BigJohnT: I got a complaint about overly long messages on the list - would you be sure to trim unneeded stuff, especially when responding to john domville - his messages are really bad.
watching the iditarod
yea, I did a huge snip on the last one
what is iditarod?
not sure what caused all the space
dog sled race in Alaska
his steaming-pile of a mailreader I assume
"Microsoft Office Outlook 12.0"
yea, it had tons of extra spaces
just noticed it in the last reply
long ago Nome Alaska was in a tight spot and the dog sled teams mushed 1000 miles to bring the needed medicine to save the town
being from Alaska (born on King Salmon Island) I'm interested in the race
do you have any idea on when 2.2.7 might be released?
it's up to jepler, I haven't heard him say what his plans are
ok, just wondering out loud
it has an important bugfix now - touch off of rotary axes on metric machines is wrong
interesting my chevy Z71 will coast farther than the goldwing
personally I'm starting to feel ready for 2.3.0
BigJohnT: bike tires low?
yea me too
no I checked them
I think it is the mass of the truck vs the windage
the bike is 800 lbs the truck is what 3000 lbs
drag to inertia ratio
kinda surprised me that my truck would coast 1.2 miles down a stretch of road and the bike only .75 miles
I wonder what the weight per square foot of frontal area is for both vehicles
I would think the truck is 3 or 4 times the bike in weight to frontal area
I should ask Hawkins about gravity...
anyhow I just want to make sure the docs are as up to date as possible before 2.2.7 is released
jmkasunich: did you see http://cvs.linuxcnc.org/cvs/emc2/src/emc/motion/command.c.diff?r1=1.113;r2=1.114
I think it was you who always complained about that stupid error
I saw it
I dunno if I "always" complained
I bet I did at least once tho
ok maybe not
I wish I knew which firefox update broke it
for the last month or more, whenever I click on a link in here (or in an email, or whatever), instead of starting FF and opening the page, it simply starts FF
I have to click again to open the page
yeah that broke for me too
what happens if you drag a link into FF?
some time ago it seems
if FF is already open, it works fine
I noticed a problem appear when I upgraded to 3.x
I'm using 1.5.-.12eol
I suppose that "eol" is their way of telling me to upgrade
1.x is older than the hills
(in internet time anyway)
this is the one that came with 6.06 (and has been updated from the 6.06 repos)
ah. I'm on Windoes or 7.10 or 8.04
I'M FREE!!! (...from calculating tool offsets)
this is SO cool
the gui is great. I thought it would have to be more complicated, but to the user it's exactly like touch off. the user wants to change the tool table such that the dro becomes a certain thing.
oh, did you write a tool table GUI? thingie?
no, it works just like touch off
instead of changing the coordinate system, you can change the tool table
ok. I believe I'm lost, but good deal! :)
say you zeroed any old active system to the top of your vise. load a tool, move Z so you can just roll a .5 dowel pin under it, touch off Z, pick tool table on the menu, enter 0.5
it write the tool table entry and the gui shows Z=0.5
ok, so "tool table" is now one of the coordinate systems you can touch off - nice!
it does a somewhat different thing, but the user interacts with it the same way
right, that's nice
this kicks butt
the things I think BOSS does better are quickly running out
* jmkasunich doesn't follow, but I guess I should update and try it instead of asking stupid questions
cradek: did you see earlier where swp said he'd do his bport retrofit before I get my gantry built?
I think I asked if I should ;)
what about you?
the question in my mind is when will I do mine?
I'll start mine the minute it stops working, or maybe sooner
jmkasunich: if you can design and build an entire machine faster, that's pretty sad for us
I got rails and a screw today
one more tiny step
so does this new feature that you wrote add "tool offset" to the list of coordinate systems that you can select when you hit the touch-off button? or does it add another top level button?
adds it to the touch off screen
(when it's appropriate)
meaning whenever tool length offset is in effect?
= when a tool is loaded, and you're touching off an axis that makes sense for TLO
the TLO does not have to be in effect, but after you do the touch off, it will be
ok, so it always operates on "the current tool number"
tool is loaded, meaning TnM6? or lengths in effect meaning you issued G43?
