Model Engine Maker
Supporting => Tooling & Machines => Topic started by: kvom on June 13, 2016, 10:25:11 PM
-
Over 3 days last week I attended a workshop at which 14 of us converted a Little Machine Shop lathe to operate under CNC control. While everyone attained axis motions under control of Mach4, there remains a bit of fettling and adjustment to get it working to its utmost.
The design of the mechanical parts was done jointly by Ron Ginger and Joe Katona, while the electronics were designed and produced by CNC4PC. I believe that drawings and BOM for the mechanical components will be made public at some point, and CNC4PC will sell the electronics, so it should be possible for home builders to obtain the same machine. Cost for the lathe and everything needed to build it was $2770, which included a Mach4 Hobby license.
We started with a LMS 5100 Hi Torque 7x16 mini lathe:
(http://littlemachineshop.com/products/images/480/5100.480.jpg)
After uncrating, the first task is to remove everything that wouldn't be needed. This included the saddle, both leadscrews, and all the change gears and associated mechanism, and the backsplash. We also removed the compound temporarily to expose and remove the cross slide.
The first operation was then to ream out the hole for the new X-axis ballscrew to 13 mm, and then to drill and tap two holes in the rear of the cross slide to mount a bracket for the ballscrew nut. Here's a photo of the bracket with the nut attached. A drill jig was used to ensure accurate placement of the holes.
(http://www.pbase.com/kvom/image/163448383/large.jpg)
Next the Z ballscrew is attached to a mounting plate, which in turn is secured using the same mounting screw as was used for the saddle. The right end of the ballscrew is supported by a pillow block containing a pair of flanged bearings. This pillow block uses the same holes in the frame as the original lead screw. The left end uses the original pillow block.
(http://www.pbase.com/kvom/image/163448385/large.jpg)
(http://www.pbase.com/kvom/image/163448386/large.jpg)
The right end of the Z ballscrew is secured with a nyloc nut and a coupling to the shaft of the Z stepper. The motor itself is attached to a bracket that itself is screwed to the frame. The bracket itself was used as a drill jig for drilling and tapping the holes.
(http://www.pbase.com/kvom/image/163448384/large.jpg)
The X-axis motor mount consists of the Z ballscrew plate, a second parallel plate which also uses a saddle mount screw, plus a pillow block that spans these two plates that carries the near end of the ballscrew. The motor itself is mounted to a plate that bridge the two support plates.
(http://www.pbase.com/kvom/image/163448380/large.jpg)
The ends of the motor shaft and ballscrew have toothed pulleys joined by a belt as seen above. Tension on the belt is set by the motor mount having three screws, with two in slots, allowing the plate to be rotated slightly.
On the left side of the spindle housing is located an encoder, driven by a belt and two pulleys. The encoder is mounted to a bracket, which in turn is held to the spindle housed by a screw. The bracket itself is used as a drill jig for this hole. A slot in the bracket allows adjusting tension on the belt. The spindle pulley is placed between the two locknuts.
(http://www.pbase.com/kvom/image/163448387/large.jpg)
The control box has estop and power switches on the front, and all connectors on the rear.
(http://www.pbase.com/kvom/image/163448388/large.jpg)
(http://www.pbase.com/kvom/image/163448389/large.jpg)
With the lathe power cord plugged into the control, starting and stopping the spindle is under control of Mach4. When I left the workshop, I was missing a cable that will be inserted into the lathe's PC board and thence to the control's VFD socket. This will allow speed control of the spindle by M4; until I receive the cable the spindle RPM will need to be set using the stock rotary switch. The speed control option, when present, will allow M4 to command constant surface speed when desired. With spindle operation delegated to the control, the oem start/stop switches were removed, and the hole covered with a plate.
(http://www.pbase.com/kvom/image/163448390/large.jpg)
Note that in the photos above one can see two homing switched, one for each axis. These are the orange proximity switches. For most users these are unlikely to be needed. However, using repeatable tool mounts such as a gang tool plate they may eventually have some use.
Here's the lathe as it sits:
(http://www.pbase.com/kvom/image/163448379/large.jpg)
The major adjustment that it needs now is making the Z ballscrew parallel to the ways. Presently the left pillow block is not attached to the frame, and securing it with screws as is causes some binding of the ballscrew and limits rapids. With the pillow block floating free movement of the end is obvious with carriage movement. Measurement and shimming is needed.
-
Nice info kvom. Thanks for the post.
I've been looking into a CNC lathe, so this is helpful.
I hadn't thought much about an encoder on spindle.
Where I'm stuck is finding a decent (and free or low-cost) program to convert CAD to g-code for lathe turning.
