I wanted to challenge myself a more delicate part, so I decided to machine the impeller from Pocket NC's tutorial section. I was thinking about going through the instructions and follow along but decided against it as I wanted to see what I could come up with on my own and only reference it when in a pickle. 

I didn't have large enough stock at the time, so I just scaled the part down by 20% so it would fit in a 1.35" square block. I then created seven angled planes along the Z-axis, each facing a different arm of the impeller. I used adaptive clearing on each of those faces with a 1/8" bull nose bit to remove the bulk of the material, leaving 0.1" stock to leave. 

Since I had some weird angles, the spindle was coming in at I ran into a slight issue I've never run into before. When loading the toolpaths, I noticed when coming in at a corner; it would cut as if it was a face perpendicular to the end of the bit, this meant it would cut a bunch of air before touching the stock. I was able to fix this by going into the rest machining area. There I changed the setting to "from stock setup," which meant it would only cut from the stock setup and pass through the air without spending time cutting it. Thankfully it wasn't an issue for the rest of the toolpaths as all I had to do was select the "from previous operations," and it wouldn't recut the same areas from different angles. 

Then I took my 1/8" ball endmill for final finishing passes. I used a parallel toolpath and set the stepover to 0.01" and 30"/min, which ended up being a bad idea as the fins of the impeller where so thin that one of them shattered from the vibrations. I then cut the speed in half and reran it, the same outcome. Finally, I decided to just cut the stepover in half at 0.005," and that worked much better. I ran it at 30"/min and had an immaculate finish.

One final tolerance test I wanted to do was to make another 3D object. I was getting tired of making the same cube over and over again, so I went to Thingiverse to find some more interesting 3D models to cut. I found a penguin pretty quick and liked it for its low poly (amount of geometry faces), and that the emperor penguin is one of my favorite animals. 

The other day I ordered some new bits from @mscdirect (found through an online recommendation for their price and quality) and had some new tools to use. I got a 1/8" ball bit, 1/8 square 3 flute endmill, and a 1/8" 3 flute bullnose endmill. For this project, I was only going to use the ball endmill as it would leave an immaculate surface finish. 

The first three faces it cut everything went fine. However, when it went to the forth face, the one where the clamp protruded, I ran into an issue. The spindle hit the workholding and threw off the machine which ended with my endmill crashing into the stock, I quickly shut it down and had to reboot it to work correctly. 

Unfortunately, due to the wood I had at the time, I couldn't just raise the entire penguin to be higher so the spindle wouldn't crash again, so I had to make due and set in some additional restrictions on where it could machine. Since I restricted the toolpath, I couldn't reach the feet and so had to use another operation from the side. 

Other than that, it was pretty smooth sailing, and I was able to get a very clean and seamless transition from each face. I did two different kinds of toolpaths for this operation, adaptive clearing for the bulk of the material removal, and then a scallop toolpath to get the surface finish. Overall I'm delighted with the outcome, and it's perfect not counting where the ball mill crashed into its arm. 

After doing some extensive testing of the tool length offset formula, I found it to be almost dead on every time with a consistent 0.0015" tolerance. 

Next, I wanted to test tool collars and using different endmills. I went to my local ace hardware and found small plastic nuts (I couldn't use washers as they didn't have any that would sit tight enough) that I could use as tool collars. These nuts were just tight enough that you had to push down hard on the business end of the bit to get it to fit, which was good to prevent it from sliding around. 

I measured each bit and inputted the information into my formula to find the tool offset. After doing that, I set each one as a different tool with a unique ID for each one (i.e., #1, #2, #3... etc.). I then went into fusion and created a series of facing operations with only one difference, the tool. I ran each one and had near dead-on accuracy with each tool almost touching the face. To be sure I wasn't just running too long of a tool first and all the other bits just running over a large void, I switched the order of the tools and had an identical result as before. 

Now that I have tool collars on each of my endmills, I needed a way to organize them as they no longer fit in their cases. I went to fusion and spent about 13 minutes coming up with a simple design to display and keep them in proper order. I am only able to use 0.125" or 1/8" tools in the Pocket NC v1, so I did not need to make any smaller or larger sections. I did offset each hole by 0.15" to make sure they would fit.

I then sent my design off to the 3D printer with a 40% infill and 0.5mm wall thickness for a total print time of 1 hour and 52 minutes. It turned out nice, and each hole gave the bits a snug fit. I only have like five tools I will actively use for the Pocket NC, so I only made seven slots. 

