[Wheatie]

Hi all. I joined quite a while back, but other things have distracted me from this project. I have a Grizzly G0484 (RF 45 clone) I bought used about five years intending to convert. Now I have the time and money to do it, so its in progress.

I have most of the design work completed and adding the one-shot oiler is in progress. Already had an OOPS of putting the holes in the saddle for the oil in locations that hit the table. But recoverable.

I've read through a bunch of threads here and other places on converting these mills. My plan is to use the Keiling (Automation direct) 840 in-oz 72 volt NEMA 34 motors on all axes. GT2 timing belts on all three axis. On X and Y I'll use a 2:1 reduction. This is the best I can manage using SDP-SI GT2 pulleys and keep the pulleys below the table and saddle. Should I consider a different type of belt or using a smaller motor pulley and drilling out the bore? The smallest 1/2" bore GT2 pulley is 28 grooves and the biggest one that fits below the table on the X is 56 grooves. There are slightly smaller 3/8" bore pulleys that would give a bit more reduction. On the Z I'll use a 72 groove pulley on the screw.

I'm planning to use the Kflop for motion control and SnapAmp for drivers. I'm slightly concerned about the lower peak amp rating of the SnapAmp vs. Gecko G320X drivers. However, I found some detailed information on Baldor servos that are similar to the Keilings (http://www.baldor.com/support/Litera...umber=BR1202-F) and did some analysis and measurement on the mill and I'm pretty sure the actual current requirements aren't anywhere near the 38 amps peak for those motors or even the 15 amp limit on the SnapAmp. But no practical experience, and the cost difference is minimal for smaller motors. NEMA 23 motors would be a better fit, though, as far as clearance.

I have screws, FK/FF type mounts and two ballnuts for each axis on order from linearmotion2008. And all the parts for the oiler are here (and waiting for me to back away from the computer). Haven't bought the other parts, yet. I'm using 2005 screws on X and Y and 2505 on the Z. I'll use bellville washers to provide preload between the pairs of ballnuts.

I'm planning to use the Kmotion CNC software, at least for a start. I've exchanged several e-mails with Tom and he's been great about answering my questions. Seems like the feedback from people who have switched to the Kflop has been uniformly positive.

I'm planning on using the US Digital optical differential encoders, per Tom's recommendation. Not quite sure about the proper count. My math is that using 2:1 belt reduction, the 5mm/rev lead of the screw, and a 500 CPR quadrature encoder I would get 20,320 pulses per inch of movement. Or, 0.000049" per pulse. Too much? Too little? The peak speed on those motors is 4200 RPM, so that translates to 140,000 pulses per second. Tom says that's well below the capabilities of the Kflop/SnapAmp. That all translates to 413 IPM, which seems way above what other folks report on these machines. Is the only source for these to buy them direct? Their pricing is pretty brutal for quantities less than five.

Still need to decide what to do for some counter balancing on the head.

[CS900]

My 45 mill is pretty much exactly what your doing with the exception of the motors/controller. Hardware wise, I'm using the GT2 belts with a 2:1 on x and y, and 3:1 on z. I'll have to look up what size pulleys I used, but I did have to bore out the ones that mounted to the balscrew (for all 3 axis). I'm also using the linear motion ballscrews with belville preloading (20mm on x and y and 25mm on z).

I used gas shocks to counterbalance my head. Seems to work pretty good. I can rapid at over 200IPM without any problems. If you'd like I can get you the McMaster-Carr part number for the ones I used.

edit: mcmaster part number is 4138T63. I have 2 of them

[ninefinger]

I just looked up US Digitals pricing for a differential output, 500CPR, with a 1/4" bore - I nearly fell off my chair!
Before laying out $300 for those I would recommend you try what I and many others have been using - the CUI AMT-102V encoders AMT102-V KIT CUI Inc | 102-1307-ND | DigiKey. Available from digikey for under $24 each. I've had mine for years and no problems - and I didn't even protect them from coolant at one point (got ahead of myself with coolant before splash protection - oops).
Another nice thing about them is that they are configurable so you can find a CPR that works for your application after you have it in hand - not before you order (works really well for experimenting with high CPR...)
Just checked and Digikey is out of stock - another vendor may have them though - I certainly recommend them
Mike

