FILENAMES: G&L1.jpg G&L2.jpg G&L3.jpg DESCRIPTION: These are pictures of a CNC conversion project underway on a large Giddings & Lewis lathe. Thanks to Kevin Carroll . Kevin provided the following description: =========================================================================== I am retrofitting a 1978 Giddings & Lewis / Gisholt lathe and thought I'd post a few pictures. The lathe has an 18" hydraulic closing chuck powered with a 30hp DC servo drive through a 2 speed gearbox. The lathe is a "Slant Bed", this means the ways are not directly under the chuck, but are at the back of the machine. The carrige is a large inverted "V" shape with the cross slide working near vertically on the face of the carrige. This arrangement not only provides 100% protection of the ways from chips and dropped parts, it also allows a chip conveyor directly under the chuck at floor level. (The chip conveyor is not installed yet, it had to be removed to transport the machine.) In spite of it's age the G&L is in near perfect mechanical condition, the only repairs required are to replace the Z axis chip curtain and clean the greasy dust from the electrical cabinets. The ways are hardened "box" section lined with teflon, there is not one scratch, not even any wear haze. Notice the incredible beef of the cross slide and tool post turret, this is one solid machine. The tool post is trick, it has 6 positions on the upper deck, 8 positions on the lower. Should be interesting to program.... G&L built some really good iron back then (we also have a 1978 five axis horizontal boring machine), but the G&L controls were some of the worst, and since they were rare (most G&L's shipped with a different brand of control) they are very expensive to repair. The control worked on this lathe, but all things considered, including the fact that the operator would be new to CNC, we decided to retrofit the machine. I used a Delta Tau PMAC in the boring mill 2 years ago and the operator likes it a lot, so the lathe got a similar (but much simpler) setup. We are also converting to new digital scales for X and Z axis, Delta Tau's PMAC can be made to work with the old analog Farrand Inducto-Sine scales, but the cost of the signal amplifiers and super precise analog to digital converters from Farrand is about the same as the cost of new scales. As the first photo shows, I built my own enclosure for the control panel. This is just one example of the "open architechture" philosophy that Delta Tau has, you CAN buy a near turn key CNC package from them, or you can buy just the minimal components. If you have the time to build your own you can save a pretty good buck, and as you will see, you can build the exact control system you want. The enclosure is 1/8" polished aluminum deck plate (also called diamond plate or tread plate) and mostly was sawed with a skillsaw and finished with a disk sander. It is screwed together using 1 x 1 aluminum angle and plated 1/4-20 flat head screws. I've seen brand name retrofit cabinets just a few years old that looked, well, they did not look good. The deckplate should hold up for decades, that is the time frame I plan on with large machinery like this. It is already 20 years old and still in excellent condition, it should have at least twenty more years left. Electricians will give me a lot of grief, the cabinet is not NEMA approved water tight. I am aware of the environment this machine will work in, it will get the ocaisional splash, but mostly it will be subject to some pretty severe cast iron dust. Rather than go with an industrial grade keyboard I went with an $11 no-name, we can replace it 10 times and not equal the price of the industrial unit. The one on the mill is fine after 2 years, you might have to pry off a key to fish out some swarf now and then... The second photo shows the control panel. Because the operator will be new to CNC I wanted the panel to be as simple and non-threatening as possible. (The original panel had 32 knobs and switches.) About the only thing unique here is the 8 position function switch at the lower left. Normally 2 switches are used, one for jog distance and another to select function: manual axis selection, auto, MDI, etc. Since this is only a two axis lathe and I have the joystick to jog with (salvaged from the old control, it's a real nice unit) I was able to combine these functions into one switch, and indulge in the luxury of a separate Rapid Traverse overide switch, a real boon to newbie CNC programmers and operators (really lowers the stress level when test flying new programs...) Additional control panel tricks are all programming, With the selector switch not in Auto pressing the CW or CCW buttons jogs the spindle, holding in CW or CCW and pressing the Start button causes the spindle to run continuously.