2-axis Sharp knee mill revival - Now with 50% more axes!!

[FlySox]

Back in 2011, I bought a Sharp knee mill with a broken 2-axis CNC controller (X & Y). I'll never forget taking this beast off the flatbed truck with a rented forklift, using strategically placed railroad ties as a lifting point. That was a nail biting day, especially as I heard one of the ties crack with the mill 5 feet in the air! Managed to get it down a long slope to the garage door, and pipe rolled it into place. Hooked it up to a VFD and had it running as a manual mill that afternoon. Joy!

It was bought with the intention of an eventual CNC revival (nice to have someone else mess around with ball screws, thank you very much), but raising kids and starting a new business got in the way. The Mach3/Mach4 universe at the time was good, but complicated and required a time commitment I just wasn't able to make. Plus, I found that I could get most everything my projects required accomplished with just manual milling, so it never happened.

But I recently stumbled across the Acorn (and Marty's videos) and have decided the time has come. I'll document some of the process here, since I have enjoyed such threads from others. Plus, I hope putting this out there will put extra pressure on me to actually finish this project.

Here is the patient, as it sat just prior to my first weekend's efforts (needing a good cleaning):

The motors are Servo Dynamics MTS30M4-38 with the following specs:
Stall torque: 3NM @ 8.6A
Max: RPM 3700 140V 47A

Here's the old control box, jam packed with early 90s technology. Note the 3.5" floppy in the upper left!

Here's the old operator interface, a Newall SmartPath, complete with cathode ray tube.

Control box, mostly gutted. I'm going to keep the power supply. The toroidal transformer is clearly marked as having a secondary output of 70V @ 850VA. That should be around 70V * 1.414 = 99V (minus some small loss from the bridge rectifier). Too high for Geckos, which is why all those years ago I bought Dugong drives from Balazs Klincsik at CNC Drives (the updated version is now known at the DG4S-16035).

After drilling a VESA hole pattern, I mounted a 24" touch screen on the old swing arm, along with a waterproof keyboard and an Intel NUC:

Couldn't resist hooking up the Acorn to see what it's going to look like:

...but the dream of using that interface is way premature. First, I need to tune the drives to the motors. This is where the project is now stuck: When I try to PID tune via CNCDrive's Servoconfigurator3 software, the motors just spin forever. They will stop when issued the "Reset" command via the software, so there's obviously some sort of communication happening, but it's as if the encoder signals are not getting through.

I think the encoders are just incompatible. I've tried both X & Y motors to see whether it was maybe just one bad encoder. I've swapped out the Dugongs, in case maybe one of them was bad. I carefully traced the encoder wires using the continuity signal of my multimeter, and for good measure have tried every possible encoder wiring configuration, too.

The encoders *seem* to be quadrature incremental encoders with A/A- B/B- & Z/Z- signals. They are SEM 1000ppr encoders with a designation of 747-1-46417 # 7B1032. These things are so old and obscure they have somehow avoided any sort of interwebs capture:

I hooked up a different motor with a different encoder to the drive and *was* able to read it correctly using the Servoconfigurator3 software. This convinced me that it wasn't a wiring/software/firmware issue, but rather likely the encoders. I broke out my oscilloscope, and tried to get a plot of the encoder signal as the motor was endlessly spinning, but I can't be sure I even had the scope set up right: the signal was just a bunch of noise, not the sharp offset square waves I was expecting to see.

Anyway, I've ordered new encoders that will hopefully get me back on track. Meanwhile, I'll work on a control box layout until they get here. I'll update this thread with progress.

[cncman172]

Looks like a fun project. Those old encoders look like they came from England with that 44 international phone number. The label seems to imply they work from 5V to 24V. You should be able setup a scope on A or A/ or B or B/ and turn the encoder slowly and you should see a square wave output on the scope.

Russ

[FlySox]

I've got a roll of Ram Board (dense cardboard used to protect floors during construction) left over from a recent remodel, so I decided to try a different kind of CAD: Cardboard Aided Design.

