[mcardoso]

Edit: Double Post Removed

[mcardoso]

So I switched gears last night and started work on the actual machine

I started by removing the touchscreen, stepper motors, motor mounts, cables, and the old PC. It occurred to me that this might be the last time I ever see Windows XP. Kinda a sad thought.

Attachment 413894

She looks a little naked.

Attachment 413896

Next the old electrical panel was removed by loosening the 4 studs that were attached to the sheet metal base.

Attachment 413898

Attachment 413900

Mind you that I cleaned this machine spotless and there was still a million chips hiding everywhere. I removed the fixture base plate from my table and gave it another cleaning.

Attachment 413902

I got the solenoid valve for the power drawbar in the mail from AutomationDirect. It is a 5 port, 4 way, 3 position, center exhausting valve. It uses 2 24V outputs and fits 1/4" tubing.

Attachment 413904

And here is a quick shot of the fixture base plate. It is 3/4" aluminum and has a 1" grid of 3/8-16 tapped holes and and offset 1" grid of reamed .251 dowel pin holes. I made this when I had access to a large kneemill at school and it has been one of the most useful things I've gotten for this machine. The cast iron table of the mill is getting a little stained from the coolant that pools in the holes. This used to bother me, but now that I keep this plate on the machine 100% of the time, it is no big deal. It mounts with 6-8 Tee nuts and flat head screws.

Attachment 413906

Being unsure of the alignment of the panel, mill, and steel base plate, I maneuvered the control panel around to the back of the machine and roped it in place. Holes were laid out onto the steel baseplate.

Although I have no way to actually weigh this panel, based on my CAD model and how hard it is to move, I think it weighs between 150 and 200lbs. I certainly can't lift it , but if I go slow I can "walk" it on the bottom edge to wherever I need it to go. That being said, I was soaked in sweat by the end of the night. Didn't help I went rock climbing right after work.

Attachment 413908

I reused the square aluminum tubing from the old monitor mount to hold the panel. One was screwed into the steel bottom plate (man that was hard to drill and tap), and the other screwed into the bottom of the machine column.

BTW, I'll be cleaning up the original CNC components and put them up for sale at a really good price. If anyone wants anything, just let me know.

Mike

Edit: Not sure why everything posted sideways

[mcardoso]

I think I am 100% done with panel internal wiring. All that's left from the controls point of view is to hook up the proximity sensors and mount the touchscreen PC.

Here is the pendant I purchased. It is surprisingly well built. I've gotten one of these cheap ones before for a school project and it was super flimsy. I'm pretty sure I could drop this and it wouldn't be damaged.

Here is the listing (it was $49 when I bought it a month ago): https://www.ebay.com/itm/Universal-C...72.m2749.l2649

I can stand about 6' in away from the front of the machine with the included pigtail.

The blue cable is the CAT 5 RS485 network cable.

Attachment 413918

With the panel mounted, wiring is a bit of a pain but not too bad.

Attachment 413920

The flying lead end of the pendant. I landed the shield (wrapped in heatshrink and a ferrule) on the ground connector of the 3 tier terminals. Wires are pretty thin, probably 26 AWG.

Attachment 413922

Also I got the timing relay (white box, red dial).

Attachment 413924

I am concurrently working on an AutoCAD Electrical drawing package. It is very much a work in progress, but I may consider posting it for people to reference if there is interest.

The original length was a touch short, so I stripped it back further and cleaned up with more heatshrink.

Attachment 413926

Timing relay wired to the power supply next to it. Power supply turns on, then 3 seconds later (one the output voltage is very stable) the relay turns on the voltage to one of the M12 panel connectors. The PC I'll be using had some issues with its power circuit where it wouldn't want to turn on if you plugged the power supply into the wall, but if the PS was already on and you closed a switch on the DC power, it had no issues. I'm just automating that process with the relay.



I did a quick power check to make sure the relay worked and it is done!

Next up is either mounting the servos or the PC.

-Mike

[mcardoso]

I got the panel PC mounted to the machine. I previously had an industrial monitor that was hooked up to a desktop PC on the floor. Its resistive touchscreen stopped talking with the PC at some point and it wasn't very pretty to look at.

