[the_canuck]

I'm taking all next week off to hopefully spend quality time with my beast. I hope to get most of the parts I need to make for the conversion finished. My day job has been crazy lately and haven't have much motivation to spend my little spare time working in the shop.

Andrew

[the_canuck]

Progress on the ball screw nut brackets. I have a bit more work before tearing down the machine and installing the new parts.

Andrew

[the_canuck]

Here's some info on my setup for AL cutting.

I used a decently sharp but used 1/2" 2 flute HSS end mill running the spindle at 10K and mostly climb cutting. I was taking 0.1" depth per pass.

Cut speed was 20ipm.

Surface finish came out pretty decent. On cuts where I did a finish pass it was excellent. These parts are to be functional so I didn't do finer finish cuts on the pockets that will be hidden once installed.

I used WD-40 as a lubricant which I sprayed on every couple of minutes. It did make a difference over dry cutting on surface finish. It does smoke some so I had to run my air cleaner during the cuts to keep the shop air from getting too foggy looking.

Yes, my machine is really loud when cutting AL. as the frame and table seem to amplify the cutting noises. Hearing protection was a must. You could easily tell if you were getting chatter which I was trying to keep to a minimum without slowing the process down too much.

Andrew

[the_canuck]

Drilled the holes in the pulleys tonight. Still need to trim down the lip/shoulder and then finally install on the nuts.

Andrew

[Arbo]

Nice work. I need to get back to using mine. Though I will probably never do aluminum, though who knows.

[the_canuck]

Pulleys are ready. I figured out if I use a very thin o-ring the pulley will self center to the ball nut as I torque in the mounting screws. The other spacers and mounting brackets for the shafts are also made up and ready to install. Next step is to take apart the machine and hope everything fits in place.

Pulleys were a challenge as my 1/4hp harbor freight drill press was barely up to the task of a 51/64" hole. I progressively drilled out the center hole with bigger bits to make it doable with my flimsy drill press. I would have drilled them on the machine but I didn't have the z clearance for the bit. I guess I could have just milled it. It would be probably better centered than my drill press effort. Lesson learned I guess but they came out usable.

The flange on the pulley OD is smaller that the ID of the flange on the nut by 0.015" so to make them easy to install and self-center I came up with the o-ring idea. If I'm careful torquing them on the o-ring should help keep the pulley centered on the nut.

Andrew

[the_canuck]

Machine is moving under its own power again with new ball screws installed. I have a few more things left to finish up but it was a good milestone to finish the main part of the new parts install.

[Arbo]

Was over and saw this machine today. A very serious and solid machine.

[the_canuck]

A bit more progress.

I added some acrylic covers to the belt reduction as well as chip deflectors on the gantry. Pretty easy upgrade. I also added screw covers made from 6mil black plastic tubing. Should hopefully keep things a bit cleaner.

[the_canuck]

With the hardware now updated, I decided to bite the bullet and upgrade LinuxCNC to take advantage of the new joint/axis approach to be able to take better advantage of my dual drive on the x-axis.

Recap of my hardware configuration for the axis movements:
Mesa 5I25 with 7I85 breakout.
DMM Tech servos with built in rotary encoders. Servos are 1HP with older DYN-2 controllers.
DMM Tech breakout board.
X1, X2 and Y are rotating nut ball screw setup. Z axis is a traditional ball screw.
X1, X2 and Y have magnetic linear encoders. Z does not.
DMM DYN2 servo controllers are setup to take step/direction signals and would normally run open loop to LinuxCNC. Rotary encoders close a servo loop between the controller and servo but not through LinuxCNC.
DMM breakout is connected to first output of the 5I25 card. Its setup like a stepper driver for all 4 servos. It also has all the hardware connections for home, estop, etc.
7I85 connects the linear magnetic encoders to the second 5I25 output.
In LinuxCNC for X and Y I have it configured as a closed loop stepper utilizing the magnetic linear encoders to close the loop in LinuxCNC. The z axis is setup like a open loop stepper.
I also have a chinese blue tooth pendant hooked up to be able to remotely jog when working around the machine.

Dual driving the X axis is complicated and LinuxCNC 2.8 has new capability to make it work better. Previously I was slaving both servos together and using only one linear encoder signal to drive the PID loop.s for x1 and x2. It didn't really work that well but there is not an easy way to do better in LinuxCNC 2.7 that I could get working.

So after researching 2.8 I decided to make the plunge this weekend.

