[00:26:12] <Jymmm> http://cgi.ebay.com/2U-Server-B11s-2x-Xeon-3-06Ghz-1TB-RAID-3WARE-8506-4LP_W0QQitemZ370229344682QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item56336379aa&_trksid=p4634.c0.m14.l1262&_trkparms=|293%3A1|294%3A30
SWPadnos: local to me
I have NFC on xeon's though
AFAIK they're just the P(3 or 4) multi-chip version
ie, so you can use more than one CPU on a motherboard
they also were the first to get 64-bit extensions and virtualization, I think
small power supply
to small for 4 hdd's?
no, it's 350W. that's just fairly small these days
I'm out of storage here. the 1U's I have are loud and can only hold 2 hdd's each. I'm thinking 4 x 2TB in RAID5 config
Maybe FreeNAS or jsut debian
that's a reasonable price for the rackmount case, and the computer inside is just a bonus
but it will probably be loud and slow
plus the 3ware card
that chipset was introduced in 2002
that RAID card is a PCI-X card, not PCI-E
you won't be able to put it in a newer machine later
I just need storage snif sniff
These ext Fw drives have no cooling and get hot
the 4U case I have is just too heavy, evenwhen empty
yeah. there are some I was looking at that are 65 pounds empty
trade ya mine for a 8 port 3ware sata card =)
wow. $229 for a 2TB drive
much more per GB than 1TB or even 1.5TB
just wait till 3T come out
huh. 1.5TB have surpassed 1TB as the best price point now
oh. maybe not. they're about even
I think I'lll go down tomorrow and look at them at least.
Hell, i can't even buy a standard PS for the 1U's I have for less tham $35
How do I invert the analog signal to the amp. It drives the wrong direction for the command and gives a large jump and a following error.
make the OUTPUT_SCALE negative
(or positive if it's already negative)
OK thanks, I'll try it.
jymmm: wasnt cable, was 2.2.8
anyone heard of this "nanocrystalline" motor? supposedly has zero iron losses: http://nextbigfuture.com/2009/07/china-has-100-million-electric-bikes.html
the nanocrystalline part is marketing bullshit.
sounds like they are using ferrite rather than laminated baked beans tins to me
ebikes exist, even when the thingies from china ride... theyre mostly crap (the cheap complete bikes) .. but there are some chinese and taiwanese vendors who build proper motors too. just more expensive
the most expensive part is the battery nontheless.
thundersky (and similar) have some good batteries for that kind of thing
spendy but light
Although with the prices of A123 cells now you could make a kick ass drag bike without blowing the budget
the 'magic' on ebikes bases on 2 things: 'rare earth magnets' also called neodym permanent magnets which are cheaper and more avail now than 20 years ago, and igbt mosfet technology. the rest iss just brushless bldc motors and batteries like we use sinve 20 years for modelcrafts
Valen1 i still ride around with 10kg of SLA ... nothing beats the price
10kg worth of SLA isn't going to get you that far?
btw.. ebikes was the reason for buying our cnc mill :)
we wanted to make torque arms
you can get A123s for ~$5 a cell now
Valen1 about 30km distance per charge.
heh we are CnCing our mill to make wheel motors for bikes and diff motors for cars
i need abour 400Wh at 48V for 30km distance. costs less than 100E per pack with good high-rate sla cells.
on A123 cells i would need atleast 15s4p packs
which makes atleast 300$ just for cells.. + BMS + specialised charger
in reality one drives these packs from 0 to 40 deg celsius ambient temperature... which lithium based batteries really do not like... even less than sla (which loose capacity, but dont die aslong as they dont freeze through)
means one also needs insulation and some package... the energy storage is still not worked out to satisfaction.. same with electric cars. the expensive and complicated part to get right is the energy storage system
tesla motors uses water cooling right?
tesla is more like a field-test than a product... but still a cool one
we dont cool packs atm. we just dimensionise them properly so they dont get hot on charge or discharge
also makes cell live more cycles
i get about 500-600 from one sla pack which i wasnt friendly to.. and driven the 2 packs every day for over a year now.