The new unit is something we scrapped out at work thanks to someone plugging it into a 50VDC power supply (it only runs on 24VDC). I had to do some surgery and replace some fried components right off of the motherboard, but it works (and I'm kinda proud of that). This unit is an integrated display, capacitive touchscreen computer running Windows 10. It isn't crazy powerful so I'm keeping my fingers crossed that it is enough to run Mach 4. If not I'll work something else out. It has a quad core Intel Atom E3845 CPU (1.91 GHz), 4 GB of RAM and a 128 GB solid-state drive.
Attachment 413932

The computer is mounted on a $10 vesa monitor mount. Its a touch flimsy, but it seems to do the job. The mount was bolted straight to the column.

Attachment 413934

It's alive!

Attachment 413936

I keep my keyboard and mouse on a tool cart along with my tool holder rack (put away at the moment), and Tormach presetting surface plate. I pressed magnets into the side of the cart to hold my prints while I'm working. The cart was a Christmas preset from my fiancee along with the stool. She knows what to get me

Attachment 413938

And the whole basement setup. Its pretty messy thanks to this conversion work. I'm normally a neat freak about the shop. That is an ATC prototype on the floor...

Attachment 413940

Servo motors for the axes! The two small ones are .75kW (1HP) and the big one on the right is 1.8kW (2.4HP) same as the spindle. This will be for the Z axis. On the left is a .75kW motor mounted to an Alpha Wittenstein SP+ gearbox. It is 11:1 and moves pretty quick!

[mcardoso]

Close up. This will be for a 4th axis that I'll get to building once the rest of the machine is done.

Attachment 413944

I'll be using this 8x8x8 ground angle plate for the base of the 4th axis. Imagine a large hole bored clean through the face of the plate to receive the gearbox.

Attachment 413946

All the motors are hooked up. I'll be testing like this for a long time until I feel comfortable about mounting the motors. At first I plan on torque limiting these guys to what my steppers could do (~425 oz-in). This way a crash won't damage the machine. Eventually I'll open up the limits to get some crazy accelerations.

I have a Mach 3 setup that I have used to test all this in the past and I used it to quickly verify some signals, the mpg, and jog the motors. There are a few issues that might exist, but nothing too big of a deal.



My plan for the weekend is to re-image the PC, install Mach 4, Putty (terminal emulator for RS485 testing), Ultraware (drives cfg.), and the ESS configuration. I want to get all the I/O checked out and working, and get the motors tuned. I may start some installation of the motor mounts and home/limit switches.

[mcardoso]

I don't have as many pictures as I'd like to share, but our furnace went out so any time I spent in the shop was not focused on taking a million pictures. The bedrooms got into the low 50's and the basement into the mid 40's (it is low 20's outside). Not terrible, but much less than ideal.

I re-imaged the computer back to a bare Win 10 machine. From there I installed Mach 4 and configured it for use with the Ethernet smoothstepper. I am very impressed with the menus in Mach 4. They seem to be laid out intuitively rather than the haphazard placement of settings in Mach 3. I did identify a few potential wiring errors on the I/O during some initial testing that I need to dig deeper into, however as it stands right now, each drive correctly receives the control input pins, as well as an enable signal. Each drive also sends back a fault signal and the encoder index output. This output gives actual speed feedback to mach for the spindle (pretty cool) as well as enables home to switch then to encoder marker which is more accurate than the switch alone. All the limit/home switches work properly, the ESTOP is working, and the pendant MPG is reading into Mach (although it seems to count by 4's?). I know that my "operator station" input (start, hold, and stop buttons) seems to only have the start button operational at the moment, and the pendant resolution selector switch seems to have the X10 and X100 inputs stuck on. I have not tested any of the outputs including the electronic gearing relays, the relay outputs for the 120V outlets on the side of the panel, or the power drawbar solenoid outputs. The panel internal outputs such as the drive enable and fault reset pins work perfectly.

I loaded Putty, a free terminal emulator, onto the computer as well as my laptop, and after an hour of fiddling with it, got the two PCs talking to each other over RS485. From there I connected the computer to the drives, but when I tried to connect, Ultraware (Allen Bradley utility for servo drive configuration) would just freeze up. It took me a little while to realize that every drive had its network node address set to node 0. When Ultraware tried to talk to the X axis drive, they would all answer at the same time. I had to power each one on, one at a time, to set them to different node addresses. Now whenever I open Ultraware it automatically connects to each drive that is powered on. This is hugely helpful for troubleshooting since every piece of data from the drives is exposed for monitoring and adjustment. Since I got the network working I have not had a single issue with it. If I can figure out Mach 4 LUA then I feel very confident that the networking of the drives will be very reliable.