I first purchased a new SSD HDD to get a fresh install of Linux Mint 18.2 64bit as 2.8 needed to run a newer version Linux. Install and real time Kernel compile went smoothly but does take some time. I'm using Prempt RT Kernel. Install of 2.8 Pre Master went smoothly as well. My latency numbers are better than my original 2.7 and I've been able to turn on the second core of my dual core I3 processor with no latency issues.

Now to the not fun parts. I pulled over my .hal and .ini files from my old 2.7, ran the included auto upgrade to 2.8 on the files which seemed to work for the most part with only a few errors to solve, Converting to the new separate joint and axis was not so easy. Partly because it took me a few hours to realize I had linear encoder 03 physically connected to joint 00 (X1) and a encoder 00 physically connected to joint 03 (X2). I was pretty confused for a while as it gave me really weird results as you could imagine. Once I had the encoders figured out, I was finally able to home all the axis without following errors and with smooth running servos. Ran a couple air cuts just to make sure and everything appears to work well. Servo's on the x-axis aren't fighting each other as much as before now that they are fully closed loop with independent linear encoder feedbacks.

Now the problem is the Chinese pendant. Jog isn't working for x or y axis but is working for z axis. I'm sure I have something in .hal file not quite right. I'll mess with it some more later this week.

Once I get the pendant working again, I'm going to finally make a dust boot.

Andrew

[Arbo]

Nice.

Am looking forward to see what you cut with it. molds first? Or something else?

[the_canuck]

New implement for the machine is the first project completed after the upgrades. Dust boot with magnets to allow for easier manual tool changes.

Hole for the spindle measured from +0.001" to +0.0015" over designed diameter measured several times across the hole at various spots around the perimeter. 10mm magnets press fit into their holes perfectly.

Success finally! Ball screws, linear encoders, and an upgrade to LinuxCNC 2.8 machine control software were the big upgrades. The hard work is already paying off. Only took about 300 hrs on the upgrades. Total cost for the new parts approx $3,500. Ball screws are expensive!

Material is 1/2" Corian. CAD/CAM done in Fusion360. Climb cut with 1/4" carbide 2 flute end mill @ 40 ipm with spindle set @ 15k rpm. Depth of cut max per pass of 0.10".

[the_canuck]

Dust boot initial tests were mixed. I can overwhelm it with too much dust when fast roughing MDF and it eventually fills up in the bristles and then the vacuum starts to suck the bristles up in to the hose making it work even worse. On finish passes and lighter cuts it works great and the shop vac can keep up. Finish passed on my latest MDF project came out very clean and very little dust or chips escape into the air.

[the_canuck]

I've been fighting random servo drive faults that occur during the cuts at any time.

What I've tried so far:
- Checked connectors and wires for some connections being loose but none were found.
- Checked CNC belts and pulleys to make sure everything is tight and not set screws are loose. I did find one loose drive on the x-axis which was fixed. (thinking there might be a loose belt or pulley causing following errors)
- Re-positioned linear encoders to make sure they weren't missing steps from being too far away from the magnetic strips. No evidence this was happening but I'm back within manufacturer's recommended tolerance again just for good measure.
- Replaced 2 blown of the 4x 48VDC power supplies that were powering the servo drivers (thinking maybe I was having voltage sags causing the servo drives to fault)
- Removed all auto Power Management stuff off the computer and wrote a custom mouse wiggler program (thinking computer might be going into an idle state and shutting down the 5i25 card or shutting down USB 5V output.)
- Replaced USB 5V with external 5V supply fearing I was pulling too much current off the USB (Confirmed I am pulling 0.75A when USB max is 0.5A. 5V is powering 4 rotory encoders, 3 linear encoders, the main DMM BOB, and 7i85 breakout board.) Jury is still out, but fingers crossed this is the REAL root cause.

I've been fighting this for 2 weeks now. When not randomly faulting I'm getting great cut results. Parts are coming out super nice but it's really frustrating to have the machine fault and shut down with 5 mins left on a 4 hr program.

Andrew

[Black Forest]

What model servo motor are you using? It seems to me a 1hp/.75kw servo on a Dyn2 is pushing things isn't it? How many volts are you supplying your drives? Sorry if all this is in the previous million pages of this thread!!!!

OK. I went back and read the first page where you stated what you are using for motors and drives.

[the_canuck]

Black Forest.

The drives and servos I have were the ones that came with the early DMM package. Today I wouldn't recommend what I have. I would use the DYN4s.

The Dyn2 version I have are definately limiting my servo max rpm as I have the input voltage at 48v to the drives. I'm getting max 1k rpm out of my servos with my setup.

