Build Thread- 2x2 Table-top DIY CNC

[JD_Mortal]

PURPOSE: PCB, Plastic, Wood, Soft-stone, Soft-metals

WORK SURFACE: 2-foot x 2-foot, (24"x24") table-top unit with an extension support that extends working area to 2-foot x 4-foot. Registration alignment and remounting the flipped piece to get the other 2x2 half milled.

STRUCTURE: 3/4" MDF (4'x8' $28.00)
RAILS: Aluminum 1/8" thick square 1.25"x1.25" (10' $36.00)
GUIDES: Aluminum 1/8" thick "T" 1"x1" (20' $28.00)
DRIVE: All thread 3/8" 16-tpi (8' $8.00)
BEARINGS: 608ZZ ABEC-7, Metal shielded, 8mm(ID) x 22mm(OD) x7mm (40qty $25.00)
XY MOTORS: Minebea, 5.1vDC, 1.7a, 4-wire Bipolar, 1.8deg, 97oz/in. (2qty $31.00)
Z MOTOR: Minebea, 5.0vDC, 1.0a, 4-wire Bipolar, 0.9deg, (Guessing~180oz/in) (1qty $19.00)
DRIVER/CONTROLLER: 3-Axis TB6560 (Toshiba TB6560AHQ chip driver) 12-36vDC, 3.0a to 3.5a peak ($53.00)
SPINDLE/ROUTER: Unsure, at the moment. Leaning towards air-powered tools.
POWER: X-Box 360 power-brick, 280W @ 12vDC ~20a (For now.) ($Free)

Total damages so far: $228.00 USD

SUPPORT DETAILS: Fixed frame, Traveling 2' table (X), Traveling 2' span (Y), Plunger 8" span (Z). Square tube span guides and supports. "T" braced table and plunger guides/glides. 6" x 2" linear-guided support on (Y) span. 9" plunger depth support. 12" x 2" linear-guided support on (X) table. Force-glides on (X) roughly cover 8 points on four "T" tracks.

I am converting my drawings into Sketch-up 3D, so I have more precise measurements and a guide to start cutting MDF. The aluminum square tube is all cut, and just needs to have the ends trimmed to square. The aluminum "T" will not be cut until I know the exact measurements of all rails, which can't be done until I get something in Sketch-up 3D. The bearings and controller are still in the mail. The all-thread is still on the store-shelves, waiting to be purchased.

Various other components are still, "at the store". More details and photo's coming soon.

[JD_Mortal]

I have a few images of the Sketch-up design, which I threw-together last night. This design is minimal, and is designed to allow structure upgrades, related to bearings and rails, without major reworking.

The bearings ride along the rigid edge of the square tubes, for reduced wear and greater linear support. The tubes can be rotated in the event that the bearings have worn-down a groove too deep to compensate with bearing tension. This "new surface" is the unused corners of each tube. One setup should last for four to six rotations, before I need to purchase new guides. This is 6061 aluminum, I believe it is T4.

The first image is the complete setup, minus the table-track details.

The second image is a side-view showing the square-tube with bearing placement.

The third image is another side-view which shows bearing placement for the plunger and the side-rails.

The fourth view is from the lower mounting plate, looking up, towards the stepper motor. This shows the use of the "T" tracks, which have also been cut to "L" tracks for the center linear guides. The rails are not floating, there will be a solid piece of 3/4 MDF or 1/4 aluminum plate mounting and supporting the tracks.

Also missing is the bearing-spacers, axles and mounts. For simplicity, they are not in sketch-up. Simple threaded axles which will be mounted into the MDF, via brackets.

I recently hurt my back, so no power-tools or heavy lifting until I get better. However, jig-saws and drills aren't realy power-tools, so I can use those.

[JD_Mortal]

Just a recap with some progress on my CNC. I had to make a table-saw to cut some decent linear lengths. It took a few hours to shim-up and sand the surface of the table-saw so it is completely flat. However, the blade has a little play from the rubber vibration mounts. I still need to re-cut the pieces, once I have a better linear guide for the home-made table-saw and a longer right-angle.

About 85% of the parts are ready. Now I just need to get the mounting hardware, and I can begin putting it all together. Also note, I had to purchase more bearings. Turns-out that I needed about 52 total, so I purchased another 40 bearings. The precision motor also arrived today, as I was uploading the video. 400 steps per turn, 0.8-deg per step. Getting paid tonight, so more parts coming soon.







Hope that works... The video's should be in the post.

[JD_Mortal]

Well, after waiting a month for my driver/controller combo-board to arrive...

Finally it is here, but not ready to use yet. The board is the "HY-TB3DV-M" version, 3-axis combo-board.