I tested some jogging with Mach 4 and found it to be a bit unresponsive to keyboard inputs compared to Mach 3. There was a barely noticeable delay when starting and stopping a jog compared to the key press. It probably doesn't matter, but coming from industry it makes me a little uncomfortable when you can "feel" the software between you and the machine. My other, much more major, issue with Mach 4 is that it hangs after 3-5 minutes of running code. This is the demo version right now, but it is not the timeout (I have seen the timeout message and know what it looks like). The machine will be running and BANG, all the servos stop dead but remain enabled. The software ESTOP does not shut them down (thankfully I also have a hardware ESTOP to kill drive power). The Mach 4 greys out and the cursor is the blue Win 10 waiting circle. This happens repeatably both running Artsoft sample GCode as well as a program I used to machine the spindle housing. The program never recovers and has to be closed with task manager. The Smoothstepper remains waiting for Mach signals and won't reconnect when Mach opens leading to me having to power cycle the machine.

At first I thought that this was due to an underpowered computer. The VersaView 5400 computer that is on the machine runs on a 4 core Intel Atom E3845 chipset @ 1.90GHz. It has 4.0Gb of memory and runs Win 10. This is very much on the bottom edge of the system performance requirements for Mach 4 (especially the memory), but there are posts in MachSupport about people not having issues running mach on much less powerful computers without issues. During loading of Mach 4 the CPU is between 70-80%, at idle it is 30-40%, when loading a Gcode program 70-80% and while running the machine it holds quite steady at the mid 60's%. The memory utilization never got above 2.0Gb (50%). This all seems like the computer was doing just fine, but I was convinced that this computer just didn't have the guts to run Mach 4.

To test the situation, I installed the same version of Mach 4 and ESS utilities onto my laptop (Intel i7 6700HQ 3.5GHz, 32Gb of memory, Win 10). I copied the configuration file from the CNC computer to my laptop and connected to the smoothstepper without issues. I ran the same test files and sure enough, Mach 4 crashed about 5 minutes into the code same as the other computer. I tried this several more times and had the same results. I then tried running the machine from Mach 3 on my laptop and it finished the 2 hour and 20 minute program with no issue whatsoever.

I am posting some questions about this on the Mach support forum, but if anyone here has experience with Mach 4 I'd appreciate any input. My next test will be to install Mach 3 on the VersaView computer and try running the machine that way. I know that the programs are very different, but if the computer fails to work with Mach 3 then I would believe it to be underpowered.

Here are some Mach 4 screenshots. Not super interesting, but hey, it's all I have.

Attachment 414200

Attachment 414202

Attachment 414204

Attachment 414206

Attachment 414208

Attachment 414210

Attachment 414210

Attachment 414212

[mcardoso]

Since it seems like Mach 4 is up and running, I am switching gears back to the hardware side of this build.

First off, I got the matching buttons for my operator button station. I have labels for each of these. They are the Allen Bradley 800FP series of operators (22mm). The plastic ones, while plenty sturdy, feel a little cheap. There are metal versions of these but I couldn't find any cheaply on ebay. The back side of the button is modular to select which contacts you want as well as lamps if the button is illuminated. Top hole is for an ESTOP which is temporarily setup elsewhere.

Edit: I guess cheap feeling is relative. They are super nice compared to most operators you get from automation direct or hobby grade stuff, but are a long shot from the mil-spec buttons and knobs I work with day to day.

Attachment 415816

Thanks to the folding mount, the computer tucks neatly behind the machine for cleaning. The mount is already getting bent, so I am looking for something a little more sturdy.

Attachment 415818

Loosening 4 bolts removes the top bearing plate from the Z axis.

Attachment 415820

From there the 50lb gas strut can be disconnected.

Attachment 415822

The gib can be knocked loose. Note that all sides are flat (no milled pockets). The way facing side is ground and flaked, the opposite side is just ground. The spare gib just looks rough machined, not ground or flaked. That being said these things bend like a noodle so I'm not sure how much value comes from grinding.

Attachment 415824

The way cover is completely shot and I got a replacement from Grizzly for $22.

Attachment 415828

I used some Dykem Hi-Spot blue to see how bad the gib was (hint: terrible). I don't have a great way to spread the ink so I use a tightly balled paper towel. I feel like half the ink sticks to the towel, but a tube is cheap.

I covered the edge which was rubbing on the flat way with steel blue and ground it on my bench grinder until the ink was cleaned up. I was fairly conservative with material removal so I may need to take a few more passes later on. Dummy here didn't cover the surface plate with a sheet so all the exhaust and grinding dust landed on the Hi-spot blue and stuck. I can be a little dense sometimes. I will completely cover the plate in the future.

The first spotting of the gib shows very little contact. I was able to bend it a little straighter by hand and get better contact along the length. I ground the end of a file with a slight negative rake and a wide radius and used it as a make shift scraper. I'm sure it doesn't work anywhere near as well as a real one, but I'm able to remove a touch of material. I'm not trying to scrape this into perfection (nor am I really sure how), but I'd like to get a touch more contact than it has. This is my first attempt at any kind of scraping.

Question: I've seen Sandvik, Biax, and Anderson scrapers for sale but they are large and fairly expensive. Are there any smaller sub-$100 scrapers for things like gibs and touchups that you could recommend. Maybe one day I'll take the leap and get all the scraping essentials, but for now having something to do some touchup would be great.

[mcardoso]

Only spent an hour last night on the gib and Z axis slide.

An initial (and excessively heavy) bluing of the slide shows the contact. Its honestly not terrible, but the heavy blue was hiding some very tall high spots.



After 5 to 6 cycles of scraping. Removing material with a file/scraper is really slow going. I'm going to do just a touch more and call it good. I can revisit later if I want to actually learn to scrape. I would rather not ruin this due to lack of knowledge.

[mcardoso]

I spent one more hour doing a touch up job of scraping on the Z Axis slide. It obviously isn't perfect, but it is better than when I started. If I decide to take the leap into actual scraping, I will come back to this project. At the the contact points extend to nearly the full length of the slide now.

After last pass:

Attachment 415834

I also went to refit the gib into the way but found that even a little work removing high spots caused the gib to seat much lower in the way, beyond the range of the adjustment screws. I had purchased a spare gib from Grizzly but there is a section right in the middle of the way side of the gib that was dished out during machining and never got touched by the grinder. I'm going to call and see if I can get a replacement...

Attachment 415836

For the time being, I reinstalled the Z axis slide using some shim stock behind the gib. I got this part done and then had to go back because I forgot the 3 bolts to mount the spindle. Later I found I also forgot the new way cover so I have to disassemble again. Ugh.

Attachment 415838

Fit of the gib is MUCH better after grinding it. That gap is smaller than it looks in the picture. Maybe somewhere between .030 and .060. My way oil actually sticks on the ways now which is nice.

Attachment 415840

Next up is the Z axis servo mount. It is a roughly 3.5" cube that was machined from billet maybe 75% on the lathe and 25% on the CNC.

Attachment 415842

Original billet made 3 mounts plus some leftover. 4x4x18" 6061. Parted off on the lathe.

Attachment 415844

Mounted to the Z axis bearing plate.

Attachment 415846

The servo mounts to the shaft with a zero backlash jaw coupling. These run pretty smooth at high speeds compared to some other types.

And the beauty shot with the servo attached. Cables route down the back of the column. It is a little tall but no issue. I'm hopeful that the direct drive makes the servo setup more reliable and easier to tune.



I still have to go back and put the way cover on, add the cable chain for the spindle, and shim the gib a little more. I also want a new monitor mount.

[mcardoso]

Much of what I have built has been acquired from various places and not purchased new. If everything was bought new, this panel would be prohibitively expensive. I have been very fortunate to be able to use some extra parts from work (with permission of course).

I'm sure I've forgotten lots of things but here is a ballpark of parts that I needed to buy:
Enclosure (24x36x8) - $100
Automation Direct (solenoid, M12 ports, disconnect, misc. components) - $200-250
ebay odds and ends (voltage converter boards, pushbuttons, RS485 converter, monitor mounts etc.) - $400 (This is a huge guess)
Missing cables for servos: $400
Raw Aluminum Stock for motor mounts: $120
SDP-SI (pulleys and belts, including scrap): $150
Assorted McMaster Carr (bolts, gaskets, etc.): $300
Replacement bearings, VXB: $120
MPG, ebay: $50
Mach 4: $200
2 years of having fun building this: Priceless
Honestly I keep the monthly shop purchases within budget and I try not to stress over the cost of an individual project. I think of my shop like a monthly "fun" membership. I certainly could be doing much worse things with my money.

In addition, I was able to use the Smoothstepper from the previous version of this build (came in around $1200 for the panel and electronics if I remember). I have more than paid off the machine + CNC conversion with customer projects, and I intend to do that again. I love doing customer projects, especially when I get to be invested in the creation and design of their ideas. I just never want the hobby to feel like a job, so I am selective on what kinds of work I am willing to do.

