Converting a CNC boring machine to full 3-axis

[kmagyoyo]

Here’s the Z motor and ballscrew assembly. This is the simplest place for it but it might interfere with the spindle location if the spindle is too wide. Guess I need to settle on a spindle.

[kmagyoyo]

I drew up the spindle mounting plate, which was a great way to learn Fusion 360. Sent it off to my machinist buddy and should have it back in a week or so.





Have most of the parts collected. I’m getting excited about putting this contraption in motion!

[kmagyoyo]

Had a buddy mill the adapter plate for the spindle. Dude makes parts all day long for Lockheed and SpaceX. I didn’t ask for aerospace tolerances but he doesn’t know another way.

Tapped a few holes in the Y-axis car to allow mounting the ball screw bearings.

The hardest part of all this was measuring the existing hardware to design the spindle adapter plate. Getting the offset on the ear that connects to the ball screw nut was particularly tricky. I don’t have anything except a digital caliper so it’s a near miracle that everything fits as well as it does!

I can already tell that tramming the spindle is going to be a PITA due to access to the mounting screws. Guess I’ll burn that bridge when I come to it.

[kmagyoyo]

Z-axis is complete, video shows auto-tuning of the Clearpath SDSK servo. That software is some CRAZY magic!

Weight of Z car and spindle is supported by the machine's pneumatic ram that originally was the only z-axis motivation, now it makes a perfect weight compensating spring. Pressure at the regulator is set to 5 PSI, although it doesn't bleed off excess pressure when the cylinder compresses..is the regulator malfunctioning or just the wrong regulator?

The spindle is also hooked up to the VFD and running. Wired per the instructions, it ran backwards..but no biggie, just swap two phases. Noticed some odd bearing noises during some break-in runs, hope they go away with some run time. Spindle documentation is sketchy at best.

If I can sort out the gremlins with the X-axis encoder, this machine will be ready to be tuned.

[kmagyoyo]

Started tuning the X and Y servos, the plots in Tuna were a mess. A couple things were wrong here (RTFM) but mainly it turns out that because the encoder is on a belt driven screw drive NOT directly mounted to the motor shaft, to quote Machdrive "to the high frequency servo algorithms [the belt] appears as about as rigid as a wet noodle." Not sure yet if the encoder is the right size to mount to the motor, but that's what we need to do..

[kmagyoyo]

IT'S ALIVE!!

I can now jog all 3 axes via the computer. A couple gremlins have appeared:


-The Y axis doesn't stay in position, it will randomly move ~.050" in either direction at idle. Possibly a tuning issue?

-The Clearpath servo on the Z axis shuts down intermittently and throws a step timing error. Suggests noise or loose connection. I improved grounding from console to machine chassis (that is, I remembered to reconnect the big 10 gauge ground wire), we'll see if that helps.

[kmagyoyo]

Long story short, I decided to abandon the DC servos and switch to Clearpath SDSK servos on all 3 axes. Tuning the servos was challenging to say the least, and I suspect the low-resolution encoders I was trying to reuse were to blame.

The Clearpath servos are really amazing, I can't say enough good things about them. However, I have a big problem with EMI on the step signal. I was able to eliminate the issue in my previous post by adding 1k resistors across the st/dir signal circuits per the Clearpath instructions for long connection wires.

Unfortunately as soon as I turn on the spindle, all 3 servos shut down due to a "step signal timing error". The step signal indicator in the Clearpath config software blinks erratically as well.

All signal wires are shielded but the spindle wiring is not. Grounding appears to be very good and all wires connect to a single ground bar in the cabinet. I've been careful to isolate signal grounds from earth grounds. Would a shielded cable between the VFD and spindle solve this problem?

[skrubol]

Would a shielded cable between the VFD and spindle solve this problem?

It couldn't hurt, and done properly has a good chance of solving the problem. You will want a cable with 4 conductors twisted (3 phases plus earth,) and a shield. I can't recall off the top of my head which kind of shield is best for this (braid is heavier duty, but foil has not gaps.) IGUS has some really nice continuous flex motor cable, but it ain't cheap. Also you'll need to make sure to connect the shields properly. There's some debate whether to connect only at one end, or at both, (I think the consensus these days for motors is to connect both,) but the connection needs to be robust, There are shield cable clamps that are optimal for shunting that noise to earth.

[kmagyoyo]

AutomationDirect has this cable specifically engineered for VFDs. I checked out the IGUS, no kidding it's expensive! The AD stuff is about 1/3 the price.

https://www.automationdirect.com/adc...WG%22)&start=0

[skrubol]

AutomationDirect has this cable specifically engineered for VFDs. I checked out the IGUS, no kidding it's expensive! The AD stuff is about 1/3 the price.

https://www.automationdirect.com/adc...WG%22)&start=0

Depending upon your usage, the advantage of the IGUS chainflex (and there are others,) is it's rated for continuous flex at a relatively small bend radius. If your spindle motor doesn't move, this doesn't matter. If the cable goes through a cable chain, and it's going to get much use (important consideration for a hobby machine,) it might be worthwhile to spend a bit more. The data sheet for the AD 14ga stuff shows a bend radius of 7". You can assume that to mean static. The IGUS stuff has derating tables to show how many cycles in a cable chain a specific cable is rated for vs. bend radius.

I've never actually used the stuff, but I've got a router build in early phases (collect random crap at scrap prices and planning phase) and I plan to go with something that has dynamic bend ratings.

[davida1234]

The DA cable is also 14ga which I don't think is necessary for a 2.2kW spindle. Could be more difficult to install into the connector.

[klaas123]

This is 16 awg and very flexable: https://www.aliexpress.com/item/4000...383d2e0eXyZadt

[kmagyoyo]

The DA cable is also 14ga which I don't think is necessary for a 2.2kW spindle. Could be more difficult to install into the connector.

I was on the fence between 16 and 14ga. Not sure what the max amperage on my spindle might be but if the voltage is 220 then it must be 10A or less. OTOH the wire run is long, I estimate 30-35'

AD does have the same cable in a 16ga.

[kmagyoyo]

Just got the 16ga shielded VFD cable installed from Automation Direct. No change. All 3 SDSK servos immediately throw a "step timing error" fault as soon as the spindle is started. VFD cable is shielded end-to-end within 1.5" of termination and grounded at both ends. I moved it away from any signal wires as much as possible, of course it will always share the cable chain.

I tried lifting the earth ground connection to the Acorn board, no change.

[skrubol]

Do you have any electronics lab equipment (oscilloscope or something similar?)
It could be something to do with supply power as well (power to the VFD).)
I assume you are controlling your spindle from your motion control? Can you disconnect it from it and try enabling the spindle manually from the VFD and see if that makes any difference?

[kmagyoyo]

I have attached an "as built" schematic. VFD control is hardware only.

ALL 3 drives instantly fault when the spindle is started. This tells me that there might be a spike or dip in the 24V when the spindle starts, I am going to try a different 24V supply and moving the AC supply to the 110VAC leg that is not used by the VFD.

I did consider that the earth ground on the Acorn board might be the source of a ground loop, I lifted it to no effect.

[handlewanker]

I would have thought the first strep would be to drive the spindle from a completely separate power supply, that is completely separate......I get the impression from your characteristics that if you get a problem then address it at that particular event.......the spindle is obviously causing a power dip on your supply line, and that might be a factor of your power supply to the house.

We had this problem with our NC machines in Bristol UK in the 70's and it was caused by a large capstan lathe braking and throwing a spike on the line.....this was cured when the power company installed a new heavier transformer for the shop supply from the grid.
Ian.

[kmagyoyo]

Here's what I've done to mitigate the "step timing error" on the Clearpath servos:

1. Switched unsheilded VFD cable (VFD to spindle) for a shielded cable from Automation Direct. Separated cable from signal wires everywhere except the cable carrier. No change.
2. Added line filter and ferrite rings to VFD per this video. Instead of stacking multiple rings on the wires, I looped each wire through one ring once. This solved the issue with the VFD shutting down all servos.


3. Added 1k pull-up resistors (total 6) to each step and direction output. This made the machine more stable regardless of spindle on/off.
4. Wired the Meanwell DC supply for 220V input, eliminating the need to use the earth ground as a neutral. No change.


There is still a stability problem, the Clearpaths will still throw an occasional "step timing error" even if the spindle is not running. At this point, I'm out of ideas. Hoping someone will catch something I didn't see. I have attached my "as built" schematic.

[kmagyoyo]

The machine is now stable. I replaced all control wires to the Clearpath servos with stranded CAT6, preserving the +/- pairing for each signal within each twisted pair. I did not reinstall the pull-up resistors at the Acorn outputs, if I have any instability in the future I will add them back. Here’s the machine making chips like a boss, using the machine’s original vacuum pods for hold down.

[nlancaster]

nice