Andrew

[the_canuck]

Disaster tonight.

I was trying to troubleshoot the new magnetic encoder I added to the x-axis a few months back. I keep getting noise on the encoder lines which is causing position errors but only on the one encoder. The other x encoder is working fine as is the y-encoder.

I was going back through the encoder wiring and checking the grounds, shields etc. and found I forgot to wire the shield to ground like I did with the other encoders. Duh! So in haste to fix it, I broke out the soldering iron and started to fix the cable on the encoder end, thinking I had it disconnected. Main power was off of course but I had forgotten to switch off the 5v external dc supply and I had the wrong encoder plug pulled on the 7i85. I had the y axis pulled not the x-axis I was soldering on. Then I heard the inevitable pop and smoke poured out of the main control cabinet and from the 7i85 enclosure. Crap! Crap! Crap!

Damage I confirmed so far:
Smoked the 7i85s board (possibly repairable but smoked at least one IC, maybe more)
Smoked the DMM breakout board (beyond repair as I'm pretty sure I smoked some traces as well as a couple ICs)

Appears I shorted 5V to some of the encoder pins back to the 7i85 encoder inputs. My guess is that the 7i85 bit the dust first and then shorted the 5v input which took out the DMM breakout. Hopefully the damage is just localized to the directly connected stuff. No way of knowing until I get a new 7i85 and breakout board installed. I have new parts on order that I know are bad as well as a spare 6i25 just in case.

Hard way to get a fresh reminder to always double check to make sure all power is off. I was tired and frustrated, and I was rushing. A bad combination to be messing with electronics.

Of course I'm in the middle of a big customer job that's only 1/2 done which I promised to be complete before the first of the new year.

I've already been down for almost a week due to a broken belt on the rotating nut on the same x axis side with the encoder issue. New belts arrived today and I had just got them installed and machine appeared functional again, but then the gantry racked again 30 mins into a program after what I'm assuming was encoder acting up. The racking gantry is probably what broke the belt as well in the first place. The reason I had the external 5v supply was that I thought that maybe it was a power capacity issue with running off the 5v usb, which I did confirm I was pulling over 700mA from a USB designed to supply max 500mA. Normally I would have had a fuse installed on the 5v line to limit damage. I also had the current limit on the supply probably set to a couple amps.

I haven't smoked anything in years, so I guess I was overdue. I used to work on high power RF power supplies where cutting corners could get you burned, shocked or worse. I totally know better. I've gotten sloppy and paid the price this time in money and repair time.

Andrew

[pippin88]

Sorry to hear about the bad luck.

Easy to make mistakes! I certainly do on a fairly frequent basis. A good project is one I get through without any MAJOR stuff ups.

[the_canuck]

Update from Spring/Summer/Fall 2019.

Machine has been running all out since Feb cutting MDF and HiLoad60 EPS foam for a couple larger customer aerospace related projects. I did have a couple minor issues with the linear magnetic encoders which I finally tracked to a bad connection in one of the encoder DB9 connectors on the x axis. Other than that gremlin it's been very sound and reliable tool.

I think my MDF dust removal accumulated to over 360 gallons of dust that left the shop. I lost count on the foam dust but it was over 100 gallons as well. I don't have an hour meter on the machine but it ran a lot of hrs this year so far. Some runs on one job were over 24hrs of continuous tool on part at 200ipm. That's a lot of surface inches and millions of lines of gcode in a single file.

The rotating nut ball screws are working great!

Now that the big jobs are done, its time to dedust the shop some more and get ready for the next big job.

Andrew

[mactec54]

Update from Spring/Summer/Fall 2019.

Machine has been running all out since Feb cutting MDF and HiLoad60 EPS foam for a couple larger customer aerospace related projects. I did have a couple minor issues with the linear magnetic encoders which I finally tracked to a bad connection in one of the encoder DB9 connectors on the x axis. Other than that gremlin it's been very sound and reliable tool.

I think my MDF dust removal accumulated to over 360 gallons of dust that left the shop. I lost count on the foam dust but it was over 100 gallons as well. I don't have an hour meter on the machine but it ran a lot of hrs this year so far. Some runs on one job were over 24hrs of continuous tool on part at 200ipm. That's a lot of surface inches and millions of lines of gcode in a single file.

The rotating nut ball screws are working great!

Now that the big jobs are done, its time to dedust the shop some more and get ready for the next big job.

Andrew

Great to hear you have made some great progress with your machine