[wheeliecake]

They were rather uncooperative with me as well. If qc issues like this persist, there's no way I'd recommend this mill. Of course, there aren't a lot of options for the price.

Thanks we shall see, currently waiting for the new mill to arrive, they promised me an update, either email or by phone, to tell me when it had shipped (two days) and once again no communication from them.

Nice build by the way, the enclosure looks excellent. I was initially going to buy an old used Bridgeport but in my price range the ones I looked at were in really bad shape and I don't (yet) have to knowledge to fix/correct things. I'm now regretting buying from grizzly and wish that I had sourced from someone else but there seems to so many positive reviews how bad could it be, right...

[manton]

As for the Z adjustment handle coming off, just push the pin in. These work loose frequently. Mine does it as well. If you want to remove the pin, you can thread a bolt into it, and pull.

The Y axis ways look identical to mine. I suspect your new mill will look like that too. I suspect it is done for oil retention.

[zamazz]

I've been steadily plucking away at the mill, and have reached the electronics phase. I've mounted everything in the box, which is a challenge even with a 18 x 24 box... Linux is all installed and I've played around with the configuration for the Mesa 5i25 and 7i77. This is my first venture into Linuxcnc, so it will take some time to learn about the hal files, configuration, etc. So far everything has been straightforward, except for the driver for my elo touchscreen. The touchscreen works, but the alignment is off. When I attempt to run the calibration, it throws an error. I'll have to get someone a little more Linux savvy to have a look at it.

The servos all work fine in the tuning software, but I'm having issues controlling them via the Mesa board. When I purchased the system, I checked compatibility and found that the Mesa output analog signals and the servos could be set to receive analog signals. The plan was then to close the loop in Linuxcnc with linear encoders, and have the DMM system close itself with its angular encoders. However, when I went to wire it, the Mesa board output +/- 10v analog while the servos run off of 0-5v analog in addition to a clockwise/counterclockwise signal.

What would be the best way to remedy this? My current thought is to use a PIC (or Arduino, as I have a bunch lying around) to read the +/- 10v and scale it to 0-5v and also send a signal depending on the sign of the voltage. Is this a good approach, or will it cause issues I'm overlooking? Also, will PWM work on the 0-5v input to the servo controller or should I use a DAC or low pass filter?

[SCzEngrgGroup]

What you're proposing will work horribly, because you will add considerable latency to the signal, which will screw up the servo response. I don't see any way to fix that other than by using a precision op-amp to scale the signal properly. Even that will have some issues, as it will be extremely difficult to generate a proper direction signal when the control voltage input is close to 0V. Really, if you want this to perform properly, you need either different servo drives, or a different interface board. Personally, I wouldn't even waste the time to try to make it work with what you have - it'll never be right. If you have any programming ability, I would consider swapping out the Mesa board for a KFlop, which would be able to drive the DMMs properly.

Regards,
Ray L.

[andypugh]

However, when I went to wire it, the Mesa board output +/- 10v analog while the servos run off of 0-5v analog in addition to a clockwise/counterclockwise signal.

That's a little awkward, but should be solvable in HAL.

You need to use the HAL "abs" function: ABS to keep the output in the positive range, then use the abs.N.is-negative output pin to drive a GPIO pin that controls the drive direction pin.

Setting the analogue output scale to a max of 5V is probably also wise.

[Connor]

If you have any programming ability, I would consider swapping out the Mesa board for a KFlop, which would be able to drive the DMMs properly.

Regards,
Ray L.

I don't think that's good advice.. He's wanting to use it with LinuxCNC which doesn't support the KFlop. Mesa is perfectly fine for what he's doing. He'll need to check with the mesanet guys or log on to the IRC channel and ask some questions.

[zamazz]

What you're proposing will work horribly, because you will add considerable latency to the signal, which will screw up the servo response. I don't see any way to fix that other than by using a precision op-amp to scale the signal properly. Even that will have some issues, as it will be extremely difficult to generate a proper direction signal when the control voltage input is close to 0V. Really, if you want this to perform properly, you need either different servo drives, or a different interface board. Personally, I wouldn't even waste the time to try to make it work with what you have - it'll never be right. If you have any programming ability, I would consider swapping out the Mesa board for a KFlop, which would be able to drive the DMMs properly.