SWPadnos: I don't know what it should do in the T1M6 G43H2 case
so people who use G43 without M6 won't see this
I know what it *does*, but not what it should do
yeah, that's a tough one
jmkasunich: that's true.
right now, the offsets come only from G43
jmkasunich: I don't see why you would use G43 without M6
(note I'm not saying nobody does)
I do, but I probably shouldn;t
I write things like:
(MSG load 5/16" endmill)
G43 Hwhatever is that endmill
why not use hal_manualtoolchange?
I should probably be doing that
then you can just write Twhatever M6 G43 (5/16 whatever)
but I don't think it can tell me "load 5/16 endmill", it will just say load too N
oh I guess I just look at the gcode display - that line or one near it will be highlighted.
I write code for stupid operators
I mean to see a comment that describes the tool
is there a comment field in the tool table?
but passing it to halmanualtoolchange would be a challenge
I wonder if hal_manualtoolchange can display that
you can put whatever you want after the useful numbers, but if emc writes the tool table, they are gone
oh, we digressed - my other question about this functionality
that could use improvement (including showing the tool description on the screen, etc)
you hit touch-off, select "tool offset" instead of G54, etc, enter a number, and hit return
does that immediately get written to the tool table and the dro goes to the desired value, or do you have to take some other actiona
cradek: Thanks, that's a great feature. And would entering 2.0 be in-range if using one of those 2" light-up height setters? http://www1.mscdirect.com/CGI/NNSRIT?PMPXNO=1757370&PMT4NO=52060023
your first description a few mins ago make me think otherwise
tool table is written, new offset is loaded
KimK_IA: sure, any number
actually any gcode expression. you could enter atan/ if you want
on the lathe I often enter something like 2.7532/2
... because I measured a diameter of 2.7532
I have to remember that
I'm often walking over to the other computer to calculate measured diameter / 2 + dowel pin diameter
it says "= 1.3766" at the bottom of the screen
then you hit OK and the DRO will say X: 1.3766 DIA: 2.7532
KimK_IA: I'd save my $127 and use a dowel pin, I bet it's just as fast
lift slowly with the jogwheel until the dowel slips under the tool, you're done
the only way the electronic thing is a win is if you can park it in the corner of the table and wire it to EMC
then you load a tool and run a subroutine that goes over there and sets the length automatically
now you can write gcode to measure all the tools in the turret
... and write the tool table
dang, that's cool
how would you do that on the lathe?
not so easily
renishaw makes a thing that mounts to the lathe and swings down where the tools can poke it from several directions
when you said turret I thought you were talking lathe
of course it costs a billion dollars
maybe two billion - I didn't look too close
I actually meant a mill with a tool changer, but same idea
The little tip on the top of the thing has about 0.1" of give after the light comes on.
KimK_IA: I've seem a similar thing with a dial instead of a light
so for a machine like the mazak, if you mounted the widget on the table at a certain spot, you could run "set_tool_len.ngc" and it would cycle thru the entire toolchanger setting all of the lengths
but, I still think the dowel is probably just as fast
jmkasunich: wouldn't that be slick?
the doodad is safer for the operator
dowels have to be rolled around near the tool tip while the servos are active
Might take a trick or two for hollow tools (no center) like shell mills, boring bars, etc, but very doable.
(also hard with big-ass tools like the Maxak could have)
it obviously depends on the class of machine
KimK_IA: yeah some tools are very hard to measure in the machine...
you mentioned Boss, Dynapath/Autocon has that feature too. Also, entering .001/ will increment the current offset value by .001
BOSS is very basic, it has a TLO=Z button. there are no work offsets, so it's easy
current offset value = whatever value you are *specifically* pointing at in the tool table, it doesn't assume anything based on currently active offsets/fixtures, etc
you can go edit the numbers manually too, but I rarely do
The only way Dynapath edits their offsets (manual or auto) is by what tool table entry the operator is pointing at when he says "enter", no assumptions are made
SWPadnos: I see the stat buffer doesn't tell you which offset is loaded, only the interpreter knows
you mean from which "slot" the offset comes?
well it may not have even come from a slot
yeah, it's an interpreter thing
the interp knows that (G43.1 is slot -1 I think) but nothing else knows
so if it should be smarter, we'd have to do some stuff.