means since i want 2 packs (to swap at home) its about 200E/year investment.
on lithium-anything i havent seen similar chargecycles in real world
(since this is all higly OT, just holler and i'll shut up about it)
* fenn is asleep anyway
Valen1: which kind of motors do you wanna build? model stuff or bigger ones?
ours are too big for the milling table X-]
i think the case gets moulded usually.. just no idea how to mold aluminium
hi, can anybody help me with a problem i'm experiencing - i'm not sure that EMC2 is communicating with my machine (Fireballv90), and i suspect it may be the parallel pci card
it's a Sweex 9865, could anybody tell me if EMC2 supports it?
not a number Ive heard of
apparently the PU005 card is supported, but this one i've got is a later version: PU005V2
test the port pins with a meter
or better still with a scope
ok, i'll try that out, thanks archivist
andyzammy: how did you try setting it up in emc2?
alex, i have no clue how to set up parallel ports in EMC2, i was under the assumption it was plug and play, as i couldn't find an (obvious) menu in AXIS for it. i recieved the config files for my machine on the website i bought it from.
gah just read what passes for the manual!
andyzammy, what motor drivers do you have
you don't have to do much
but.. you need to figure out the address port for the pci card
or else it won't work
emc2 (and the config you got) probably uses 0x378 which is the default address for the onboard port
andyzammy: this page describes what to do for PCI cards: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?NetMos
dang string broke again
I thought you were getting some new string
a wireless one
people with wifi seem to suffer more drop outs
I'm putting up a tower at the top of the hill so the my receiver can "see" the broadcast antenna
this is line of site wireless to another tower
line of sight
no cable service out here and the mail rider still uses a horse to deliver the mail
archivist: does the EMC map show the flags for you?
not at the moment
login seems borked, like where is the login
archivist, i bought the Probostep Motor/Driver kit
I'm automagiclly logged in and I don't see the log in either
alex, thanks for the link, i'll try it out when i get some time! i'll stick around idle till then
andyzammy, as alex_joni said set the port address, there is a simple prog to show probed addresses (cant remember name)
that sound right
BigJohnT, we both have to nag alex_joni then :)
BigJohnT_ is now known as BigJohnT
ok, i have read the doc the link took me to and would just like to confirm what i have to do before i make any changes
there is a line in the .hal file: loadrt hal_parport cfg=0x378 and i have to append the address of my pci card to it
if you don't have an onboard port, then you need to change 0x378 to your address
the example uses a dual port card, and my card only has one - there are 2 "I/O ports at xxxx" lines (rather than the 6 lines in the example), is it safe for me to hit and miss which ones to use?
lspci to find
i do have an onboard parallel port, but lspci only shows the card i installed
yes i use that to get the address, i think i'm ok with that part then. i'm still unsure about the next part (about the "read" and "write" functions). do i have to alter those lines also?
only if your adding another parallel port I think
depends what the functions are reading
if your lines are reading something like "addf parport.0.read .."
then you need to add the same lines with parport.1.read
ok, i tried it out and it didn't work, here's what i tried:
my lspci gave me 2 addresses: "bc00" and "c000". my lines were "addf parport.0.read base-thread" and "addf parport.0.write base-thread"
try the other address
i changed the lines to match those on the instructions, and appended the addresses to the line loadrt hal_parport cfg=0x378 (so it became "0x378 bc00" and "0x378 c000")
no combination worked
are you going to use both parallel ports?
then just load the one your going to use
ah, if the "0x" matters then i didn't do it right (mind EMC2 didn't crash or complain about them) - i assumed it was ok not including it because the whole number was in inverted commas so emc2 understood it was one big one - i'll try with "0x" though.
still didn't work
andyzammy: what does "didn't work" mean?
did you get an error? did it start but no outputs on the port?
there were no errors, the scope is being used elsewhere at the moment so right now the only way i can see what's happening is to plug the port into the controller box. the entire machines are tested before they're sent out (box included) so there's not much chance that there's a fault at that end. though ofc not impossible
i did have a do with the scope before i tried these things though, i did get +5v out of the data lines but no fluctuation at all, just a streight 5v.
