Re: [CAD_CAM_EDM_DRO] Re: Servo torque requirements

siado01 wrote:

>Well, after more consideration and a little more time to think about it...
>
>I still seem to be as lost as before. I have a little more clear idea of what I want to do now,
>so here we go...
>
>- I am designing a 3 axis (for now) mill which will be used for machining small primarily
>aluminum, but occasionally harder (likely inconel) parts.
>
>- I will be using a cast iron table measuring 9x18"
>
>- My spindle is yet undecided, but will either be an R-8 mini-mill spindle, or a modified
>Sherline unit.
>
>- I am using TKH LM Guides on all axis.
>
>- 3 Parker Compumotor SM232B-N-FL-N Servos (1000 line encoders)
> http://www.compumotor.com/literature/pdf/pg190_sm_sz23_encdr.pdf
>
>- I think that I'm looking for around 300ipm max traverse rate but any feedback would be
>great from you guys.
>
>- I had originally planned to use the servos directly mounted to ground ballscrews. I have
>one ballscrew with 8mm pitch, but decided to do a little research before buying any more.
>
>- I will be using a PC based controller most likely with EMC. I think this is where the loop
>gets closed.
>
>- I would like to use the Parker Compumotor OEM770T Drivers for the Servos. They are
>designed to operate together and I can pick them up reasonable cheap.
> http://www.compumotor.com/manuals/OEM770T/0_OEM770T_Rev_A_Entire.pdf
>
>
>Now, here's my dilemma. I want to produce parts for model gas turbines, so the ability to
>work inconel will be a must. However, precision is a must. I have looked at some 4mm
>lead ballscrews, but I still wonder what sort of feed rate I can get.
>
>

You really need to think seriously about this. Hard to machine
materials like Inconel
suffer from work hardening, I believe. So, you need to make agressive
cuts while
not suffering form deflection of the machine. Cutting parts out of
Inconel on what
is essentially a wood router is just not going to work, unless these
linear rails are
different from what I think they are. You need a stiffness of something
at least
approaching 1 million pounds force per inch (1000 Lbs force gives .001"
deflection).
Now, a Bridgeport knee mill in good shape approaches this level of
stiffness. My
old one is a bit worse than this, and I'd approach a difficult job in
Inconel with some
trepidation. I'm afraid your machine sounds like it would be at least
ten, and possibly
100 times less stiff.

>I think the thing that I'm getting most confused about is how to figure out what my feed
>rates will be limited to. Now, I understand that most cards, such as ServoToGo or Vigilant
>use some sort of quadrature something in processing the encoder signal. What confuses
>me is that if a card says that it can handle PID bandwidth of 20KHz, does that mean that I
>can only drive my servo at (1000lines x 4) = 4000 ----> 20000/4000=5 rev per second
>or 300RPM? Somehow, I think I'm way off on this. Help!?
>
>

This is the least of your problems. But, servo bandwidth has nothing to
do with encoder
count rate. It is related to the rate at which the encoder position is
read and a new
velocity is sent to the DACs. If you can do the servo update cycle
(read pos, calculate
PID, send new velocity out) at 40,000 times a second (that's 25
microseconds for the
whole thing, for ALL axes) then you can get almost 20 KHz servo
bandwidth. This, of
course, is totally insane on a machine tool. Do you really need to
reposition the machine
table 20,000 times a second? Most EMC users are running their systems
at ONE KHz
update rate, providing less than 500 Hz bandwidth. (Now, my ancient
Allen-Bradley
7320 CNC control had a 100 Hz update rate, and the servo bandwidth was
limited to
about 5 Hz. This led to somewhat sluggish acceleration.)

Jon