Cut out to approximate dimensions of my control box:

Added placeholders for the existing toroid and capacitor, and folded it up to check size:

Unfolded and laid out main parts for four axes (gotta plan ahead, right?). The Dugong drives are the big boxes on the sides, along with their braking resistors (gold colored). The four small squares at the corners of the Acorn are dual RJ-45 breakout boards, since the Dugongs take encoder and step/direction inputs via RJ-45 jacks. I've got a DB-25 breakout board attached to the Acorn to take advantage of the 5VDC signals provided there, since the Dugongs are allergic to the 24VDC signals that come via the screw terminals (this was confirmed in a different thread; yay search function). In the photo, I placed a spare toroidal power supply at bottom as a placeholder, but the plan is still to use the original power supply, which is still in the controller box.

I've now got a better sense for what size aluminum sheet to cut for permanent mounting, and will size out cables and wires once I finalize things (e.g. home/limit switches). Doesn't look like I'm going to be able to fit panduit in there, so I guess there will be a lot of zip ties in my future.

[martyscncgarage]

Did you do your motor testing and tuning with the CNCDRIVE software FIRST?
When you say the motor just runs, if you used the software without power to the motor you should be able to see the encoder counts forward and backward.

You MUST follow the manual and use the software to set them up first.
The encoders sound ok. What you describe is a runaway. You can try and reverse the motor leads on the drive and see if the motors stay put.

[FlySox]

Hi Marty, yes, my tuning attempt did not involve the Acorn at all. It was done purely with the Servoconfigurator software, a bench power supply, and the drives. I did swap the motor leads, since I had read that this can fix runaways, but that didn't help.

I'm curious about your suggestion: "if you used the software without power to the motor you should be able to see the encoder counts forward and backward." Are you saying just power the encoder (5VDC), leave the motor unpowered and spin the shaft by hand to try to see an encoder signal? I can try that. It's weird, though, that I have been able to get a different motor and encoder combo to work, which suggests my setup is correct. If I saw encoder signals using your suggestion, I'm not sure where that would leave me. I guess it would leave me searching for something I've somehow overlooked. If that does turn out to be the case, I'll take photos of the entire setup and see if the helpful eyes and brains of this forum can point me in the right direction.

[martyscncgarage]

Honestly it's been awhile since I set those drives up. I did have to change encoders on my motors before things stabilized.I think Chaz has a similar setup. Do a search for him.

Marty

[converterking]

I have used 5 volt differential encoders and never had a problem. I believe the 5-24 volt encoder is your problem.

[FlySox]

Theoretically, that 5-24V just means the encoder is happy to be powered with that range of voltages, meaning if it is fed 5VDC (as I'm doing), it shouldn't be a problem.

However, I do think this encoder is bad. I hooked up the motor and encoder that I know works (unfortunately, not a motor suitable for this conversion) to my oscilloscope to confirm that I can indeed read a known good signal with my set up. Here's what it looked liked after a spin of the shaft by hand, with the familiar offset A (blue) & B (red) square waves:

And here is what the SEM encoder looks like:

It's just not sending any pulses; it always reads a straight 5 volts regardless of whether or not the shaft is spun. I ruled out any possible wiring issues from encoder to control box (e.g. a short of some kind, even though continuity tests suggest all is good) by pulling the encoder off of the motor and wiring it directly to my test equipment. Still no joy.

It's *super* suspicious that both encoders are bad, so there's still a chance I am just missing something obvious, but for now I think this part of the conversion is on hold until the new encoders arrive.

[martyscncgarage]

Odd indeed. But do you know the history of the machine? Was the old control working?

Try automationdirect.com for encoders. All depends on your servo shaft size. Centroid does stock a good variety.
Marty

[FlySox]

I don't know the history of the machine, except that the control was not working. You're right: for all I know, they put mains powers into the encoders and blew them both out, and the subsequent runaway is why they decided to get rid of the machine.

Thanks for the encoder supplier suggestions.