Re: [Emc-users] dynamic feedrate control

[Thomas J P. <tom...@sb...>]

Ray,
 I cant understand your reply. I read it forwards, I read it backwards. I think it even smelled it.
No one has ever done dynamic feedrate and shown it to others. I've asked to see the code
or any docs, photos etc many times, and never saw any. I'm sure people have tried and
have wanted to, but if there are examples or attempts, ( and I'd  gladly be wrong ),
please show us where to look. Please, no offense Ray, I like your work, but I hate for Mac to think
there's some prints or documents for him to work with. 

Mac, 
the 'currrent sense' is classicly a 'voltage input' that works makes cnc work like a classic servo. 
Moving forward along the path at a rate proportional to the 'gap value'. BUT, allowing reversals,
and these reversal are foreign to EMC ( and Mach2 etc ).You know all this if you work in EDM.
 It will be easy to use input a gap signal to HAL. It will be very very difficult to use it in the way
EDM works. The modifications to the control system that allow maintaining the gap distance 
is foreign to this project. The authors use feed systems, where the path is only forwards and 
not dynamic in the sense you want.

 If you want a simple circuit for the gap signal ( say +/- 10V and all > 500Hz removed ),  just
email me.

  I work in EDM and have managed to use EMC to position work, but not to EDM it. 

 I hope you all the success possible, but be aware that most EMC people will have blank
looks at the concept of dynamic feed or dynamic position control.This is not centered on EMC,
but most control system manufacturer fail to understand. 
Only DeltaTau, INAV, Aerotech, MachineMate, Beckhoff and Heidenhain have off the shelf solutions. 
None of these are < 10K$ for just the control (  no motors, amps, i/o, machine tool, magnetics). 
ISG Stuttgart has some links but I suspect these are Beckhoff's TwinCat.

 The TP (trajectory planner) is where this work is done. The present TP is in flux. It is getting stronger,
but the developers have not been interested in EDM, at least until 'simpler' issues like backlash and
2d ( 3d) axis compensations and blending are stable. 

 The developers work hard and dont vary from thier goal. Until EMC becomes more stable for mills,
then lathes, it wont be worked on fo EDM ( or force feedback polishing, or even torch height ). I hope 
the system can allow for EDM once it gets stronger, but worry that the omission at the beginning will
hurt it later. (the code has and is being re-factored, but the factoring may not make EDM easier later).

 I think there's 3 levels of EDM 
  Hole Drill are simplest becuz its single axis ( dynamic control along a vector in 1 dimension )
  Sinker ( Senkerodieren ) are next, they only need to know a couple point to retract to.
  Wire, is the most complex, as you ned to retrace geometry.

  (Graham Stabler of UK has pre-run Gcode for his stepper edm, and recorded every 
   step into a CF chip. He then uses the GapVolatge like parsing a string, like *cp...
   he merely goes fwd or bwd in this file of steps according to the GapV.)

 You'll also require on the fly position changes ( jump flushing and inspection ). These haven't been
allowed for either. Fred P of NIST suggested that these can 'simply' be implemented by injecting the
new positions into the segment queue, but I cant see how it's a queue if an element can 'butt in'.

 Also, the ability of the system to control will be tested when dynamic control is tested. This is very
different than what ECM does now. Recently I tried to bring up that the machines being tested did not
have the mechanical characteristics neccesary to test a control system. Most of these machines
do not use linear rails.
 This means they are not stiff,
 That means they cannot be tuned 'tight'.
 That means they cannot respond to a dynamic control, the slop inthe mechanics will destroy the process.
 The whole idea of dynamic response tests is foriegn, EMC setup doesnt even include 'step response' tests.

 There's a lot of .... non-interest in EDM, and something like resentment to such assertions that EMC
is not as good as some commercial applications. BUT , i hope it can  be.

 Sorry for the harang, but EMC can do what you want only if you write the code.
 I'd be very glad to help in your efforts.

regards,
TomP
tAG

EDM since 1973, operating, servicing, designing, modifying

copley 423's? I use Baldor M55's