Regards,
Ray L.

I had wondered if delay would be an issue. I considered Kflop initially, but reading your posts told me it would require significant programming, which I could probably do, but would rather not. Also Kflop seems pricey when you put it next to a Mesa board, and there is far more documentation on Linuxcnc than Kflop. No doubt they're both good products, but I've got the Mesa, so I'm going to get it to work.

That's a little awkward, but should be solvable in HAL.

You need to use the HAL "abs" function: ABS to keep the output in the positive range, then use the abs.N.is-negative output pin to drive a GPIO pin that controls the drive direction pin.

Setting the analogue output scale to a max of 5V is probably also wise.

Ah ha! I knew I could scale the output down, but I didn't know if there was an abs command that would work. I also thought about setting an output to control direction, but once again had no clue how. I'll look into what you suggest, you may hear back from me in about a week if I can't figure it out, or I'll post on the Linuxcnc forum. I've been going through the documentation, but there is a bunch of it, and some of it is only half complete... Thank you for the suggestion though!

[zamazz]

That's a little awkward, but should be solvable in HAL.

You need to use the HAL "abs" function: ABS to keep the output in the positive range, then use the abs.N.is-negative output pin to drive a GPIO pin that controls the drive direction pin.

Setting the analogue output scale to a max of 5V is probably also wise.

Took your advice, and after a little reading and playing around with hal I got the card to output 5v, always positive, and set a pin when the output should be negative. Got it wired up and the drives behave as they should, apart from the fact that the scaling is way off (I'll fix it once I get everything on the machine).

Two problems:

The CW/CWW output pin is running off of field I/O, which is supplied by a separate 12v supply. I'm assuming (the documentation could be better written) that the servo drives require a 5v signal for direction. Currently I have a transistor type voltage regulator cutting the voltage. Anyone think this will cause an issue?

Finally, with the servo drive in velocity mode, everything works correctly when the computer is powered on. However, if the computer is powered off, the servo turn the full 3000 rpm, which is obviously less than ideal. After combing the documentation (several times) I realized that the speed can be controlled by a potentiometer across the 0-5v terminal and ground when in analog mode. Apparently a low resistance means no motion, while a resistance of about 1000 ohms will give full speed (just speculating here, nothing is mentioned about it in the manual). I noticed that with the computer on, there is no resistance between the terminals, but one it shuts off the resistance goes to about 800 ohms, causing the unwanted rotation. To try to fix this I installed a lower value resistor between the terminals. To get the motion to stop, however, required a 25 ohm resistor! With a value that low, if 5v were to pass through it, it would exceed the power rating of the resistors and draw 200 milliamps from the card per axis, which is risky considering a 2A maximum for the drives and encoders (of course, all three axes would have to be running at full speed simultaneously). I'm not certain how the board works, however. Will much voltage be going through the resistors when it's powered on? Is this something I should be concerned about? Anyone got a better way of dealing with this?

[andypugh]

The CW/CWW output pin is running off of field I/O, which is supplied by a separate 12v supply. I'm assuming (the documentation could be better written) that the servo drives require a 5v signal for direction. Currently I have a transistor type voltage regulator cutting the voltage. Anyone think this will cause an issue?

I think I would use an opto-coupler in that situation, driving the DIR pin with 5V sourced from the drive itself.

Finally, with the servo drive in velocity mode, everything works correctly when the computer is powered on. However, if the computer is powered off, the servo turn the full 3000 rpm, which is obviously less than ideal.

Page 6 of http://www.dmm-tech.com/Files/Specification_DYN2.pdf shows the input circuit.
There is a 4.7k pull-up on the analogue-input pin, so if the analogue input pin is open-circuit the motor runs at full speed.
I think it might be worth contacting DMM to ask if there is an on-board jumper or config option to disable that.
The other option is a small normally-closed relay that gets opened by a drive-enable signal from LinuxCNC.

[andypugh]

The DYN3 drive has a conventional +/-10V input, and a proper drive-enable pin too.
Not having a drive-enable and also having a mode that runs motors full-speed when not connected seems like something of a quirky design.

