Wells Index conversion

[Dave]

I've been contemplating the retrofit of my old Wells Index mill. It's a 9x49 vertical knee mill, which came from the factory with big NEMA 42 steppers and a Bandit II control, circa 1979. I bought it when the shop that was using it blew a stepper board in the control, and after a bit of effort to repair the control I decided to give up on the wire wrap boards stuffed with discrete components and 7400-series ICs and run it from a PC. Life intervened, though, and now five years later I'm finally at a point of being able to think seriously about it.

My initial thought was to reuse the original steppers - they'd need around 8A at 100V to run at their full performance, but I could potentially derate them and run them from a Gecko drive. My reading suggests, though, that old round steppers often lose some of their field, and that they sometimes don't microstep well, both of which could lead to performance problems. That, in turn, pushes me to think about replacing the motors.

If I'm replacing the motors, my inclination is to go with servos. The machine is set up with 5 tpi ball screws and 2:1 belt reduction drives, so it seems like an ideal application for adapting a servo directly to the original mounts - 2000 RPM on the servo would yield 200 ipm rapids, which is probably about as fast as I'd want to push this machine. Adapting nearly any servo face to the NEMA 42 mounts seems easy enough, and pulleys and bushings are available with pretty much any bore size and tooth count that I'd be likely to want.

The question, then, comes down to control method. Originally I'd been thinking of Gecko G320s and brushed DC servos controlled by EMC; the main limit I see with this setup is that getting 2000 RPM from a PC serial port (~25kHz, if I recall) requires 750 CPR servos; rounding this up to 250-line (1000 CPR) encoders, I'd get 1500 RPM, good for 150 IPM rapids, which would probably be enough to keep me satisfied. On the other end, 1000 CPR times 10 revs/inch (5 tpi, 2:1 reduction) is .0001" resolution, which is certainly acceptable. However, there's not much wiggle room - if my serial port winds up only managing 22 kHz, my rapids drop to 132 IPM, which is getting awfully slow for my liking. I could go to 500 CPR encoders, but then my resolution goes to .0002", which is getting toward the high end of what I want.

So, to make a long story short, I've started thinking about brushless servos. The ones I've priced so far seem shockingly expensive (~$1500/axis), but there is such a dizzying array of servo amps and servos out there that it's hard to nail down just how much I'd really have to spend. Some of the amps take step/direction inputs, although that would run me into the same limitations as the brushed servos and Gecko drives; it seems to me that, by the time I'm paying all the costs for brushless servos, I ought to be looking at a servo card to go into the control computer. This approach has the most appeal to me, since it closes the loop at the controller rather than at the servo amp, but it's also the most expensive - $500-$600 seems to be the starting point for servo cards with +/-10V output for the amps, plus whatever the amps and servos cost.

I have to wonder if I'm over-spec'ing the servos. I've ballparked 850 oz-in continuous as a reasonable torque figure - does this sound about right for a machine very comparable in size to a 9x49 Bridgeport? That pushes me into a 1500W brushless servo if I expect to maintain that torque up to 3000 RPM (and more power still if the motor can spin faster!), which seems awfully big. Off the cuff, I'd expect to be able to get by with a 1000W servo, although I don't have any basis for making this judgment other than it sounds about right. What have other people used on their machines? What does Bridgeport/Hardinge put on their CNC knee mills?

Also, is a +/-10V velocity signal the best way to go? Many of the servo amps I've seen have multiple other modes - torque mode and PWM come to mind. I don't mind spending the money on a servo card, if there's really a performance benefit to it; I do, however, object to spending money just because I don't understand enough to know what I should be doing.

Thanks in advance for any input or insights that any of you can give me!

[skullworks]

Kirk's Shizuoka Bandit to EMC2 conversion.

Then...

You are right that the 850oz-in servos will do fine for that size machine with about a 3:1 to 5:1 reduction.

In my younger years I did run a Wells Index with the Bandit. - Crude CNC, but very effective with a skilled operator.

Todays operators don't know the pain of being 19ft into a tape and have a typo....

[boilerbots]

I don't know where you are on this conversion but I thought I would jump in and also point someone who has sent me a private message to this thread. He was also asking about Wells Index conversion help.

I would rather put some effort into answering peoples questions once here in a public place.

I decided to take the safe and cheap route and it has been working fairly well for me. I threw out everything in the original bandit controller except for the power transformers inside each of the 3 stepper drivers.

The original steppers are 80V steppers from what I can tell and what information I got from Wells Index. Pull the transformers out of those drive amps and you have your 80V supplies almost done.

Next you will need a big cap and a bridge rectifier diode to complete your 80V supply.

I am using Gecko 203V stepper drives.

I control the 203V using a Smooth Stepper and an interface board that I modified.

For the Spindle, since it has a real nice 3-phase motor I purchased a Hitachi SJ200. These are nice because they have a modbus serial interface built in, just remove the control panel and make an adapter cable using an RJ45 cable.

I use a USB to RS485 converter to control the Hitachi VFD with a few tweaks to Mach 3. This gives me full digital control over the spindle.

I locked the variable speed head into it's highest setting because the VFD should not be used to spin the motor faster than it was designed. This gives be approximately zero to 4200 RPMs. Sometimes if I need more low speed torque I might use the back gear and then just compensate mathmatically.

The next trick is the wiring. Boy this was a real pain in the ass. I am an electrical engineer by trade and I never thought I could have so many problems with this. The biggest problem is electrical noise. Everybody talks about this and those that don't either got lucky or really really really got some secret sauce that they are using. After it was all said and done, I think the thing that really fixed my problems was placing a choke on the USB cable between the smooth stepper and my computer. You got a ways to go before worrying about this. As far as the wiring on the Mill, if it looks good then use it. My mill had this small metal box mounted on the side with barrier strips. All the motors, lights and switches where terminated here. If your cables are all run in the metal flex conduit then I would say everything is good to this point.

Get a big metal box and install everything new in it, run a cable between that new box and this box on the Mill. Strongly consider a metal conduit for this.

Last big item to think about, the spindle brake. Mine was setup for 3 phase and required some fairly high voltage to hold the brake off. A VFD isn't probably going to work well with the brake. I ran my mill for a while without any brake and it sucked. It really sucked because I still use the quick change adapter and it isn't so quick without something to stop the spindle. I called the manufacture of the motor brake and they did not have any alternatives. I though maybe they would have made a pneumatic brake but I was told they do not. I decided to try to modify mine using some junk I had laying around, basically two short air cylinders, probably 1" stroke 2" diameter. The brake has springs to engage but requires something to push it down to release. I was able to modify mine by attaching the air cylinders to the top cover in such a way so that they use air pressure to push the brake pad down. I programmed the SJ200 to control the air valve through one of it's IO pins. The only downside is that if I forget to turn the air on and the pressure is low, I can start to burn the brake (I did this once).

Since the conversion I have used the mill for hundreds of hours and it is working just fine minus software bugs and programming mistakes.

I am still using the spray mist lubricator and auto lube system that was on my machine.

After some use I found that the spindle was cracked up inside the quick change holder. Wells Index was able to find me a salvage spindle and fully rebuild it. It is very awesome and tight, although it did cost some $$$ to get that spindle.

I am sure you have questions, ask and I will try to answer. I also have lots of pictures I can post if you want to see something specific.