I’m somewhat of a DIY cheapskate and so when I was studying the xCarve and looking at the major kit options, there were a few items I thought “I can do that cheaper”, and so ended up ordering a kit that was not fully populated. This is somewhat of a blog entry to describe what I did, what was successful, what was not so much. As you will see, sometimes trying to be cheap ends up costing more.
I did recognize this last point as I was considering building my own CNC. At the same time as looking at the xCarve, I was looking at the various open builds projects, predominantly the Ox CNC, and looking at the Shapeoko 3. I priced out the Ox and piecing that kit together would cost more than the xCarve, fully populated. There were some intriguing features on the Shapeoko 3, but when I was ready to buy, they still hadn’t published the price, nor did they have an estimate on when it would even be available.
Although somewhat of an oxymoron, in addition to a cheapskate, I’m also somewhat of a tool snob. So I settled on the xCarve in the large size and started with the fully populated kit. I certainly left the ACME lead screw, but I deleted the waste board, the stepper motors, replaced the spindle with the DW611 mount, and then kept deleting the power supply, the wiring, drag chain, limit switches, tool kit, and all accessories, except vCarve. This took the price of the kit from $1279 down to $765, as well as it would drop the shipping charges significantly due to the elimination of the waste board. I did add in some extra hardware for the waste board so I could bolt my home made board to the machine and the pulleys that attach to the stepper motors. I then sourced the deleted components from other suppliers, predominantly, surplus electronics vendors.
Finally, all the parts came and it was time to begin assembly. As I was working through the various steps, I realized that those options I deleted above, didn’t just contain the titled component, but also had all the necessary nuts, bolts and brackets required to attach that to the machine. Being in the US, metric fasteners still aren’t that available. Instead of purchasing them by weight like I can for SAE fasteners, the hardware stores all have them in the little plastic boxes packaged in individual plastic bags. Purchasing small nuts bolts at 30 cents each rapidly eats into your projected savings for that kit option.
Attaching the NEMA 23 stepper motors, each require four 5mm bolts, washers, and lock nuts (or 16 each). At least I had thought ahead enough to get extra 5x25mm bolts and nuts, but on the X axis end plates where the Y motors attach, those end up being too long and interfere with the V Wheels. I was able to change my assembly order to install the longer bolts, and then cut them off with a dremel to make them fit. If you are going to go this route, find some 16 and 20mm bolts to make it easy.
For the power supply, I read through the specifications for the GShield board, and it has a voltage range of 12 to 30 volts. I had a 200w laptop power supply available that put out 19.5 volts. Since I’m not driving a spindle, this is sufficient. When it was time to wire it, I cut off the laptop end and simply connected the appropriate wires together. This isn’t as neat and compact as the integrated solution, but it has been working well.
As far as the wiring, I’m an electrical engineer doing industrial computer support for a living. I have boxes of miscellaneous wiring laying around. Surely I have enough scraps to connect everything. Except this project takes more wire than it looks like. You end up needing about 35 feet of 4 conductor wire to connect the stepper motors, After reading about stepper motors causing noise, and also being subject to noise themselves, I decided I wanted shielded wire. I probably could have got by with Cat 5 ethernet wire and doubled up the pairs, but I just want this to work. You would also want to use patch cables instead of in wall wire to have flexible stranded wires. Likely the solid wire installed in buildings would fail after only several hours of use (if that). So… what is the wire that inventables uses? Best I can tell is it is 4 conductor 18 gauge shielded wire typically used by alarm companies. You would think this would be common and inexpensive… it is… IF you want to purchase 500 feet of it. If you are looking for 35 to 50 feet of it, I could only find people basically selling it by the foot.
In addition to the wire, the wiring kit also contains terminal blocks used to connect stepper motor wiring together. No problem… I’ll just order some inexpensive ones from electronics surplus and pick up 3 limit switches while I’m at it. Turned out shipping cost as much as the parts did so now I’m $17 into the wiring cost without wire. After driving all over town, I finally found wire by the foot at Lowes, but at 60 cents/foot it turns into about $25 after tax. Oh… don’t forget a handful of 30 cent screws to attach the terminal blocks and the aggravation of blocks being slightly different size and having to modify plans. Overall, I probably spent $40 to $45 plus gas trying to save on a $30 kit item. My advice: Unless you have alarm wiring in your posession Buy the wiring kit from inventables. You can’t make it cheaper yourself.
For the drag chain, I ordered two 1m pieces of 15x20mm chain from Amazon for about $8.50 each. The on-line instructions for the drag chain appear to be focusing on the 500mm kit. The photos showed the chain going up and over the top of the steppers in X and below on Y. In addition to the metric hardware, there are several brackets necessary to attach things. I started with the X axis on the gantry. Being the cheapskate I am, I’ve held onto some 1/16" aluminum plate I salvaged from some junk years ago. Taking some measurements, I came up with brackets to hold things in place. During the mock ups, found that 1m isn’t long enough for the X axis so I had to steal some links from the Y piece. Finally got things set up on the X and it looked pretty good. After running the machine for a bit, I’ve determined I want to re-work this setup so it loops to the back and doesn’t drag on the motors on the gantry and stays away from the heat they generate.
