[Liberty4Ever]

I'm a month and a half into a lathe CNC retrofit, working part time. I was going to buy a Grizzly G0602 10X22 lathe and retrofit it, but a machine shop friend talked me into an old American made lathe. Those guys all look down their noses at the import machine tools, even though I think the parts I'd keep for the CNC retrofit would have been good (spindle, taper bearings, induction hardened ways). So I haunted Craig's List and found an ugly but lightly used Clausing 4902 lathe about 100 miles north of me for $1200. Buying it and getting it into my basement were epic adventures, but I won't bore you with that. The lathe is in the basement and I've been degunking it. The original owner painted it blue and it's now rough looking with age, but it's still tight and should be a good CNC retrofit.

Getting the lathe:




The lathe is finally in the basement:




I've been researching EMC2 quite a bit, and studying other people's CNC conversions. I'm an electrical engineer and I've designed many custom machines, so I'm not too put off by the electronics and wiring. As I researched EMC2 CNC and found good choices for subsystem solutions, I purchased the hardware. I'm well along the way. I installed EMC2 on a $100 used Compaq D51S Evo PC I bought off Craig's List for $100. It's running the EMC2 lathe simulator. I added a small Logitech K400 wireless keyboard with touch pad ($38 on eBay) that I think will be very handy and might take the place of a pendant. I have some nice Italian 900 ounce inch servo motors left over from my 2005 CNC mill project (never got a round tuit, but the mill CNC retrofit is next, after the lathe). I found some great $23 CUI encoders at Digi-Key. I have a couple of Gecko 320X servo drives. I took a chance on a 2HP Chinese VFD on eBay for $106 (looks OK, packing material smelled like Harbor Freight, Chinglish manual could be better but looks reasonably easy to set up and use). I have E-stop switches and shaft couplings and inductive prox limit/home switches, etc. I just spent $115 at Lowes on the wire, plug and outlet to wire the CNC lathe into the 30A dryer circuit in the basement. I bought a very heavy duty roll around tool cart at Sam's Club for $200. The PC sits on top, various measurement tools and lathe tooling go in the top five drawers, and the CNC electronics will go in the large drawer on the bottom with a lot of forced air cooling. Connectors to the lathe will be on the back of the roll around.

The roll around PC stand, tool cart, and electronics enclosure:




I still need to make some servo motor mounts and buy some fuses/breakers, connectors, etc. The only remaining big piece of the puzzle is the hardware to connect everything to EMC2 and the PC. I have a parallel port, but I don't think it'll have enough I/O. Should I add a second parallel port and use simple optoisolated breakout boards? Is there a better single board I/O solution that works well with EMC2?

I want the PC to be optically isolated from the rest of the electronics. I need to control the X and Z servos via the Gecko 320X drives and the three phase spindle via the VFD (0-5VDC, 0-10VDC, 0-20mA, 4-20mA, or RS-485). I think I'd like to use 0-10 VDC, and I may need a couple of outputs to signal RUN FORWARD and RUN REVERSE? I have four limit switches I need to read, as well as three quadrature encoders with index pulses (X, Z, and spindle). I plan on running the encoders single ended and not use the negative signals for A and B, so I'll need three inputs for each of the three encoders (A+. B+ and Index). I'd like some extra digital I/O because... call me crazy, but I eventually want to build a 5' bar feeder and collet/chuck closer so I can run semi-unattended operation for my short run production (up to 500 parts a month).

Got a good recommendation for an EMC2 friendly I/O solution, that's hopefully reasonably priced? Something that'll work well with my 24VDC digital I/O, the single ended absolute encoders, the Gecko servo drives, and the VFD to control the spindle?

I'm documenting this entire project, and when it's working, I plan on making a web page with a bill of materials, tips & tricks, and a series of YouTube videos to "pay it forward", and make an EMC2 conversion easier on those who follow me along this path.

Thanks in advance for any EMC2 I/O hardware recommendations you might have.

[Tkamsker]

Hi,
i did recently the same (even if i now doing an EMC2 update)
i retrofitted an new Drehmaschinen und Frsmaschinen Hersteller - Klippfeld Turning lathe which also has an Mill head. Thats why i bought it with special Linear gear for the Toolhead.
I am an it specialist and know electronic quite well (do some micocontroller stuff so i personally think when you want to do it easy use the standard parallel ports and if you have to put 2 or 3 of the parallel cards in there.