[Wheatie]

Yep, I agree about the pricing on the US Digital encoders. $115 each for the differential ones and another $30 or so for a shielded cable. I was initially planning to use the AMT102's. But in discussing it with Tom from Dynomotion, he recommended differential and optical. But over $300 more to use the US Digitals vs the CUI's. The US Digital's are cheap compared to some of the other alternatives. But its not a huge gamble, so maybe I'll give those a shot. Automation Direct says they have them in stock, but I have a ways to go before its urgent.

Thanks for the pointer on the gas springs. What force did you use?

Started getting things reassembled and got the oil pump mounted tonight.

[CS900]

I used 2 of the 50lb ones.

[dracozny]

I am purchasing the Us Digitals, E5-1000-250-IE-L-D-D-B. might be over kill but they are $112 and I am making my own cables. thankfully two of my motors already came with Encoders, these will just match them for the other two axis.

[Wheatie]

Some progress. Milled oil slots in saddles and have the oil pump mounted. But not connected, yet.
Attachment 256866

When getting everything back together I may have broken the mounting bracket for the existing x-axis motor drive, or it was previously stressed. Perhaps from moving. In any case, fabricated a new bracket.


Attachment 256868

Made all the parts for the x-axis. Ballnut mount. I don't have accurate measurements to let me drill this for bolts, so that will have to wait for disassembly.

Attachment 256870

Fixed end bearing mount, motor mount, and purchased parts. Free end mount is cut, too.


Attachment 256872

Then the on/off relay for the spindle gave up the ghost. I had noticed some buzzing previously, but everything worked and it went away quickly. I guess that was the early warning. Of course the part is not available from Grizzly. But a on/off switch off ebay and a relay from Newark and its working again.

Attachment 256874

Unfortunately this arctic blast moved in and its way to cold to be in the garage. I've moved all the electronics, motors and encoders inside. I'll try to get that working while waiting for the weather to improve. I did end up going with the AMT encoders, but was able to find line driver cables from Digikey. I have all the motors, encoders, Kflop, SnapAmp, and motors and power supply from Keiling. Also have bolts (Bolt Depot) and Bellville washers (McMaster-Carr). Forgot to order the gas shocks, but I don't need those right away, so I'm waiting to see what else I might need from them. I have an old, large desktop PC. A 12 year old Best Buy VPR brand monster. I'm planning to pitch the motherboard and storage devices and put the power supply, Kflop and SnapAmp in there. Also bought a refurbed windows laptop to be the controller with the KMotionCNC software.

[CS900]

lookling good.

So I have my X motor mounted just like you, and may I suggest moving it so that it faces the other direction. The problem I'm encountering now, is that once I put a chip cover on the Y axix, I can collide the motor into the chip cover when jogging the table left. It will clear the table just fine, but the added height of the chip cover screwed me over. I've lost a LOT of table travel due to that fact, and I'll be making a new bracket to flip mine very soon.

[80gus]

nice looking build so far keep the pictures coming.

[Wheatie]

