Slant bed lathe, tool changer, pneumatic 5c drawbar, DYNOMOTION control, PICS!

Lets start a machine that gets finished! I know funny right, are they really ever finished?

The machine is being built to machine some parts that I need to make a few hundred pieces of, a couple times a year. You can see the pics attached, about the size of a pencil eraser and made of delrin. I would also like the machine to be able to turn out some small aluminum parts if needed in the future, as needed by the same project the delrin parts are for. Pics of the delrin part, and its drawing are attatched.

The design travel will be 14" in the Z axis an 8" in the X axis.

Found a great deal on a ELGIN lathe 5c spindle head with Hardinge taper lock spindle nose! $125. and in good condition!


I started out the build with a lathe bed made of a 6" x 6" x 36" section of surplus structural steel that was rusting away. Its got a 3/8" thick wall and I filled this section with a polymer concrete. So far I have welded support rails down the length of the bed to support the INA linear rails that will make the z axis ways, and welded on the spindle mount plate. All this was machined in a single setup with the bed "delicately" clamped to the milling machine so as not to induce any twist or stress during machining. The last thing I needed was the bed to be held in a twist while machined, and then allowed to untwist when un-clamped from the milling machine bed.

Cleanup, welding and machining of the bed section was rather time consuming but it turned out well and is ready for paint. I was delighted to find that after machining the bed to accept the rails, that there was no rocking of the INA rails in the machined slots. After bluing, the rails and slots I found there was adequate contact between the bed and rails so no scraping was needed to mate them.

I had green paint that I use for the Fence business I run, so green it is.

The machine enclosure has been started as well, and is designed to be able to wheel around the shop floor. There is no room for a machine that needs a forklift to move, so I consider this an "office" sized lathe build!

Some crude design outline drawings.

Nice headstock. What kind of a machine did it come from.

It came from an elgin lathe, elgin was partnered with hardinge, the company split hardinge continued elgin tried for a while but did not go on for long, at least that's the story as I have discovered it.

The headstock came up on ebay from a dismantled machine, no other parts were available from the seller.

Attachment of the bed to the frame. Attachment 340190Attachment 340192

Made a bit more progress on the carriage and ball screw mounts.

I must say, the travel down the length of the z axis is so smooth! These INA rails and bearings are great!


I got the ball screws from linearmotion2008 on ebay, and I am pleased with them so far. The angular contact bearings in the fixed end support are NOT preloaded so a spacer was made to set the preload within the bearing set. Once this was completed the screw has a very stiff feel. I dont have servos mounted or drive couplings completed yet so I cant tell exactly what the backlash is. I dont expect to get a real reading on the actual backlash until I get the system ready to fire up electronicly and thats a far bit off in the future.

Just locking the rotation of the 20mm screw and pushing back and forth on the carriage revealed about .0001 - .0002 of deflection with what I estimate to be 50 - 60lbs of force applied. Before the spacer was added to the bearings this movement was .005"! These screws are both the double nut variety, and I am glad I went with that option.

I went with 20mm srew on z axis and 16mm screw on the x axis. The 20mm was very staight, the 16mm not so much. There appears to be about .007 runout at the free end of the 16mm screw. Wont know how much this affects operation until I get the linear rails cut down to fit the top of the carriage, and the cross slide mounted up.

Got the servos in, and they look great. I got 4 of these at $25 each. could not pass up getting some extras as i only need 2. at this price I took 4!



The encoder is 200line, so that will come off, and a 2000 line heds unit will be adapted to the shaft with an am26c31 quad line driver board to carry A/_A, B/_B and index/_index channels back to the kflop control.

Awesome build Andrew! I love watching CNC lathe projects progress. Yours is going to do great work and is the perfect size for making lots of useful widgets.

Are you going with gang tooling, turret or a QCTP? I have a QCTP on my little converted ORAC along with 20 QC tools dialed in to the Mach3 tool table. I'd love to have more X-travel so I'd have room for a 2'd QCTP or a dedicated 4 position gang tool holder on the back side. It's something to think about in the early phase of your build while you can still add more X travel. Can't have too much!

Also may I suggest that you design as accurate an X-axis homing system as possible. It'll save you much frustration later.

Keep the pics coming!:)

Thanks for the good words Dickeybird.

The plan is for the turret to hold 9 tools and be a rotary index mechanism with a hirth coupling or cirvic coupling for index position lock. I am thinking a 4" diameter pneumatic cylinder to lock with ~1000 lbs of clamp force, with about 100 psi air pressure.

Turret will be 6" dia with 4 od turning tool positions and 4 I'd turning collet holders in the er16 size range. These would be mounted around the outside of the 6" turret with an additional I'd collet tool position in the center of the turret.

I am building in enough X axis travel to get the center of the turret in line with the center of the spindle, even though most of this travel will not be used on any of the tools, except the one in the center of the turret. I envision this being primarily a center drill location.

Thanks for the heads up on the precision homing route.