I searched long but finnaly i found a controller called tripple beast and verry strong steppers ( Benezan Electronics) i dont have the ballscrew so i have to fight backslash.

i found out the hardest stuff was to get an good CAM Programm for what i have to "lathe" (for milling i have an heiz Portal mill and use different software)
my next step is to build an "Handwheel" which i want to use arduino base like
sansbury95's Channel - YouTube so i can then use the lathe and mill as it was on hand or at gcode .

And then my winter project is doing an Turret. Where i bought already the plans but i am not happy with the air compressing stuff and the way it works.
so i think i will do an second arduino based turret. i want to use the design i bought for the shaft and tool holding plate the rest i will do it controller based. my idea is to also use it for manual mills. where you press button tool x and it jumps to tool x or center drill and so on.

And the reason i want to use arduino based stuff because the software support is cool and no parallel board i only need an usb board for it.

so thats my 5 cents
thx
thomas

[dlktdr]

I am a fan of Mesa's products (Mesa Electronics). Buy a 7i43, 7i33 and a 7i37. The 7i43 connects to the parallel port providing 48 programmable io via FPGA. The 7i33 connects to the 7i43. It has four +/-10V Analog ouputs and 4 quadrature inputs for encoders. The 7i37 is an isolated IO board providing 8 outputs and 16 inputs. The drivers are all included with EMC and they are a reasonable price.

On my machine I bought a 24IO G4 Opto-22 Relay rack instead of the 7i37. It plugs directly into the 7i43 and allows you to pick and choose what IO you want. It's a little more money this way but ebay usually has good deals..

[Liberty4Ever]

Thank you both for your comments and suggestions. I'm leaning toward a Mesa solution. A couple of weeks ago, I talked to Peter at Mesa (nice guy!) and it seems that they're realizing that a significant part of their market is now CNC conversions, and they seem to be developing specific solutions for that market, although their products still seem mostly directed toward general purpose PC I/O. They reportedly work well for CNC, including EMC2, but I think they're working on a more integrated CNC solution, hopefully a product that doesn't require three cards to get all the features needed for CNC.

I had the initial impression that EMC2 prefers interfacing directly through the parallel port hardware, and that's certainly the simplest concept. I figured I could get simple low cost optical isolator breakout boards for the digital I/O, and with the STEP and DIRECTION servo controllers I have, the digital I/O would be most of what I needed. I do have a VFD and I'll probably use 0-10V to control spindle speed, so I probably need at least one analog output. I thought maybe someone might make a fairly simple parallel port breakout board that's mostly digital I/O with perhaps a frequency to voltage controller and maybe EMC2 could provide a pulse train on an output pin that could create the voltage I need easily and inexpensively, or maybe a simple integrator circuit to convert a pulse width modulated output to an analog voltage.

I'm fairly cheap, but the expense of the very nice Mesa hardware isn't much of an issue. Mostly, I was wanting to keep it simple. Putting an FPGA card and a couple of daughter cards into the mix didn't sound as simple as I was wanting. On the other hand, maybe it will be easier to let Mesa do all of the work and treat the I/O hardware as a black box. I'm just worried about EMC2 config file problems trying to get it to recognize the Mesa cards. Based on the recommendations I'm getting for Mesa as opposed to using the parallel port hardware, I'm guessing that Mesa is the easiest path to success.

I'm going to try to get my I/O hardware on order today so I can get this project over the hump and on the long downhill run to making some chips. I'll read the hefty EMC2 documentation a bit more to see if I can get a better understanding about the I/O issues. That's still fuzzy to me, and that's keeping me from making the I/O hardware decision.

Thanks again for the info and assistance!

[Liberty4Ever]

I/O Electronics Ordered!

Mesa Electronics

5i25 $89.00
7i76 $119.00
10' DB25M to DB25M IEEE-1284 cable $12.50

The 5i25 is fairly new. The goal is a product that's more optimized specifically for CNC applications, so it has the features we need, not a lot of capability we don't need, and the cost is a bit less because we aren't paying for features and added flexibility in excess of what we need. It's a solution that's a bit more integrated, and down the road that should translate into less complexity getting it to run with EMC2 or the controller of your choice.

