Machining CNC mill conversion project begins

That's a seriously nice vice!

Into the "punch list" phase near the end of the build. This weekend:

1) I suffered a return of the sporadic stops. The ESS claimed it was an E-stop condition, but by watching the lights it really looked like the 24V power to the breakout board was blinking. So I went to Home Depot for the 723435th time and got some 22-2 shielded alarm wire, replaced the former twisted pair, and paid particular attention to the security of the connections. Problem solved. I had this same kind of cycle when everything was still out on the tray, and one thing needed when installing the tray was to disconnect/reconnect the power leads to the two digital boards (ESS and BOB). After realizing this, I looked again at everything with a very suspicious eye, and homed in pretty quickly on these push/latch connectors on the BOB (the one with the red/black power leads).

Some experimentation showed that these connectors are problematic in a couple of ways - the most insidious being that you can have the wire fully inserted without the gripping part of the connector being fully engaged. There is no obvious visual indication that it's bad. This problem became a lot more observable after I tinned the ends of the stranded wires, which made it much easier to tell if the wire was not secured. The second issue is that even with the latch properly engaged on the wire, it doesn't take a ton of force to yank them out. The next time I take the tray out of the electronics box, I'm going to remove all of those connectors from the BOB and either replace them with screw type terminal blocks, or put on hard-soldered leads and a JST connector. I'm now pretty confident that an intermittent contact in this connector was the source of the fault events both times. Along the way I also replaced the ESS 5V power lead with the shielded wire, but that one has screw type terminal blocks, and I never saw its LEDs go out anyway.

I don't have a picture, but when setting up the shielded power leads, I was careful to solder the shield braid lead to ground at the power supply end of each power cable.

2) Installed the rubber way covers since I'm starting to make a lot of chips fly around. Aluminum chips will travel amazing distances.


I changed the routing of the oil tube on the far side a little bit so the rubber sheet wouldn't push it around.

3) I found a dimensional problem with the X table - not a total showstopper but kind of serious. In short, the middle T-slot width is out of spec, narrower than the top/bottom slots, and will not admit hold-down hardware that is accurately sized to the nominal 14 mm slot size. This showed up when I installed the Kurt "sine key" alignment fixtures on the bottom of the vise, and found they would not go in the middle slot. Basically I need to have the slot widened...will be contacting Precision Matthews about this tomorrow.

4) Finally I used the mill a bit to work on the limit switch brackets. In the picture the Z bracket is at top, the -Y (aft) bracket is bottom left, and the +Y (front) bracket is at bottom right. That one is longer since it has to extend out about 4 cm beyond the front of the base casting.


The limit switches have already been electrically integrated and tested, so the only things I really have left to do on the build are to fab the X brackets, drill/slot/install them, and charge up the oil system and fix the inevitable messy leak.

BTW it's hard to say enough good things about the VistaCNC pendant, and I've still only used half its features. I almost bought the version without the screen, but I'm REALLY glad I got the version with the display. With the pendant and a few lines of hand-hacked G-code the bracketizing is super easy. At slow feed rates I'm getting fantastic surface finish too. You can kinda see this on the chamfer cuts at the lower side of the two Y brackets. I did the left one at 500 sfm but a pretty slow feed rate and you literally cannot see any mill marks. I did the right hand one about 30% faster and you can just barely see some mill marks, but you cannot feel them.

Lastly I ordered the Vertex 6" rotary with chuck and tailstock. A couple of the pieces have to come on a monthly shipment from Taiwan, so it's going to be a couple of months before I have to work on them.

Tonight I bit the bullet and charged the oil distribution system with the Vactra 2. Everything worked as planned and I only found one tube that wasn’t pushed in all the way. As expected, the Z side filled up first and I had to keep pumping a few more strokes to fill the XY side, and wipe up some excess from the Z axis.

The clear tube was totally essential to see what was going on, and I’d strongly recommend it for any one-shot oil project. With opaque tube it would be really hard to see when oil reaches the X ways since they are never exposed to view.