make sure you have the proper lines for parport.1.read and parport.1.write in your hal file
if anyone would think it might help to see the /hal file (or any of the config files for that matter) they can be found at this site: http://www.probotix.com/cnc_downloads/
click V90 EMC2 Configs
alex what do you mean by proper?
there were only parport.0... so i changed them to 1. also unsure of what to add on the end as there weren't any numbers on the end.
so you don't want to use the onboard parport at all?
no, i have no IDE connector handy so i just used that parallel pci card.
how are IDE and parallel port cards related?
the shuttle i'm using has an onboard parport, but it doesnt have a socket on the back - it has some pins on the motherboard that i must connect to if i wish to use it. so i chose instead to install a pci card.
(they're usually called IDC connectors :) )
oh, sorry, i'm forever getting my abbeviationss mixed up.
too many TLA's
er.. (after a google).. yep, damn right!! :P
* BigJohnT hums Hi Ho it's off to work I go
anyone in here using a touchscreen in ubuntu? =)
I have one but not got it working yet
I'm going too :)
i borrowed one from a friend and having som issues :P
when i pull my finger left to right the cursor goes up > down P
and I have no idea how to fix =) im green with linux
google all you can
what brand touch screen?
it seems the touch is turned CW kind of
archivist - that conversation we had about G00 the other day, I checked the actions on heidenhain and haas controllers and neither controller blend the moves with G61 or G64
the other interesting behaviour is the G00 rapid traverse, on the heidenhain when u G00 to say X100 Y50 it will go flat out on both axis so Y gets there and stops in 1/2 the time of X
yes some old controls do that
G0 isn't guaranteed to be in a straight line on those (on EMC it is a straight line)
ahh ok thanks for clearing it up! btw how about the G00 when in G61 and G64 ?
i think its when in constant velocity mode it can be a problem
as it could hit the job if u say G00 the Z up then G00 the X along
I agree it can cause a problem in some cases - in those cases you can program G61
I think usually it is fine and it saves time
[14:38:01] <pjm> http://pjm.dyndns.org/cam/g00.gif
was the path plot i made
is it possible that in a future release it could be settable in the ini as to G00 action, exact stop or constant velocity?
I don't have plans to do that, but of course it's possible to do
the fix is just to G61 each G00 ?
or make your rapid plane a bit higher, or use G64 Pxxx to give a path following tolerance
ah yes ok that makes sense, thanks! I'll do some more testing to get it in my head
sure, you're welcome
how to save xorg when edited btw
say what =)
trying to use gedit but that doesnt work well
too bad about gedit sucking
I wish the default gnome editor didn't suck
what to use then
the two editors I regularly use are not 'modern gui' editors so I don't know what to recommend
maybe I only think they're good because I've been using them for 15 years
frallzor, I assume your problem is permissions, I would be in the dir in a terminal sudo vi filename
EDITOR=gedit sudoedit filename
how to find the "folder" via terminal
Was someone looking for a fiber channel card? SWPadnos?
frallzor: I have the same problem with my touch screen
it should be easy to fix
now food, bbl, pm if you fix :P
I've kinda been liking SciTe actually
thar we go
back to hell
how to tell what your device is called btw?
so you put the right identifier in xorg
tom1 is now known as tom3p
LawrenceG: How is the weather up there. Cold here
very nice at the moment.... sunny and beach weather but not too hot
we could use some rain... fire hazard is pretty high... we had a few days of sprinkles last week but that has all evaporated
beach weather?! What 54F ?
hey Jymmm no brass balls here... cast iron all the way!
oh, so easier to crack with a single blow
[17:45:06] <Jymmm> http://supermicro.com/products/chassis/3U/832/SC832T-550.cfm
photo shows 8 bays, desc says 6
Anyone know what "SAF-TE" is? SAS / SATA Backplane (without SAF-TE)
cable quadrupod: http://www.youtube.com/watch?v=rWgPH1vMFTk
ahh a few hours later i still dont got touch working in ubuntu, great :P
how to tell which version of X you got
dpkg -l xorg*
xserver-xorg, I think
should I combine what you both said or something? =) non did any good
thar we go
try this: dpkg -l xorg
can that be right
thats pretty old?