Lot's of stuff for milling but lathes seem to be scarce.
-
Mach4 currently has a set of "wizards" for lathe work that will generate g-code for simple shapes. CamBam will also do turning generation.
-
I just finished up (wait, are we ever finished?) some upgrades on an HNC Hardinge Lathe with Mach4. For G-Code, I'm using AutoCAD Fusion 360. It's licensed for a year at a time for free to enthusiasts and has some pretty nice CAM for Lathe and Mill. I've posted some code so far and done some edits to the post processor. Also some hand coding to allow a bar puller to advance the stock I'm using.
Too much fun! The lathe looks great and I'm sure you'll be churning out some cool parts soon! Great info!
-
Kvom, thanks for the description, it is a very nice review of the lathe. It was a very interesting project for me, and very satisfying to see 14 guys go away with working machines.
I was a bit surprised at the variability of the machines mechanical fit- some had saddles that slid freely, some had serious binding. I guess its just the way of these low cost machines- they are a nice kit of pre-machined castings for the user to finish.
Mach4 hobby now has a set of 'canned cycles' built in. The release for the class was the first release, these will soon be in the mach installer on the net for all hobby users. A more complete set of wizards, in the style of the NfsMill wizards will be done 'soon'.
We did make up a few extra mechanical kits that will be offered as soon as I can produce some instructions for assembly. If anyone wants a set let me know. We are considering a couple options for making these more available.
-
First step in aligning Z axis ballscrew is to remove the lathe from the chip pan and mount on the Bridgeport table. Then a scan of the ways showed that the headstock was .01" low. After placing a sheet of .01" shim stock under the headstock scanning the ways showed a max variation of .001" over the entire length. Thus the mill table will provide a height datum for aligning the ballscrew vertically.
-
Next I wanted to measure the vertical variance of the ballscrew over its travel. A parallel on the screw was held flat by my height gauge, and measured at both the carriage and the right pillow block. The difference was .002"; at this point I think that can be ignored.
-
To measure the horizontal variation relative to the ways, I employed a 123 block against the screw and used a caliper to measure the space between the block and the side of the way at both ends. Here the difference was .008".
-
With the power block attachment screws removed, it's possible to slide the left (OEM) pillow block on and off the end of the ballscrew. It is visible that the tapped holes are several mm too low, and that there is a significant gap between the frame casting and the bottom of the pillow block, probably 20 thou or so. Therefore the next move is to enlarge the holes in the pillow block to allow it to be screwed to the frame, and for shim stock to be inserted between the block and the frame.
At that point I can decide if I need to shim the right pillow block to bring the screw more in parallel to the ways.
-
Mach4 currently has a set of "wizards" for lathe work that will generate g-code for simple shapes. CamBam will also do turning generation.
Thanks. Good to know.
Now I'm wondering, when it comes to lathe turning, what constitutes 'simple' shapes versus 'complex'. :thinking:
-
Before taking the lathe off the mill, I measured the runout of both the chuck body and a ground round clamped in the chuck. About .001 or less TIR on the outer surface of the chuck, and .002" on the rod. This measurement was taken when manually rotating the spindle.
-
I removed the left Z-axis pillow block and milled out the mounting holes as slots. Then added some shims until it "seemed" in the right position. Also added a .005" shim to the right pillow block. Then after some tweaking of torque on the mounting screws, I managed to get 65 ipm rapids across the length of the bed without losing steps. I might get more with more fettling but for now I'm going to leave it. I did reduce the acceleration from 8 to 6 since it's less important for most lathe ops compared to mill.
The mounted the DTI to measure backlash (4" rapid away followed by 50 ipm return). Backlash was consistently less than .001". I intend to reverse the setup and measure closer to the headstock next time out, and also measure X backlash.
-
In an effort to get rapids faster than 55 ipm I've tried various tweaks without success, including milling the end of the frame square to the ways intending to make the motor shaft parallel to them.
(http://www.pbase.com/kvom/image/163526452/large.jpg)
Coming to the conclusion that 55 ipm rapids will have to do, and having received a set of index lathe tools from Shars, I wanted to determine if the 4-way tool holder would be useable, or whether I'd need to fork out for a QCTP setup. First I chucked a length of ground rod, measured it with the mic, and then use the height gauge to find the distance from the ways to the top. Subtracting the radius of the rod gave the height to the center of the spindle.
(http://www.pbase.com/kvom/image/163526453/large.jpg)
Then I measured the height from the ways to the tip of the turning tool, and determined that it was .049" higher than the center.
(http://www.pbase.com/kvom/image/163526454/large.jpg)
Milled the bottom of the tool holder in several increments.