Its continued to be cold here, so I brought the electronics and motors inside. I have the electronics built. Ready to power it up and get the motors to turn. I want to double check the wiring piece by piece one last time. Any warnings before I power this up?
Attachment 262788
Motors with extended cable and circular connector. Encoders are attached. I ended up using the CUI AMT-102 encoders with the line driver (i.e., differential) cable.
Attachment 262792
Cables for the encoders. These were messy. The output of the line driver cable is cat5 type wiring. Four twisted pairs, for A, B, X (index) and power. But only about 3' long. I found some male to female cables on amazon with a bulkhead mount on the female end. So, I cut these in two and mounted the bulkhead jack in the back of my case. This gets connected to the SnapAmp. Then I soldered the male end onto the line driver cables. I knew making all those tiny connections would be a pain. Turns out that the wires inside the cables I cut don't follow standard color codes. Color codes are always an issue for me because I'm red-green colorblind. The standard ones are good, though. The wacko scheme in these cables were much harder to distinguish. And no easy matching up to the standard ones. So numerous trips to show each step to my wife for checking. We will see soon.
Attachment 262794
Inside of electronics box. This is just an old computer case. Ripped out all the guts and fabricated a plate in the back for the motor and encoder connections. Power supply is Automation Technologies Unregulated Linear 1440W/72VDC/20A Toroidal PSU (KL-7220). Kflop and SnapAmp for motion control and motor drivers. Motors are the NEMA34 850 oz/in Dual Shaft DC Servo (KL34-170-72) from Automation Technologies. The connections for the SnapAmp are concentrated on a single IDC 50 type connector. So I found the breakout board in the middle of the picture on ebay. That connects to the SnapAmp with an 50 pin ribbon cable. Internal parallel SCSI, if you ever messed with that. Then there are screw terminals for the 50 pins. Those are wired to the bulkhead end of the encoders. Eventually limit switches and other controls would go here, too. There's a IDC 26 type connector on the Kflop that could go to another breakout board if needed. And a 8 pin RJ45 style connector. So plenty of room for expansion with this combination. Common ground point above power supply. The power supply is actually on the bottom of the case, the picture is looking at the side. Temporary wiring for the incoming power. I'll change it to 220 when I get this out to the machine and add fuses and switches. I need to fabricate a plate for the front, but that needs some warmer temperatures.

Attachment 262796
This is the back of the enclosure. RJ45 jacks for the encoders and military style circular connectors for the motors. I only needed three pins, but couldn't find them in the right gender and mounting type.

Attachment 262790
This is the base with the saddle all the way back. The Y screw is protected. I've designed the bracket so the motor SHOULD clear the base and the limit switch (black box with the sticker) and still be below the top of the table. Its a tight fit and I'm concerned about exactly where the mount ends up. As far as I've been able to figure out, this is a trial and error process of clamping the mounts in place and sliding the table back and forth until the mounts are positioned so it moves freely. I've designed in some clearance in the mounts so nothing sticks up or out, but if its significantly off center vertically the motor could interfere with the base or table top. I considered using a NEMA 23 motor for the X axis. That would clear easily. I did some measurements and calculations and I think it would work. But too much uncertainty about forces required for cutting and the real coefficient of friction for the ways led me to go with the larger motors.

[Wheatie]

Well, checked everything over, convinced myself its all wired correctly, and powered it up. No smoke or anything bad. But also not really good motions on the motors, either. I have the motors clamped to my table and I'm playing with KMotion to make them spin. I'm able to get one of the three motors to turn cleanly when wired up to one of the four channels on the SnapAmp. Any other combination has problems following the commanded curve. The combination of parameters that works is:
P=5
I=0
D=0
max limits
output=400
integrator=1e+6
motion profile
V=100000
A=2e+7
J=4e+6
feed foward
V=0
A=0
dead band
range 0
gain 1

Same parameters on any other motor or channel results in the commanded motion, but a lot of oscillation when the motor should be still. Is it reasonable that the motors should behave well when completely unloaded?

[bhurts]

You will get nowhere with your current acceleration value. Your velocity may be to high as well. Lower your acceleration to 2-3 times your velocity and try a few more plots. If your commanded position falls behind your actual positive then your velocity is to high.

Ben

[Wheatie]

I played around with the settings some more and have better behavior on two of the motors. The third still has a lot of oscillation with the same parameters. Here's a picture for axis 0. Axis 2 is similar.
Attachment 263276
Axis 1, OTOH, looks like this:
Attachment 263278

[bhurts]

You will most likely not get any ideal tuning out of the motors until the are attached to the machine. If you are successful on the bench then the values you use to do so will be nowhere near correct once they are attached to the axis. I did the same thing your doing and it's a good learning exercise I just don't want to mislead you into thinking it will also work on the machine.