There's been a gap in updates on this thread owing to the demise of my home MacBook. All the photos from my phone are finally backed up onto Google Photos so I can post from any computer now

Here's a situation I ran into that was kind of instructional, though it cost me a couple of evenings of work. First a bit of a clue...


and what happened soon thereafter:



Mistakes were made. Approaching the lower limit of main Z travel, the bellows style way cover piles up into a *very* solid stack, which slows down the motion enough to completely conceal the fact that you have reached the hard end of motion. Combine this with the shaft coupler halves being really tight on both the leadscrew and the motor shaft, and the burly servo exerted enough force to completely mangle the end face of the angular contact bearing as it pushed the large nut right through it. So I once again rendered the mill headless, taking off the motor and electrical box, and pulled the Z axis. To my surprise it went a LOT faster than before, taking somewhere under 90 minutes.

ArizonaVideo99 was kind enough to send another bearing (I only needed one, the lower one was intact - size is 15 mm ID x 35 OD x 11 long). Installing it was a bit interesting. The nut is intentionally very tight on the threaded end of the screw, and you have to clamp the screw in a vise (with brass shim stock around it!) to unscrew the nut. My old cheap bench vise did not have anywhere enough clamping force. It took me a shamefully long time to realized there is this nice new Kurt 6" mill vise with idiotic amounts of power...



Problem solved. It took about 2-2.5 hours to reassemble the Z axis and put the head back on. To try to prevent a recurrence I did a few things: a) got rid of the Z bellows, to be replaced with a flat rubber sheet that won't interfere with motion; b) slightly bored the shaft coupler half that goes on the ballscrew and reduced its tightness so that it may slip before trashing the bearing; and c) marked the actual travel limit on the column, which is around 15.1" on the graduated scale.

This was a good chance to measure the Z travel fairly accurately. The column travel is 15.4", and the quill goes an additional 2.98" (based on its DRO), so the absolute Z envelope of the machine is over 18" (!!). You do have to remember to extend the quill if you're going to machine something that is right down on the table, otherwise you can't get there with the column motion.

In a small bit of forward progress, I cut a pair of X limit switch bracket blanks and did some testing with the switches detecting the bolts I installed on the Y and Z axes. They work fine, but you have to have them within ~1mm of the sensor to activate. I also found they'll trigger on a piece of aluminum almost as well as a steel bolt.

I'm heading for NARAM-61 tomorrow, so there won't be any more updates for the next week+. While I'm there I'm going to work on the CAD for my first set of custom av-bay lids for Airfest.

One more update before I go to Muncie - I received and installed the thinner Y motor mount plate. It's quite a bit thinner than my hacked original one, and the motor shaft now fully engages in the coupler.

Old fashioned mechanical limit switches... very reliable! Nice Z travel.

caveduck said:

Here's the PM-30 steel base on the completed Grizzly mobile base. It seems pretty stable even when I get up and stand on top. You need to adjust the height of the deployable rubber feet carefully so that the base is close to level when they are deployed. It doesn't have to be perfect, the floor is not totally flat anyway and fine tuning will have to be done at the final location.

View attachment 381505

Click to expand...

Are you happy with the mobile base? Is it pretty solid? I am thinking of doing something similar with mine. Also is there some reason you oriented the wheels on the back the way you did?

Thanks,
Brendin

@xnaron The mobile base is OK; there is a little bit of wiggle but it doesn't noticeably affect the results of the mill. I put the back wheels in the way that favors the most frequent way in which I need to shift the machine (laterally). If I had it against a wall I'd have them the other way so the machine could be pulled straight out.

A more general update - I'm prototyping a Centroid Acorn based controller rig because my breakout board for the ESS is kinda flaky and the vendor (CNC4PC) is showing signs of fading after many years of being the gold standard.

There is not really enough room in the existing enclosure for modifying/debugging the digital system wiring nor expanding it in any way, so I'm planning to pull the existing digital rig out onto a DIN-rail-on-plywood and do a flyoff before repackaging into a secondary enclosure. I'm also hoping to put in a VFD to drive a new Marathon spindle motor with way more RPMs (should get me to 10,000 rpm) and encoder for rigid tapping, and I think the VFD will go into the existing enclosure where the 240VAC lives once the digital stuff is relocated.

