tomp is now known as tom3p
did anyone try alex's url ? http://lmgtfy.com/?q=linuxcnc+AXIS+fullscreen&l=
Yeah, I love lmgtfy, but I've never been able to use it
i think archivist pointed to http://engrishfunny.com/
that's spot on
Guest305 is now known as skunkworks_
anyone foresee any problem running emc on 9.04?
jaypro: you need to build it from scratch
that's a bit of work..
(you need to build a RTAI-patched kernel, etc)
damn... so it would be better just to use 8.04 then huh
guess ill have to do that
you can also try to use the 8.04 packages on 9.04.. but that can be problematic
it's worth a shot if you have 9.04 installed, and plan to reinstall using 8.04 ;)
[12:39:46] <skunkworks_> http://www.youtube.com/watch?v=7TowJZQi-qY&feature=channel_page
Hi jepler - I forget you usually go to bed at a decent hour ;)
I didn't last night, but I still woke up at 6:30
now I'm wondering when someone will be at the shop to let me in..
heh - we are trying to get motovated
stuart seems like an early riser also
I should probably get going
skunkworks_: as predicted, stuart was here already
rob__ is now known as robh
SWPLinux: You slummin around?
no, I'm hob-nobbing with the hoity-toities
actually, I'm at the Red Carpet Club in Chicago for a couple of hours
then on to Wichita
Nice, though a tad early to have a drink, unless you've been up all night, then carry on!
SWPLinux: oh you heading to fest?
just drinking coffee and OJ at this point
cool, I thought you weren't gong to make it.
finally got the camera stuff working, so here I am
wha was the fix?
remember to watch MTV this Saturday night
they should have a behind-the-scenes or pre-show show for the Movie Awards (which are on Sunday
Unless it's available via inet, not gonna happen, so if you have a copy let me know
nope. I don't get MTV ;)
or strem it to us =)
I'd be using their connection to do it ;)
that would be funny now, wouldn't it
SWPLinux: Works for me =)
not that I'll have any access to their feeds. we're just doing Red Carpet stuff (no relation)
SWPLinux: I have 18Mbps here and can easily setup an inbound server
somewhere there will be a website where the stuff we're doing gets uploaded, but I don't know the URL
I'm not sure they know the URL yet actually
SWPLinux: ssh -d ftw
so, I finally figured out how to run a command on a bunch of computers (listed in a file) from a script
SWPLinux: how's that
ssh does silly things to stdin, so you have to set up an extra file descriptor to read the list from, or it only executes on the first computer in the list
exec 3< somefile
then read -n 3 the_computer
ssh $the_computer $@
and '3' being ?
the file descriptor
0 is stdin, 1 is stdout, 2 is stderr, so I used 3 :)
[14:27:05] <SWPLinux> http://pastebin.ca/1432012
took a while to figure it out. tried various quoting options, using sh instead of ssh, echo only ...
I think you can probably even have "user@computername" in the computers file
of course it's easiest if you have the "host" in the authorized_keys file, so you don't have to go typing passwords all the time
(thanks to cradek for that tip)
I chose a fine password, which becomes annoying when you have to type it 100 times in a day
SWPLinux: Actually, If you ssah from A to B, Then from B to C. ssh will grab the keys from A for you automagically too
for C that is
C has to have A's key to do passwordless login though, doesn't it?
or is it sufficient for B to have A's key, and for C to have B's key?
Yep, but you dont have to do anything, C will grab them from A for you
ok, that makes sense
which is a good thing, because I have huge keys =)
it was kind of fun having a really fast PC (running Jaunty) next to the really fast Mac (about the same speed)
with the total cost of the PC, including monitor, being less than the monitor alone on the mac :)
actually the PC was faster, but who's counting
on that quad box?
yeah, it was a phenom 955
8G RAM, 1TB disk
8GB? Seems "low" for a dual quad core.
single quad core
still, that's 2GB per core
the motherboard is nice, though it looks like the second gigE controller isn't working
yeah, it's enough
oh that sucks, you going to rma it?
I need to check something in the BIOS first, but I probably will
no access until Friday, so I'm not worrying about it
or maybe Thursday
too bad NewEgg won't ship to a hotel. I could just buy another one
hmmm. maybe I should have one shipped to MPM
SWPLinux: I'm sure you could
are you flying in this afternoon?
do you think there's a PC shell that I could use to test it?
yep. should be there at 2:something
I was answering about shipping something to mpm
are you talking about the shell thing you were discussing?
PC power supply :)
ie, "something in which to test a motherboard" :)
or on which
or next to which ...