(Press Hold to stop the spindle.) Similarly, press and hold the Hold button and press CW and the tool turret will rotate to the next clock- wise position, ditto for Hold and CCW. The display is a 10.4" LCD with a touch panel to replace the mouse. I like this setup, but the technology is still immature. The display has too many random colored pixels, I blame the special video card. Back in the old days some Commodore 64's had this problem, called "sparkle". Connecting my digital camera will crash the touch screen mouse driver every time. There is just not enough competiton in this specialised hardware, but the situation should improve in the next year or so as flat panel monitors come of age. We will use a stylus to work the screen, not the operators gritty finger. I've got some 3/8" Derlin / Teflon rod to fabricate them from. We are going to use Virtual Gibbs as our CAD / CAM software, our needs are very modest in this area, until now all coding in our shop has been by hand. Should be something...draw the part right on the screen, fill in the dimentions and Presto, G-code file. The control panel is fabricated from the same deck plate, but reversed. The tool to engrave the lettering is a broken 3/16 carbide end mill, sharpened to a long 3 sided point in a whirly-gig. The deckplate was polished with 600 grit paper in a DA sander (with lots of wax type tapping fluid) followed with buffing with Simi-Chrome. The panel was washed with solvent (which dulled the finish somewhat) then painted with expensive automotive enamel clear coat. The third photo shows the inside of the control enclosure. The clone PC can be seen at the lower right, with the PMAC motion controller card and dual port RAM cards instlalled in two of the three ISA slots. Note the PMAC is a full length card and won't fit in every Pentium motherboard. I had to trim one corner of the Pentium's cooling fan to clear the cable to the dual port RAM board. Delta Tau sells a clone Pentium board, if in doubt pay the extra and purchase from them. The PMAC features a socket for an optional, simple control panel. Changing a setup variable converts this socket to twelve general purpose inputs, I have the (4) eight position rotary switches connected to this port. Note: the selector switches are some really nice units, the cams are custom made to produce 3 bit binary output, yet they were only $30 apiece. In the back left corner is the standard terminal block assy for the PMAC. It connects to the (new) digital scales, the limit and homing switches, the amplifier (drive) fault circuits, and to the amplifier analog input. The middle circuit board on the left is the (dual) Resolver to Digital converter board. This lets me retain the old fashioned analog resolvers for spindle and tool turret position. The one advantage of resolvers is the position can be read at power-up, so there is no need to home the tool turret. Delta Tau's R-to-D converter yields 4096 counts per revoulution. If the DC spindle drive is capable of reasonably precise motion I will have a custom M-code to rotate the chuck to the same point after each part, this will speed the operators work on attatching the hoist to the large castings that will be most of this machines work. It is hard to get precise motion from a spindle because the gear ratio is so high and the inertia so great and the friction so low: the spindle wants to overshoot the destination, often to the point where the system becomes unstable and it begins to oscilate violently. De-tuning the system to the point where it does not overshoot reduces precision so much the spindle won't orient within even 5 or 10 degrees. (at least thats been my *limited* experience.) The last board is the 32 input 32 output module, used to drive the various solenoid valves, motor starters, and the lamps in some of the control panel switches, and read some of the control panel switches and various limit switches in the tool turret, saftey, hydraulic and lubricating circuits. The outputs from this board are only strong enough to drive solid state relays, the original output modules from G&L are retained. There is a wad of wires at the bottom of the control panel, these were spares on the original control and when the machine is up and running I'll trim these back to 2" or so. I'd rather not get into a pissing match with fans of Metalworking.com about my selection of Delta Tau over Fanuc or AutoCon, etc., or our choice of Virtual Gibbs over the many other brands of Cad /Cam. Using 1998 technology it is well within the average motion control companies capabilities to make an adequate CNC, heck, I wrote my own steppper motor CNC, the theories are old news by now. Ditto for CAD / CAM, the differences from one entry level product to another is not great. Our choice of Delta Tau is based on their philosophy of open architecture, they had absolutely no problem answering dozens of my calls on how to get my home-made control panel to work or helping me with the logic to program one of the most complex tool changers ever, the one on the '78 G&L boring mill, in spite of the fact that I did not attend their training school. When push comes to shove I will say that Delta Tau has some technological features that are unmatched in the industry for less than 5 times the price, so many it is almost a shame to use this card in something as slow as a CNC lathe. Also, the cost of a Delta Tau system is quite competitive, and if you build your own panel and / or have to work with old hardware like resolvers they may be your lowest cost retrofit. The 8 axis PMAC is only $1000 more than the four axis version, and they sell a small amp that will control four servo valves for just $200, if your machine uses hydraulic servos. As far as Virtual Gibbs, they have a post processor that works with Delta Tau's CNC program, and a very attractive purchase price, that was enough to warrent a trial. Kevin Carroll (stepper@npcc.net) Haberle Machine 3202 N. Kenmore St. South Bend, IN 46628