Homemade CNC 2 spindle lathe question

  Quoted Text Here  
    [ ... ]
  Quoted Text Here  
    That's O.K. As you have now seen, the debate went on anyway. :-)
  Quoted Text Here  
    I agree.
  Quoted Text Here  
    Good! The clutch will make a difference -- absorbing any possible jitter in relative position with slip in the clutch. Don't make it too tight a grip in the clutch to allow for the slip when both chucks are gripping at the same time.
  Quoted Text Here  
    Good enough.
  Quoted Text Here  
    Hmm ... this is a place where I would want the amplifier-driven servo with tach feedback -- syncing to another similar tach generator on the headstock spindle. But your clutch may handle a multitude of sins. :-)
    Good Luck,         DoN.

wrote:
  Quoted Text Here  
Greetings DoN, The clutch is a magnetic disc type that is pretty easy to control by varying the voltage. I'm a rank amateur when it comes to electronics but am able to make things work by following directions and advice. The things I don't know enough about are the speed at which the electronics work and the time it takes to accelerate and decelerate the spindle. There is the time it takes for one chip (LS7184) to read the encoder and output the step and direction signals to the G320 drive and the time it takes for the step and direction commands to be executed. Then the time it takes to bring the spindle motor to speed. I'm afraid that if the master spindle speed varies too quickly the slaved spindle will constantly lag. I'm probably overthinking this but since I don't know, and won't know until I try, I keep thinking up new problems. Like how to accomodate the inevitable part length variation. The slaved chuck will need to grab the part when it is bottomed out in the chuck. But as the part length varies it may not bottom out or it may crash. So I need to figure out how to move the slave spindle to the part gently but firmly and then retract the spindle and lock it in place with virtually no variation in X or Z when the slave spindle is in the machining position. Cheers, Eric

  Quoted Text Here  
    [ ... ]
  Quoted Text Here  
    [ ... ]
  Quoted Text Here  
    That is what matters, of course.
  Quoted Text Here  
    Well ... the timing of the signals out of the headstock spindle are going to be defined by two major things:
1)    The RPM of the spindle itself.
2)    The number of encoder positions (including whether it is     set up to generate pulses on both rising and falling edges of     both signals).
    Let's say a spindle RPM of 2000 RPM -- which becomes 33.33333     RPS.
    Now -- let's say an encoder disk capable of producing 1000 PPR     (Pulses per revolution).
    That is 33,333.33 PPS (Pulses Per Second).
    This is *way* too fast for a real stepper motor to handle.
    But -- it is well within the capabilities of a DC servo motor     with tach feedback -- and apparently the Gecko G340 driving a     DC servo motor.
    If you're using the Gecko G340 driver, you certainly don't want it jumpered to move 2, 5, or 10 steps per input pulse. But looking at the manual for the Gecko, it *will* handle a step pulse rate of 45,000 PPS (for a motor speed of 3000 RPM and the prober line count encoder), so presumably it can keep up.
  Quoted Text Here  
    This makes sense -- but the question is "Lag by how much?".
    And another thing which I have not yet seen considered is whether the orientation of the part in the second spindle will matter. In other words -- can it simply catch up and then close the jaws and open the jaws of the headstock spindle, or do you have to orient it properly -- for some kind of custom fixture, plus the possibility of drilling or milling from the side while in the tailstock chuck. If that matters, you need to not only know that they are going the same speed, but also that they have the same face towards you at the same time. This means that you'll need to use an encoder on each with not just the lines to generate step pulses, but also have a single index pulse per revolution so you know where they are.
  Quoted Text Here  
    And things like orientation of the workpiece in the chuck.
  Quoted Text Here  
    You need something to sense the projection of the part into the chuck. But can't you handle this by programming -- you face the workpiece in the first chuck before transferring, and thus you can be sure of the length of the workpiece.
    Good Luck,         DoN.

<SNIP>
  Quoted Text Here  
I don't think programming will work because as the tool wears the part will vary in length. So as the part gets .001 longer from tool wear it means that the part will be pressed into the bottom of the chuck a little harder than if the part was the programmed length. And when a new insert is used the part will be a little shorter. So yes, the program will handle the chuck position but I wonder if maybe the chuck needs to have a little give when in the clamping position in order to assure the part is always well seated in the chuck. Eric
  Quoted Text Here  

  Quoted Text Here  
    [ ... ]
  Quoted Text Here  
    [ ... ]
  Quoted Text Here  
    Hmm ... my feeling is that if an insert has worn 0.001" it is dull enough so you should replace it anyway -- or at least index to the next point. Yes, the rake is pretty much the same for that little wear, but you are losing relief angle on the face of the insert, so more friction heat, and more torque required. (Hmm ... monitor the spindle motor's current with the computer and flag a worn insert based on that?)
    But -- do you need to truly bottom in the chuck? If the grip is hard enough, you should not need to. The usual reason to do so on a manual machine is so the faced surface against the chuck face serves to eliminate tilt of the workpiece. But since this is already being held in the other chuck until this one has a good grip, it is probably not a problem.
    Again -- you could also monitor the current in the servo motor feeding the tailstock chuck into place -- if you are using a servo, not a stepper.
    Good Luck,         DoN.