I'm sure that could be arranged
ok, I'll probably do that then
oh hmm. I'd need an AM2+/AM3 CPU to test with also
I wouldn't expect to find one of those here
too funny. the whole PC I ordered < 2 weeks ago is now $75 less than I paid for it then.
SWPLinux: How much L2 cache ?
or per core
[14:47:12] <SWPLinux> http://www.newegg.com/Product/Product.aspx?Item=N82E16819103674
oh, that's L3
hmmm. that doesn't look right
in the description at top
yep, it's right (confirmed at AMD site)
2M total L2, 6M L3
return it all, and get the 2MB per core version =)
too funny. the Mac was a quad-core Xeon at 2.66 GHz, 8GB RAM, unsure about the HD
cost $4000 or so
the PC was $1350, including monitor
($4k is without monitor)
and maybe without keyboard/mouse too
Jymmm: there isn't an 8M one. I could go down to 4M though ;)
well, you pay for the name too
it's a Black Edition too
SWPLinux: 1MB L2 is far better than half
SWPLinux: Many times Ive seen the diff between retail cpus and like XEON's are the L2 cache
it's a little different in the Athlons
I have no clue on AMDs
the Xeon shares the L2 between cores, the Phenom doesn't
isn't the sharing a "good thing" (tm)
I think it's two dual-core chips, and each of them shares its L2 between those two cores
it means there can be contention between cores for access
So they have instant access
Right, well if you are running a SMP aware app
of course, there are Xeons with 12M L2, so who knows
no, this has nothing to do with applications
unless they're specifically locking things in cache (Which is rare)
k, it's too early to go into SMP crap
this is CPU-level, not application level
in any case, I can tell you that the PC ran GLXGears about 4x as fast as the mac, and at a higher resolution
incidentally, the Samsung 23" 2048x1152 monitor is very nice. if it had an HDMI port, it would be just about perfect
$219 (on sale at the moment)
[14:57:03] <SWPLinux> http://www.newegg.com/Product/Product.aspx?Item=N82E16824001317
the only pet peeve I have is that the (very slick) controls aren't backlit, so they're hard to spot
day glow paint!!!
since they're touch-sensitive and don't break the lines of the bezel
SWPLinux: 15" touch screen atom PC http://usa.asus.com/products.aspx?l1=24&l2=169
I noticed that
too bad it has no PCI slots or parallel ports ;)
brb. gotta get more coffee
phooey. I guess it is the coffee
darned dark roasts
they just never turned it off yesterday =)
oh wait, yu said 2048x1152... doens't that require Dual DVI ?
dual-link, not dual connections
dual-link can do up to 2560x1600 at least (that's the common 30" reolution)
Yeah, well most video dont have dual link
oh, that could be
you can use analog too, and it doesn't even look too bad
the D945GCLF2 drives it at full res, using onboard video
that's the $89 including CPU+MB thing, that runs EMC very nicely
Ya know, I hate digital
DTV, Digital video.... analog rules!
DVI does look nicer. there are some color artifacts using analog
SWPbut were you looking at it on a digital display or crt?
it looks like the subpixel hints for font rendering are wrong - like the old Apple II machines
on the Samsung
but using the analog input from the D945GCLF2, which has no DVI
Yeah, I have a 24" Sony crt here so...
err - s/input/output/
does it do 2048x1536?
(assuming it's 4:3)
ok. that was about the highest resolution that was ever (commonly) available in CRT monitors
It has dual switchable inpts... d15S and RGBHV
err D-Sub15 & RGBHV (5 BNC)
given that afast but cheap video card (something like a 9800GT) is $50 now, I don't think the "onboard DVI isn't dual-link" problem is really a problem
If that' nvidia, there might be =)
nvidia's ass is in trouble
I'd be surprised if any motherboard from the last couple of years that has DVI doesn't have dual-link (unless you really shoot for the bottom of the barrel)
it is NV, but there are ATI cards in the same price/performance range
did the $94 one come with dvi?
no, it has analog only
SWPLinux: I want a Dell 1900
built in VM
so get one
I said WANT, not NEED
oh. well in that case
9.4 wasn't too happy running in VM
oh, you mean 9.04?
(the zero is significant :) )
dont you speak typo?
ok I'm back
Yeah, I love you too!
so I did a search at NewEgg for 512M+, PCIe2.0 x16, HD4000 or 9500GT (or better) video cards under $60, and there were 24 matches
about half of them were under $30
so dual-link DVI *really* isn't an issue :)
several were low profile capable (came with full size brackets attached)
I don't know if the "standard" low profile is 1U or 2U though
I have SO many machines here that I don't know what to do withright now
I have a T60, MacMini bth are not even in use
3 1U dual core boxes, one is Xeon other two are Opteron 100
the mini only has 4 USB ports, right?
or is it 4+2
yeah + 1 FW
ok, that's why I couldn't use those
I put together 15 small computers based on the D945CLF2, in the mini-box M300LCD case
why 6 USB ?
the customer would have preferred macs, but each computer had to connect to 4 cameras and allow for a keyboard and mouse as well
plus the minis are $800 for a dual-core, and the machines I put together were $300-ish
I still would have installed Linux on them, but they'd look like macs ;)
ever heard of a hub?
um, in what context?
