I just noticed that the back plane is not connected electrically to the cabinet. It seems this enclosure has isolating bushings around the plate mounting bolts. Should I install a jumper between that mounting plate and the outside enclosure?
I am also wondering if there i any reason to NOT use a DMX or XLR type connector for the VFD cable into the cabinet?
Thanks for the help,
Robb
Last edited by rcmcdee; 03-06-2018 at 06:51 PM.
My cabinet has metal standoffs. I haven't checked continuity, but it should be connected to the back plane. I did add a wire to connect the door to the main cabinet. There are studs on both for that purpose.
No reason to NOT use a connector for the spindle cable. Just make sure it will handle the current. It would be best if the connector body was metal and grounded to the cable shield, and chassis. I avoided this by just passing the cable through the chassis wall and keeping the unshielded section as short as possible.
-Robert
Here is an image of how I did the motor cable. You just remove about 2.5-3 inches of the outer jacket and cut the internal wires as shown. The 2.5-3 inches allows the ground wire to connect to the motor chassis and gives enough length. to terminate the motor power wires.
Robert
Neat trick but I can't help but think there is a more elegant way to do this. It turns out the 4m cables I ordered are too short for the 4 x 4 machine. I am imagining some sort of connector at the exit from the end of the cable chain that would provide red/black/green to the near motor then red/black/white(green) to the second motor. I'm not sure but it seems the primary issue is to have separate grounds for the motors? If I crimp spliced through do you think that would be a problem?
As a side note we did not receive the large 2 stage drive gears and cams for the Saturn 2. I left a message for Nate but have not heard back from him yet.
thanks for your help,
Robb
Fine Line Automation
www.finelineautomation.com
Last edited by rcmcdee; 03-09-2018 at 08:27 PM.
I'm not an expert, but my understanding is that safety grounds should generally not be spliced. If some failure causes high voltage to short to the chassis, you want the ground wire to have a very low resistance to ground. The splice is a point where the resistance is a little higher and theoretically would be a place where the wire could burn up in a high current situation. Like I said, I'm not an expert, but this has been a requirement for every machine I ever designed.
There are alternatives to how I did it. My main concern was having enough space in the cable chains, so I didn't want to have to run 4 separate cables. You could run 2 conductor cables for power to the motors then run individual ground wires. These would have to connect to chassis ground somehow.
Your connector idea is probably no worse than the fact that the ground goes through a connector at the panel anyway. My only comment is that connectors are the least reliable part of an electrical system.
I understand your comment that the solution I used is not the most elegant. With thinner insulation on the cables, you could probably fit 4 cables, but you might have to solder custom connectors to them...trade offs.
-Robert
Thanks for the input Robert. Your concern about the splice on the safety ground being a potential hazard makes sense. I am now imagining a soldered splice that lays within the upper end of the cable chain and branches out with 3 conductors to each motor. That would use the two 4 conductor cables to near the end of the cable chains then branch to the motors. I would think that the soldered splice would reduce the resistance concern compared to the crimp.
Is there a trick to opening the cable chains? I am having a really tough time with that.
Robb
I would avoid soldering the splice. I've been reading up on this a little tonight, and I have found a couple of references to NFPA 79 and OSHA regs that indicate that grounds should not be spliced. Soldering cable makes it less flexible and prone to breakage at the solder joint. If the air heats up during an "event" the solder can melt before the wire would. I would rather see you use wire nuts if you splice. But even then, you need to use bigger wire for the main wire that splits out to the spliced wires.
The multi-conductor splice you are indicating also will create a situation where all of your motors would be exposed to the high voltage short in the event of a problem. It is general best practice to ground all of the metal on a machine. You can create "star" grounds where ring terminals are attached to studs on the chassis with nuts and toothed washers, but the chassis also needs to be grounded with thick enough wire to carry the current of a potential failure. A CNC router is challenging here because so many parts move relative to each other, and ball bearings are not good conductors. In theory, each part that moves relative to the others should be grounded, not relying on bearings to conduct. In my machine, I am relying on the connections between the motors and the chassis and the grounding on the motor.