wrote:
  Quoted Text Here  
Greetings DoN, I do need to bottom in the chuck to assure accuracy. I'm thinking about using air to extend and retract the chuck. When the chuck is retracted it will be clamped in position before machining is started. As far as .001" being a lot and time for an insert change this is not the case. Machining steel takes more tool pressure than aluminum and negative rake tools also have higher tool pressure than positive rake tools. As the insert dulls the tool pressure climbs and more than .001" difference in part length from a new insert edge is common. Even on small diameter parts. Cheers, Eric

On Sun, 17 Aug 2008 16:19:17 GMT, snipped-for-privacy@whidbey.com wrote:
  Quoted Text Here  
Why two motors? I was imagining a modest sized spindle. If it's large enough that you're considering an induction motor for machining, how about drop the servo and run the spindle with a vector drive (closed loop VFD)? Vector drives are capable of speed regulation of .01% or better and some models (some Yaskawas, for example) will follow a pulse input.
  Quoted Text Here  
I think Karl Townsend mentioned Galil controllers earlier. I'd encourage you to take a look at them. Your system is getting involved enough that you'd almost certainly have a greater chance of success, and save yourself a lot of headaches, with a general purpose motion controller. I've used several brands of controllers over the years and have a very strong preference for Galil. One of their older units, assuming it has enough axes, would be more than adequate for your app.
This is a fairly recent stand-alone: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item )0252936481
A couple generations older: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item20283862778
They're also available as cards for a PC, for example: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item (0256859431

wrote:
  Quoted Text Here  
Greetings Ned, I already have the Gecko 320 servo amps and the servo motors. I also have the encoders from US Digital. I didn't even think about using a VFD for matching spindle speeds. That's a good idea. I will look at the Galil stuff because of Karl's suggestion and you backing it up. Thanks, Eric

On Tue, 19 Aug 2008 03:54:56 GMT, snipped-for-privacy@whidbey.com wrote:
  Quoted Text Here  
The older Galil controllers were strictly +/- 10V analog out. Some of the newer models can be set up to output step and direction data, so, though it seems rather convoluted, you could connect to your Gecko drives if you wanted to.

wrote:
  Quoted Text Here  
Greetings Ned, After looking at the Galil and what it does I have decided to stick with my original plan. I don't think I need the Galil. At this time I just want to be able to machine both ends of a part and not do any cross drilling or milling. I think I can use M codes to perform the part swap. I may need to have some type of basic PLC. If so I'll probably use a Basic Stamp because I'm a little familiar with them having bought and used a teaching kit from Parallax. Cheers, Eric

wrote:
  Quoted Text Here  
Tachs are so 70s <g>. Modern servos calculate velocity digitally from the encoder signal. I've been designing and building equipment with servos for a long time and haven't seen a tach on a new installation for at least 20 years. If an analog voltage proportional to velocity is required there are black boxes available that will do so by monitoring the encoder.
Incidentally, Jacob Tal - the founder of Galil, mentioned elsewhere in this thread - gets much of the credit for the commercialization of digital motion control in the early 80s.

wrote:
  Quoted Text Here  
Greetings Ned, What do YOU think of using the master spindle encoder as the source of the step and direction signal to spin the second or slave spindle in synch with the master? Thanks, Eric

On Fri, 08 Aug 2008 15:18:27 GMT, snipped-for-privacy@whidbey.com wrote:
  Quoted Text Here  
I just finished responding to your original post, Eric. In short, I think synching the spindles is a no-brainer. My concern is whether the Gecko drive can be tuned to give acceptable performance under the various load conditions. It's an easy test if you have the hardware available - say a motor, drive, and pulse generator and the means to hang various loads and inertias on the motor.
Why do the spindles need to rotate for the transfer?

wrote:
  Quoted Text Here  
So I don't have to wait for the spindle to stop. But the transfer could be done with the spindles stopped. I do already have the hardware. It's time to run a test. Eric

On Fri, 08 Aug 2008 01:47:37 GMT, snipped-for-privacy@whidbey.com wrote:
  Quoted Text Here  
I don't think it would be a problem swapping the part as you're suggesting. What would concern me is the fact that you'll also need to drive the second spindle with the Gecko servo drive at high speed when you're machining the part, and there's no easy way to remove the step and direction translator and associated PID loop from the system. It may be difficult to tune the PID loop to get acceptable performance at both high and low speeds with varying inertia and cutting loads.
With a discrete motion controller and amp, it would be easy to switch control modes, or even tuning, in software when changing from handoff to machining.
Not saying the Gecko won't work, depending on your requirements, but something to keep in mind.