Basically, the 5i25 plugs into the PC bus and has a DB25F port on the back of the computer. The cable connects the 5i25 in the PC to the 7i76, which will be in the electronics panel, which in my case is in the big drawer in the bottom of a roll around cart. The PC and monitor and keyboard are on the top of the cart, and lathe tooling, manuals and measuring tools are in the other drawers. There will be connectors on the bottom back of the cart with cables running to the lathe. The cables power the X, Z and spindle motors, carry encoder signals back to the CNC electronics, carry the end and home limit switch signals for X and Z, etc. The 7i76 card has screw terminals so making connections should be easy.

I've designed a lot of machines, including some big systems with a lot of PC based I/O (Opto 22, etc.), and I don't remember fretting over the hardware decision like I did this time. I entered this thinking EMC2 liked bit banging on the parallel port and the I/O would be easy. Just get one or two inexpensive optically isolated breakout boards for one or two parallel ports and program the EMC2 configuration file to know what hardware was where. Not so.

I probably made it harder than it needed to be. I'm glad I finally have the I/O hardware on order. That's the last big unknown checked off the list. All that's remaining are many little details - custom motor mounts, sensor brackets, lots of wiring, and later on, some ball screws... PLUS, all the big gotchas that I can't see from here.

Lots of work left to do, but I'm well under way. Thanks to Peter at Mesa Electronics for some hand holding and pre-sales technical advice in a couple of phone calls. Thanks also to those online who took time to help me.

I'm meticulously documenting this project, so hopefully I can provide a step-by-step guide that will save those attempting CNC retrofits some effort and a lot of the head scratching I did. I may not produce the very most optimal solution, but hopefully it'll be a good solution that others can emulate if they like.

The 5i25 I/O board is brand new, and Andy is apparently just finishing up the drivers. Thank you Andy! I'll need to do a little double clutching to get EMC2 to like it. I'll need to upgrade from 2.4.3 as installed on the live CD to the new version 2.5, but I'm fairly comfortable with Linux. I wasn't going to network the EMC2 machines, but in anticipation of the need for EMC2 version 2.5 and probably a couple of profile downloads, I ordered a three pack of Ubuntu friendly USB WiFi dongles from eBay last night. Those have gotten cheap while I wasn't paying attention. They're about $5 each! The next live CD version of EMC2 should support the 5i25 board without any upgrade hassles.

I need to devote some serious time to this project. I've been piddling around an hour or two a day, and adding more to the To Do list than I'm crossing off the list. At this rate, I'll never finish.

[Tkamsker]

Hi,
i think important is that you can get support if you need to so if mesa is helping you out go for it.
I did use Triplebeast from germany for that and i am happy.

Bzw i will now (in next 2 Months - add the threading functionalty
And i found a nice link of one who did it successfully. And he also has a kind of parllel buffer board which may be of interest for you
Lathe Spindle Encoder

take it as a kind of my 5cents

hello from vienna

thomas

[Liberty4Ever]

I went with Mesa instead of Triplebeast largely because I'm in the USA and international shipping and support seemed like an unneeded complication. I did bookmark their website from your previous post, though.

Speaking of bookmarks, your link to the homemade spindle encoder was also added to my bookmarks. I've been wondering how I'll attach the tiny encoder I have to the spindle. The 7i76 has inputs for a spindle encoder, but mounting that encoder may be difficult. To be honest, the lathe is still up on the dollies and I haven't been able to open the end to look at the back side of the spindle. I'll come up with something. I have access to a laser and it would be easy to cut a spindle encoder disk from black acrylic as shown on that web page. I hope I can use the little CUI encoder I already purchased. Whatever I do, I'll document it so others can share in the idea.

[jmelson]

I've been wondering how I'll attach the tiny encoder I have to the spindle. The 7i76 has inputs for a spindle encoder, but mounting that encoder may be difficult. To be honest, the lathe is still up on the dollies and I haven't been able to open the end to look at the back side of the spindle. I'll come up with something. I have access to a laser and it would be easy to cut a spindle encoder disk from black acrylic as shown on that web page. I hope I can use the little CUI encoder I already purchased. Whatever I do, I'll document it so others can share in the idea.

I have a Bridgeport mill, and there is no place in the head to mount a standard
encoder. The spindle moves up and down and the spline that drives it has a
large diameter. I mounted two gear tooth sensors inside the head to pick up
the gear teeth on the bull gear, and drilled a little dimple to trip a 3rd
sensor for the index pulse. This has worked remarkably well. See
Adding a spindle encoder to a Bridgeport 1J head
for the details.

The CUI encoders have a problem responding to acceleration, but should work
fine as a spindle encoder. They rely on whatever is being sensed to supply the shaft bearing, however, so unless you provide a shaft with bearings and a toothed
belt or something, then it will be hard to use on a lathe spindle.

Jon

[Liberty4Ever]

Hi Jon,

Thanks for the info. That Bridgeport spindle encoder is ugly but good! The gear tooth detector was very clever. For the index pulse, I'd just chisel off a tooth and wire up a missing pulse detector. Just kidding!

I just spent four hours leveling my new lathe, because I had to make new PVC pads for the leveling feet and they had to be pretty. Stuff like that is why this project will take me a long time. Similarly, I've been researching a lot of details and loose ends, and I just added a gallon of way oil and a gallon of spindle oil to my shopping cart, about $60, more like $80 with two nice oiler cans, and stuff like that is why I'll be way over budget, too.

I hadn't read of any problems with the CUI capacitive encoders not handling acceleration well, and Gecko seems to recommend them for their servo controllers. Are we talking about the same encoders? They ship with bushings that adapt the encoder to nine different common shaft sizes. Digi-Key part number 102-1308-ND

Digi-Key - 102-1308-ND (Manufacturer - AMT103-V KIT)

Next up for me - run the wiring for the 220VAC 30A service from the clothes dryer in the basement and temporarily wire the spindle through the $106 2HP Chinese VFD to get my motor running and head on down the highway. Lookin' for adventure, in whatever machining may come my way. I'm born to be wild. I also have a little manual lathe project to do, and I'll probably use the lathe to make parts for the lathe CNC conversion. How cool is that? I'm also looking forward to verifying the alignment of the tailstock and other lathe calibration and setup stuff.

[Liberty4Ever]

Jon,

I finally pulled the cover off the upper headstock section to look at the spindle, with an eye toward attaching an encoder. I knew I couldn't just slap an encoder on a hollow spindle. I was originally thinking I'd put an encoder on the motor shaft and program EMC2 with a spindle encoder count that would adjust for the pulley reduction ratio. I was hoping for a 1:1 ratio, but the highest ratio was 17:18. The original maximum spindle speed was 1700 with an 1800 RPM motor, and while I'd like more spindle speed for turning small diameter parts, I don't think I should go much faster than the original because the bearings and other parts were specified for 1700 RPM max. I could use the 900 RPM pulley selection which is a 1:2 reduction ratio, which is a better ratio than an irrational number like 17/18. With the 1:2 ratio, I could run the motor at twice speed (or slightly more) with the VFD and should have good low end torque and good speed. Then, I'd need to tap the motor shaft, epoxy in a dowel, and press on the encoder.

From a control system perspective, I'm not crazy about measuring the motor RPM and inferring spindle RPM. I might decide to change the gearing ratio. Belts slip. I'd rather directly measure what I want to measure, so I'd like to have a spindle encoder measuring the spindle RPM directly.

I might do as you did on your Bridgeport, only it looks easier in my case. The index divot is already drilled, and access is fairly easy. Here's what the Clausing 4902 looks like with the headstock cover removed.



I may be ordering three gear tooth sensors from Digi-Key. In addition to directly measuring spindle speed, it looks like it may be easier to install than trying to drill the motor shaft.

Thanks for the tip!

I'd like to put the Z axis servo inside the headstock, but that will be a lot of work. The gear box isn't easy to access for modification, and it's not big enough for the servo motor, which would need to extend into the back of the headstock. It'd be a lot of work to gut the gearbox and cut a large hole in the cast iron for the servo motor. I may be forced to hang the servo motor for the Z axis off the tailstock end of the lathe. I'll eventually need to use a ball screw for the X and Z axes, and I'm not looking forward to that.