the current stable debian has the same
im trying to get evtouch functioning and the latest is for X 0.8.7
what am i missing? =)
hmm nm =)
how do I stop X from starting then
or kill it
you have to mess with the init scripts to prevent it from starting at bootup
seems I have to do that to calibrate
to stop it, you can go to a text console (ctrl-alt-F1), log in, stop gdm (sudo /etc/init.d/gdm stop), then switch back to the X console and exit from X (ctrl-alt-backspace)
to exit, you can also kill X, but it's just as easy to switch back (oh, it's ctrl-alt-F7 for the X terminal) and exit
it shouldnt be this hard to get a simple touchscreen working
you may be able to do all thsi without switching consoles actually
just open a terminal and do the sudo /etc/init.d/gdm stop
if X doesn't go away, then ctrl-alt-backspace
fenn: the hangars wont come off the bar, the hooks are opposed. cable bot is cool, but 1st thing i think of for a party aint hangars ;)
I think you have to use the left hand ctrl and alt keys also
tom3p: i think that was on purpose
how to start a commandline or what it said
like terminal without X
fenn: the prototype in lego is neat too. in the 80's i had an isreli plotter that ran loose on a tabeltop, it visually found the edges of the paper, 'zeroed' itself & plotted
yeah i think the lego thingy does that too, not sure
frallzor, you should get a login prompt from a text console
if there isn't one, press enter
if nothing appears, press ctrl-alt-f1
that should bring you to console 1
Now you can start the calibration program "calibrate.sh" from a text
how to do that?
i know its path
um. type the path the /path/to/calibrate.sh
like you would run any other program
hmm I forgot 1 foldername
can i search?
you may need to make it executable first, by executing "chmod +x /path/to/calibrate.sh"
use the tab key
it completes names for you
command not found
i can just type the name when im the folder right?
you need ./calibrate.sh
well now it "worked" but it didnt find a file needed =)
have to fix that first I guess
how to start X again?
like /etc/x11/xfree86.conf or something?
startx usually does it
this time too
you'll be able to exit by using logout or ctrl-alt-backspace this time
no need to mess around with gdm, since you've already stopped it
the file it wanted is where it is supposed to be
baaah I hate this
any more tips or shall I give up? =)
search the web for people who have had similar issues, and try their solutions
been doing that for 6 hrs now =)
does the error message hint that its looking some other place
why waste that investment. try it for another 6 hours :)
the file is in the same folder
is it supposed to be in the same folder?
i got it via apt-get so I suppose it cant be wrong
it does say ./ev_calibrate when I do the run
is this thread applicable? http://ubuntuforums.org/archive/index.php/t-644258.html
following that as we speak
read the entire thread
then It should work
then do things
there are some rules that have to be set up for udev, then apparently a reboot (or some other udev restart)
then add an option to the xorg.conf file
then run the calibration script
also note this: "Instructions for using the calibrate prog are in /usr/share/doc/xserver-xorg-input-evtouch/README.calibrate. It's very finicky and unintuitive. Also check out /usr/share/doc/xserver-xorg-input-evtouch/README.Debian"
also looking at that file
can i do the rules add from X?
the way I read that last comment, you need to reboot afterwards anyway
so you might as well use nano (a text mode simple editor)
woow im getting confused =)
i tried doing it from X
but it didnt want to
permissions once more
how to copy with permission then? =)
skunkworks, all 2.0
but of course i get an error I cant see now when I run calibrate
pipe stderr to a file eg ./progname 2> err.txt
or i might see enoughm but I dont understand it
dont see why ppl want linux when it takes +6hrs and countint to get touch working =(
I didn't check them all at once, but every one was 2.0 when I tried them individually
frallzor: is your screen supported by linux?
