[Mill] VF-25MV Build Thread

[tkalxx]

The X-axis has 2 angular contact bearings back to back on the left side of the table which provides all the axial support for the ball screw. The motor is mounted on the right with a small needle-bearing support. The needle bearing provides radial support and prevents screw whipping - this was actually an afterthought, but provides a lot more support to the ballscrew. Both the bearing mount on the left side and the motor mount on the right side utilize the locating pin holes that come factory with the PM25. The factory pins are 6mm OD - these holes were drilled larger and tapped for M8 bolts. The spacing of the factory M8 mounting holes were too narrow to utilize. The X-axis has 18.75" of travel.




X-axis bearing, motor, and ball screw mount:



I decided to bead blast the X and Y-axis mounting parts before anodizing. They turned out looking pretty good. Someone with a keen eye will notice that the bearings in the following two pictures are actually installed the wrong way Don't make this mistake or you'll have a bad time...
Here is the X-axis bearing mount:






All the X and Y-axis parts:



Installing the X-axis ball screw:



X and Y-axis motors and ballscrews installed:



Here is the machine in its home position (positive travel limits):



-Adam.

 

[tkalxx]

I spent quite a long time tuning steps/in, motor acceleration, velocity, and tramming the head. One issue I ran into right away was the column perpendicularity to the X and Y ways. I noticed on some test parts that drilled holes were nowhere close to their intended location. I spent days trying to figure out the cause of this issue before realizing that I had never squared up the column to the base.

You can see in this picture that the drilled holes are all offset some distance from the counterbore which was interpolated with an endmill. If the column of the mill is not perpendicular to the X-Y plane, when you change to a longer tool with a larger tool offset (i.e. a chuck and twist drill), the mill head moves up, but also side to side/front to back depending on how out of alignment the column is.



I managed to borrow what I like to call a "4-6-12 block" that we use at work. It is a 4"x6"x12" block of aluminum that was machined to 0.0005" perpendicularity. I used this block and a dial test indicator to square up the column in both the X and Y direction. It's a very repetitive process of measuring, shimming, and repeat. I had to add nearly 0.020" of brass shim stock under the column to get the Z-axis square within 0.001" over a 10" travel. I suspect the column was that much out of square due to disassembly so that I could machine the base when installing the Y-axis ballscrew.




Like any wise and logical person, the first part that was machined on the CNC was a 4 hour run time head spacer. The head spacer is 1.375" thick and centers the spindle in the middle of my extended Y travel.

I am 2 hours into having a functioning converted CNC and it appears I have already outgrown the size



Op 1:



Op 2:






Overall the results are OK. The head spacer works as intended, however, I ran into a host of problems with the PMDX board having connection issues which turned this 4 hour run time into more like 6 hours. In addition, I had the Y-axis proximity limit switch triggered by some chips midway through the program.



The head spacer is secured to the head with 8 M6 bolts and mounts to the Z-axis slide like normal with some extended M10 T-bolts that I purchased from McMaster. I had to add a 0.0015" shim between the head spacer and the head in order to get the tram of the head in the Y direction perfect.

-Adam.

 

[tkalxx]

As stated earlier, I wired my control cabinet with only the essentials to get the mill under power for testing and tuning purposes. It was now time to redo and finalize the control cabinet wiring.

Throughout the early stages of the build, I noticed issues with the PMDX-424 motion control board I was using. The PMDX is a usb controlled board and the usb connection proved to be an issue for me. No matter what I tried, I couldn't eliminate connection issues between Windows 10 and the PMDX. I went through weeks of diagnosing, posting on Mach4 forums and testing different solutions. Here are some of the things I tried and had no luck with:

• Shielded wiring for all wires coming from the PMDX board and an abundance of ferrite chokes
• Fresh install of Windows10 and Mach4
• Different versions of STM32 usb driver
• Different computer
• Running Mach4 without power to the stepper motor PSU/drives
• Verified proper grounding of all electronics and any potential grounding loops

I tried contacting PMDX multiple times and got no response, most of the input I got from various forums was to ditch the usb connection. I finally gave up and decided to purchase a Pokeys57CNC board. I have nothing against PMDX; their board appeared to be of good quality and their Mach4 plugin is well written, but I just didn't have good luck with it. For anyone trying to decide on a motion controller my only piece of advice is to get an ethernet board rather than usb. There are just too many things that can go wrong with usb and if you find yourself in a similar situation to myself, you'll be chasing your tail for weeks trying to find the source of the issue.

The Pokeys57CNC has been a great board. It offers a lot more expansion possibilities than the PMDX, up to 8 motors at 125kHz kernel speed and backlash compensation (something the PMDX was not capable of). For those who are not familiar with Mach4, Mach4 itself does not handle backlash compensation. There is no core code in the Mach software for backlash comp, and thus it is up to each individual motion controller manufacturer to implement backlash comp into their controllers. This is a critical detail I was not aware of when I initially purchased the PMDX board.

