table4 that I have has a 5/1 ratio timing belt drive. It is pretty good and you could do some light machining on it. But it would be possible to add an idler gear between the driver pulley and the driven pulley and probably increase it to at least 10 to 1.
Dan Mauch


OOPS I just noticed this was an old posting.

QUOTE=Maglin;1407586]I think that .005mm accuracy might not be achievable with a worm gear drive. I would think it could on a belt drive with a lot of reduction like 5:1 to give you maximum holding torque. I also like the idea of duel closed loop servo system to ensure positional accuracy. LinuxCNC will definitely be better suited for control until Mach4 comes out.

Sent from my KFTHWI using Tapatalk[/QUOTE]

[MichaelGolden]

The low profile Chinese trunnion table4 that I have has a 5/1 ratio timing belt drive. It is pretty good and you could do some light machining on it. But it would be possible to add an idler gear between the driver pulley and the driven pulley and probably increase it to at least 10 to 1.
Dan Mauch


OOPS I just noticed this was an old posting.

QUOTE=Maglin;1407586]I think that .005mm accuracy might not be achievable with a worm gear drive. I would think it could on a belt drive with a lot of reduction like 5:1 to give you maximum holding torque. I also like the idea of duel closed loop servo system to ensure positional accuracy. LinuxCNC will definitely be better suited for control until Mach4 comes out.

Sent from my KFTHWI using Tapatalk

If anyone else is interested in the same setup I have pm me for details. I'm starting a list of interest and may be going ahead with a kit/instructions.

https://instagram.com/p/zti2IfgZwX/

[MichaelGolden]

Oops! Typo there. I meant 0.05 mm.. And its referring mainly to the rotating parts and spacers.

But anything needing to be concentric/round and accurate, such as a turbine would first be done as a blank on the lathe.

The ballscrews for the x and y will give me .001 " increments per step. Actually 200 steps per complete revolution. So I would assume to try and have similar increments for the rest.

Thanks for the input. I think I will definitely be going with closed loop servo's once everything is said and done. But for prototyping I'll be using the steppers. It will be a while.

[Fastest1]

Oops! Typo there. I meant 0.05 mm.. And its referring mainly to the rotating parts and spacers.

But anything needing to be concentric/round and accurate, such as a turbine would first be done as a blank on the lathe.

The ballscrews for the x and y will give me .001 " increments per step. Actually 200 steps per complete revolution. So I would assume to try and have similar increments for the rest.

Thanks for the input. I think I will definitely be going with closed loop servo's once everything is said and done. But for prototyping I'll be using the steppers. It will be a while.

If you are using a G540, it is configured to have 10x microstepping (it is not adjustable as some controllers are). You will have 2000 steps per revolution instead of 200. If theoretically you were at .001 per step prior, you will now be .0001. In actuality it wont be that accurate.

[SCzEngrgGroup]

If you are using a G540, it is configured to have 10x microstepping (it is not adjustable as some controllers are). You will have 2000 steps per revolution instead of 200. If theoretically you were at .001 per step prior, you will now be .0001. In actuality it wont be that accurate.

And, in actuality, MANY other factors will will much more significantly limit the achievable, real-world accuracy....

Regards,
Ray L.

[MichaelGolden]

If you are using a G540, it is configured to have 10x microstepping (it is not adjustable as some controllers are). You will have 2000 steps per revolution instead of 200. If theoretically you were at .001 per step prior, you will now be .0001. In actuality it wont be that accurate.

Yeah I see what your saying. But can we tell it to move just 1 microstep at a time? I thought it was just a way to smooth out the signal for a full step.

The x and y kit is coming from cncfusion.

[LongRat]

Yes, the microsteps are the smallest movable increment. You have to treat the motors on a G540 as being 2000 steps/rev.
Although many will point out that this resolution in no way represents the true ACCURACY you will be able to achieve, if you don't have high resolution you will never be able to get good accuracy, even with the best mechanical components on the planet. For a true 5-axis system, especially one with quite large rotaries, you will find high resolution is actually quite important, because a tiny rotary increment will lead to a somewhat large linear move when at a significant radius from the rotation centre. This is where I found the Smoothstepper so important. On my benchtop 5 axis I wanted a linear effective resolution of <10µm on the table, which meant quite fine encoders on my rotaries. This led to the requirement for VERY high step rates to achieve a decent speed which isn't a problem for the SS, but you'll not get good performance from the parallel port.
I would also add the the trunnion table I was using was equipped with two harmonic drives. I think you may find it difficult to get low backlash with normal worm-drive rotary tables, but you might get lucky with good quality units.