From your experience how does the following method sound?

Axis moves positive until limit switch is hit, then moves negative slowly until shaft encoder index mark is hit. Axis zero is set at this position.

My plan is to have this method on both axis for the homing procedure.

Quote:

---

Originally Posted by kn6za

Thanks for the good words Dickeybird.

The plan is for the turret to hold 9 tools and be a rotary index mechanism with a hirth coupling or cirvic coupling for index position lock. I am thinking a 4" diameter pneumatic cylinder to lock with ~1000 lbs of clamp force, with about 100 psi air pressure.

Turret will be 6" dia with 4 od turning tool positions and 4 I'd turning collet holders in the er16 size range. These would be mounted around the outside of the 6" turret with an additional I'd collet tool position in the center of the turret.

I am building in enough X axis travel to get the center of the turret in line with the center of the spindle, even though most of this travel will not be used on any of the tools, except the one in the center of the turret. I envision this being primarily a center drill location.

Thanks for the heads up on the precision homing route.

From your experience how does the following method sound?

Axis moves positive until limit switch is hit, then moves negative slowly until shaft encoder index mark is hit. Axis zero is set at this position.

My plan is to have this method on both axis for the homing procedure.

---

Oh wow, 9-tool turret. Quite a project in itself but certainly worthwhile...way above my skill level!:eek:

I forgot you have servos, awesome. Servo index homing as you've described will work great. I have stepper motors on mine & had to put a slotted disc/opto sensor on the other end of the double ended stepper motor that was switched on by a microswitch close to the end of axis travel. Not as good as servo index homing but the next best thing.

Made some progress on the encoder driver boards.

I ended up milling the circuit board traces in the cnc milling machine, this made drilling the holes and cutting the outside shape easy too.

These boards are attached to the HEDS encoder module and function as differential line drivers. The chip is the am26c31 quad line driver. This converts the single ended signals from the HEDS module to differential signals that the motion control needs to see. The advantage to this setup is that a differential signal has greater noise immunity than a single ended.

Twisted pair 8 conductor shielded cat 5 cable will be used to connect the encoder signals to the control board.

A,A-,B,B-,Z,Z-,+5v and GND

Removed the old encoder from the servo and got the new one adapted to the shaft.


The old one is a 200 line low resolution metal disk unit. I want a 2000 line encoder, so I went with the us digital HEDS style units. Conversion was pretty strait forward, I drilled and tapped 4-40 threads to hold the HEDS module to the back of the motor, and turned some press fit delrin disks to pinch the encoder disk to the shaft. Time will tell if this method of fixing the disk to the shaft will work long term, but seems like a solid connection with little axial or radial run out.

The math on the resolution will be 2000 line x 4 in quadrature = 8000

2:1 belt drive servo to ball screw 8000 *2 = 16000

5mm ballscrew lead = .1968" or 5.081 turns / inch for 16000 * 5.081 = 81296 encoder counts per inch.

This gives a design resolution of 1 / 81296 = .0000123007" :)

The motors are 170w 2600 rpm at 32volt no load. I will be using a 48 volt axis drive power supply to start out with. This will give approx ~3900 rpm no load speed. 300 ipm is my target rapid speed on this machine, and with a no load top rpm on the motor of 383 ipm, I think this will be a stretch, but a 60volt power supply can be obtained if necessary. The 48volt is what I have, so I will start there.

Very interesting Andrew; looks like you have a strong electronics background. Please keep posting & I'll keep learning.:)

Your project is coming along very well!

Thanks Dickeybird!

So Far, so good. Its been fun too.

Ok, so I have done some more work on the cross slide. Cutting down the linear rails to fit the carriage was tough! I used an abrasive cutoff wheel in my circular hand saw. Not sure if there was a better way to cut those things or not, but considering how hard they were, I could not think of any other way. The rails mounted up very nicely to the top of the carriage and the cross slide moves like the Z axis, very smooth! The servos are mounted up with 1/4" aluminum angle motor mounts.

I checked and rechecked how square the machine work was done to the carriage, as the machine work on the bottom relative to the top will determine how square the x and z axis are to each other. Now that the cross slide is assembled, mounting a machinist square to the cross slide and indicating the travel revealed about .0005" deflection in 5" of cross slide travel. Im happy with that!

I got more work done on the enclosure too. I wanted the section of the machine enclosure that covers the lathe work envelope to be removable for maintenance, or what ever? This part went well and I got some sheet metal applied to the outside of the removable enclosure section, and the internal sheet metal way covers put on. Way covers are all 1/8" alum plate.

Still needs a door and door track installed on the enclosure cover.

The block that is on the cross slide in the photo will be the tool changer base, but for now its a solid piece of fortal aluminum.

I also stated machining the operator panel. Not sure how to engrave the nomenclature of the respective buttons yet. I don't have a cnc machine large enough to mount the operator panel so direct engraving of the panel will not be an option.