Ok having said all that the next step to stability for me and I think most others is a 2nd order low pass filter. On the irr filter screen choose filter 3 (the third from left). Change the dropdown to 2nd order low pass and for values insert 1000 in the first box and a Q of 1.4. Then press compute to fill in the bottom values. Then you can try some more moves.

Increase p until you go unstable then back off 30-50%. Then increase d until you go unstable and back off 30-50%. Go back and forth between p and d several times then add a small amount of I gain. Some where on the order of .0001-.001.

Ben

[bhurts]

Also you should be able to keep the output and integrator values at default. That should be like 230 for output and 200 for integrator.

Ben

[Wheatie]

Ben, thanks for the assistance. I figured this would be something of an exercise in futility until I actually have some mass to move around. Still interesting to play with the software and learn how it works. I continued to mess with this after posting and have a bit more stability. I'd like to get it stable enough to start playing with KMotionCNC and get some experience with that. My goal is to have at least two of the motors moving at once. If it warms up, I want to get back to the garage and finish cutting the parts. Overall, though, I'm pleased with the progress.

[bhurts]

If you want enough stability to play with coordinated motion build a fast and dirty test fixture. That should help the motors enough to get them stable. A fixture is easy to make to. Just get a couple of pieces of wood and screw them together. One piece will have three holes spaced far enough apart to mount your motors with the shafts coming thru. It can be made with wood screws or lag bolts. Hardwood would be better but 2x8s or 2x10s would work fine.

Ben

[Wheatie]

I ended up just clamping them to the table. I did figure out how to get the tuning parameters and initialization statements put into a C program and hooked that into KMotion CNC and have the motors spinning all at once. Never could get the Configuration screen's "Export All To Open C Program" button to work, though. Had to do them one at a time and copy and paste them together. Then searched through the examples and discovered the functions to set up and enable the axis. Dynomotion's documentation leaves something to be desired. Its kind of an adventure game to search through all the different pages to find useful tidbits, then put them together.
Here's a short video of all three motors spinning.
https://www.facebook.com/video.php?v=10152629468643354

[fluvannabear]

Hello Wheatie:
I am in the process of converting my G0484 but I am using NEMA23 425oz-in steppers. Will you please tell me what length 2505 (Z-Axis) and 2005 (X and Y-Axis) ball screws you used for each axis? I am saving my pennies for some but do not want to tear down the mill to measure what leadscrews are in there now. Starting out just moving the Quill in Z but will upgrade to be able to move the head after I get gas springs in place. Working on electronics now, thinking about where to put the limit/home switches. Hope to hear from you or from anyone that has done this conversion about the ball screw lengths. Thanks and take care.
Burt

[Wheatie]

I honestly think you should tear it down and measure carefully. I did do that and I'm still not 100% sure everything's going to work smoothly. Despite the lack of updates, I have been making progress. X and Y are all cut, electronics is ready to go. I'm working on the Z and I'm planning on taking the head off again before I finalize the ball nut mount.

I bought the screws and nuts from Chai (linearmotion2008). Here's the specs I sent him:

2005 ballscrew C7, 926mm overall length, fixed and floating end machining
2005 ballscrew C7, 446mm overall length, fixed end machining only
2505 ballscrew C7, 736mm overall length, fixed and floating end machining
4@ RM2005 ball nuts (for first two screws)
2@ RM2505 ball nuts (for third screw)
3@ FK15 fixed end bearings with nuts (as in Ball Screw Bearing Mounts ballscrew Bearing Blocks FK15 FF15 End Supports 3set | eBay)
2@ FF15 floating end bearings

Those lengths are, in part, dictated by the bearing mounts. So, if you use a different design you may need different lengths. Here are my end machining specs: