CNC Mill Re-Re-retrofit

[Jerry]

About 5 years ago I got my Supermax YCM-16VS cnc mill from a friend. It had come out of Tektronix's model shop. It had an old control that was failing so I did the mach re-retrofit as logged in this thread:

http://www.cnczone.com/forums/vertical_ ... grade.html

Its been a good mill. A few little quirks. If you powered up the servos with the drives enabled it would jump and sometimes fault out or trip one of the breakers inside of the control for the drives. This was caused by the Pixie step/direction to analog control boards. These boards would take step and direction inputs and control standard analog servo drives with encoder feedback. They worked pretty well, but tuning was a nightmare. Not only did you have to tuned the analog drives, you also had to tune the position loop. It had some decent gui software but I really never got it tuned like it should. I would get random follow error trips when the machine changed direction.

Rather annoying. That and the X axis motor is rather long and I dont know how many time I have ran into it.

With this last week being a production break from Paranorman I figured I would finally have some time to do something about these problems. A couple years ago I got that Mitsubishi injection molder. Among the servos on it were two Mitsubishi MR-H series 1kw servo drives and motors. I also had another Mitsubishi 1.5kw MR-J series drive and motor from back when I was gearing up to do the servo retrofit on my 10EE. The 1kw motors were much smaller than the old ones and about 5" square. The 1.5 is a touch bigger than the old motor. Both models of drives have step and direction input native.

In anticipation of doing this project I had machined adapter plates and motor mounts for the motors ahead of time. I also hand some small circuit boards made up through the dorkbotpdx circuit board service. These boards broke out the 40 pin MDR connector to give me easy access to the step, direction, and alarm out lines.

This is what it looked like after the first retrofit. The gold parts in the control box are the servo drives, the main power transformer up top, aux transformer below that ran the old control, chokes below the drives. Also three contactors that controlled coolant, servo power, and a spare. The three Heidenhain boxes on the bottom are interpolators for the linear scales it had mounted for feedback on the original control. I initially tried to used the scaler but the performance was terrible.


DSC03386.JPG by macona, on Flickr


DSC02756.JPG by macona, on Flickr


DSC02758.JPG by macona, on Flickr

[Jerry]

Wednesday night I decided to dig in. My goal was to finish it by sunday as I need it for work this week. I put it on rollers and slid it away from the wall.

Thursday I gutted the machine. Pulled out anything to do with the servos, transformers, drives, scales, contactors. Yanked of the servos and pulled off the pulleys. Bored out the pulleys to 24mm and broached a 8mm keyway for the motors.

Y axis motor installed without a hitch. Did have to use a rotary file to remove a little more of the motor mount casting but I expected that. It bolted right into place. The Z motor was similar. Mounted in place of the old one. The X axis was a little more painful, literally, I managed to get myself really nice like while trying to get the pulley off. Since the new motors have a larger face on them I had to mount the X axis to the sides of the casting.

Friday I did the motor and encoder wiring. Ran all over town to try and find the wire I wanted, ended up getting stuck with MTW. The machine has sealed cable carriers and I managed to get the new motor cables and encoder cables in them to the control box. The Z axis encoder had to go externally. It was just too fat to fit through the conduit.

Saturday I mounted the drives and support pieces. I wired in mains power to the drives breakers, contactors, and finally to the drives. The newer drives control the buss power and shut it down if there is a fault. I finished the connections to the motors and encoders. The Z axis encoder amphenol was a pain. Must have been about twenty wires. Differential A,B,Z plus hall tracks for UVW. X wasnt so bad, by that time they went to a serial interface encoder and there are only a handful of wires. I tied the motors into the drives and the encoder cables as well. I got initial power to the Z drive and was able to jog the motor from the test mode.

Originally the machine was ran off 4 wire 208 so it had 120v off the lines. I did a no-no and was using the ground as a neutral. With the extra space I installed a .5kva 240 to 120v control transformer to run the computer, contactors, and lube pump.

That night I also got the cables ready. I reused the drive cables that came out of the molder. They have 50 pin MDR connectors on one size and 40 on the other. I took one of the cables and put the 36 pin connector for the MR-J drive on the end for the Z axis. While doing this I found a mistake on the board. So I cut that trace and jumped to correct it.

So, finally, this morning I did the control wiring. I needed to mount the interface boards. For some reason I neglected to put in mounting holes so I had to add those. I too a piece of 3/8" white PVC sheet and made a plate to mount the three boards to. This mounted next to the drives. I wired up the contactors and estop interlock for the drives.

I went into the computer cabinet and yanked out the old pixie boards. I reused the analog signal wires for step and direction signals.