Time for a real update to consolidate what I've done in the last few months.

• I went ahead to pull the digital control system out of the big box and onto its own plywood board to get room for expansion and access for debugging / experimentation. That went very quickly and it's done and working.
  • Everything is now DIN rail mounted. I used DIN clips from Electronics Salon together with some strips of .090 fiberglass and #4/#6 standoffs to adapt the various boards
  • New rail-mount DC power supplies from Automation Direct.
  • DIN rail terminal blocks "DINnectors" (also from Automation Direct) to eliminate the rat's nest of soldered DC power leads. They are very nice.
  • Added a Weidmuller terminal block for unswitched AC neutral to eliminate a Y cable.
  • The future VFD is not going to fit comfortably in the main servo box, and there's not enough room inside the PM mill head e-box either.
  • TODO: bump the 20A breaker to 30A - the 20A one trips sometimes on the inrush current. It has never popped during operation though, even when I've stalled a servo.
• Got an Acorn controller and SW and started making a swappable controller board for that. Will have to build 4 new custom driver cables though. Planning a "flyoff" of sorts - I'm aiming to make it possible to swap the ESS and Acorn controllers in a matter of a few minutes. I think there are going to be about 10 connectorized cables to hook up.
• Built a vertical mounting plate for the Vertex rotary, with a milled-out arc so you can actually swing a 9.0" diameter object on the fixture plate.
• The last bits of flakiness (C62 BOB goes crazy once every hour or two) are getting eliminated with an AD FC-33 signal conditioner/isolator on the analog spindle output. Turns out the Chinese DC motor controllers like the ones in Precision Matthews machines don't provide any isolation of the analog speed control (driven by a pot) and the rest of the DC. I noticed this way back when I was reverse engineering the controls but didn't think then it would be too bad, and it *almost* works Likewise, neither the C62 nor the Acorn has built-in isolation on the analog output. When I started researching the Acorn I finally stumbled on a video on Franco's YouTube channel where he talks about this, in Acorn context. Meanwhile I've been running the spindle speed/dir manually with 100% stability, though it has made me really appreciate having the spindle automatically shut off at the end of a long gcode run. A side benefit of the signal conditioner is that it will trivially convert the 0-10V standard speed control output to the slightly wacky 0-5V used by the DC controllers. Right now I have the C62 adjustment pot cranked down as far as it will go and it's still 2-3% off.
• Put an active line conditioner in front of the 120VAC inputs to the digital PSUs, and totally separated the power on/off controls from the servo box. This turned out to be handy because if you fault a servo, the C62 doesn't have the ability to control the ENA signal and the only way to reset the servo is to power off its driver. Formerly this would also shut down the ESS and you would lose a lot of state and have to restart Mach4. Now that they're separated I can just toggle power to the servo box and leave the digital section running.

I have revised block diagrams almost done; they should be ready to post in a day or two.

Here's the updated block schematic of the digital section for the ESS+C62 controller.

Notable changes:

• Change to DIN rail mount +5 and +24 VDC power supplies
• Add AC line conditioner in front of DC PSUs
• DC PSUs are no longer gated by the main 240V breaker/switch, now independent of the servo power
• Show new E-stop interface box for easier switching to/from Acorn controller
• Add signal conditioner on analog spindle speed control output
• Show separate wired Ethernet link from Mach4 computer to ESS
• Show VistaCNC USB jog pendant, remove placeholder for old style pendant
• Removed C55 dual 20A relay card; it needs 12VDC and it's not needed yet

And here's the updated drawing of the 240VAC power section. Sadly, on TRF "full image" doesn't mean it's full size.

Changes

• Add terminal blocks for unswitched line/neutral. This lets the aux 120V outlets be always on, handy for the computer charge cord.
• Add ground bus bar. It's always been in the build but wasn't on the drawing
• Correct the wiring design for the optional logic-controlled power relay bypass switch
• Add DIN rail terminal blocks for the DC PSU outputs
• Show case fan 12V wiring
• Show inline fuses for DYN4 logic section as not implemented. They don't protect against a very common scenario and I hate fuses.
• Add note about needing to bump the breaker/switch to 30A.