One step at a time, and the next step is to temporarily wire the spindle motor to the VFD and start using the lathe in manual mode (albeit with full spindle speed and direction control).

[Liberty4Ever]

I ordered four of the Allegro ATS667LSG Hall effect gear tooth sensors from Digi-Key.

Digi-Key - 620-1328-1-ND (Manufacturer - ATS667LSGTN-T)

One sensor is for the encoder index pulse, once per revolution. Two sensors are for the A & B quadrature encoder signals, in my case, 33 pulses for each of the two sensors. The fourth sensor is a spare.

Trying to infer the spindle speed from the motor speed wasn't a good way to go. In most circumstances, belt slippage wouldn't be a problem, but without a 1:1 gear ratio, there was no way to get a true spindle index pulse. I'd eventually like to make a chuck closer, and I'll need a true spindle index signal to precisely locate the spindle position before unlocking the chuck.

[jmelson]

Hi Jon,

Thanks for the info. That Bridgeport spindle encoder is ugly but good!

Thanks! Really, in the confined space of that head, there really is NO
other way to do it.

I just spent four hours leveling my new lathe, because I had to make new PVC pads for the leveling feet and they had to be pretty. Stuff like that is why this project will take me a long time.

Rebuilding the bed of my 15" Sheldon lathe took me 22 months, but it was worth
it!

I hadn't read of any problems with the CUI capacitive encoders not handling acceleration well, and Gecko seems to recommend them for their servo controllers. Are we talking about the same encoders? They ship with bushings that adapt the encoder to nine different common shaft sizes. Digi-Key part number 102-1308-ND

Yes, these are the ones. I could never get the motor to run well under EMC2, they were
always twitchy and seemed on the verge of instability. I finally put an HP optical encoder on the same shaft so I could compare both encoders at the same time.
See this picture
http://pico-systems.com/images/compare_encoder2.png
The white trace is the HP (good) encoder. The red trace is the CUI AMT-103
encoder. These are velocity traces, in raw encoder counts per millisecond. You can clearly see when there is acceleration, the red trace lags behind, and then the velocity has to overshoot for the position to catch up. This makes servo tuning very hard to do.
It may be that the Gecko servo drives handle this more smoothly, but this much lag has to worsen following error. Of course, with a Gecko 320-series drive and no other measurement being made on position, you have no idea what the real following error is.
Anyway, the lag in responding to acceleration seems to be about 3-5 ms. The time scale in this plot is 20 ms per division.

Jon

[Liberty4Ever]

Well, that doesn't bode well for my success with the CUI encoders.

I have them, so I'll try them and see how well they work in my application. If nothing else, I'll add to the body of EMC2 knowledge... good or bad.

My current strategy, in the order I plan on implementing the servo motion control:

1) CUI encoder on the rear servo shaft, with the encoder signal fed only into the Gecko 320X servo drive. I'll set the Gecko following error as small as possible without generating nuisance trips and do my best servo tuning on the Gecko. With no EMC2 position feedback, I'll measure the actual following error by measuring the parts to see if they're accurate enough for my needs. I'll measure servo stability by watching and listening.

2) Pair of CUI encoders on the servo motor, with the encoder signals sent to the Gecko 320X servo drive and the EMC2 controller - small servo control loop inside a bigger servo control loop.

3) HP or comparable encoder on the servo motor shaft, fed only to the Gecko 320X.

4) HP or compatible encoder on the servo motor shaft, fed to the Gecko 320X and the EMC2 controller.

5) Sell the Gecko 320X and find a more directly controlled servo amp that uses analog voltage to control velocity, with the encoders fed directly to EMC2.

Based on the observations and more thinking about this issue, I may alter or revise my implementation plan.

Thanks again for your very helpful suggestions and info. I look forward to being in a knowledgeable position to help others soon. This is the reason I'd like to publish a detailed How To guide describing a known CNC lathe solution. There will be some machine dependent hardware choices, for example, the spindle encoder mentioned recently, but most of the hardware and software should be plug & chug. Off the shelf solutions will free EMC2 hackers to be more productive and spend their time on the important custom stuff, rather than reinventing the wheel our scouring various online sources and comparing different implementations to see what might work.

[giz]

Any updates?

[Liberty4Ever]

Any updates?