(http://www.pbase.com/kvom/image/163526455/large.jpg)
Finally made my first cut in anger - facing some brass. ;)
(http://www.pbase.com/kvom/image/163526456/large.jpg)
I had placed the various Shars tools on the surface plate, and the cutting edges are all very close to the same height, so hopefully this holder can work for all of them. Time for more experimental cuts.
-
Hi Kirk
Looks like good progress; have you considered gang tooling? I run a Sharp CNC Tool room lathe at work and we have it set up with gang tooling; it is relatively easy to configure and teach all the tools for the job. With a gang set up you can also put your drills reamers etc. in ER holders. If you are running repetitive parts you don't want to be having to change tools for every operation. Once your tools are set up you also don't' have to worry about errors created by your quick change tool post. Have you given any thought how you are going to drill holes?
If I were to have a small CNC lathe at home this is probably the way I would go.
Dave
-
This is an interesting project.
If you get the time, it would be interesting to hear some background on just how you intend to use this setup.
Jim
-
Gang tools are great for production, something i don't intend to do. In general, i expect to use it for decorative turning for model engine parts, and also for threads that are too long for thread milling.
-
I have started to build a small gang tool holder for this lathe. I had hoped to have it at the class but didn't make it.
I really think a CNC lathe without quick and repeatable tool changes is not very useful, even for one-off parts.
-
I agree Ron
I work in a R & D shop and we rarely ever make more than a few parts of anything; definitely not production quantities. I just thought I would throw that out there as not everyone might be familiar with this type of tooling set up.
Omni-turn makes some nice tooling and its what we are using on the Sharp, along whit ER holders form Maritool
http://www.omni-turn.com/Pages/Tooling/Tooling%20page.html#Anchor-OmniTurn C-8655
http://www.omni-turn.com/index.htm
https://www.maritool.com/
Dave
-
That looks like a very nice, clean design. It's good and bad that you can put it on the milling table. It's good that you can lift it and it fits. The bad part is it won't take "large" parts. The Little Machine Shop 7x16 lathe seems a good candidate for conversion. I would hate to discarding all the irrelevant parts though.
I'm still converting a Sherline lathe to CNC. (https://forum.linuxcnc.org/forum/26-turning/30181-sherline-lathe-conversion) The basic machine is functioning but I need to mount the encoder. The Sherline CNC kit was easy to install but they have no mounting scheme for an encoder. Have been taken away by travels and now "restoration" of a basement wall. (Had a friend and structural engineer look it over. Expected to hear it's ok OR call these guys and get it fixed. He said you should put in some steel support beams but, oh, you could do this yourself handing me a set of plans. More time away from the shop!)
I'm mainly doing the Sherline to see how useful a CNC lathe will be in a hobby shop setting. It is a little small, but should answer the question. If I'd known about this 7x16 design I'd have been real tempted. The Little Machine Shop 8.5x20 looks like a good size if CNC is found to be useful. I've stated to use CamBam for lathe G-code generation. They have an experimental plug in that's marginally useful. I expect it will rapidly get better. Till then I expect a good bit of hand coding.
My plan is to use a quick change tool post. This, I think, will be similar to the Tormach mill tooling, consistent manual tool changes. I don't think gang tooling would be useful for a one off parts. But hey, I have no experience with CNC lathes. I'll be real interested in your experiences.
My reasons for a CNC lathe include fancy round profiling and threading. Round profiles like columns for model engines. Also threading a dozen parts or so, like studs or special screws. It'll be interested to see what uses others find.
Please keep posting and let us know what uses you find. Thanks.
Hugh
-
That looks like a very nice, clean design. It's good and bad that you can put it on the milling table. It's good that you can lift it and it fits. The bad part is it won't take "large" parts. The Little Machine Shop 7x16 lathe seems a good candidate for conversion. I would hate to discarding all the irrelevant parts though.
I'm still converting a Sherline lathe to CNC. (https://forum.linuxcnc.org/forum/26-turning/30181-sherline-lathe-conversion) The basic machine is functioning but I need to mount the encoder. The Sherline CNC kit was easy to install but they have no mounting scheme for an encoder. Have been taken away by travels and now "restoration" of a basement wall. (Had a friend and structural engineer look it over. Expected to hear it's ok OR call these guys and get it fixed. He said you should put in some steel support beams but, oh, you could do this yourself handing me a set of plans. More time away from the shop!)
I'm mainly doing the Sherline to see how useful a CNC lathe will be in a hobby shop setting. It is a little small, but should answer the question. If I'd known about this 7x16 design I'd have been real tempted. The Little Machine Shop 8.5x20 looks like a good size if CNC is found to be useful. I've stated to use CamBam for lathe G-code generation. They have an experimental plug in that's marginally useful. I expect it will rapidly get better. Till then I expect a good bit of hand coding.