the annnoying thing is that the motherboards seem to just die for a while, from time to time
no idea, but it works but the axis are turned CCW
frallzor: did you see SwapX This option is in the driver since V0.61. When set to 1 the X-axis is swapped. This option is applied BEFORE the rotation-option. false -- no swapping
SwapY This option is in the driver since V0.61. When set to 1 the Y-axis is swapped. This option is applied BEFORE the rotation-option.
now the calibration is allmost running but a file cant be found =)
[19:38:30] <anonimasu> http://www.conan.de/touchscreen/evtouch.html
I got that link and a millions others but no go
followed it 1:1
frallzor, could you put the errors you see onto pastebin.ca?
atm it seems the author made some change in the path
so just trying to find the correct one
hi all, anybody here that can answer a pyVCP question?
in the wiki adding custum midi buttons there is an example to add a rapid to home button
when I modify the the files, emc2 won't load
doesn't like the post gui file command
you probably have a typo somewhere
run emc from a terminal and look at the error(s) you get
I did a cut and paste
(please cut and paste the error, don't transcribe :) )
halhalui.mdi-command-00 doesn not exist
using a java plugin won't let me paste
I get a paste option in the right-click menu (on the edit line at least)
[15:47:10]<SWPadnos>I get a paste option in the right-click menu (on the edit line at least)
(I'm SWPadnos, if you didn't guess)
I'm using the (iassume SUN java) plugin
it will paste into gedit
just go to http://pastebin.ca
and paste in the text of the terminal session
and give us the link here
how do I copy from the emc2 error window
run from a terminal
select the text with the mouse, then middle-click in the edit window on pastebin
I'm pretty new to linux, what command to i need to run from terminal
ok it's a patebin, /m196b86e0
can you also paste the contents of your custom_postgui.hal file, your ini file, and any other hal files in your configuration?
it seems likely that either halui isn't loaded, or somehow the mdi command pins aren't being created
[19:59:37] <Guest745> http://pastebin.com/m5375306a
not being able to paste sucks
you could always try a different IRC client
like chatzilla, an add-on to firefox
there's also piudgin (I think) installed by default
some people like xchat
or you could ask skunkworks how he manages to paste using the web client
(I'm on a Windows machine, so the JVM is guaranteed to be different than yours)
you can only do 1 line at a time. pastebin is much better
i loaded chatzilla but will have to close firefox to restart, be back in a second
* frallzor is giving up
hmmm. trouble with chatzilla?
just trying to get a grip on it
you need to connect (they suggested using chat.freenode.net some time ago), then join #emc
I use a plain irc client xchat
Guest301: did you have line in your inifile:
[HAL] HALUI=halui ?
the commands are connect and join, and you always use a / at the beginning of the command (which must come at the beginning of the line
I was going to ask about that
add HALUI=halui to the [HAL] section of the ini file, and that should fix the problem
note that your change will be deleted when you re-run stepcon, and I'm not sure there's a way to get stepconf to put that in for you
that fixed it
Guest301: options to add halui is on 'Advanced configrations options' in Stepconf
micges, is that true in non-TRUNK versions?
or non-MASTER in git parlance
that fixed it and it works, thanks
I've run it from menu
I wanted to play with pyVCP and just started with the wiki info figuring it was the safest way
hello at all
SWPadnos: it is in 2.3.2
yes I just updated my machine to 2.3.2
btw, atom_ is me, depends on which window I'm in
was there a better place to start with pyVCP?
[20:16:02] <micges> http://www.linuxcnc.org/docview/html//hal_pyvcp.html
the wiki has examples of things people have done, but the individual pages aren't meant to be comprehensive
ok, thank you for your help, off to learn more
I added a note about HALUI=halui on the page you started from, but there's still a lot of information missing from there (which makes sense)
I'm back, sorry I'm trying to understand the connection between a hal signal like parport.0.pin-10-in and the widget. For example the LED widget described here: http://linuxcnc.org/docs/2.2/html/hal_pyvcp.html
defines the halpin as my-led in the xml, but then below says the default is led.n
Aloha everyone. With tuning, I have an axis that runs well except for when it gets close to the commanded position it takes a few seconds to remove the residual .002". What term in the tuning will help this?
sed_: does it overshoot or undershoot?