Well, I was waiting until I had something substantive to post. I've been plugging away an hour of two a day, but work has been busy lately and it's cutting into my lathe retrofit time.

My CNC interface electronics arrived from Mesa Electronics last week, but I haven't plugged the PCI card into the Ubuntu PC yet. I need to upgrade EMC2 version 2.5 for the new Mesa board to work. I received a WiFi router that can be configured similar to a bridge, so it'll connect to the existing WiFi network as a WiFi device to supply wired networking in the shop via the upstairs WiFi network so I hopefully won't need to run an Ethernet cable, but I haven't configured and installed the new WiFi router.

I have the lathe headstock apart and I've figured out where I want to mount the three gear tooth sensors that will implement a spindle speed absolute quadrature encoder, but I haven't built the brackets and wired the sensors yet. It should be a nice installation, well protected inside the heavy cast headstock enclosure, but easy to get to if needed. I'll route the cable through the hole in the back of the headstock that's now available after I removed the FORWARD/REVERSE lever, interlock and drum switch. Those are all spare parts once I get the VFD connected to the spindle. I'll mount an EStop on the hole in the front of the head stock that's open after removing that FOR/REV lever. I toyed with the idea of leaving the lever in place and connecting it back to the CNC electronics so it would switch the spindle direction, but if that's possible, it'd probably be a hassle. I think I'll keep the lathe mechanics as simple as possible and keep the EMC2 controls as stock as possible, dividing the project into lathe mechanics at the lathe, and all control functions at the PC. Similarly, the tumbler and all of those threading gears are going to be surplus when the CNC is working. The lathe mechanics will be a spindle, and X & Z servo motor driven axes with limit switches. The spindle should be single speed mechanically, with all RPM controls at the VFD under control of the EMC2 CNC software.

I ran the 10-3 with ground cable from the clothes drier outlet up into and across the basement ceiling joists and I mounted an outlet for the CNC lathe to plug into on the ceiling, and I ran a short piece of cable from that over to where the milling machine outlet will be, but I still need to buy and wire that outlet. It's a 30A service, so I can run the milling machine and the lathe at the same time, or the drier, but not both. I lucked out on the ground. There was no ground wire to the drier outlet, so I figured I'd get a ground from a cold water pipe. Turns out, the #2 bare copper wire that grounds the breaker box on the other side of the basement (that was per code when the house was built in 1982 but not now) originates directly above the drier outlet, so I can grab a ground there. Too bad the breaker box wasn't accessible, because I'd have liked to run a separate 220V line for the lathe and milling machine so they didn't share with the drier or each other, but this is the next best thing. Have you priced copper wire lately? $50 for 25 feet of 10-3 w/GND. Ouch.

I replaced the cheap plastic knob that someone put on the tail stock hand wheel after breaking or losing the original. How can you break a steel handle? This was a simple 10 minute job (including drilling out and tapping the cast hand wheel) that turned into a 90 minute hassle because the Jergens handle I purchased new from McCmaster-Carr had a manufacturing defect and I basically had to remanufacture it. A lathe would have been very handy for that. Heck, maybe I'll turn my own steel handle.

I found what I believe to be a great buy on eBay. I bought a 1.5 HP 220V 3 phase Baldor motor, brand new and still in the shrink wrap, for $53 delivered. It's a heavy duty motor that I think is intended for large spa pumps or similar pumping applications. MSRP is $567. There are heavy duty bearings on each end and rat wire screen in the cooling vents. The motor shaft is a little odd but nothing too weird. Baldor qualilty, at a treadmill motor cost! I bought it for the upcoming CNC milling machine upgrade, and went ahead and bought another 2HP Chinese VFD on eBay because it looks like the prices were going up from $106 to $140 or so. The dollar will continue to drop against the yuan.

I've spent a lot of time staring at the lathe, thinking where the inductive prox limit switches will go and where the servo motors will be mounted. I also spent a fair amount of time cleaning up the lathe and getting to know her. It looks like piddling, but I prefer to think of it as an important subconscious design process.

