Large Lathe Retrofit
Here are some pictures of my lathe I have retrofited with linux cnc. This is a 72" swing lathe, about 10 feet between centers. I am not sure of the brand or the year. I am guessing the year is around the1960´s. Somethign interesting about this machine is that the toolpost slide is backwards (maybe its British :-D ). The spindle motor is 20HP and runs at about 53 spindle RPM. Currently the servos are 750W 3Ph running 50:1 gear boxes on Rockford ballscrews running linuxcnc step and direction open loop. Its kindof an ugly setup, but hey, its payed for itself multiple times since the retrofit. This is a work in progress... Next step is to cast a proper foundation so that I can make a REAL third way instead of the current railroad rail. We mostly use it to turn railway wheels, both locomotive and freight. The tooling for this job is .75" RNMG inserts left and right that are used both for roughing and finishing.
On the todo list in approximate order:
Add an encoder and VFD to spindle
12 Inch x 10 Foot x 20 Foot cast reinforced concrete foundation (The current foundation is 6 inches and flexes under the third rail.)
New 4340 Ground Third Rail on New Foundation
New Heavy Duty Extended saddle - 4340 w/turcite on ways with space for heavier preloaded ballscrew on X
Double ballscrews on Z Axis on both sides of the lathe
Linear scales on z and x axis and closed loop operation
New headstock and tailstock (fabricated,not cast) to increase swing about 12" add a morse 8 on the tailstock and 8" spindle axle with 4" bore.
So far all the work on this lathe has been done in house and I think we'll be able to do all the upgrades inhouse as well.
Any questions, comments, suggestions and/or opinions are always welcome.
Here is a link to the picasa album with some pictures. I was never able to upload the pictures directly to the forum.
Do you build up the tires and axles then turn them back down?
An old friend of mine had a 20' lathe of 50's or so vintage sitting on a garage floor... he had to crank the X in and out as he went down the rails to get a straight part.
We were lucky that the lathe, beside having hardened ways, the wear on the ways is very little to the point that we were unable to measure it. According to my father, when he purchased this lathe in the late 60s it was a Panama Canal surplus unit sitting in a dry-dock shop and had only been used to balance stub shafts on tug boats. After he bought it, it sat in one of our warehouses assembled the entire time until I decided to put t back together last year.
The railway wheels can only be turned down, they cannot be built up because the wheels could crack under the stress of the weld as it cools. On a freight train traveling 100mph this would be catastrophic. Typically a wheel set comes with discs that are about four inches thick at the tread which usually allows you to turn a wheel down three times or so. After that, if the axle is still good, the discs are pressed off of the axle and a new set of discs is pressed in. The discs are pressed in using a 600T press. The rule of thumb is that if it takes less than 150T to press the new disc on the axle then the whole thing is scraped.
Any questions, comments, suggestions and/or opinions are always welcome
I like to see LinuxCNC making money
I want to know: what electronic are you using ? I mean board, drive and servomotor.
Also i want to know what ballscrew are you using.
Thanks you and ,
The discs usually last about 2 years before they are scraped. One of the local scrap yards usually picks them up, though I don't think they classify them. I'm not sure what kind type of steel they are made of, however I can tell you the new hardness on both the axles and the wheels is about 60 HRC and the wheels quickly harden way beyond that within a few hours of being on the rail. The inserts we use are a special grade coated carbide by Kennametal and we get pretty good wear and performance out of them (about one insert per wheelset) but anything else below that the wheel machines the insert, not the other way around.
I know what you mean with your friends long lathe. In our case because the insert is cutting basically right over the third way, almost 100% of the load rests on the third way and the concrete visible flexes under load. You can put a straight edge on the ground and watch the concrete move away from under it. The new foundation is going to weight approximately 16.8 metric tons, so that should take care of that!
We can't afford to go out and buy new cnc equipment, so linuxcnc has sure been a life saver for us! My goal is to have a complete CNC shop using only linuxcnc. I am very thankful to the linuxcnc team and community and hopefully I'll be able to give back to the project by helping with the development soon (I'm a computer scientist by education).
The current setup on this lathe is very basic:
The X and the Z axis are driven by 750W Panasonic Minas servos and drives with 50:1 zero backlash reductions. The X ballscrew is a .75" diameter 5mm lead unit. The Z Ballscrew is a 2" ballscrew with a .5" lead.
The control is linux cnc through a parallel port card and an optically isolated breakout board that I designed that shifts the step and direction pulses from 5v TTL to 24v single ended. They are driven open ended: the only inputs into linuxcnc are the limit switches and the E-Stop. Soon I'll be upgrading to a MESA setup with linear scales so I can take all the outputs from the drivers back to linuxcnc. Specifically I am most concerned with connecting the watchdog from linux cnc to the "servo enable" pins on the drives and then the servo "error" communication and positional info back to linux cnc. Right now, if we get a drive error (input voltage etc.) where the drive stops, the other joint keeps going... we've had some close calls. If not for performance, I'd at least like to be able to stop all movements when a joint isn't doing what its supposed to. The increase in positional accuracy would be an excellent bonus. I'll probably add a VFD with a spindle encoder at the same time so I can add the ability to turn threads in this lathe.
A motion card will pay itself in a short time.
Soon I'll be upgrading to a MESA setup with linear scales so I can take all the outputs from the drivers back to linuxcnc
Instead of linear scales i will consider to use the encoder output from the drive controller conector. If the drive doesn't have it, i will use an external encoder attached to the ballscrew.
Specifically I am most concerned with .... positional info back to linux cnc
One of the other things I was thinking about doing, not for motion control purposes but for improving our cutting processes is to mount an accelerometer and a vibration sensor on the tool post so that I may measure chatter in order to identify situations that cause vibration or chatter and improve our cutting parameter selection. It would be interesting if I could for instance identify the vibration signature of when an insert wears out or breaks and then possibly output an error message on the user interface, retract the tool post or even stop the job immediately if the case warrants it.