Ripping out all the "temporary electronics"



Adding some additional receptacles to the back of the tool cart, along with an Ethernet input.



I began to lay everything out in the cabinet. Items are positioned in a fashion such that wires with a large differential voltage are not grouped together and don't interact with each other. All AC wires run in down the far right side supplying power to the 3 PSU's along with some front panel switches, and the power bar mounted to the rear of the tool car which supplies power to the computer, monitor, etc.

Low voltage DC wires run through the middle of the cabinet and any low voltage signal wires are isolated to the bottom left corner where the Pokeys57CNC is mounted.



Midway through building out the cabinet, I replaced my 48V switching power supply to a 76V toroidal power supply and added an 8 relay expansion board. I would highly recommend using a non-regulated power supply with the highest voltage your motor drivers can handle (especially for a stepper system). The jump from a 48V to 76V power supply made a massive difference to the torque curves of my stepper motors and I was able to achieve 300ipm rapids on all 3 axes with 40 in/s^2 acceleration.






The wire ducts really clean everything up nicely!



And finally, a little reminder for myself every time I turn on the power switch



-Adam.

 

[Spider55]

Very nice build and thanks for sharing. Can I ask how you calculated the thickness of the head spacer and how much extra Y travel did you get?

 

[tkalxx]

Very nice build and thanks for sharing. Can I ask how you calculated the thickness of the head spacer and how much extra Y travel did you get?

I used a dead center chucked up in the spindle of the mill and jogged Z down close to the table to take some measurements. From those measurements, a suitable thickness was calculated so that the axis of the spindle was centered within the Y-travel. One of the flaws with the PM-25MV is that the spindle axis is not centered on the Y travel from the factory... I found that kind of odd.

The total Y travel I achieved was just shy of 7.5" (if I recall correctly, it's about 7.41"). I have my soft limits set at 7.125" of travel. With the 1.375" head spacer, the spindle axis reaches both extremes of the table since the table only measures 7.125" in Y. I can't remember what the original Y travel is for the PM-25MV - Precision Matthews states 7" but I remember measuring less than that when the mill was manual.

 

[macardoso]

Sweet build man, very impressed with how clean those electronics look!

 

[tkalxx]

I built my enclosure out of t-slot framing. I very well could have built an enclosure out of sheet metal for probably the same cost as all the t-slot extrusions, however, there are a lot of features and functions in terms of eclosure ergonomics that are a mystery until you build one, use it for a while and determine what you like and don't like. The reason for the t-slot extrusions is that it is very versatile, and nothing is set in stone. I figured if I don't like how the enclosure turns out, I can always disassemble it and use the extrusions for a different project.



The enclosure is secured to the stand with some aluminum flat bar that I painted black to match everything else. I put some riv-nuts in the square tubing of the stand which allows me to separate the entire enclosure from the stand with approximately 20 screws. I was a little worried about the rigidity of this construction method, but it turned out great.



The walls are made from corrugated plastic board that I picked up from my local home depot. I used some "t-slot dust covers" from mcmaster around all the edges of the plastic board. This "dust cover" fit my corrugated board perfectly and is a tight fit into the t-slot's of the aluminum extrusions. It basically pre-loads all the plastic walls so they do not wobble or vibrate when installed in the t-slots.

McMaster-Carr

McMaster-Carr is the complete source for your plant with over 595,000 products. 98% of products ordered ship from stock and deliver same or next day.

www.mcmaster.com

I don't have flood coolant (probably a future project) so I haven't been able to test how waterproof the enclosure is over long periods of time. I did take a garden hose and sprayed the inside of the enclosure just to see what would happen and I believe this setup would present very minimal leaks with flood coolant. Worst case scenario, I can patch some of the small open gaps with some silicone.



I made a bi-fold door out of t-slot framing and some clear polycarbonate. The door slides in the t-slot of the extrusions on the top and bottom with some small Delrin bushing that I made. One of the requirements I had with the enclosure is that I wanted as big of an opening as possible. This is the reason I didn't make a traditional 2 piece sliding door. Working on the mill is super easy with the large opening. In addition, I can remove the table and saddle from the mill without having to disassemble the enclosure.



Of course, I had to add the VF-25MV logo with some simple stencils and spray paint.



I added some accordion style bellow covers to the Y-axis with two small mounting plates that I made - they utilize the existing holes for the x-axis gib locks on the saddle. Other than protecting the dovetail surfaces from chips, it also covers my y-axis homing/limit switch which was prone to being accidentally triggered.

-Adam.