I got in contact with Pocket NC's support team, they suggested another possible solution. That was to recalibrate the tool setup, which required manually finding how far the z-axis tool traveled and then offset it with a 123 gauge block. I didn't have a steel 123 block, so I used a piece of wood, cut it down to roughly 3" making sure all the faces whereas square as they could be, and then used a pair of calipers to find the exact thickness. Using wood is not the most precise way of doing this, and I am well aware of that, I just needed to come close enough to find out if the tutorial would get close enough. 

I then followed the tutorial and had similar results as before, which was frustrating. Since it kept measuring consistent results, I decided to try once again and find how far it was off by dividing it by two to account for both sides and adding that to the tool offset measurement. I was shocked to see that it worked! It measured out to 1.0015," and the 3D design measured 1.0", 0.0015" off wasn't too bad and was an acceptable tolerance. I then faced another side of the cube to make sure it was consistent, and I got 0.9980," which was 0.002" off, also an acceptable tolerance. 

I then reverse-engineered this into an equation I could use for inserting other tools. The previous equation was -(3.6-tool length)= TLO tool length offset; with the new equation, I got -(3.573-tool length)= TLO. I had to confirm then this equation would work for other tools and get consistent results. I inserted a new endmill, measured it, and input the information into my equation and got near identical results. I was delighted with this outcome, and now all I need to test is the b-table offset and make sure that it is consistent and accurate. I didn't have enough time to cut another cube, so it would have to wait until next time to do more extensive testing.

Third attempt at making a "perfect" cube

I was able to get a decent looking outcome, no problem the third try. However, the dimensions didn't quite line up with the actual measurements. Each side was off by about 1/32", pretty consistently which I thought would be as easy as adding that to the tool length offset. In the second image above, you can see how bad this is; the area where the bottom arrow is pointing is the before.

It would shave off that much of the face on the side operation after completing the top. It should look more like where the top arrow is pointing where it shaves next to no material off the face. I initially thought it was a stock to leave issue, but when I turned that off it still produced the same problem. I was able to get it to shave very close to the face but not entirely on it by adding 3/64" to the equation when finding the tool length offset. I then spent nearly 8 hours attempting to find the perfect offset, I finally gave up and messaged the support team at Pocket NC. They replied within an hour and gave me some new information. 

Since I did not set up the PocketNC when it first arrived, there was some calibration you had to do to get it all set up. The b-table had a specific offset for this machine, meaning two of the same models may not have the same offsets. The b-table is the bed that your material stock attaches to. 

I did have a little trouble in offsetting the machine properly as the model you're supposed to use to set up your CAM operations wouldn't let you edit certain aspects of it. So I manually created a sketch and offset it and just left a point in the air to set up the origin. Then I set a block of wood in the machine and set it running with a couple facing operations I could then measure to see if I had the correct TLO (tool length offset). Unfortunately, I did not and spent a few hours trying to see if I could calculate how much I was off by and then narrow it down to a specific number I could add to the equation.

The upside is I was able to fix the issue where it would shave part of the top of the cube when making its passes, however that was not the one I was looking for in this instance.

This morning I received an email from PocketNC with another possible solution to my problem which I will be trying out later this afternoon.

_________________

Summary

I realized the last post was a spew of information and probably didn't make a lot of sense. 

I was having an issue where after I machine my part and measured it, it wouldn't line up with the measurements in the design. This is related to the tool length offset or TLO. When inputting a new tool into the machine you have to measure it and put it into a basic equation to find out how far it sits from the origin. This means if you measure it slightly off, you won't get the proper dimensions.

As you can see in the image above, it measured at 1-13/128" instead of the desired 1-1/8". In the third slide, you can see me using a pair of calipers to measure each face on the cube then comparing it to another one. I was able to get reasonably close to an acceptable amount of tolerance at 1/128," but it wasn't consistent. I got in contact with @pocket_nc, and they responded within an hour with some possible solutions. I tried them but wasn't able to get anywhere. This morning I got another email from them giving me another possible solution, which is to recalibrate the machine by hand with a 123 block, which I will do this afternoon.

I had much more success in my second attempt and was able to create a half-decent cube. I say half decent because I messed up one of the faces of it. While I was machining the second face in operation the tool suddenly popped loose, and I didn't catch it right away as it only shifted a couple of thou. I numbered each face to show the sequential order of how it was machined. 