Finally it was time to make it move. Powered everything up, grabbed the pendant, and turned it. Nothing. Huh. Eventually after some probing I found I had mislabeled the silkscreening on the interface boards. I connected the signal wires to the correct spots and X and Y moved! But Z didnt. On the newer drives you can enable the drive in the parameters. I had forgotten that the old drives have to be hard wired. Jumped that and still nothing. Oh yeah, left and right stroke inhibit. Jumped those. Now it moved.

But it moved very slowly with the native encoder resolution the X and Y were at 81920 pulses per inch. The Z axis was 64000 pulses per inch. I used the internal electronic gearing to divide those numbers by 8 on the X and Y and 5 on the Z. Much better.

All while messing around with it I managed to get it to run through the Z bottom tie rods on the balance cylinder. The had been welded together from shorter screws. I repaired one, still need to do the other two. I also made a couple other mods, installed a SSR running off the computers 5v line to kill power to the servo contactors when the computer is off. Also tied the Z axis servo-on line to the e-stop loop. The older drives dont have a dedicated e-stop line like the newer ones.

So far it looks like it is going to work great. The Z (Knee) is a whole different creature now. Moves quick. I had to have the old motor's acceleration all the way down to 3 in/sec/sec. I can easily run the new motor at 50.


IMG_0939 by macona, on Flickr


IMG_0940 by macona, on Flickr


IMG_0941 by macona, on Flickr


IMG_0944 by macona, on Flickr

[Doug Coulter]

Dunno about anybody else, but I find this pretty amazing. The only time I work with such large and complex blocks of functionality is in the software domain -- opsys-specific libraries, or hardware interface modules. And in that world, you really have to dig down inside the libs and find out how they do things, plus, you'd never start on Wednesday job you had to finish by Monday unless it was a darned trivial job. For me, this is a tough place to work - hardware modules that are supposed to abstract functions and be simple on the outside -- but rarely are when this complex a system with this many of them is involved. If a little grinding and filing was all that was needed -- that's almost miraculous to me -- you expect that, but then I'd expect a fight on getting brand A box to talk with brand B box that might require a full re-do, not just a little tweaking here and there.

I'm usually operating at some different level -- either at the bottom -- bits and low level software (self written opsys for example) or much higher -- I buy a car and just drive the thing.

Though I once did help a guy transplant a chevy small block into a ford pinto -- back in the day, the systems were a lot simpler and more self contained. That only took (other than the basic sweat) welding driveshaft pieces together and making some motor mounts, and a tiny bit of wiring -- it only took a few wires to get ignition and starting going on the new engine....

Hats off, Jerry.

applause.gif (99.66 KiB) Viewed 25740 times

[Jerry]

Servo motors are pretty simple, only a few different methods to drive them. Old systems used a +/-10v analog signal to command the drive and an encoder or resolver somewhere in the loop for feedback to the main control. Many drives made now have positioning controllers internal to the servo drive themselves. They will accept pulse inputs (Step/dir, quadrature, cw/ccw) or proprietary industrial network interfaces. Some are even programmable basically have a internal plc.

Oh yeah, Yesterday I inadvertently tested out how powerful the new Z servo motor on the knee is. Left the jack I was using to block the knee while I was working under it and homed. Snapped one of the two 3/8" tie bolts, pulled the threads off the other and bent the 3/8" steel end cap for the pneumatic cylinder. About a half hour with a hydraulic press and welded up the bolts.

I turned the max torque down to 50% after that...

Other than that goof it is working incredibly well. Its like a new machine. With the higher acceleration possible with these motors there is very little dwell time at the 4 reversal quadrants when doing a circle move. Where I used to get a little mark I get almost nothing now.

This is really phase one of the retrofit. Next step is to replace the spindle motor and varispeed drive with a 3.5kw servo I have and HTD belting to the spindle.

My ultimate goal is to eliminate that massive control cabinet. It takes up a lot of shop space around the mill.

[Jerry]

Little bit of CNC Porn (High Def on youtube):

http://www.youtube.com/watch?v=nmdSIzvh3gI

[JonathanH13]

I find this pretty amazing too. A truly enviable skill, the ability to translate software into precision cut metal! That's an incredible retrofit; looks rock-solid reliable.

[Doug Coulter]

Yup, made me sweat too -- must be pron. As you might guess, my wife still remembers those nifty Ti rings you made so it's dangerous to show her things like this

[Jerry]

BTW, the thing in the spindle in the last half of the video is a Nikken spindle speeder. It multiplies the spindle speed by 5. So 4000 RPM in, 20000 out. Its a high precision planetary gear box, basically. Another ebay steal. I think I paid around $300, they are about $3000 new.