 

[tkalxx]

I would love to start a discussion about ballscrews, specifically rolled vs. ground screws for hobby applications. As stated in my first post of this thread, I have C5 ground OFU ball screws on my mill, however, this was a recent modification. I started with C7 rolled ballscrews from Chai on ebay - I assume most folks who go down the route of converting a mill start with ebay screws or something similar. During my very early research on ballscrews, I found hundreds of threads discussing ballscrew accuracies but I never once saw a direct head-to-head comparison of C7 vs C5 screws. The majority of people in the threads that I stumbled across concluded that rolled ballscrews were the best option when comparing cost to performance for a small hobby machine.

A wise man once told me "the only difference between science and di*king around is writing stuff down". With that in mind, I've attempted to log data and gather information specifically in regards to the lead accuracy of my C7 and C5 screws for direct comparison and will share my findings in the following couple of posts.

Some background information:

SFU (or single nut) ballscrews are single nut ballscrews where the ball tracks have been machined into the nut to match the lead of the screw as best as possible - SFU ballscrews will always have backlash because of this. Many manufacturers will load SFU ballscrews with slightly oversized balls to combat the inherent backlash issue. Ball nut preload is defined as a percentage of the total dynamic load rating of the screw itself. When manufacturers preload SFU ballscrews with oversized balls, typically you will find a preload in the range of 1-3%.

DFU (or double nut) ballscrews have two single nuts back-to-back on the screw with a spacer in between them to attempt to preload the nuts. Typically, DFU ballscrews are not loaded with oversized balls as the spacer in between the nuts is meant to remove any backlash. A major advantage with DFU ballscrews is that the nuts can be seperated and the preload can be adjusted with shims, a new spacer, or even belleville springs.

OFU ballscrews are a combination of SFU and DFU ballscrews. The ballscrew nut is a single-piece construction where the ball tracks inside of the nut are ground with a small offset to preload the balls. Advantages to OFU ballscrews are rigidity, due to the single-piece construction, and a smaller overall footprint compared to DFU nuts. Specific preload may also be specified depending on the application. A major disadvantage to OFU ballscrews is that they cannot be simply taken apart and re-packed like DFU or SFU ballnuts due to the offset ball tracks.

I purchased my C5 ballscrews from Terry Machinery on Alibaba and spoke specifically with Sophia@ntl-bearing.com. The ballscrews came from TBI-motion. From what I can tell, Terry Machinery deals directly with TBI-motion, and Sophia acted as a middleman. The entire transaction was very smooth and the ballscrews were delivered within 15 business days of payment. Everything was ground within my specified drawing tolerances. I specified a 5% preload for all three screws. Some of the following pictures have grease still on the threads - in retrospect, I should have cleaned them up better before taking pictures.









View attachment PXL_20210330_012433133.mp4

The X-axis screw had the worst run-out out of the 3 ballscrews and is shown below. Even though everything was packaged really nicely, I assume because of its length and size it got tossed around during shipping.

View attachment PXL_20210330_013844081.mp4

View attachment PXL_20210330_014159691.mp4

 

[tkalxx]

As a disclaimer before I share my findings: please take the following information with a grain of salt. I did these tests for myself because I was genuinely curious about the differences between rolled and ground screws. There are probably many aspects of my testing method that are incorrect and could be improved upon. I am not a metrology expert, nor do I have the proper testing equipment to test ballscrews like a proper lab would - I am just a guy in a garage trying to build a CNC. I'm sharing these results because I feel it would have helped me choose ballscrews. Had I seen data like this earlier, I may have saved a good chunk of money by not having to experiment. There are so many variables that haven't been accounted for, your results will vary.

Something to keep in mind is the price difference between the C7 and C5 ballscrews. C5 ballscrews cost me $640usd +$100 shipping.

My procedure went as followed:

Zero test indicator and mach 4 on a 123 block.


Command mach 4 to move 3 inches, and record the actual moved distance (actual moved distance = 3.000" in this scenario).


Shift the 123 blocks 0.5" from the "home position" so that I am now measuring move distance on a new section of the ballscrew.


Re-zero the test indicator and mach 4 on the 123 block


Command mach 4 to move 3 inches, and record the actual moved distance.


Repeat previous steps multiple times for each axis and for both C5 and C7 ballscrews. The same 123 block was used for both ballscrews. In addition, I verified the 123 block against a gauge block with a 0.00005" micrometer and saw a negligible difference. All the results are tabulated and graphed below. Unfortunately, I missed a couple of measurements for the C5 ballscrews in the Y and Z axis, hence the 2 fewer data points.



--------------------------------------------------------------------------------------------------------------------------------------------------------------





--------------------------------------------------------------------------------------------------------------------------------------------------------------





-Adam.

 

[macardoso]

Nice work Adam, looks like a nice set of screws you bought there!