I had a new parts and service manual that came with the old Clausing 4902 lathe, but I thought I was missing an operator's manual. I called Clausing, expecting a 16 page manual to be $30. They emailed me the PDF for free. For a 1968 lathe! Great company! It was the electronic version of the manual I already had. I guess they expect that anyone buying a lathe is already trained in basic lathe operations, so the manual only covers operational issues specific to this model. It can't teach new lathe owners how to operate a lathe. I have zero practical experience operating a lathe. I have a healthy respect for them, having seen some lathe carnage videos, but I realize that I can't learn to use a lathe from watching YouTube videos... try as I might! I'm looking for a book, and will either get a machinist friend with several lathes to trade me some training for me running some parts for him one day, or I'll get some vo-tech training.

I'm almost completely finished buying lathe CNC conversion parts. I still need to buy aluminum for servo motor mounts and some piddly stuff like cord grips and probably some Lexan guarding and coolant shield. I spent another $650 today on lathe tooling. I bought a few basic carbide index tools (AR8, E8, and a grooving/cutoff tool). Almost half of the cost was a nice 1" micrometer and set of 6" calipers, both coolant proof, both Mitutoyo. I also got a magnetic base dial indicator for setups. I could have shared some of that tooling with the tooling I already had for the manual milling machine considering they're next to each other and I'm the only operator, but I thought a little redundancy would be good. I'm gradually replacing Chinese tooling for better quality tooling, generally carbide index tooling.

I fixed the crooked cuts on my cheap Harbor Freight horizontal band saw that my dad gave me. Pretty much, clean it up and move the blade guide rollers in a bit. I had put on a much better bimetal blade but it was too stiff to make the quick turnaround at the big end pulley. I haven't used the band saw much, but now that I'm going into CNC lathe production mode, I'll need to chop the 10' bar stock into pieces I can feed through the spindle - hopefully 5' lengths (almost 3' of overhang out the back of the head stock), but I may need 3.33' lengths to be safe. I designed a flood coolant system for the band saw. It'll have an ON/OFF switch, and when in ON mode, it'll only run when the saw is cutting and will automatically turn off when the saw finishes the cut. I ordered a gallon of concentrated band saw coolant as part of today's order. I already had the tubing and magnetic base flexible wand and nozzle. I ordered an 80 GPM Little Giant magnetic coupled pump for $21 on eBay (new, delivered). I designed a chip filter and coolant sump using two nested plastic cat litter buckets - free!

I also bought an electromagnet on eBay today ($7 delivered) that I plan to use to transfer a few pounds of steel chips at a time from the chip collector into the plastic recycling bucket. I hate scooping wet chips, and I think this may be easier and faster. If it works well for the saw, I may do something similar to clean out the swarf tray on the lathe, where there will be a LOT more chips.

The lathe CNC project seems to be at a pregnant pause. I've been gathering up a lot of parts, learning a lot of stuff about a lot of stuff, and doing a lot of piddling. I'm about ready to jump in and start wiring, but I think it's better to think things through and let it percolate and bounce around in my brain pan for awhile so I'm not making too many costly mistakes. I'm sure I'll still make plenty of mistakes, without rushing in where only fools dare tread.

I am enjoying the project, even if it is progressing much more slowly than planned. I should have the spindle turning under VFD power soon. Every day lately, I thought would be the day, but something always comes up. I have a couple of simple turned plastic parts to make, and I probably won't hurt myself on those projects. Now that I have a lathe, lots of projects look like lathe projects. That'll only get worse once I have a WORKING lathe. :-)

[will gilmore]

I entered this thinking EMC2 liked bit banging on the parallel port and the I/O would be easy. Just get one or two inexpensive optically isolated breakout boards for one or two parallel ports and program the EMC2 configuration file to know what hardware was where. Not so.

I'm in the planning stages of a mill retrofit and the above is exactly what I was thinking. Why not? I haven't read up on the Mesa products yet. If you could point me in the right direction I would appreciate it. --Will

[Liberty4Ever]

Given that this is my first EMC2 retrofit and I'm just getting started, I'm far from an expert, but I get the feeling that most retrofits are complex enough that a parallel port will quickly run out of I/O pins. For example, each absolute quadrature encoder will need three input pins. You can add a second parallel port, but even this isn't all that much I/O. For little more cost than a couple of optoisolated breakout boards, you can get a lot more I/O, as well as some intelligent I/O that's optimized for specific machine control functions.