Here's the Vertex Taiwan 6" rotary stack assembled. Nice gear at reasonable prices; was worth the wait. At left is the A axis servo motor that will eventually drive it.

And here is the Haimer 3D sensor in action, being used to measure the dimensions of an oval slot. It does edge/surface finding in all 3 dimensions to high accuracy. Combined with the VistaCNC pendant, you can sense and zero the X and Y offsets for a part in 30 seconds or less.

The new DIN rail mounted ESS+C62 control rig after extracting the core of it from the big box. The all-important signal conditioner for the 0-5V analog spindle control signal is at lower right. You can just barely see the corner of the in-work Acorn rig at upper left. Flakiness is down a couple orders of magnitude.

Here are some recent updates

• Put on a Noga mist cooler unit and plumbed air back to the compressor on the other side of the garage. If it's too misty I'll spring for a Tormach Fogbuster spray cooler, but the Noga was not very expensive and I figured I'd give it a try.
• Added a chunk of 5/8" ID water hose (the kind with the diagonal helix braid reinforcements - found at HD) to the coolant drain barb.
• Added an LED worklight.
• Changed out my C62 BOB to the newer generation C82 after concluding that it had probably gotten damaged back when I blew out the original ESS. That turned out to be correct as all the flakiness has entirely disappeared. The digital schematic is now slightly out of date; the ESS is now powered from the C82 and thus there is no need for a separate 5VDC supply.
• Changed from native R8 tooling to TTS indexable holders using the short R8 adapter from Tormach. Seriously I should have started this way. Combining with deterministic homing, Z touchoffs are eliminated except when initially finding the height of the fixtured part. An unexpected side benefit is that tool changes are *drastically* faster since you don't have to fully unthread the drawbar on every change. Now it's maybe a half turn...I no longer feel any need for a power drawbar.

So the remaining hit list is down to

• CNC the rotary
• Finish off the Acorn digital rig and do the flyoff
• Replace spindle bearings with ones rated for 10k rpm and make a video, because no one has ever posted a complete one
• Change motor to a Marathon 5400 rpm (max) 2HP inverter duty + VFD. With the 2:1 pulley it should hit 10000 rpm and go thru aluminum much faster

11 years after Dave's last append above, I just wanted to say "wow"! I have a PM-30MV on order and looking to CNC-ize it, stumbled across this thread (in a rocketry forum?!) and have spent the past 2+ hours reading every line, top-to-bottom. This was an amazing amount of detail and tremendous help. I leaned a TON reading through this. THANK YOU, Dave.

@gregreid It isn't 11 years, that was 2020! I posted it here because its main purpose at the time was to make rocket parts and I don't have a presence on any machining forums I basically spent most of 2019 building it.

I've made some upgrades lately - split out the digital electronics into a 2nd Wiegmann cabinet, built up a breadboard with a Centroid Acorn controller (planning to ditch Mach4, it's totally unsafe with probes), fixed the janky way I originally made the cover plates for the power hydra, got some great tooling, and other stuff that I've forgotten already. I really need to do some updates to this thread.

Good look on your PM-30MV conversion!! I hope Dave at ArizonaVideos is still making those great ballscrew conversion kits.

caveduck said:

@gregreid It isn't 11 years, that was 2020! I posted it here because its main purpose at the time was to make rocket parts and I don't have a presence on any machining forums I basically spent most of 2019 building it.

I've made some upgrades lately - split out the digital electronics into a 2nd Wiegmann cabinet, built up a breadboard with a Centroid Acorn controller (planning to ditch Mach4, it's totally unsafe with probes), fixed the janky way I originally made the cover plates for the power hydra, got some great tooling, and other stuff that I've forgotten already. I really need to do some updates to this thread.

Good look on your PM-30MV conversion!! I hope Dave at ArizonaVideos is still making those great ballscrew conversion kits.

Click to expand...

Interesting observation on Mach4. My CNC router (wood, not metal) seemed to be pretty decent with Mach3. What specifically is the bugaboo with Mach4 and probes or is it just the fact that your machine is more heavy duty compared to CNC routers? Just curious as I was considering a project that would have likely used Mach4! I might need to reconsider!

Sandy.

I know Dave personally. I also know he went to school on a 'special' bus. Not like mine. His was a stretch Rolls complete with chauffeur and wet bar. He can get any shade of Poupon he wants. To that end, his third grade teacher was Thomas Edison. I have also heard him speak fondly of Paul Dirac and Richard Feynman. He once gave Schrodinger some catnip. This is the only CNC in the free world to operate at the quantum level. The main power source has been upgraded to fractionated Dilithium Crystals thru sub-etherian worm holes. I struggle just to say hi to him. I just figured out the different colour of sticks in Tinker Toys are different lengths.

Sandy H. said:

Interesting observation on Mach4. My CNC router (wood, not metal) seemed to be pretty decent with Mach3. What specifically is the bugaboo with Mach4 and probes or is it just the fact that your machine is more heavy duty compared to CNC routers? Just curious as I was considering a project that would have likely used Mach4! I might need to reconsider!

Sandy.

Click to expand...

The issue I had with Mach4 probe macros (and I found other references out there about it) is that it would sometimes make a sudden unconstrained downward rapid movement. I've had it happen twice and except for a piece of good luck I would have broken my KP-3 probe already. This is some kind of bug in the software. Also Mach4 has no provisions for limiting probe movement based on a hardware probe detect input like Acorn does. Mach3 is a completely different piece of software; Mach4 was an attempt at a rewrite but it has never really settled down to a stable release as well as Mach3 did. I'd recommend looking at Acorn or LinuxCNC for new projects, or continuing with Mach3 if they are still selling licenses.

Centroid Acorn is adapted from an industrial grade product that is pretty widely used to retrofit old heavy duty CNCs where the electronics have aged out. It has good probe protection features like *never* allowing a rapid move when the probe detect signal is active.

@fyrwrxz Scout's honor I did *not* give Schrödinger that catnip. But I happen to know that his cat is alive and well sipping umbrella drinks somewhere in the tropics and smirking about the mice he stuffed into that box.

Reviving this with some updated design info. My original art for the block diagrams died with a previous computer and I'm re-doing them in draw.io by looking at the old ones in this thread. First one done is the as-built power section. There are a couple of corrections to the part numbers, removal of the unused bypass switch, added the line isolator in front of the digital section power, added missing connections to the relay coil, and more pictures.


Edit: I think I never mentioned before the rationale for the DC power supply triggering a relay to power up the servo motors. The reason for that is to allow for a future upgrade that will permit switching off the motors while leaving the DC enabled for servo brakes. The way things have turned out, a brake would be very helpful for the Z axis which has a tendency to start sinking when the servo holding torque is turned off. Next time the Z servo dies (I already had one die about a year ago) it will get replaced with one that has the brake.

Here is the block diagram for the digital section, updated to show the CNC4PC C82 breakout board. The C82 is very different than its predecessor and when fully populated, a majority of the C82 is covered by daughtercards. The Warp9 Tech Designs ESS (Ethernet SmoothStepper) plugs in atop the left half of the C82, and is powered by the C82 - a major improvement. The right side of the board carries a CNC4PC C78 "connection card" that brings all of the I/O ports out onto RJ-45 connectors. A full ESS+C82 rig looks like this:

but I don't have an overhead view of the combined assembly, so the diagram below just uses a pic of the bare board, with annotations for the daughterboards; their connectors are shown in light blue. Besides switching to the C82, things that are different than the previous diagram are

• No dedicated home switches. Homing Y to the front limit and Z to the top limit has worked out fine. It's slightly annoying that X has to home out to one end but it's not that bad. I rarely do much in absolute coordinates anyway.
• In the limit switch array, I ended up using A+/A- for the Z axis so that I could have limits at both ends.
• Added the touchscreen monitor on the column. It has been really handy.
• No more separate 5VDC power supply; the ESS is now powered from the breakout board.