As per all the web-sites, forum-postings, blogs, and general rants online, I have deduced that the actual stats for absolute operation of this board are more like this...

Normal operation voltages: 15vdc-29vdc
(It can handle spikes of 33vdc before it fails.)
(Voltage needs to be over 15v, for the 12v linear regulator to operate correctly.)

However, for now, I am using an x-box power-brick. I could simply remove the 12v regulator and the 5.5v regulator, which runs off the 12v regulator, and use the 5.5v from the x-box and power the fan directly with the 12v power. I do have a 24v @ 6a power supply on the way, though the x-box power will still be used for all my other goodies down the road.

Strong note: With computers now, printer ports are all virtual. Besides bios settings, you ALSO need to setup ports in the control panel. Enable legacy support, use EPP, or ECP, NOT AUTO!... It will NOT detect the connection, because there is no real controller/device/language on the other end. Like a printer or a scanner or a camera. These are hacked data connections, as most all CNC control cards are.

So far, the only problem I have with the board, is the crappy pin-connections. They used pin-1 (Strobe as x-step), and pin-14 (auto-linefeed as x-enable). Why is this a problem? This has to do with most new LPT/PP cards, cards made in the last two decades for consumers. Pin-1 has severely fast-speed ability, and severely low-amps. Since it is a "strobe" and intended to match high-speed hardware, it uses lower amps so the average voltage shows as 0.14v, when the other data-signals show as 0.40v average. EG, it does not trigger the optical-IC with enough power to show actual output. This was tested by applying 5.5v directly to the pin, and even 1.1v, both triggered movement in the motor, but nothing happens when mach-3 tries to STEP on that pin.

On the opposite end, pin-14 is a low-speed, high-amp pin. Since printers only line-feed at the end of a line, it has a slower and stronger pulse, which causes "issues", like missing steps and over-heating, since it keeps enabling and disabling the motor. The high-amp line is not intended to "stay on". The LPT card will turn it off, to stop it from frying, or it will toggle the line, which would not interfere with a "line-feed", but it will interfere with the "enable motor" signal. Eventually, it will destroy the LPT card, on some cards.

Soooo...

Resolution...

I have to rewire those two pins, and possibly relocate some others to more logical locations, like where they normally belong on a printer-port. Just a small hurdle.

As for the other lines, they seem to function fine. The Y-axis didn't like to spin reverse alone, with odd stepping, odd amp-values, and odd ramping/decay. Funny, running any direction on the X-axis instantly corrected the jittery reverse motion on the y-axis.

As for speed, I can get these motors up to 2000 Hz, which is about the stated maximum for the motor specs. While in operation, I have the ramp max setting to 2400 steps/in @ 48 in/min and 5 for step-pulse setting. That is the best I can get. Then the motor starts to bind. Though, in all honesty, I have no clue what I am doing for half the settings. They are not as transparent, even with some explanations. Add that to the funky, blind, settings I am throwing into the driver-board, and disaster is just around the corner!

Just for good measure, I also picked-up a 4-axis controller board (the red PCB ones) made by the same guys, which is said to have better/correct PCB wiring and response. Also another two stepper motors, which I might use for dual-rail-drive. More power!

Also, my second set of 40 bearings came-in, as well as my 8 thrust-bearings. Digital calipers on the way, and so is the new power-supply mentioned above. Now it is time to dissect my computer, a sacrifice for the offering to the CNC GODS at this forum, all of whom, this would not have been possible! (No, you don't get the computer, it is an offering, not a gift. It stays with the lamb!)

[RomanLini]

This was tested by applying 5.5v directly to the pin, and even 1.1v, both triggered movement in the motor, but nothing happens when mach-3 tries to STEP on that pin.

I remember reading here on the forum somewhere that Mach3 can produce very short (too short) output pulses which can cause issues. You could check if that is the case, and you can get Mach3 to send a longer pulse.

If you post a picture or link to your controller board someone more familar with that board or its driver chips might be able to suggest power supply and setup options.

[JD_Mortal]

The link to the board I have here, HY-TB3DV-M
The link to the other board I purchased is, RKCNC-6560-3V
* Correction, it is not manufactured by the same guys... I saw the comparison, and assumed. Lol, once translated, I see they are making fun of my other board! Oh, those competitive Chinese... hehe!

Yes, I found the mach-3 settings for the pulse-lengths. I found that the controller I have, likes pulses at 3-5 for timing. (Micro/Nano-seconds? I am unsure what the actual value relates to. It was in the motor tuning page, the last two boxes on the left side.)

As for the board settings... It is temperamental since it uses the TB6560AHQ chip. (That is a hybrid controller.) The ultimate settings depend on the combination of voltage, available amps, motor coil-type, ohms, load, motor limits, and resistance of drives. EG, tuning it now is actually useless.

