The G-Rex has been a huge disappointment for me. The development time
took 5 years and untold $$$. By comparison, the G203V drive took less
than a year and only $25,000 in direct expenses to develop. It is an
idea I probably loved too much and I loved it for too long. It is
expensive to build, expensive to test and very expensive to support.
In a word, it is nearly unprofitable.

The mistake was misapprehending what our strengths and weaknesses are.
We design good hardware, we are not so good at firmware. The G-Rex
needed both. Second, the G-Rex is being squeezed by improving PC-only
solution step pulse rates at one end and me-too, less expensive
similar products from the other end.

I cannot justify investing any more effort and expense into further
development, hardware or firmware, of the G-Rex given its past market
performance and future potential. Instead, that effort and expense is
going into new motor drives. The Rabbit fiasco was the last nail
driven into an already expensive coffin.

We will always have the G-Rex available. We ordered 500 more G100
boards yesterday. We just aren't going to develop it any further.

Mariss



--- In geckodrive@yahoogroups.com, "Mark Vaughan" <mark@...> wrote:
>
> That hasn't been written yet for the G100, and if you are going down the
> mach3 road, I doubt at present whether it will be issued.
>
> Mach3 is the only CNC machine control program I am aware of that
works with
> the G100, this required a special plugin and firmware to be written
for the
> G100 and a license paid for the firmware.
>
> It is to the point where a machine will operate with mach and a
G100, there
> is no spindle sync for rigid tapping, code consisting of many short
moves
> runs very slow, it cannot handle very hi speed moves with short code
> segments, and probing is limited.
>
> Maris has spoken about closed loop stepper control using a G100, but
this
> relies not on correcting a positional error, but slowing all stepper
motors
> should one start to lag behind too much, being slower it has more
torque and
> the ability to catch up.
>
> There is no closed loop control via the G100 in mach3, we can connect
> encoders back to the G100 and see the value change, but nothing is
done with
> that data, and you cannot reset the count values in the G100. You could
> write a vb script of brain to look at the difference between the
mach dro's
> and G100, enough to signal an error, but not enough to control backlash.
>
> There are other control devices being offered on the mach group that
tend
> towards this possibility, a DSP controller, and a smooth stepper
control,
> however backlash control by these means generally isn't very effective.
>
>
>
> Mach does have some inbuilt backlash control, but the best solution with
> back lash, especially on a mill, is to engineer it out. The trouble when
> milling is that the cutting force on the workpiece and bed can be in the
> direction of motion, or against it depending on whether you are
conventional
> or climb milling, or coming around a corner, software cannot
estimate where
> this occurs and where a backlash jump, overcut, gouge or undercut is
going
> to occur. You won't really achieve much if you try to hardware
control this
> as you are suggesting, the cutting forces usually result in the bed
jumping
> far quicker than your control will be able to respond when working with
> steppers. Servo's use much higher resolution than steppers, steppers are
> expected to move to exact position, whereas servo's work with an
error, but
> that error is so small it doesn't bother you, so a servo might get some
> improvement with this method. You can often get away with this well on a
> lathe since most of the time you cut in one direction so load on the
screws
> is in one direction.
>
> Linear slides are used on mills and lathes in the commercial world
in closed
> loop, this is usually to confirm motor encoders, allow for thermal
expansion
> of the bed, and allow for errors in the screw linearity and thermal
> expansion, backlash correction is minimal.
>
>
>
> You engineer backlash out usually by providing a second nut. Some
machines
> will have one nut mounted on the bed, and the other by slotted holes
with a
> spring between. Loosen the mounting screws in the slotted holes,
allow the
> spring to push the nuts apart so taking up backlash, and retighten,
a common
> technique on lathe cross slides. Many mills will use two nuts that are
> racked up against each other and clamped, usually via a precision ground
> washer between then, we call this a preloaded assembly. Other techniques
> involve using a mixed size of balls in a ball nut.
>
>
>
> I am presently running a G100 with my mill, with the above
limitations it is
> still a pretty good machine. I do have mega bucks commercial zero
backlash
> ball screws since the conversion was based on a commercial mill with bad
> electronics, running with G100 and mach we have 0.6 micron step
resolution
> on big 4.5Nm hi voltage servos with no step multipliers, high speed
rapids
> of 15m/min, cutting at 5m/min before the G100 falls over, spindle speed
> control form the G100 analogue ports, just no rigid tapping; a
floating head
> covers that, slow 3D work, and poor probing ability due to delay between
> mach and G100 communicating. The thought is however heavy in my mind to
> change to something else since there appear to be better products
now, plus
> support and continued development for the G100 seem to have all but
> vanished. Sorry if that statement hurts Marriss, but if it does,
please kick
> someone back into action, your closer than me, the product still has
promise
> but needs more work before alternatives supersede it.
>
>
>
> Hope that helps
>
> I would suggest you try joining the mach1mach2cnc group.
>
>
>
> Regs Mark
>
>
>
> Dr. Mark Vaughan Ph'D., B.Eng. M0VAU
>
> Managing Director
>
> Vaughan Industries Ltd., reg in UK no 2561068
>
> Water Care Technology Ltd, reg in UK no 4129351
>
> Addr Unit3, Sydney House, Blackwater, Truro, Cornwall, TR4 8HH UK.
>
> Phone/Fax 44 (0) 1872 561288
>
> RSGB DRM111 (Cornwall)
>
> _____
>
> From: geckodrive@yahoogroups.com [mailto:geckodrive@yahoogroups.com] On
> Behalf Of bustermick72
> Sent: 06 December 2007 09:26
> To: geckodrive@yahoogroups.com
> Subject: [geckodrive] g100 rex
>
>
>
> just looking for drivers to make machine full closed loop.
> the machine is a bridgeport with standard lead screws, with steppers
> direct coupled to them, screws have some backlash, if I were to use the
> G100 with linear encoders on table slides not on motor shafts, the G100
> would be able to work out the precise movements on the table
> considering backlash on the screw and motors would would be turning but
> the table may not due to backlash. if it can do that compensation I
> will gladly go and buy one. Also it seems they best run useing MACK3 is
> this correct. can anyone help
>