I recall some discussions about whether tool comp belongs in the interp or CANON or wherevere
this is a sure case of 'make the general case easy'
you can always edit the tool table and then file/load tool table
its funny how messing with vismach makes me think I know a bit of python
till I actually try to _program_ something
I want to write a loop to put 12 boxes in a collection
actually, it would be "foo = Collection([" and so on
so, with a loop, do I do something like "foo = Collection([ foo, Box(blah))" inside the loop to add each new box?
ah, I think I want to build a list of Boxes, then stuff that in the collection
jmkasunich, cradek, jepler: would you guys consider giving a seminar on how to use vismach at next year's CNC workshop?
not sure if a seminar is the right way, but I'd be happy to do one-on-one
Many of us have seen your cool art and animations and have absolutely no clue how to do that, even just making a simple groove in a block or something
a lot of it depends on the ability to visualise geometry
vismach models are made by adding together simple shapes
mostly boxes and cylinders
to make a block with a groove in it, you use three boxes
one is the portion of the block under the groove
the other two are the portions on each side of the groove
can you subtract? that would only need two boxes.
that is a real weakness of vismach, but its inherent in the openGL methods used to do the rendering
since you can't subtract, you can't make something like a round bottomed groove, or drill a hole in a box, etc
povray can do subtraction (they call it either intersection or difference), but it is much slower - no way to do realtime panning, zooming, etc, or to make parts of the object move in response to HAL pins
there's got to be some GL meshing tool that will let you use CSG
Wow. So that cool 5-axis-ball animation was built "backwards"(?) (ball and chips coming together to form a cube) and then shown "forwards(?)" (cube coming apart to form chips and ball)?
the yellow one?
that was POVRAY
I think so
the yellow one was povray
which can do CSG I think
subtrecting the tool shape from the blank
Oh, so POVRAY can subtract, then?
frame 1: calculate initial tool location, subtract from blank
frame 2: calculate tool location a second later, and subtract both tool shapes (1 and 2) from blank
POVRAY can do a lot, but it can't animate (unless you want a 50x50 pixel image)
frame 3, subtract 3 tool locations from blank
each frame took povray a minute or two to render
then they were stuck together to make the video
there is a couple _days_ of rendering time in that video
povray can subtract, add, rotate, and do all kinds of neat stuff - its a very cool program
some people have WAY too much time on their hands
[03:57:07] <jmkasunich> http://hof.povray.org/office-13.html
there's even smoke rising from the cigarette in the ashtray (front desk)
Ha. Must be a european office?
plants, crumpled paper on the floor
[04:00:20] <SWPadnos> http://hof.povray.org/
Now we need to do a time-lapse with the sun moving the blind shadows
cool, the bonsai uses only POV's language, and the source (zipped) is 34K
no imagemaps or other bulky stuff
wow pov has sure come along since i last looked
Approximately 370 millions of polygons are present in this image. In spite of the radiosity and the area lights, it rendered rather quickly at 6000 x 8000, in a couple of days
POVray is the reason I got a 387 coprocessor for my old 386 computer
yep. PORay and fractint were the coolest things around at the time :)
Reminds me of that classic Dilbert... Dogbert: The company has decided that you only need a 286. Besides, how many times in your life are you going to do a 3-D rendering? Dilbert: Once, if I hurry.
got my loop working, my boring machine table now has 6 t-slots
and on that note, bedtime
good idea. night
jepler: how difficult would it be to make stepconf use non-doublestep, either by checking (or unchecking) a box, or automatically for step lengths greater than some threshold?
jmkasunich: I've held off on doing it because I fear I'll introduce bugs
If I did it, I'd do it automatically
I just noticed an email from someone who got stung by that, and someone else who managed to figure out how to make it work (perhaps without actually knowing why)
probably I do need to do it
if(port->reset_time > period/4) port->reset_time = period/4;
IMO that is a bug
^^^ I think this accounts for the reason that a fairly short maximum was seen for reset-time
did we ever get a clear report of what numbers are wrong in the sherline timing defaults?
cradek: no, I looked but didn't find any real timing numbers
cradek: I don't think so
jepler: that line silently violates the timing the user requested
step space/length 22/22, direction hold/setup even bigger
is what I saw in emc1 configurations
jmkasunich: yeah it does
I expected values like 1us, not like 22us
quick "fix", that line should say "Sorry Dave, I'm afraid I can't do that" and bail
btw, I do think that period/4 is a good choice for the upper limit of busywaiting
it saves me having to figure out the cut-off in stepconf, I can just use that one.