My plan is to use a quick change tool post. This, I think, will be similar to the Tormach mill tooling, consistent manual tool changes. I don't think gang tooling would be useful for a one off parts. But hey, I have no experience with CNC lathes. I'll be real interested in your experiences.
My reasons for a CNC lathe include fancy round profiling and threading. Round profiles like columns for model engines. Also threading a dozen parts or so, like studs or special screws. It'll be interested to see what uses others find.
Please keep posting and let us know what uses you find. Thanks.
Hugh
Hugh, I hope that maybe you will start a thread dealing with your Sherline CNC conversion.
Jim
-
Hugh, I hope that maybe you will start a thread dealing with your Sherline CNC conversion.
Jim
A CNC build thread is a little off topic here. I also hate to document the retrofit in several places. I am documenting the process on the LinuxCNC forum. (https://forum.linuxcnc.org/forum/26-turning/30181-sherline-lathe-conversion) I hope anyone interested will follow the thread there. It includes all the nitty gritty ugly details showing my ignorance. I will post the final result here and, certainly, how I end up using it in making engines.
It's good to see interest in CNC here. I know how useful a CNC mill is. I have more doubts about a lathe but there's some interest in this also. I'll be interested in everyones experiences, and hopefully add a little.
Thanks.
Hugh
-
Hugh, I hope that maybe you will start a thread dealing with your Sherline CNC conversion.
Jim
A CNC build thread is a little off topic here. I also hate to document the retrofit in several places. I am documenting the process on the LinuxCNC forum. (https://forum.linuxcnc.org/forum/26-turning/30181-sherline-lathe-conversion) I hope anyone interested will follow the thread there. It includes all the nitty gritty ugly details showing my ignorance. I will post the final result here and, certainly, how I end up using it in making engines.
It's good to see interest in CNC here. I know how useful a CNC mill is. I have more doubts about a lathe but there's some interest in this also. I'll be interested in everyones experiences, and hopefully add a little.
Thanks.
Hugh
I can see your point Hugh. Thanks for the link..............I skimmed over your thread and will look at it in more detail later. CNC seems to be a whole subculture unto itself.
Jim
-
After some trial and error, I finally did a test turning job using the lathe. I have had some Mach4 issues that have gotten in the way, and as well I have found a number of problems with the lathe operation offered by CamBam. This week I discovered a reasonably priced and simple CAM program for lathe that I am trying out: eCam.
http://ecam.altervista.org/
The hobbyist price for lathe functionality is 75 euros, which at current exchange rates is about $83. The full download gives a 2 week trial period, and except for a few minor issues I had little problem getting working gcode is less than a day.
Attached are a photo of the part and a screen shot of the program window.
-
The workflow for external turning is pretty simple. Draw the profile in the upper half of the screen, either using eCam's CAD facility, or else import it as a DXF. Then select a tool and fill in machining parameters for roughing and finishing.
The program has a pretty good visual postprocessor facility, so I was able to get proper code for Mach4 easily.
My test part was machined using .01" steps, but because lathe is in diameter mode each cut was only .005". I will need to determine what cuts the lathe can stand in various materials, but for a small one-off part I'll likely tend to be pretty conservative.
I did find a reasonable workaround for the fact that M4 doesn't allow direct setting of the X DRO but instead requires you to rely on tool table offsets. Since the tool post I'm using isn't consistent across changes, a tool table makes no sense. So the first solution is to not specify a tool change in the g-code or else tool 0. Next I either take a skim turn off the stock with the tool or else jog until it touches. Then zero the X DRO, jog it away, measure the stock, and then issue a G0 X-nnn command, where nnnn is the diameter measured. The tool will then move to the spindle centerline. Now zero the X DRO again, and issue G0 Xnnn to move back to the stock diameter. Obviously in radius mode you'd use the radius and not the diameter in the moves.
-
After several tries I managed the first complete profile turning. In the case I a frame spreader from the Ouzof Coventry as the example. A photo is attached.
Next I attempted to cut a thread using a series of G32 passes since I couldn't get the G76 canned procedure to work as documented. In attempted to cut a 5/16x24 thread using 10 passes at 1500 RPM, I got good looking threads but the TPI appears to be less than 24. Unfortunately I can't tell whether M4 or the lathe electronics are to blame.
-
Looking good to me :ThumbsUp:
-
It appears that the motor tuning steps calculate for the Z axis may be off. I spent some time tonight trying to tune, but didn't get repeatable results across all Z lengths. More tomorrow.