Since I stole links from the Y axis, the chain won’t lay with both ends at the front of the machine, but instead terminates at about the middle of the machine. I set the chain up to drop below the axis, but as it is off the edge of the table, it droops down. It works, but doesn’t look right. I’m going to re-do this as well so it loops towards the side. I’ll probably pick up another length of drag chain as well to fully enclose the wiring bundle. So… I’ll end up having about $25 into the drag chain plus about 4 hours of tinkering. You’ve got to decide if you have the ability to fabricate brackets at no cost and if your time is valuable to determine which direction to go. If you do it yourself, don’t forget those 30 cent screws (that should be 2 cents). They have different heads on them so even if you’ve bought pan head screws in bulk, you will be back at the store to get screws to make it right.
For the limit switches (really should be called homing switches), I was able to get them individually for $1.30 each and I had enough small gauge speaker wire laying around to make the connections. Just about any flexible wire would work as they are only signal wires carrying no power. You need some miniature 2mm screws to attach them. X and Y also require nuts, so you will spend almost as much on hardware as on the switches themselves. My only advice… buy extra switches. You are going to end up smashing them while getting everything working. I’m going to try and come up with some hard stops that will allow the machine to trip the switches, but not smash them. I ended up saving quite a bit here.
The waste board is probably the biggest area where you can save money. I had dreams of using the machine to mill the board itself, and even draw the grid onto it by attaching a sharpie. This is not as easy as it may appear because the board extends beyond the millable area of the machine. You can’t slide a 3/4" board sideways to get all way to the edge either as it hist the Y axis V wheels. Moving front and back is blocked by the Y axis end plates. Finally the board is an integral part of the structure of the machine and is used to hold everything square. I still haven’t given up on drawing a grid but I’m picky and want it to exactly define the millable area. I need to design a sharpie holder in place of the router.
Constructing the board itself was pretty simple. I cut a 37x39" piece of 3/4" MDF (probably really some metric thickness). I then measured and laid out the edge slots to somewhat evenly space them around the perimeter. I used a router with a 1" bushing and a jig to cut the slots and screw head recesses. The jig was simply a piece of plywood with a 1" slot cut into it. Looking at the technical drawings of the waste board, both the slot, and the recess have the same center point… just simply a different radius (or diameter). The router with a bushing will be the same concept. The 1" bushing has a 1/2" radius so add the half inch radius to the length of the slot and you can cut both the slot and the recess with the same jig. Just change the bit in the router. Clamp the jig to the board even with the edge of the board, and cut all the slots with a 1/4" spiral bit and then switch to a 1/2" bit for the recesses. Moving the jig around the board. I used a forstner bit in a hand held drill to mill the recesses for the center holes, and a standard drill bit for the through hole. Even counting that I purchased a full sheet of MDF, I still saved $90 plus shipping here.
Finally we get to connecting the wiring. The surplus stepper motors I purchased didn’t have any technical information with them. I assumed they would use standard color codes I wasn’t worried, but no such luck. The motors were 6 wire motors and even though 4 of the wires had the same colors as the inventable motors, those aren’t the correct ones. After some Google research on stepper motor theory, I was able to identify the correct motor leads and get them connected. Even though I purchased NEMA 23 motors, they didn’t have quite as much torque as the ones in the inventable kit, although they had significantly more than the NEMA 17 motors they offer. Once I got everything connected, I found that I was losing steps during rapid movements, most evident in the Z axis. Thinking I had binding issues, I completely tore the Z axis apart and reworked it paying very close attention to details. After about 6 hours of troubleshooting, I was convinced the issue wasn’t mechanical so started looking at electronics. You can read my entire thread in the troubleshooting section, but basically I found that the stepper motors I have can’t run as fast in rapid motion as the stock motors can. During rapid motion, the controller was out running the motors and so steps were being lost. Ended up I needed to set the max speeds to about 80% of the defaults. You can tune that by ear. During rapid travel, the steppers should sound musical. If you are losing steps the motor noise will sound distorted as you will hear the square edges, instead of a nice sinusoidal sound. Overall, this was my second biggest cost savings in the project, saving about half the cost of the kit. However, I probably spent 2 hours working through the wiring of the non-standard motors, and another 6 diagnosing the missing steps.
I ran the first half dozen or so projects using carpet tape to secure the work pieces. Finally last night I used vCarve to draw up a grid of 100 hold down holes for 1/4" inserts and used the machine to mill the holes within the field. I then used those holes to manually lay out holes outside the field and manually drilled them. Ended up the holes in the front corners were too close to the frame brackets and interfere but otherwise things went well. Today, I’m planning on milling some hold down clamps and then begin working on projects in earnest. Overall, I’m very pleased with how things turned out. I do have some modifications in mind but the machine is very capable as it sits. I would wholeheartedly recommend the xCarve, whether you go the cheapskate route, or simply purchase the full kit. From what I can tell, the machine is just as capable as the $4000 CNC shark sold by Rockler, at less than 1/3 the cost. I can’t comment on customer support, or replacement part availability as everything from Inventables was perfect and just worked.