I agree that we should release sooner than later. It would be great if the
issues mentioned by Rene were fixed prior to a release, but realistically some
of those may take a while.
Run-from-line for instance has more problems than that mentioned in #246, it
also does very bad stuff if there is a probe move in an external subroutine
prior to the line you try to run from. I implemented what I called "simple run
from line" which gives an option to simply run from the given line ignoring
everything that came before. This works well if you are careful where you run
from, and works around various bugs. My "fix" was done on 2.7, but I intended
to push it to master and have not gotten back to that.
Rene asked:
>> should I start adding stuff like that to the 2.8 milestone on github?
>> do we care about milestones?
That sounds like a good idea - looks like there are a few issues there now.
Moses
On 09/19/2018 06:50 AM, Gene Heskett wrote:
On Wednesday 19 September 2018 05:37:10 Rene Hopf via Emc-developers
wrote:
I agree that 2.8-pre is very stable at the moment, and works great
with stretch. But as far as Im aware those features have already been
tested by other people. A release is probably a very good idea, there
are a few things I would like to address before: -
https://github.com/LinuxCNC/linuxcnc/issues/246
<https://github.com/LinuxCNC/linuxcnc/issues/246> -
https://github.com/LinuxCNC/linuxcnc/issues/470
<https://github.com/LinuxCNC/linuxcnc/issues/470> on trivkins machines
there should be no axis setting. having 2 sets of vel/acc can be
confusing(to new users), and lead to very strange behaviour if they
don't line up. on stepgens, you also have to adjust the stepgen
limits, so you end up with 3 values... - increase of tooltable
limit(until my fix is finished), to something like 1000
https://github.com/LinuxCNC/linuxcnc/issues/374
<https://github.com/LinuxCNC/linuxcnc/issues/374>
https://github.com/LinuxCNC/linuxcnc/issues/314
<https://github.com/LinuxCNC/linuxcnc/issues/314> merge of spindle
stuff like
https://github.com/LinuxCNC/linuxcnc/pull/354
<https://github.com/LinuxCNC/linuxcnc/pull/354> and robs various fixes
for spindle sync moves.
This one I've seen on my g0704. Doing a tap with a tap big enough it
can't pull it to full depth in one g33.1 with only a 1 hp spindle, I do
in a loop, by pecking it a fraction of a turn at a time, but the
miss-tracking if the backlash isn't set right on the money, results in
enough additional cutting to loosen the thread noticeably.
As for the speedup in reverse, can that not be calculated from the axis
in questions MAXVEL modified by the tpmm or tpi?
"Axis in question" can of course be other than z as I've turned the head
horizontal and used X on a few occasions.
On the stepper machines I've built, I have observed that the axis's speed
limit can be exceeded, generating a following error, stopping the
machine with the tap still in the hole and needing a bunch of manual
twisting and jogging of the axis to extract the tap w/o breaking it.
This was a serious problem for the original kit Z motor which was a very
slow 1600 ounce/inch 4 wire nema34. A 940 oz/in 8 wire and a $170 AC
powered driver took the Z MAXVEL from 22in/min to the middle 140's. A
much more usefull machine resulted.
I have also observed that reverse speed, given the same S speed is
actually slower by 10% even with very high Pgain values, something I can
use without oscillations now that I have a very high resolution encoder
on the spindle, equ to a tad over 7000 in high gear, and a tad over
14000 in low gear. Running master, this effect is seen in G33.1, or in
M3/M4/M3. Top speed doesn't seem to be similarly affected. 1450 in low
gear, 2900+ in high gear, which should eliminate the motor. What counts
is that it gets the job done. :)
should I start adding stuff like that to the 2.8 milestone on github?
do we care about milestones?
Rene
On 19. Sep 2018, at 06:45, Kurt Jacobson <[email protected]>
wrote:
I agree with Chris. I don't think there could be a better time to
release master, 2.8, or whatever it will be called. At this point it
has been stable for some time and has been VERY heavily tested and a
LOT of people are using it in production. If anything it is more
reliable than the current stable. I am a strong proponent of merging
Multi Spindle, External Offsets, Revers Run and QtVCP, but I think
we should seriously think about releasing master beforehand. If we
don't it will probably end up being another year at least before it
is released. That would not be bad, but I do think a lot of people
would benefit from on official release the supports J/A. (trentster
on IRC this evening being a good case in point)
I also think that jepler's Stretch ISO is ready for prime time and
should be put in the spot light as it deserves :) Cheers,
Kurt
On Sep 19 2018, at 12:14 am, Chris Morley
<[email protected]> wrote:
I too agree with merging branch such as you named.
I also am interested in merging qtvcp soon.
My question is about merge strategy.
Merging invariably will cause/show a couple bugs... that's really
the point of master before release. It's been a long time since a
release from master.
We could release master before merging a bunch of stuff in - as it
seems pretty stable right now. Or we probably should wait six
months after merging major code projects, to clean up any bugs.
my 2 cents.
Chris M
________________________________
From: andy pugh <[email protected]>
Sent: September 18, 2018 12:05 PM
To: EMC developers
Subject: [Emc-developers] New Stuff
If I am going to merge Multispindle, perhaps we should also look at
adding two other things.
1) External Offsets
2) Reverse Run.
As far as I know neither of these have any effect unless actively
enabled. So I see no reason _not_ to merge them.
I find myself advocating both branches on the forum reasonably
regularly. External Offsets is potentially extremely useful for
Plasma THC (and also potentially for PCB engraving, following an
STL surface like probekins tried to.)
Lots of users are using Plasma with THC, with varying degrees of
success.
Reverse run looks useful for wire spark erosion. At least one user
is building one of those.
--
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils
and lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1916
_______________________________________________
Emc-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-developers
_______________________________________________
Emc-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-developers
_______________________________________________
Emc-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-developers
_______________________________________________
Emc-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-developers
_______________________________________________
Emc-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-developers