Wow, it's looking really good and you are really making fast progress. Looking forward to seeing the tool turret design.

Thanks Hezz!

I have been collecting parts for a year or so, finally got started on this thing and it's moving along well.

I have all the turret components ready to machine into parts, so it's coming soon.

I got a steal of a deal on the turret blank on ebay. It was advertised as 6" dia x 3" long steel round for $25, when I got it, it was marked 4340! I decided it was too nice a steel for the intended use and would be saved for this turret! 4340 will make this a heat treatable component if I decide to do that in the future.

The plan is to get the op panel finished, then turret, then 5c pneumatic draw bar. Oh and electronics in there somewhere;)

After getting the operator panel test fitted with the control components I think I figured out how to identify the buttons without directly engraving the operator panel itself.

Attachment 340526

I had some 1/16" thick dark bronze anodized aluminum plate and I decided to cut it out as a border around each control or set of controls. The darkness will give a great contrast to the rest of the panel back ground and make the text engraving show up well.

I have a 6"X9"X7" cnc mill that I used to cut out the knob holes, the outline bevel edge, and the text in one setup. I used super glue to mount the AL blank down to the machine bed.

The super glue held well, but made a real mess of the anodizing on the back side of the thin plate. It seems the superglue reacts with the anodizing in some negative way, leaving a mark in the surface that cannot be cleaned off. I decided it does not matter, cause it cant be seen from the front.

I think this is the best solution I could come up with.

I will post more PICs when I get some further progress made.

Excellent job.

Fwiw..
My lathe has similar resolution with 10.000 count encoders, driven at 1;2.'I went with 8 mm profile, 30 mm wide, HTD belts, and taperlock bushings.
Reason:
Smaller belts are not stiff at all, and cannot transmit a lot of torque at low rpms, under say 10 rpm.

8 mm profile HTD is about 2x as stiff as 5 mm, and 30 mm is obviously twice as stiff as 15 mm wide.
Even then, I wish they were more rigid.

Another thing to take into account-
For good stiffness, belts want to be quite tightly tensioned.
Look this up, and make sure You don´t over-tension the shafts on the motors.

Maybe You already knew all that, just pointing it out in case it is useful.

Good luck.

Thanks hanermo!

That's a good sugestion, with your setup at 30mm wide, what are your servo drive motors rated for torque?

750 W, Ac brushless servos,
220V mains,
10 Nm peak torque, so == 3.3 Nm continuous,
4 mm 32/4 ballscrews on lathe.

Spindle drive.
2.5 kW continuous AC servo, 220V,
30 Nm peak,
1:3 belt drive with HTD8-30,
so 90 Nm peak !

Servos would accelerate the 100 kg carriage to full speed in approx 20-30 ms- limit is breaking belts.
Ie 0.03 secs.
3000 rpm x 1:2 = 1500 rpm = 25 revs/sec.
I only use == 1:3 the top speed, and 1:4 the top acceleration, and limit it in machx.
Step size is == 0.2 microns, or 0.0002 mm.

At 1:2 I get about 1600 kgf push force on the screws, peak, for up to 3 secs.
I even faulted the z axis, once !
I never ever get any visible variable error on cuts on the 10.000 count servos.
There is so much torque/push they always track *"perfectly"*.
On the servo controller boxes their leds show the current error in real time.

My tuning is too soft, but it works and have not had time (need), since it works,...
Following error is only a Q of speed, at the moment.

Servos lock after 0.5 secs of no signals, zero jitter, hunting, thrumming, noise.
Have forgotten machines on for a week, powered, running M3 with CSMIO-IP-S.

10.000 counts x 3000 rpm = 500 kHz servo speed.
24 V DC, isolated, sinking, I have about 150+ lines of IO to the controller.
The CSMIO is 4 Mhz, so no issues.

750W servos were 690€, 2.5 kW 1600€, 22% VAT taxes paid all-in.
Our other company imports/exports stuff.

Note.
The 90Nm torque is 0-1000 rpm at spindle. 0.1 secs 0-1000 rpm if You wish (I use softer settings).
A 50.000€ Haas ST-10 lathe, of 11 kW, has 102 Nm peak at 1200 rpm, and quite a bit less at less than 300 rpm.

The end result is that over a tiny movement, I get 1600 kgf force to move the cross-slide or carriage.
It WILL move.
Theoretically, 0.2microns, today in the real-world, less than one micron mechanical resolution (incremental).

Screws are no-where near that accurate, but making 1 micron cuts work fine, in steel, ccmt inserts (tips wear fast at tiny cuts).
When I fit glass scale (I will), I will be able to measure the real results, and map the screws.

Info:
AC servo with differential signals.
a,b,z, with differentials, = 6 lines io.
On, error, reset, inpos, disable = +5, 11 total.
Limits, home, +3.
3x 14 = 52 lines for 3 servos, 64 for 4.
MPG === 25.
ENCoder module == 10.

Spindle -C, X, Z, B, it all adds up..