After the whole operation completed, I was able to cut it off the base as I left a small amount of material it would avoid. One huge thing I noticed was that it would shave the side face it was working on and cut it down by about 1/64", when done on each side it would make cube smaller on each face by 1/32". This may not seem like that big of a deal but when trying to get precision parts, this is pretty big. It's hard to notice from just looking at it, but with a pair of calipers, it's self-explanatory. I looked up possible solutions and may have found something here I'll look into for next time.

For the time being, it turned out pretty great aesthetically, and each face is immaculate and sharp, which was better than I could have hoped. Next, I am going to attempt to make a "perfect" cube in a single operation-perfect meaning without errors and having to shut down the machine.

I'm finally getting around to learning how to use the Pocket NC, a 5 axis CNC machine. It's been sitting for nearly three years without use and has gathered quite a bit of dust. 

Pocket NC started out as a Kickstarter in 2016 and exceeded its goal after an hour of putting it up. Until now there really weren't any affordable desktop 5 axis machines on the market. 5 Axis CNC machines allow you to machine in 5 different angles, this allows you to create some very complex objects with ease.

A few weeks back, I started a new project to mostly trying out the epoxy resin. I've seen a couple of people on Etsy make wall maps of local lakes and thought it would be a pretty cool idea to emulate with a local lake that our community is based around.

I needed an overhead view of the lake and convert it to an SVG in which I could use in Fusion360. I first went to the local community website to see if they had a decent photo I could use. The majority of the images they had were of low quality, and I couldn't use them. I then headed to google maps, after zooming in on the area I wanted I was able to capture a snapshot of the screen and bring it into adobe illustrator. There I was successful in converting it into a traceable SVG. 

Once I had the SVG, I was then able to take it into Fusion360 and setup the toolpaths. I initially ran it small to get an idea of what it would look like, but the detail was tough to capture with the tools I had, so I scrapped that and went with the size I wanted. 

I started looking for some cheap epoxy resins I could pick up at stores near me and found that my local Walmart sold small quantities in their arts and crafts department. However, once taking a closer look at it, I discovered that you couldn't pour more than a 1/8" without getting into some dicey territory.

For the wood itself, I wanted to go with something relatively inexpensive, so I headed to my local Lowes and found Poplar would fit my needs. It was cheap, and in size, I was looking for, plus it was almost effortless to cut. The only downside to poplar is it has this weird greenish-yellow color, which in my opinion, is very ugly; I decided to use it anyway and apply a stain and finish to cover it up. I picked out a piece 24"x 16"x 3/4" and cut it down to 16"x 16" stock. The size of the Shapeoko 3 cutting radius is 16"x 16" and so couldn't exceed those limits.

After finding all my restrictions, I started to set up the toolpaths. I decided on a 1/4" depth of cut and only fill it with epoxy halfway, this would give some extent to the finished product. The whole operation only took 24 minutes and turned out pretty good; however, I noticed that certain aspects of the lake cut seemed off. I took a closer look at the toolpaths and found that I accidentally set it to cut with a 1/8" square mill but used a 1/4" square mill. One might think this isn't that big of an issue as everything would be slightly larger than the original, right? Unfortunately, not, since every side was 1/8" more than what was needed, this would give some of the inside details be bloated larger than they are. I didn't have time to go back and start from scratch, so I just dealt with it. 

Before pouring the resin, I needed to laser engrave the wood with a compass and location. I was able to find a compass online and modified it a bit to suit my needs. After putting it together with the name and location, it was ready to go. I set the laser engraver to 400mm/s  with 50 power. The first pass was a bit too light, so I ran it twice more and dug into the wood.

Once all the features machined out and laser engraved, I then moved over to staining and finishing. I started with using a blow torch and lightly pass over the wood a couple of times to give it more texture. For the staining, I found half a can of red chestnut stain and finish and used a paper towel to wipe it on. I then came back with another cloth and wiped as much of the stain as I could off to leave it somewhat lighter than the stain advertised on the can. 

After it dried, I read the instructions on the epoxy resin and started mixing it with a bit of blue pigment. I was aiming for a baby blue but ended up putting too much pigment, and the outcome was a midnight blue, I thought it was just darker in the cup since it was a more concentrated area, but when I poured it, it was only slightly lighter than it was in the container. I poured it halfway, approx 1/8" and took a blow torch to it to remove the bubbles. And finally, I tacked on a simple brass wall piece to mount it.

Overall, working with resin was much easier than I was expecting.

I honestly wasn't expecting it to turn out as well as it did. If I were to redo the project I would definitely double-check the tooling and even use smaller tools to get those smaller details. Also, I probably would have added more details through the laser engraving like streets.