Here's the thing...most of the DIY machines out there don't meet industrial standards. Mine doesn't, I am sure. I just wanted to make it reasonably safe by making sure that the high powered items (mostly the motors and the stuff inside the control box) are properly grounded. The rules were written for a reason, and I tried to follow the ones I know about. The way I wired the motors is unorthodox because it's kind of a pain and prone to nicking insulation. Not a good production process. To avoid this, running separate ground wires is your best bet.
On on the cable chain, I found it easiest to take a small flat blade screw driver to open up the covers on one side. The screwdriver would be lined up along the length of the chain. There is a little space to pry up the covers. You need to pry both ends of the covers (across the width of the chain). Be a little gentle, though. The plastic is glass fiber filled, so it is a little brittle. I broke a couple of them, but fortunately I needed to remove a couple for wire routing anyway.
Robert
I routed all the machine cables and I got one splice completed yesterday using crimp connectors. I did try to solder but found the results unsatisfactory as i was trying to minimize the spacing between the staggered joints. I reverted to the butt crimps with a longer spacing on the stagger. Everything is pulled to proper position in the main cable chain and it seems to flex well without stressing the splices. I do think that next time I will simply use more connectors at the panel and run a 3 wire cable to each motor. Those yellow ZIP cables are beautiful and very flexible. It seems that 4 3 conductor cables will fit reasonably in the cable chain - just more money.
I have a Swiss engineer friend who has a company that produces super high precision positioning machines -> https://nutec1.com . He has taken an interest in my DIY project so I will chat with him soon about this grounding issue in more detail. Our brief conversation to date has indicated that grounding is of the highest importance but he does differentiate between safety ground and clean ground. I suspect Faraday may be lurking in the details. I will update when I know more.
I did the screwdriver brute force approach and broke 2 of the cable chain covers as well. They do sit back in place but come off easily. I imagine hot glue will hold them adequately at the right time.
Robb
If your friend knows half as much about EMI/EMC as some of the gray-beard electrical engineers I know, he will be pretty appalled by my layout. The biggest thing he will point out is that having the VFD in the same zip code (or at least the same enclosure) as the main controller is a bad idea. The big contactor right beside a DC power supply is also probably not the greatest, but that's why I have two 24V power supplies. This machine would not pass IEC standards for electromagnetic noise emissions.
That said, I haven't seen alot of evidence of noise issues. I also don't have that many hours on the machine, so its possible that problems haven't reared their heads yet.
-Robert
Finding time for this project has been very challenging but it is moving forward at turtle speed. Wiring in the controller enclosure is complete and cables are spliced and pulled through the chains. I'm generally pleased with the connectors & connections at this point. I'll post pictures in the next day or so.
My friend advised that "any anomalous movements are usually the byproduct of grounding issues." He offered numerous stories of silly failures in high places where you would expect the expertise to not fail in such a way. Based on his recommendations, the enclosure will not be connected to the backplate ground but will be grounded to the Saturn frame which will be grounded with a separate wire run back to the building plumbing. The backplate serves as the safety ground and the enclosure / machine frame serves as the clean ground. As best I understand, this separation of machine/enclosure eliminates electrical noise from the building power safety circuits.
Today's glitch is about my effort to AutoTune a servo motor. I downloaded and installed the Clearpath software - it appeared to install some Microsoft libraries as well. When I launched it I got a couple of error messages about missing DLLs and check the C++ installation. Now dead in the water. I assume a call to Clearpath/Teknic tomorrow to help get this sorted.
Anxiously awaiting controlled movement...
Robb
Here's a good video on taking apart the cable chain -
David
David
Romans 3:23
CurlyWoodShop - www.etsy.com/shop/CurlyWoodShop
David Falkner - www.youtube.com/user/difalkner
difalkner - www.instagram.com/difalkner
Robb,
Think ahead just a little bit before you go too much further with assembly and consider the motor tuning process. You need to be able to get the moving mass that each motor moves close to correct before you try to tune the motors. You won't be able to tune the dual gantry motors at the same time, so you have to put half the gantry weight on the motor. I know that some people have tuned the gantry motors by using the X-axis drive instead and simply adding some weight to it. However, this does require you to temporarily add some weigh to the Z-axis in order to get the weight right. You just need to know how much the various parts weigh, and that may require some disassembly. It may be worth focusing on getting the ClearPath software running before going much further.
1Jumper10 also worked through this in his build, I believe. I am currently not completely satisfied with the Y-axis tuning on mine. The motors buzz when stationary more often than I would like. I also recently tried cutting some circular bores in aluminum and I wasn't thrilled with how oval the holes were. I need to try to adjust the tuning, but that may prove to be challenging now that everything is assembled.
-Robert
Thanks Robert,
I finally got the ClearPath software running on my Win 10 laptop. For reasons un-explainable (by me or the ClearPath tech) we could not get the SW running on two different Win 7 laptops. I ran auto-tune on the X-axis (with spindle in place) and it kept failing due to losing communication with the motor. I swapped motors and succeeded. There is an issue when I jog the motor down the full length of the gantry. When it gets to the far end it stops as expected. BUT when it returns to the home end, it hits the stop then churns and grinds the gear til I manually release the jog button. I tried tightening the spring but that made no difference in the grinding but definitely demanded more torque/current for the run. It does seem there is a spot, about 2/3 the length of the gantry, that binds a bit. The motor will stall if I am jogging at slow speed 1/3 back toward home. If I jog at *5 it goes without any issue.
The Z-axis seemed to tune just fine. I was surprised to see that in neither case did the SW run the motors end to end on their axis. Is this normal?
The Y-axis presents different challenges as you point out. My current plan is to lift the gantry and auto-tune the near motor (isolated power connection - not through chain) with the proper weight loaded. I would then assign that tuning profile to the motor on the opposite side of the bed. I'm thinking that this would then get reversed through UCCNC to drive the motor correctly. It would also eliminate the need to auto tune the motor that was losing communication. Thoughts?
I've started terminating the motor power and data connections. I find the Molex terminals to be demanding and time consuming. Many MANY thanks for your great terminal hookup diagrams - 8 conductors is a lot to sort into that 1 connector. I'm glad there are only 4. This is all taking WAY more time than I expected.
Robb
I'm trying to remember what I did here, but I did not have good luck with auto-tuning the rack and pinion drives. I MAY have been able to get the X-axis to auto-tune, but I honestly don't remember. If I remember correctly, there is a setting for the number of revolutions used to tune the motor in the ClearPath software as well as a distance in each direction. That will keep you from running into your hard-stops. You have to remember that the motor is driving a 20-tooth pulley driving a larger 64-tooth pulley resulting in a 3.2:1 ratio. The gear is ABOUT 1" on it's pitch diameter (I think it is slightly larger, like 1.006" but for this discussion, this is close enough). For every turn of the motor, the X or Y axis will travel 1.0*(pi)/3.2 inches. This is CLOSE to 1" per revolution, but a little less, so you can use that to estimate how many turns you need. I remember starting the auto-tuning somewhere in the middle and measuring with a tape measure to decide how far to move it.
In theory, you don't need to auto-tune over the entire length of the rack, but the fact that you seem to have a tight spot is a little concerning. Here are a few things to try:
1) Is the rack greased? If not grease it with an NLGI 2 Lithium grease. If it is greased, check the tight spot for debris.
2) Loosen the anti-backlash spring all the way until you start feeling backlash. I usually test for backlash by trying to turn the large pulley back and forth. You would feel the backlash and also see a slight pause in motion when you reverse direction. Then tighten the anti-backlash spring until you just get rid of the backlash. Finally, tighten the spring by about 4-5 full turns. I forget where I saw it in this site, but either Nate at FineLine or one of the gents from CNC Router Parts had recommended a procedure for tightening the spring. I found that this was a little too light for my motors under hard accelerations, so I tightened things up a little more than recommended. The downside here is potentially extra wear on the rack and pinion, but skipping teeth on gears is much worse, and really makes it hard to get accurate cuts. In theory you can calculate how much force the spring needs to apply to prevent the gear from separating, but I haven't done that...it requires that you know the spring rate of the spring, and I don't know that number...I could test it, but then I would look like a bigger geek than I already am.