I totally misinterpreted that line you pasted
I thought it was IN stepconf
its in parport, right?
if it is in parport, then it is most emphatically NOT a bug
that's a line from hal_parport
I should have realise that is was C, not python
no big deal
I'll be awake later ;-)
ah, that's no fun :)
I think I need to buy more memory for this PC, and set up a few more VMs
I had my dapper desktop running for the last month or so, so the dapper farm slot would get updated
and I've had the shoptask computer on 24/7 for about the same time
it made a noticable difference in electrical usage
we should have an 8.04 farm slot anyway
jepler: you just changed the default for the inputs to be unused instead of limit... is that in response to the various folks who have been unable to figure that out?
sounds like a good change
I just didn't expect to see that in a commit fixing something else
ybin: Blessing /dev/hda2 with Holy Penguin Pee...
that was my feeling too
the mac bootloader prints that
[17:26:47] <jmkasunich> http://www.jargondb.org/glossary/holy-penguin-pee
note the gray penguin on the right edge
and what he is doing
jmkasunich: the logic seems to work on the h-bridge. So far :)
I hope to hook it to 'real' pwm and play with it today. (pluto) and higher voltage than I have been testing 50v vs 20. Baby steps. ;)
don't know if you saw this http://www.youtube.com/watch?v=eUmFKOVepYY
jmkasunich: yeah that also seems to be a common problem
skunkworks: yeah, I saw the vid
was that halscope on the screen?
I'm not used to it refreshing that quick
roll mode in the base thread, I bet
jepler: you said "that also seems to be a common problem", but I can't tell what you were responding to?
jmkasunich: the home switch thing
an hour at the dogpark is good for forgetting what I was talking about ;-)
jepler: should we do anything about tool table comments, and if so, what?
cradek: beats me
if I add them to CANON_TOOL_TABLE it will be a lot bigger (will probably have to enlarge the nml buffer again)
let's let it suck for now
also I can't quite figure out how you say "the rest of the line" in scanf
I could keep them, but not make them available to emc
wonder if that would be a useful halfway solution
let me see if I can pull that off
they are for the benefit of a human editing the file (IMO)
true, but ideally the gui would show them somehow
I'm trying to remember (cause that machine is powered off at the moment), but I think my shoptask not only has per-tool comments, it has a block at the front
reminding me that my zero-reference point is 4" from the spindle nose gage line
lines that don't parse right are just skipped over
... but they won't be preserved
and preserved when re-writing?
if I start using that feature, I'll probably stop doing the 4" thing anyway
that is mostly for offline tool measurement
oh, I just remembered something I wanted to ask you yesterday about this
it will still work fine if your tools are shorter than your reference tool
for lathes, you've said that you need to use touch-off with care to avoid breaking CSS
you'll get those correct negative numbers
I guess with a lathe, offsetting coordinate systems (in X) doesn't make a whole lot of sense
I don't really know what that would mean
well, on my lathe, the toolpost is in a different place every time I mount it
I can either set offsets to zero, then touch off each tool (setting tool offset, not coord offset)
you really want this new feature then
or I can set a single offset that (hopefully) will correct for the new tool positions for _all_ my tools
correct for the new toolPOST position I mean
I think if you don't care about css, that would work fine
but not for CSS, right?
a couple holes and dowel pins will fix your problem...
assuming that I always want the toolpost in the same place
it can mount in either of two t-slots (Z) and sometimes there are good reasons to switch
if the two positions have the same X, you're still ok though
I don't remember why css is the way it is - maybe you could fix it to work with g5x.
if that happens it will be after I have variable speed on my spindle
which is about 4 projects from nwo
if you have good homing, good toolpost locating is something you really will want anyway
yeah, I need to figure out how to do that
dowel pins and holes in the table aren't swarf friendly, but there is probably a way
EMC: 03cradek 07TRUNK * 10emc2/src/emc/iotask/ioControl.cc: preserve comments that have been manually added to the tool table