[mcardoso]

OK. Been a little while since an update so here is where we are. I've spent probably a dozen hours over the last week working on an AutoCAD Electrical set of drawings for the control panel. I originally hand drew these, but so much has changed in the panel since those were done that I needed to make a new set. They still need a lot of work, but have most of the information I need to troubleshoot the system right now.

I disassembled the Z axis again to fit the way cover. I returned the new gibs that I received defective for replacement, so I'm stuck with what is on the machine right now.



I ran into a serious issue where my Ethernet Smoothstepper lost its configuration and wouldn't respond to mach. I wasted a number of days on this but after a support call with Warp 9 Technology, it was resolved. They couldn't identify why the board lost its IP address, but the utility to reprogram it was being blocked by the windows firewall. The support from those guys is top notch.

I started to run the Z axis servo, first using Ultraware (Allen Bradley Servo Utility) and then Mach 4 once the ESS was up and running again. This thing is a beast!. I ran it slowly for a long time, but finally cranked it up to 1000ipm (5000rpm). It is loud, but the sound is normal for a ballnut with a return channel. That is about 95% of the noise. I have a few videos I will attempt to post below.

I also got a new monitor mount in. UPS may have played soccer with the box because this is how I found it:

Attachment 415860

The new mount is super heavy duty. It is rated for 70lbs but only cost $20 (plus it came with a pretty nice HDMI cable). Here it is next to the old mount.

Attachment 415854

The extra extension length of the mount gives me some flexibility in where the computer can be positioned and tilted.

Attachment 415856

And it has some nice cable routing provisions.

Attachment 415858

I will try to add videos below. Next step is teardown and assembly of the X and Y axis servos.

[stuarttaylor00]

Watching this with much interest - thanks for going into such detail.
I'm looking at a similar machine with DMM 750w AC servos, and a 1.8kw DMM on the spindle - Still deciding whether to run under a MESA card and LinuxCNC or a smoothstepper and Mach4 - how have you found Mach4?

Will you go for linear rails on your machine?

[mcardoso]

Oh good! Wasn't sure if anyone looked at this. I have probably a couple dozen posts to add, I will get back onto it next time I have a few minutes!

DMM's will be awesome. I've been a big fan of the smoothstepper from day 1. Mach 3 was a necessary evil, but I am really enjoying Mach 4. It seems like everything just works like it should. LUA scripting is a steep learning curve, but I am in my 2nd or 3rd solid week of coding in it and finally start to feel like I am getting it (I am not a formally taught programmer). I would definitely recommend Mach 4.

I think I will pass on the linear rails. The plain slideways allow me to hit my full rapids, and if I start to wear them out the it will be a good reason to buy a bigger machine. I haven't seen a linear rail conversion where the machine rigidity wasn't compromised.

[CL_MotoTech]

Not trying to crap on Mach 4, but I just don't understand why you should need to know LUA to run the software. I run Mach 3, and while I have done VB in the past I have never used it with Mach 3. I run UCCNC, and while I done some C+ I have never needed to use it with UCCNC. Why is that all Mach 4 users are having to learn LUA?

Also, I've been silently enjoying your thread, so keep up the good work!

[mcardoso]

So I've totally dropped the ball on updating this (I still want to get it finished up). If you want the full thing, it is posted on Hobby Machinist under the same title.

I was offed by having to learn LUA as well but really don't feel bad about it now. In short, if you run a stepper based machine and don't want to modify the functionality of the program at all (and it works perfectly as-is from NFS) then you don't need to know LUA. LUA comes into the picture when you are adding features to the program that aren't natively built in.

For example, my servo drives have a serial host command interface that allows you to read and write all of the parameters from the drive. No CNC control is going to have this built in, but the flexibility of Mach4/LUA allowed me to create the entire interface from scratch and add realtime spindle load monitoring and absolute homing. I used Mach 3 VBA extensvely and while the language is simple to learn, I had lots of problems with working in Mach 3 with all of its crummy variables and editors. LUA is a tad more complicated but once you are up to speed (took a few weeks for me) it is much easier to use and arguably much more powerful than VBA ever could be. I was such a Mach 3 fanboy, but I haven't looked back since I upgraded.

[CL_MotoTech]

I appreciate your in depth response!

I definitely wouldn't consider myself a Mach3 fanboy, I can speak honestly to it's qualities good and bad. The same with UCCNC. I of course have paid to be customers of these companies, I have zero interest in blindly promoting them.

I think everybody agrees that Mach 3 is a bit of a mess on the back end.