I think undershoots. What pins in Hal scope do I need to look at?
seeing the plot would help
commanded vs feedback position
if it undershoots you might just have too much D
or FF1 is wrong (if it has the .002 error throughout the move)
I'll setup hal scope and try it.
atom_ is now known as nebee2emc
the signals are so close together that I cannot tell the error at the end of the move.
[21:06:48] <Jymmm> http://pastebin.com/m7fa552af
skunkworks: Google Voice VM translation.
they even wrapped the AC +number properly
How does the FERROR and MIN_FERROR effect tuning or does it at all?
I think it did a pretty good job.
ferror and min_ferror is how far command and actual postion can deviate before the machine e-stops.
What keeps the pid going until the axis reaches the commanded position?
the requested versus actual position
I am giving it a G1X0F10. the axis moves to within .001" or .0015", then paused for a few moments then slowly moves to the position.
Also where do I change the gui to display for decimal places for the positions?
Sorry, Also where do I change the gui to display more decimal places for the positions?
how large is your deadband?
I have it set to 1 count of the encoder which is .00005"
I'm not a expert, but it sounds like your pid isnt properly tuned
so instead of being critcally dampened it's overdampened..
Overdamped is too much I?
no, too much D
D is often called damping
(convenient, no? :) )
I have very little D = 8
"very little" is relative
I = 1200, P = 2500
anf FF0, FF1, FF2?
FF0 = 0, FF1 = 0.6, FF2 = 0
what kind of morot drive do you have again?
Baldor AC Servo motors and drives
velocity mode, torque mode, voltage mode, or current-mode?
hmmm. OK. I don't know the specifics of tuning current-mode drives
I'll switch it to velocity and see what I can do.
velocity mode should be a lot easier to tune, I'd think
but the numbers will be totally different, of course
Yes, I'll start over with the tune.
you should set the DAC scaling so that sending a "1.0" to the DAC causes the motor to go 1 unit per second
if you look at the mailing list archives in the last week or two, there have been a few comments on good tuning procedures
OK, I'll check them.
Does EMC have the ability to talk to servo amps with serial at 9600 baud.
it shouldn't be too hard to write a userspace program to do that, if you're thinking about updating settings and that kind of thing
of course, you'd have to know the serial protocol, which isn't always readily available
I have that information
is it modbus or some other standard protocol?
These amps can run in a following mode where the encoders are connected to the amps and the amps output an encoder to the control. It also has RS-485 serial.
the best way to run the drive is in the dumbest mode it has, but either velocity or torque mode (not current or voltage)
RS-232/485 is not usable with EMC2 for coordinated motionpositioning
Seting the OUT_PUT SCALE to will be a challenge.
not in velocity mode - that's the easiest mode to do it in
if the axis goes 2 IPS with a 10V command, then you need 1.0 on the DAC HAL pin to result in 5V output
so the scale will be either 5.0 or 0.2 (5 or 1/5), depending on how that driver is written
I calculated .15
(some treat scale as "the highest number I should see", others as "what I should multiply the inout by" - I don't know which STG does)
mm or inch?
I use inch
so you have a 1.5 IPS max speed?
(on that axis)
I calculated by the max rpm * screw pitch / 60
I guess I should multiply that by ten to get the max.
if you have feedback working correctly, you can stick a battery (like an AA) on the amp command input and look at the velocity in HAL
OK I'll try it. I guess I'll have to bypass the enable so it wont fault out.
and of course EMC shouldn't be "on", it should be in "estop reset" or "Machine off" state
What HAL pin reads the velocity?
there should be a velocity output from the STG encoder
I have to have Lunch. I'll be back later. Thanks
possible to update xorg in ubuntu installed from the live-cd?
skunkworks, you still around?
skunkworks, opps... walked away for a Friday afternoon cocktail with the wife
skunkworks, I got a job using a 33fj128mc202 pic.. it looks like it will make a dual servo controller
skunkworks, that is one wild chip.... I/O pins for the various peripherals can be programmed to any of 16 pins