Re: Threading under EMC

[Phil Smith <phil-at-conwaycorp.net>]

Chris Stratton wrote:

> Am I right in thinking that rigid threading modes are not currently
> supported in EMC?  As far as I can tell, there is a cycle for tapping
> (I'm really intersted in single point threading on the lathe) but
> it depends on a canonical function "SET_SPEED_FEED_SYNCH" which
> doesn't do anything yet?
>
> I'm trying to figure out what would be involved in implementing this,
> and the problem I keep running up against is that everything seems to
> be based on precalculating where they machine "should be", leaving it
> up to the servo loop alone to fix any difference with reality.  If
> this difference gets too big, a following error is declared.
>
> Somehow, it would be necessary to slave the position of the z axis to
> the spindle rotation.  One possibility would be to program a feed rate
> that should be very close to what is needed, and substitute the
> spindle-derived positions in somewhere instead.  Problem is EMC will
> expect the motion to be done based on how long the feedrate and
> acceleration parameters said it should take - perhaps there is a way
> to make it wait?
>
> Alternately, the interpolater could be altered to ignore the commanded
> feedrate and simply generate points based on the actual spindle
> rotation.  The problem is that EMC's PID servo loop as most actually
> use it is really a P & Feedforward algorithm, and we can't know
> exactly where the spindle will be several trajectory (or is it servo?)
> cycles from now.  Sure, we can initially plug in numbers based on the
> projected spindle rotation (it shouldn't change much) so the FF
> calculations can work and then use the actual spindle position for the
> P calculation - but will the loop close following errors
> satisfactorily this way?
>
> Of course there is the dodge of servoing the spindle and hoping it
> will be where we tell it to be.  When not threading we could either
> disable the axis and write a raw dac value, or have a flag that leaves
> the servo loop enabled with a fixed velocity and no position
> calculations.  Problem with this is that it requires serious
> improvement to the machine, compared to simply mounting a large
> encoder wheel between two bushings and boring out the assembly to fit
> over the back end of the spindle.
>
> Thanks for any ideas,
>
> Chris
>
> --
> Christopher C. Stratton, stratton-at-mdc.net
> Instrument Maker, Horn Player & Engineer
> 22 Adrian Street, Somerville, MA 02143
> http://www.mdc.net/~stratton
> NEW PHONE NUMBER: (617) 628-1062 home, 253-2606 MIT

Basicly the feed rate is controled by the spindles RPM,  say a 1/2in 13
tap  at oh in steel, anywhere from 80
to 160 sfm depending on the tap,  so we do some math,
80/.5*3.82=611.2 rpm for 80sfm    at a liner feed rate of (drop decimals
of rpm or just round)
600*(1/13)==600*0.07692=46.1538 inches per minute(once your in feed per
minute you can calculate your
time.  im going to probably get lost here because ive never calculated it
beyond this point but lets see how i do.
hole depth is 1.0in
create a time trial of time to break and stop spindle ide do this over 20
stops at same rpm to define a
prameter to work with, then split the rpm ranges up to find time at
various rpms until you find some sort of
constant.  say this magic number for 600rpm is about .5 sec.
so what were going to have is starting point,  .5sec pre stop feed
point(this is enough to damage the threads)
and the cutting feed. Darnit all it needs is spindle encoding, I cant
figure this out well i could but its a calc.
calculation bised on as speed aproaches zero and the as speed aproaches
comanded rpm in which you let the
feed rate be a clamped function of the direct relationship between
spindle speed and feed/rpm.
im just trying to get some pennys in the pot,  Ive been a CNC machinist
since 94 working with Fanuc,
Mitsubishi, and Seimens controls, all the way to very old fanuc and ge
controls. I know a lot of neat tips and
tricks for machining but rigid tapping without spindle encoding would
wrack my brain trying to match speed to feed as a time function.   Ill
try to spend some time thinking about this for a few days, depending on
what
you guys may need, or already know.
Thanks ahead for not calling me just crazy, :)
Phil Smith

P.S. Floating Tap holders are wonderful.