The machine is a compromise of cost, performance, and durability. One of the tradeoffs is the cost of the rack and pinion components. They are nice, but not top-notch accurate...that would have cost ALOT of money. Slight inaccuracies in the spacing between teeth are inevitable over the length of the rack. Normally this is handled by allowing a little space between the pinion teeth and rack teeth to prevent interferences from causing binding. The anti-backlash spring prevents this space from happening, resulting in a "tight" gear mesh. Loosening the spring a little allows the gear to move away from the rack a little, sort of like a car suspension.
3) Use the manual tuning procedure that I link to in post #59 of my thread. It is possible that the gains set by auto-tuning are too low to overcome the perturbation you have.
4) Stuff happens. It is possible that you have a rack with some sort of defect. This could be a ding or a burr on a rack tooth that makes the mesh a little tighter. If it is minor, you can either sand or file it out (gently).
5) It is possible that the rack isn't bolted on straight. I would expect that you could see that by sighting down the rack. If it isn't you could just loosen the 5-6 screws around the tight spot, move the axis back to the high spot and see if that improves things. If it does, hold it at the high spot and re-tighten the screws.
There really shouldn't be that much variation in the resistance applied by the drive mechanisms, and because these are linear axes, the load doesn't change with position, so short distances for tuning are normal and OK.
This is how I tuned my Y-axis. The only difference is that I THINK I reversed the second motor using the ClearPath setup instead of UCCNC, but I need to double check that. Again, the auto-tuning didn't work for me. It resulted in a very "soft" controller where the axis felt springy when I pushed on it. I used the custom tuning algorithm. You have to define a custom acceleration and travel distance, and the axis will end up moving at least 10 inches because it needs to move at constant speed for at least a second. I don't remember the acceleration and speed I used, but it was close to 600 or 800 IPM and 400 IPS^2 for acceleration. It was a little scary, but I wanted to tune for high performance, assuming it would still perform well at slower speeds and accelerations.
I have recently tried cutting aluminum with a pretty long 1/8" wood bit. I was not super happy with the accuracy I got, mostly because I got oval holes. They were oval along the diagonal, though, so I'm concerned that I may have not tuned the motors quite right. A 1/8" bit also flexes alot, so I might just need a bigger bit and different feeds/speeds. My Y-axis also makes noise when stopped, so I may have to turn down some gains on that one also.
No problem. That's why I shared it. I'm glad someone is able to get some use out of the documentation. Over the years I've had to use those connectors quite a bit, so they go pretty quickly for me. I also drafted my son into stripping back the wires for me, so that cut back on a little of the tediousness.
BTW, I like the cable track you used leading into the X-axis cable chain. I may have to adopt that.
-Robert
I have been overwhelmed with work and the purchase of a new house so I've had no time for the CNC machine. Just got around to auto-tuning the Y axis. Per Robert's suggestion we lifted and weighed the gantry (= 195 lbs) and loaded 1/2 that weight onto the A side motor platform. After experience with the X axis auto-tune I had concerns about 97 lbs of lead shot in motion all over the shop but that did not occur. I now need to load the A motor profile to the B motor but I am uncertain how to proceed. There is nothing in the manual but my assumption is that I just need to power up the motor (using my temp power connection) then upload the saved profile to the B motor. The question is... How do I adjust the profile without independent movement of that one motor? Do I need have that motor free of the gantry to adjust the numbers? I still have not hooked up the full control cabinet or attempted any movement via the UB1 / UCCNC. How do i communicate with the motors once the cabinet is hooked up? Do I still need the USB cable - 1 motor at a time, or can I load the ClearPath software and communicate through the data cables via the UB1?
Silly side question... how to hook up the spindle cable? This is not about the actual wiring but about how to route and support the cable from the end of the cable chain to the top of the spindle. I am looking for pictures but everything I see in this forum tends to focus on the tool bit end of the machine and crops that detail.
Thanks for all input,
Robb
Here is a video of Auto Tune in action for those who are interested ->