-
So if it is not consistent you are missing steps maybe? Have you tried lowering your acceleration and speed?
Looks like your are making good progress though.
Dave
-
The motor tuning requires entry of a "counts per unit" field for each motor. Basically how many step pulses needed to move the avis one inch. Of course, the pitch of the ball screw figured into this. I've been using the number supplied in the class. So when just testing that moving in one direction and then reversing the same amount yields the same reading on a DTI, I have just verified no steps being lost.
So now I need to use a gauge block between the carriage and the DTI; zero both the DTI and Z DRO, and then command a move equal to the gauge block length. I found that a 3" move was short by about 30 thou. So now I need to try to adjust the count field to get the correct feed distance.
Whether the error was enough to effect the thread pitch I was cutting will be know when I retry it.
-
I have been able to adjust the count field on both axes to what I hope will be reasonably accurate. I did find a problem with the Z axis where the pillow block had worked slightly loose screwing up the results. I'll need to cut another thread to verify if the pitch will come out properly.
The adjustments took a lot of trial and error, plus setting up DTI and gauge blocks to accurately measure travel.
-
Did you find what caused the axis error? I've not seen a situation were the screw pitch and step angle didn't result in proper step increment.
-
I have tweaked the steps/unit in M4 but haven't yet tried a new thread. We've had house guests this week.
-
Ran a threading program today on brass rod for a 3/8-16 thread. Got a nice threadform but it was 32 tpi. There's a software bug here somewhere, but not sure if it's M4 or my control.
-
Are you using the threading cycle or G32 moves? I posted a thread from Fusion which uses G32's and It ran just fine on my Mach4 Hardinge lathe. I haven't tested out the G76 but it's high up on my list of things to try.
I'm really surprised to hear you coming out a a factor of the pitch. Is the encoder setup correctly on the spindle?
-
I tried it both ways and got the same result. I prefer the G32 because I can single step it when setting up to do a new thread (i.e., check pitch with a thread gauge). I also have had problems with G76 retracting too far so that first few passes just cut air.
I suspect a controller issue rather than mach4, but debugging is out of my hands.
-
Is the encoder setup correctly on the spindle?
Having the wrong number here will certainly "throw a spanner in the works" ... :censored:
Flip that coin and it can be an easy way to get the right pitch without changing anything else .... ;)
Ehh .... correction, if all other things are as should be - not a good solution.
I would check that your feed is correct - or is there another way where you can check how many pulses you get from the encoder for one full 360 degree of the chuck. I have a very old version of M3 and you can check many things in the diagnose section - I would expect M4 to be a lot better. If so - have a person counting the pulses shown on the diagnose page (if not done by M4) while you turn the chuck exactly 360 degrees in one direction only. There is a high probability that this is a nice binary value (2, 4, 8, 16, 32, or 64) - is this the same value you got in the setup for the encoder in M4 ?
-
There is no M4 encoder setup to my knowledge. The encoder output feeds to the control box, which uses Pokeys. I don't know whether M4 or the control box is the one coordinating movement, but there is also Pokeys motion control software in Windows between M4 and the control.
-
There is no way that any CNC program can do threads on a lathe without reading an encoder ....
The encoders I have encountered give between 1 and 6400 pulses per revolution - you therefor need to be able to tell the software this number (unless the software is locked to one particular encoder type).
I would be rather surprised if a professional general type CNC software like M4 should be lacking these fundametal functions .... but I have no clue where you find the correct setup menu.
-
Mach4 uses a hardware motion control board, in this case a PoKey57e. For this lathe we selected a 500 line per rev encoder. This is configured in the plugin for the PoKey device.
The PoKey does support a mode to use only an index pulse for threading as was done in mach3.
I have been trying to test kvoms code to see if I can duplicate the tpi problem, but I am having a problem with my encoder. I do know that the last time I made a thread, using the G76 it ran fine. We also had everyone in the class making threads and all came out right, as I remember even kvom made a thread then. A test machine in Brian Barkers shop, with a PoKey motion control makes fine threads. I suspect something has changed in kvoms config.
-
I didn't do any machining in the class.
I installed a newer version of M4 from the test FTP site per support's recommendation, and that required running the pokey install script supplied in the class. Other than that, the only changes were modifying the steps/unit to make the axes moves more accurate.
-
Mach4 uses a hardware motion control board, in this case a PoKey57e.
OK - so Mach has moved the control software from the computer, into firmware inside a "black box" ....
I just checked the PoKey57e - very smart device - and in this case it need to be told the encoder data ....
But I'm sure that you already knew that Ron, so I should probably just stay out off the conversation ....
Best wishes
Per