I suspect it may also be easier to create a configuration file by cutting and pasting the hardware config file from a particular manufacturer rather than assemble the lower level bit banging config file, although I'm not there yet and this is just speculation. Perhaps someone who has tried it both ways can educate us both.

Having developed some realtime microcontroller peripherals for PCs, I can tell you that a dedicated microcontroller has a much better real time response. In the EMC2 world, we're talking jitter. I once designed, programmed and built a custom data acquisition module that simultaneously sampled encoder pulse data on four independent channels with a timing resolution of .2 microsecond. To put that into perspective, in .2 us, light travels 196 feet.

Given the interrupt driven and indeterminate nature of PC hardware, I'm amazed that EMC2 running on a realtime Ubuntu kernal can do as well as it does bit banging on the parallel ports.

I was looking at one of the hobby 3D printers yesterday at SeeMeCNC.com, and they've further cost reduced the RepRap to $450 assembled and shipped. That's amazing. They use Mach2, but it can be run from EMC2 as well. It's a fairly simple three axis stepper system and they're apparently running it straight off the parallel port with optoisolated breakout boards to keep the cost as low as possible.

[will gilmore]

I have a lathe running Mach3 which I got up and running and then started improving so I have very little time on Mach3 on the lathe. I have some time running mach3 on a milling machine which I found to be a little buggy / quirky so I'm considering EMC2 for this upcoming mill retro.

I'm putting together an I/O list. Are you bringing the encoder signals back to the computer? My impression was that with the G320X the encoder signals only went as far as the drive which took step/dir signals from the PC.

[Liberty4Ever]

For my first attempt, I'm going to run the servo motor encoder signals back to the Gecko servo motor drives only. EMC2 will be running essentially in open loop stepper mode, although I am planning on having the Gecko drives generate a following error to stop EMC2 if there's a problem, so it's kind of half way between open loop and closed loop. I've essentially closed the servo loop between the servo drives and the servo motors but the loop is open between EMC2 and the servo drives (other than the following error signal).

People in the position to know tell me this is probably not a good way to do it, but it sounds better than having a competition between one encoder feeding back to the servo controller and another feeding back to EMC2, which seems like it could cause oscillation if the two control systems became out of phase. Running one encoder signal to the servo drive and getting STEP and DIRECTION outputs from EMC2 and the following error input to EMC2 certainly seems like the simplest wiring.

If it doesn't work to my satisfaction, I'll make another plan.

Ultimately, I'd prefer to measure the actual position of each axis directly into EMC2 to close the big loop for best precision, rather than measuring the rotation of the back of the motor shaft that's connected via a belt drive that stretches, to an acme screw with backlash. At the very least, there are zero backlash ball screws in my future. They're already in my McMaster-Carr shopping cart as a saved order, but I'm delaying that $500 expense until I have a better idea what I'm doing.

The spindle will be directly controlled by EMC2. I'm making an absolute quadrature encoder with three gear tooth sensors to directly measure the spindle speed and direction. EMC2 will generate an analog signal to control the speed of the variable frequency drive that controls the spindle rotation.

[jmelson]

Ultimately, I'd prefer to measure the actual position of each axis directly into EMC2 to close the big loop for best precision, rather than measuring the rotation of the back of the motor shaft that's connected via a belt drive that stretches, to an acme screw with backlash. At the very least, there are zero backlash ball screws in my future. They're already in my McMaster-Carr shopping cart as a saved order, but I'm delaying that $500 expense until I have a better idea what I'm doing.

I have a stepper controller that can generate step pulses at high speed, up to 300,000 steps/second per axis. This may be a real help, depending on encoder resolution.
I also have a device that interfaces between the encoders and the Gecko 320 drive.
It makes an opto-isolated copy of the encoder signals for the Gecko 320, and also sends
a copy to EMC. You can hit E-stop and power the Gecko drives down, and EMC still
knows the position of the machine. If any Gecko drive faults, it powers them all down, signals EMC, and records which axis had the fault.

You likely will need some kind of interface between the encoders and EMC, as the count
rate may exceed the ability of software encoder counting to keep up. (Again, depends on
encoder resolution and motor speeds.)

See Jon Elson's Pico Systems Home Page under stepper systems for more info on these products.

Jon