[zamazz]

The DYN3 drive has a conventional +/-10V input, and a proper drive-enable pin too.
Not having a drive-enable and also having a mode that runs motors full-speed when not connected seems like something of a quirky design.

I agree. I don't think they intended these drives to be run as standard servos, but rather as replacements for steppers and controlled with step/direction signals. Had I known this originally I may have gotten the larger servos with they DYN3 drives or continued shopping. It's worth noting that none of this has been the Mesa Boards fault. The 5i25/7i77 has worked perfectly since I plugged it in, and the documentation is great, its just the drives that are being difficult.

[zamazz]

Back at it! After a very busy summer (including moving) I'm starting up again. I worked several times throughout the summer, but didn't seem to get a whole lot done. The electronics are functioning well, after sorting out the interfacing issues with the DYN-2 drives. To get the direction signal to work properly, I had to implement a simple NPN transistor and a few resistors, since the drives actually needed to be grounded to reverse direction, instead of the positive signal given by the I/O card. The electronics have been mounted in the box, which, in hindsight, is not all that big. The wiring is very difficult to do cleanly, and I'm just not happy with the way it's turning out. To remedy this, I'll probably construct a larger box and mount it to the side of the enclosure once it's finished, which will give easier access and get it out from under the coolant pan. This will also leave room for nicer coolant drain and chip collection tray. For now, the box will work till the enclosure gets built.

As for the electronic box, I started off by cutting some square holes in box for fans. I quickly found out this isn't as easy as it looks. I ended up using an angle grinder with a cutoff wheel to rough the boxed in and a Dremel to attempt to clean them up. Ultimately it worked, but I was wondering if anyone has a better method. I may try a jigsaw when I build rebuild the box. I then popped all the holes out of the opposite side for the air to exit. Originally, I planned to use an air filter over the openings and fans, but since I've decided to scrap this box, I'm not going to put anymore time into this one.





Next I decided to get a little work done on the stand/enclosure. The next logical step was to create a chip tray. Given the way my base was designed, I decided to place the chip tray underneath the two mill supports. I saw this at one point on this forum, but forgot who's build it was. This configuration will allow plenty of table clearance, and chips and coolant should easily flow out of the machine to the reservoir underneath. At this point, the mill was a mess from being shoved into a corner all summer. This picture shows the state it was in at this time. You can see the linear scales, which I'll get to later. Also, on the far side is a 3" triple stack air cylinder with a 2" stroke. I'm considering using this for the power drawbar, since I got a great deal on it. I'll probably use a lever reduction, although at 100 psi it should generate about 2000 lbs for force.



To build the stand, I went to my local metal supplier for some sheetmetal. It turned out that all they sold was 16 gauge and higher, which I though was a bit thick, but it ended up working out in the long run. Without a plasma cutter (or torch), cutting sheet metal has always been a major PITA for me. I looked for alternative ways to cut sheet metal, and settled for a 16 gauge nibbler. This worked well, for about 2 inches, at which point my compressor ran out of air. An upgrade to a 60 or 80 gallon is in the near future... After this, I went out and purchased a harbor freight 16 gauge shear. Suprisingly, this works very well so long as I want straight lines, which is 90% of the cuts. They have plenty of power for the 16 gauge steel, and seemed to cut well all through the first project. Normally I'm against cheap power tools, so I may upgrade to a DeWalt shear in the future, or perhaps just spring for the plasma.









This went fairly well, all things considered. MIG welding this steel isn't much fun, but a TIG is still out of reach. There was quite a bit of warping, even after welding in 1/4" long beads, which is why it's probably good that I didn't have a thinner steel. I'm currently working on sealing all the holes so it will be water tight, then I will prime and paint it.

As for the Linear scales, the first thing I did when they arrived was rip the connectors off and screw them into the terminal blocks on the Mesa board. The 5i25/7i77 configuration for LinuxCNC has been working great, and this was no exception. I simply connected them to the correct terminals and they worked! After changing the pulse per inch value and the direction in LinuxCNC, they were set, no further calibration needed and the machine will always hit its target no matter the leadscrew error. Additionally, the max allowable error is (currently) set to 20 thousandths. If the machine deviates by more than this it will throw an estop condition. After running a Tormach for the past few months, I cannot wait to have this. The Tormach has given me nothing but trouble, as it seems to always be losing position or missing steps. This is mostly Mach's fault, I'm sure. If there were a LinuxCNC config for the toolchanger I would switch without thought. I digress...

Mounting the X axis was extremely easy, it just involved drilling a couple holes into the front of the machine. The Z axis was similarly easy, although required a couple of standoffs. The Y axis, however, took some creativity. I first fabricated a shelf for the scale to sit on. It was quick and ugly work, but functional. It will be one of the first pieces I replace when the machine is making chips. On the sliding end of the scale, there was very little clearance and little room to mount a bracket to. I ended up using the Bridgeport at work to create a .375 x .375 L bracket to slide in between the scale and the saddle, just underneath the ballnut. It's shown in one of the pictures. Once this was complete it was a matter of bolting it all together.









Next up will be the limit switches, which, although they aren't required to run, I would like before I run any programs. I purchased these from DMM, and was expecting round threaded sensors, but got these instead. They will still work perfectly, but it does change my mounting plan. This mill is quite crowded already, so mounting these will be somewhat challenging. With the column space in place, there is a little extra clearance between the saddle and the column. I will place the limit switches on the backside of the saddle, and affix the interrupters to the ends of the table. This should still leave enough room for the Z axis way cover. On the Y axis, I will simply put them on the slanted portion of the base and affix interrupters to the saddle. The Z top will go as shown, with a shared interrupter for the top and bottom located on the carriage.

I apologize for the poor pics, it is quite dark in the new garage at night.







I'd like to have a running mill by the time school gets back in session in a month. I'm not guaranteeing that it will happen, but I'm pretty close. I've had all 3 axes running simultaneously and following g code commands. A little more tuning will get it reliable, and I also have yet to preload any of the axes, which makes the feedback positioning a bit unstable. Also, I will be machining oil grooves in the ways for an automatic oiling system. I have an idea for an oiler that may significantly simplify the implementation. I'm sure I'm forgetting several other things I have done. There is a lot of work that goes into the seemingly minor details, which isn't shown here. I hope to continue this next weekend with mill set up and preloading, since the mill is totally apart right now for the oil passages. I'll try to take more pictures and clearer pictures. Let me know if you have any advice for the sheet metal work. Thanks,

Zach

[zamazz]

Well, I believe I've finished milling the oiler passages. To do this I threw a 3/16 ball endmill into the trusty Bridgeport, and put the saddle in the vise. I was careful as to not apply too much force, as I could feel the saddle flexing under compression. I laid out the grooves roughly with a sharpie, then followed the markings noting my dimensions on the DRO. Initially I started with a 50 thousandths DOC, but this seemed excessive, so I lessened it to 40 thou. I ran the groove on all four parallel faces of the saddle.





Next, since I wanted oil not only on the parallel faces, but also on the dovetails, I began looking for a suitable cutter. After some searching, I found a V-shaped woodruff cutter that would do. A radius woodruff cutter would be ideal, however I wasn't going to buy a cutter just for this. This went smoothly, as I only cut a line in the non-gib side of the saddle and the Z axis carriage.



Next I drilled all the connecting holes, which was no easy task. With the linear scales in place, there isn't a surplus of available area in which to install the oiler fittings. Especially on the gib side, the clearances were minimal. Also, I added holes to push oil through each gib, to distribute it between the gib and the dovetail. Overall, I ended with 12 oiling holes in the saddle and 4 in the Z carriage.

Next, I needed to cut distribution grooves in the gibs themselves. Here is where it got interesting. Tapered gibs aren't the easiest things to hold on to, what with not having parallel faces. I tried a couple things unsuccessfully, before finding a working method. Eventually, I used two small clamps to attach each gib to a .25 x 1.5 parallel. I then clamped the parallel in the vice on top of another parallel, as to clear the clamps. Knowing the taper to be approximately 70 thousandths over the 7 inch length, I shimmed the lesser tapered side of the parallel with a 50 thousandths gauge block (which was a good approximation, since the gauge block was not right at the end, but rather closer to the center). This left the surface of the .25 x 1.5 parallel angled with respect to the table, but the top surface of the gib nearly perfectly parallel. With this attachment, I milled each of the gibs with an oil slot on the sliding side and a little relief on the non-sliding side, so that oil could make it through the hole even if the gib was pushed in further so that the hole in the saddle and the hole in the gib were no longer co-linear. The holes were put in prior to milling, using a transfer punch and a drill press. Overall it worked nicely, and although the oil passages don't run the full length of the gib, they will certainly be better than nothing and should allow plenty of oil on the mating surfaces.







I probably won't install the oiler for awhile, but it will be nice to have the lines already in place when I do. All that will be required is the oiler itself and some distribution manifolds. Shown are the complete pieces with the oil grooves milled.

[zamazz]

Carrying on, I decided that it was time to tackle a couple of side projects important to finishing the G0704. The primary one was constructing a sheet metal brake, which would be necessary to bend the remainder of the enclosure as well as the covers for the motors and electronics. Purchasing one was out of reach, since they start around $300 (for one that is too small) and increase quickly. Someday I'd like to get a small 3 in 1 machine, but there are other priorities. I began by googling DIY sheet metal brakes, and saw a number of possibilities, from copies of commercially available brakes to light duty designs that would never handle the 16 gauge I currently have. In the end, I settled for a design that was simple, easy to fabricate, and seemed to do alright with 16 gauge material. The original build thread is linked. My build was extremely similar, only stretched to accomodate 48" wide pieces.

DIY 16ga Sheet Metal Brake - Dodge Diesel - Diesel Truck Resource Forums

To begin, I went out and purchased the steel. I got 3 five foot sections of angle, 2 four foot sections of flat strap, and quite a bit of 1" square tubing, as it always comes in handy. Originally I wanted 1/2 ID tubing for the hinges, but they didn't stock anything close, so I opted for 1/2" black pipe, as the original poster did. This has an ID of .625, which just meant I needed larger hinge pins.



Since the mill is currently in about 50 pieces, I re-purposed the stand as a welding/fabrication table. I first cut everything to the exact length needed, and smoothed the ends with a flapwheel. I then began notching the angle for the hinge bodies. This proved to be no easy task with an angle grinder, and took 3 cutting disks and a couple hours. Eventually, however, I got them roughed in. Next I chopped the pipe into 1 inch sections, four per each hinge. I quickly noticed, when trying to insert a 5/8 rod, that the pipe has a seam on the ID. To get rid of it, I simply put a sanding drum on the Dremel and ground it out. I placed the two notched pieced of angle together, as shown, and positioned the hinge bodies as shown. It is important to get the hinge directly on the bottom surface and directly between the two pieces of angle. After some finagling, the first section of hinge was in place. I then tack welded only the first piece to one of the angles. Following this, I inserted a pin and the remaining 3 sections and tack welded each of them to one side, alternating each time. Doing this ensured that the sections were all colinear and that they were correctly located on the corner of the angle. I did this for each side, and then tested the rotation. The first try was a failure, the angles would not move. With a little force I managed to break the tacks, and noticed that one of the hinge lines started out correct, but drooped toward the end. A little grinding and re-welding corrected the issue and the second try worked well.





I should also note that the brake required a bench, which I did not have. So I also constructed a simple bench, 2'x6', fairly sturdy with a 3/4 top. I notched the front beam for the bolts to pass through into the brake. Once the bench was complete, I test fit the brake. Everything was passable, so I drilled 1/2" bolt holes through both the steel and wood.





At this point, I added reinforcement to two pieces the rotating one and the one acting as a stop. These will see fairly high loads. I took a length of flat strap and welded it at 45 degrees to the angle. With a 20% duty cycle on the welder (at full power) this took some time. When finished, I built a handle. Unlike the original poster, I made mine full width, so that it would not be directly under a 4' wide piece of material. I constructed the sides from 1" square and the cross bar from 1/2" pipe. Finally, I measured the offset for the 16 gauge steel (since it needs to be behind the bend axis) and drilled the holes in the top stop for the 1/2" bolts.





All there was to do now was see if it worked. I threw in a piece of 16 gauge and gave it a pull. For as simple as it is, it works pretty well. The bend radius is quite large, but that may just take some adjusting. Also, this 5/8" rod does not spin freely in the hinges, it binds slightly. It was a very tight fit before welding, and must have gotten slightly misaligned during welding. I will machine new pins slightly smaller from steel during the week while I have access to a lathe. Overall it was a success, considering I only have about $100 in materials. How it handles 48" wide sections is yet to be seen. I have my doubts that it will be able to bend full width 16 gauge sheets. If this is the case, I will probably bend the large pieces with 6" tabs spaces every foot (ie six inches of material then a six inch gap then another tab, etc). This will still give plenty of area to affix the panels, yet only require half the force to bend. Here are the results:





What with the holidays, it could be awhile til my next post, but hopefully I'll get the chance to work next weekend. Thanks for reading!

-Zach

[zamazz]

Well, I got a few things done in the shop at work over break and also received a few new parts for the mill. Among the parts received were disk shims with a .500 ID to be used for shimming the pairs of angular contact bearings on each of the three axes. The only issue I had was that the shims were only sold in .001 increments, and I'd like to get the backlash to a lower value then this. In an attempt to remedy this, I took the .003 shims (I purchased .001, .002, and .003) and surface ground a tiny amount from them. Originally I tried to place them atop another block which I ground flat, but the electromagnetic chuck refused to hold the shims. It was like they were similarly charged magnets repelling one another. This may be totally normal, I just haven't used the machine enough to have experienced this. The shims did stick directly to the chuck, however, so I proceeded to grind them in this fashion. They were somewhat loosely held, as were apt to slide if I took too deep a pass, but after several .0001 passes I took a half thousandth off of several of the shims.



Also, while I had the lathe, I turned new hinge pins for the brake. I used a O1 tool steel drill rod of 1.25 dia and turned a four inch section down to .505, which slid nicely into the pipe hinge bodies without being overly loose.





In the couple of hours I had to work today, I managed to get some reassembly of the mill completed. I preloaded both the x and y ballscrews with the double nut arrangement. To do this, I used small belleville washers underneath the screw heads on one side. The other ballnut was left rigidly mounted. I then tightened the screws evenly until the screw was quite difficult to rotate by hand. Ideally I'd put the screws in tension, as I have heard that this is better, but it would require a different set up and take up more space, reducing travels. I doubt that their being in compression will affect their lifespan much if at all. Note that the y preloaded ballnut is only affixed with two screws and nuts. I initially wanted to use 40mm screws through one ballnut flange and through the block, but could not find full thread 40mm screws. In the end, I simply turned two of the tapped holes in the y axis ballnut mount into through holes and inserted the 40mm screws through. I then placed the spring washers underneath a nut. It ended up working nicely, since the diameter of the ballnut stopped the nut from rotating. I simply tightened the screws till rotation was difficult.



I then inserted and attached the y axis ballnuts and screw, and put the saddle in place. Also shown in the picture are the column spacers. I've had these made for awhile, but kept ordering the wrong screws to replace the stock ones. I finally got my hands on the correct ones, 260mm in legnth, and they work decently. I don't, however, have the spacer made for the head, so after test fitting the blocks I took them back off to finish assembling the y axis. I'm a little worried that this much column spacing will significantly reduce the rigidity of the mill. If this is the case I may take one of the two spacers out, or build a brace for the column.





Next I began shimming the AC bearings on the Y axis. I got all the way to .007 of shims before I noticed any difference in the difficulty of rotating the screw. Is this normal? At .007, I checked the runout, and could only get the tenths indicator to deflect .0002 with a hard push on the screw. I'll check the actual runout once I get the electronics back together. I then worked on the x axis, but didn't quite finish tonight. It's cold out there for us Texans, around 25F outside right now.



I should get quite a bit done this week, as I'm off. Hopefully I'll get some test cuts by the end of the week! Thanks for reading,

-Zach

[zamazz]

I managed to get the mill put together. This is the first time in about a year (since I bought it) that the head has been back on the machine. I ordered some collets and a couple endmills, so hopefully I'll see some chips shortly! I've put several hours into servo tuning, I've got the x and y axes tuned to within a couple tenths, so that the axes are only off by a couple encoder counts throughout point to point moves. This was all done at 120 ipm, which I figure will be about as fast as I'll ever need to cut. Rapids are currently set to 200 ipm which is close to the motor and pulley combination's limit. The inner loop, closed by the DYN2 drives, was difficult to tune, as there was no indication of what was helping or hurting. I ended up getting them pretty close, probably a bit overdamped, but it seems to have little effect compared to the tuning in LinuxCNC. As for the LinuxCNC tuning, I got what I believe are somewhat weird numbers. The P is set somewhere in the 120-150 range for all axes, while the I is about half that and the D needed to be very small. The D worked best set to around .75. Setting it to 1 caused oscillation under acceleration, and having none at all caused many problems. I also had small values for FF1 and FF2, and set the deadband to .0001. It's working quite well right now, I may get some halscope screenshots when I get the chance. Note that only the X and Y are double nut preloaded at the moment. I need to turn my one piece Z axis mount into a two piece one before I can add the second nut to the Z.



With the mill pretty much ready to run, pending the delivery of collets and a clamping set, I started to work on the enclosure. To start, I plan to add a pan with 'wings' extending to either side, angled in as to drain coolant. I'll then bend flanges up on all sides for the front and side panels to attach. Using 16 gauge should make it rigid enough to support itself. Starting on the pan, I marked the outline on a scrap of 16 gauge I had. Each piece has two 90 degree bends, one to going down into the stand's frame, the other upward to be the flange. The corners are mitered to 45 degrees, so that the four pieces can be tacked together. I then cut the out the two blanks using my handheld shear and cleaned them up with a flapwheel. Now I just needed to test out the brake. The bends were 30" and 36", so I figured I'd start with the 30". I put the piece in the brake and tightened it down after aligning it. I then bent it about 10 degrees before everything shifted. I had made the (stupid) mistake of slotting out the mounting holes, so that I could adjust the top piece forward and backward for different thicknesses of material. The problem is, there is so much force and the brake that the top piece simlply slides all the way back when trying to bend. So, after some frustration, I lined it up where it needed to be and drilled another half inch hole in each side for a dowel pin to sit in. This again prevented it from moving under force. I then put the 36" bend in the brake and pulled the lever. It took a good amount of force to bend it, and was pretty close to the limit of the brake, I'd say. The table was on the verge of tipping when I got to around 90 degrees. There was also a considerable amount of bending on the top beam, probably and eighth of an inch at the center. That's telling of just how much force this took. Bending a piece of reinforced 1/4" angle iron isn't exactly easy. I'll probably add a structure above this piece to prevent this in the future, like you'd see on any commercial brake. I also will probably make the sides of the enclosure out of a thinner material, either 18 or 20 gauge. Overall though, the bends are decent. I'll try to finish the rest of the pan today.

-Zach

[SCzEngrgGroup]

I am confused by your description of "inner" and "outer" servo loops, as I don't see any linear scales on your machine. Are you running both loops off the motor encoders? If so, that's really doesn't make any sense - the two PIDs will be fighting each other, with no functional benefit (more likely, quite the opposite). If you have both rotary encoders and linear scales, you want to use the servo drive PID to control velocity, and the LinuxCNC PID to control position. But with only a rotary encoder, you should only use the servo drives PID.

Regards,
Ray L.

[wizard]

Looking good!

It's been awhile since I stopped and looked at this thread, you have made really good progress here.

[zamazz]

I apologize ray, I thought that the picture of the full mill was the current one, but I did a bit of work after that, including putting the glass scales back on. I am running linear encoders, so there is a loop in the servo drive as well as the linear loop closed with the Mesa board and LinuxCNC. Here is the current picture.



Thanks Wizard! I'm really hoping to have it mostly running before school gets crazy again, so I've really stepped it up the last couple of weeks.

[pippin88]

My bearings from Chai have required ~0.25mm (~0.01") of shimming between the outer races.