I could have done that, but it's another thing to stick on the truss, plus power, etc.
plus another cable
this guy hates cables
(of course, we have 100 of them on the truss, but that can't be avoided)
it could even be better organised to have a hub for all the cameras
hmmm. more like 150
might result in less cables near the computer
near the 12 computers, you mean?
this is an interesting networking setup too
SWPLinux: what kind of usb cameras?
lots o' data
which is why connecting too many to a single computer isn't ideal - too much image processing work
I keep tellin ya... FSCK USB.... 10GigE + PoE for ALL devices
kybd, mouse, hdd, optical, camera, video, etc
tell Nikon to include 10gigE in their next DSLR and I'm with you
and make them lower power, they're just slightly too high for PoE as it is
yo want 48 monitors, no problem
place em all in the same subnet (IPv6)
I thought about that actually
you want RAID1.... place the hdd in promiscous mode
the long-term design is to have a box that rides where the vertical grip would be, which provides power and control, via PoE
but that's longer-term now, since we're doing this MTV thing
you said D700 right?
don't suggest the WT-4a
or WT-5 or whatever they're up to
vertical grip for 440
that's what we'd replace. the vertical grip itself does us no good
$40 then, not 440 ?
is that from mkgrip.com?
yeah $40 - this fscking kybd
looks like only D200
no, HK Dollar is a company or website or something
[15:54:17] <Jymmm> http://www.dealextreme.com/search.dx/search.nikon%20D700
ok, I noticed 2 grips at $75-80 each
(that fit the D300/D700)
Ah, some are $40, guess not the hi end ones
yeah. if it has a thing that sticks up (into the battery compartment), it's the wrong one
Hello, I am trying out Vismach, but I don't have a clue how to run it. I downloaded vismach.py and pumagui.py but running each doesn't indicate any more information.
kirk_wallace: several of the sample configs have vismach models. try running '5axis'.
you should not have to download anything that was not already in your emc2 distribution or source checkout
there's a "puma" sample configuration that uses pumagui.py intoo
this is in 2.3.1
So Vismach is invoked by running EMC2 and a config file?
(in fact I can't get the 'puma' simulation to do anything useful)
kirk_wallace: yes, if you look in the associated hal files, you'll find a loadusr line for pumagui and some nets to connect joint commands to it
loadusr -W pumagui
net J0scaled scale.0.out pumagui.joint1
I am not familiar with the specifics of the pumagui, but basically you start the visualization with 'loadusr', and then link it up with 'nte'
Somehow I go the impression that it was a standalone application. I'll look at the sample configs and give it a try, thanks.
kirk_wallace: the vismach things are standalone applications, kind of like halui :)
ie, they are separate userspace processes that need HAL present to run at all, and need some HAL connections to be useful
* archivist cant wait for vismach models to have material removal by the gcode
Okay, now I have something to go on. Thanks.
well, I suppose I should make my way to the departure gate. see you later
So are there going to be any videos made at the fest this year ?
doesn't seem that way
too bad. Hope you are all having fun out there.
i've been hacking on hm2 stepgen this whole time, haven't looked at SPI yet :-(
think ill have to book a flight for emcfest 10
me too :)
20.000 km :|
micges, i noticed a couple things wrong with hm2 stepgen, and all of them are fixed in trunk now
i'm going to commit it all to the 2.3 branch later today
if you still see any problems, let me know
probably I noticed some of them
but main problem is that I (and you) can't reproduce problem I saw
garage_seb: I'll test throughly on monday
I saw two problems you fixed on stepgens was overshooting
garage_seb: and what means for stepgen maxvel = maxaccel = 0?
if maxvel=0, then the stepgen velocity is limited only by the step timing parameters
if maxaccel=0, then the stepgen does not limit its acceleration
that means the stepgen.position-cmd must be acceleration-limited
all emc2 motions are acceleration-limited
so stepgen.maxaccel=0 is a way to not have to set it to "axis maxaccel + 5%"
garage_seb: does what your working on have anything to do with this? http://www.linuxcnc.org/component/option,com_kunena/Itemid,20/func,view/catid,10/id,286/lang,en/#335
garage_seb: this is additional info about dir swapping bug:
1. after swapping dir normal motion can't be done, stepper motor "slips"
2. during jog or program run motors works normal but machine is loosing mm
3. swapping dir occurs on both axes
BigJohnT, is that guy using steppers & encoders? if so he should probably use stepgen velocity-mode control (which i'm about to commit to trunk)
micges, what are the stepper amplifier inputs like?