Once I get the correct power-supply, and have rewired the pins to the correct locations, and have built the cnc... then I will have to play with it all over again. I just wanted to be sure it worked, and get an idea of my unloaded speed potential. For $50, I don't care if it does blow. I have another better version on the way, and this will give me some spare parts, or it can be used as two additional axis' if one does blow. Like for a tool-changer and for my 3D laser-scanner/digitizer.

Some of the data I found here, about this board, but more info was found on the RepRap websites. They are the largest consumer of these boards, besides the robot community. (Those are also low-load users, who keep the boards operating well below limits. I intend to ride the upper-limit.)

Technically, the best modification to do to the board, besides using the additional data-buffer, would be to add some signal amplification on the computer side of the optical-isolated chips. It is horrible to assume/expect that any LPT/PP is going to have enough power to power an optical-isolated chip. The computer is only expected to provide "A" signal, not "A signal at an expected volt/amp". The LPT/PP is not intended to "drive" anything, just expected to relay pulsed DC as data. All these controllers I see, demand that the LPT/PP be some industrial style regulated power-driver source. That might be true for a hand-full of industrial LPT/PP, but not for any consumer LPT/PP cards. They tend to be "power-saving", and as per "energy-star ratings", they tend to use/waste as little power/amps/volts as possible. It is a low-level data-line, not a high-level peripheral controller interface. Industrial LPT/PP don't care about saving the owners money. They are built for solid operation and have lots of wasteful amps ready to dump into anything plugged-in. (That is partially because the industry usually attaches large printers over long wires or they abuse the cards, using them for things like CNC controllers, directly, without an actual controller card.)

That is why, ultimately, I will probably build my own controller if I can not find a reasonably priced Gecko, or other industrial controller. Sorry, those $30 of parts used to make a gecko are not worth $400 to me. I can purchase a whole computer for the price of one motor-driver. That is why the china-boards are selling by the thousands, per week, and the gecko's are selling only a few dozen a year. 1/10th of the china-boards are failing, due to people pushing the boards to the limits, being newbs experimenting, having little knowledge about steppers, and failing to understand that "limits" = "boom". The china-board manufactures are also idiots for stating the specs they do. Whoever translated the board info and designed the data-in/out pin setups should be fired. I have seen more successful setups with china-boards then failures... but when you have a failure on anything that says "made in china" and is "sold on ebay", they make an added effort to broadcast that failure. Not to mention, you see so many failures due to the simple fact that there are millions of boards in peoples hands... Seeing 100 failures from millions is not actually bad. Though, I have not come across a single Gecko failure video or blog or forum posting, though I am sure some exist. They have better customer loyalty, the Gecko drivers, and thus, less broad-casted complaints. (The website suggests that Gecko-drivers do fry, as one advertisement uses the term, "Unkillable", which implies that the other drivers are prone to "killability".)

[JD_Mortal]

Another update...

I got my three drive-screws. Decided to get 1/2" 13-TPI all-thread, just so I had enough room to turn-down the ends for the 7-mm ID 608zz axle bearings, and the 10-mm ID needle thrust-bearings. I did a practice turn-down, by hand, using a grinding wheel and a 2x4 to support the far end of the screw as I turned it. Came out real nice. Barely 0.08-mm off-center, so there is a slight wobble, which needs shimming to re-center the bearing. Again, this was just a test-run, free-hand, without a grinding/turning jig setup. Next, I will build a turning jig!

The secret is constant and even pressure while gradually turning off 0.01-mm per rotation. One long spiral is more centered than one deep uncontrolled gouge. The hard part was keeping my unmounted grinding-wheel parallel, which has a horrible wobble and likes to walk around. I used that to my advantage. I would let it drift, creating a slightly tapered end, and slide the bearing on, seeing where I needed to shave-off another mm or two. I'll put photo's up once I get my laptop back into the garage. Perhaps I will make a quick video of me grinding another section, for those who are unsure how to do this without a metal-lathe.

I will also show, if I can get a honing stone, a better way to auto-center the final cut on the surface. You still want to get it ground-down with a grinding wheel first. My first mini-axle test was a long bent screw, which was ground with a dremmel-tool and no mounting to support the rod while I turned it. That is horribly off-center, but still works fine. (To me it is horribly off center, which is about 0.5 to 1.0 mm.)

Thanks, RomanLini, for kicking me in the correct direction...

I also got the funky-operating X-axis to work on the china-board, though I will still be re-wiring it as soon as I get a spare LPT/PP cable to dissect. I had to turn-up the pulse-time to 10 us from the 5 us that it was before, and the clipped-strobe pin stayed powered long enough to raise the average voltage above 0.40, from the 0.15 that it was previously. This was enough voltage to trigger the optical-isolator chip. (Seems that the LPT/PP strobe also has a natural 1/3-cycle pulse. Eg, the solid pulse sent to that pin results in {Off-on-off} when you feed it {On.......}, thus, the 1/3 average voltage. Again, this may not be all LPT/PP cards, but it is this way on all mine.)

Ok, so the board now works fine, and I had to dump my double-precision motor due to the bipolar-setup, which was a 6-wire setup. My only option for this china-board driver was to run the double-coils as one long coil, leaving the common un-tied, which raised the ohms too high. Thus, they would have required double voltage or half the amps. (Being that they were only 1A to start, now they were 0.5A, which got too hot with this 3A driver. If I had an 8-wire setup, I could have linked the coils in parallel, and the amperage potential would have been double, as the ohms would have been halved, but the voltage would have been the same. Oh well... I will save that one for my 3D laser scanner.)

Time to get working on the fixed gantry and linear slides. Perhaps I will do the plunger, z-axis, first. I really want to fold the motor around to the back, instead of poking it out the top, which would be in-line with the axle. However, I need to get two sprockets and a smaller bearing for the axle of the stepper. I can do that later. I guess I can rig-it for now, in the linear setup.

I seem to be having trouble locating machine-bolts for my bearings... Everything I find has full-threads on the axle. I need solid-axle, with partial threads. Perhaps Ace-hardware has what I need. The closest thing I could find was at home-depot, for $1.25 per bolt! However, they were not quite what I needed, they were standard inches not mm sizes. If that is all I can get, I will just have to go larger, and grind them down too. I could always go with solid-rods, since these don't exactly need to be bolted, the way I designed it. Time to think... where is the toilet-paper...

[ger21]

Yes, I found the mach-3 settings for the pulse-lengths. I found that the controller I have, likes pulses at 3-5 for timing. (Micro/Nano-seconds? I am unsure what the actual value relates to. It was in the motor tuning page, the last two boxes on the left side.)

micro seconds.

(The website suggests that Gecko-drivers do fry, as one advertisement uses the term, "Unkillable", which implies that the other drivers are prone to "killability".)

Only the Gecko G203V is advertised as Unkillable. If you wire them wrong, you can kill all the other Geckos.

I got my three drive-screws. Decided to get 1/2" 13-TPI all-thread, just so I had enough room to turn-down the ends for the 7-mm ID 608zz axle bearings, and the 10-mm ID needle thrust-bearings. I did a practice turn-down, by hand, using a grinding wheel and a 2x4 to support the far end of the screw as I turned it. Came out real nice. Barely 0.08-mm off-center, so there is a slight wobble, which needs shimming to re-center the bearing. Again, this was just a test-run, free-hand, without a grinding/turning jig setup. Next, I will build a turning jig! The secret is constant and even pressure while gradually turning off 0.01-mm per rotation. One long spiral is more centered than one deep uncontrolled gouge. The hard part was keeping my unmounted grinding-wheel parallel, which has a horrible wobble and likes to walk around. I used that to my advantage. I would let it drift, creating a slightly tapered end, and slide the bearing on, seeing where I needed to shave-off another mm or two. I'll put photo's up once I get my laptop back into the garage. Perhaps I will make a quick video of me grinding another section, for those who are unsure how to do this without a metal-lathe. I will also show, if I can get a honing stone, a better way to auto-center the final cut on the surface. You still want to get it ground-down with a grinding wheel first. My first mini-axle test was a long bent screw, which was ground with a dremmel-tool and no mounting to support the rod while I turned it. That is horribly off-center, but still works fine. (To me it is horribly off center, which is about 0.5 to 1.0 mm.)

Here's how I did mine. Close enough to call them perfect for any home built router application.
http://www.cnczone.com/forums/diy-cn...out_lathe.html

On a machine that size, you don't need thrust bearings. I use a single skate bearing as a thrust bearing, and my gantry is at least 75lbs. No problems in a year of use, and if they do fail, it's a $1 repair.

[JD_Mortal]

That is similar to how I did mine, except the only stone I have is on my grinding wheel. That is what I had in mind for my jig, except the jig would be on a pivot, with a long handle and a stop-screw to limit depth while I grind.

I didn't think about using a file as a grinding-tool. Seems like it would be easier to do it your way, with the addition of a file-jig, limiting the depth with an adjustment screw. Much better than trying to use my table-grinder.

Yea, I didn't think the thrust bearings were super-important, but they were only a few-bucks for 8 of them. My skate-bearings have "too much" thrust-play for my taste. I was hoping that they would reduce my over-all screw-drive backlash and extend the life of my skate-bearings by removing all thrust forces. (Eg, my skate-bearings can slide along the shaft, which helps extend the race-life.)