garage_seb: I think so
the mesa anyio board output pins can be in open-drain mode or push/pull mode
in push/pull mode, they swing from 0 to 3.3V
3.3V is not enough to reliably trigger some amp inputs
I gotta run talk to you later Seb
garage_seb: I'll check it
garage_seb: I noted all things to check, will se results
garage_seb: Too bad you haven't had a chance to work on SPI yet, but good to work on the stepgen code.
Lerman______ is now known as Lerman
OT: anyone know anything about the Aloris type lathe quick change toolposts?
there are wedge type and piston type, is there some reason why I should spend twice as much to get the wedge type?
I wouldn't pay more for a wedgie ;)
I have heard nothing but compaints about the piston type. I have the wedge type and it works great.
If you can swing it get a KDK setup, even better than the Aloris.
FWIW, I've been trying to get the pluto-p working with an embedded PC104 computer that has an integrated parport that is accessed via an 8 inch ribbon type breakout cable with a db-25 connector that fits directly on the pluto-p, but I kept getting communication errors. By placing 300 ohm 0603 surface mount termination resistors in series with all non-ground signals, where the ribbon cable solders to the db-25 connector (ie, as close to the
pluto-p as possible), I was able to eliminate all communication and data errors. Anyway, FYI.
skyfox00: thanks, that's what I was afraid of
hello all, hope you can help with my newbie question.............. how can i setup multiple parallel pins for same emc drive output in stepconf wizard?
you have two motors on the same axis on a gantry or something like that?
no, i want to send balanced signal (signal and inverted signal)
guessing you can't do that with the wizard
it seems stepconf won't work if the same signal is used on multiple pins
seems like fixing it up by hand after the fact should be fairly straightforward
I just want to do quick test of a servo drive
mookie: you can simply connect it afterwards (once you finished with stepconf)
or using the custom.hal file
net name pin-whatever-inverted parport.0.pin-xx-out
i'm newbie, is there any really quick info on editing file?
or you could just ground the inverted input on your drive, that almost always works
its a ssd 635, anyone know if someone has got one working with emc?
I will try the grounding
[21:08:07] <mookie> http://www.ssddrives.com/uk/servo_ac/Brushless_635.htm
I doubt anyone has tried to get that to work yet
servo drives fairly rare around here as it is
have to register to get the manual, I take it step/dir is an option?
I could swear that stepconf allows you to select the same signal on multiple pins
and you can then invert one of them to get the complementary pair
or maybe not
my mother always taught me to swear, because people know what you mean anyway
hmmm. actually, the error I got was due to some var file shenanigans (removing/moving where the var file is)
ok, yes. it does work
you can select the same function for multiple pins, and stepconf will connect all of them
SWPLinux: 2.3 ? 2.2?
grounding sort of seems to work, but I am not totally sure, I think the servo drive needs config. The stepconf test is definitly doing something to the drive, but it has so many options I think there maybe problems on the "easyrider" (the 635 windows config software) config.
alex_joni: 2.3 seems to work, but in my TRUNK checkout I get an error when stepconf tries to copy the .var file
this is RIP
because the .var file is gone
[22:05:02] <mookie> http://dcnz.com/resources/files/Servo_Drives/635.pdf
I am out of the workshop now, so can't tell all the details.. will try again tomorrow.. but it definitly seemed not to work when I tried multiple pins on same signal. The emc2 version was probably the last live-CD release as I downloaded it around end of feb or begining of march. I will post details tommorow.
that sounds like 2.2.x
mookie: if you installed from that LiveCD you should probably upgrade to 2.3.x before you try to set up the machine
I will download the latest version tonight and try tommorow.
you don't need to download the whole ISO for that
just upgrade emc2
[22:10:58] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingTo2.3
I cannot install to harddisk, I will see if I can upgrade to 2.3, but think I may have to go for the whole new live-CD.
if you didn't install to harddisk, then downloading a new LiveCD is probably easiest
I'll go for the live-CD, should be done by tomorrow morning.. It's past 11pm here, so i'm off for the night......... many thanks for all the help.
think im going to weigh in that big horizonal mill
clear the space and get a big ass ABB arm
mookie: it's past 1am here ;)
so I guess I'm off too
i tried going to bed
got bored and came back to the computer
tomp is now known as tom3p