for the folks interested in the 5i20 buy... O
I'll volunteer to make the buy and reship
I just found out I have 2 whole rolls of 50 conductor ribbon cable (the kind that has 18" of twisted pairs and then 2 inches of straight)
I'll toss in whoever much each person wants for their cables (within reason)
steves_logging is now known as steve_stallings
thanks JMK, I have cable etc., let me know how much for shipping, how to pay, can you take PayPal?
checks are fine
why am I now surprised 8-)
man, 50 pin IDC connectors are expensive
$3.47 for the ones that digikey doesn't have in stock, and $9 for the ones they do have
the old ampenolg connectors sound cheaper
another reason to love the Mesa interconnect system
look in computer junk boxes for old single ended SCSI ribbon cables
heh, thanks for the reminder
my box of cubix crap (the now retired compile farm box) has at least 10 of them
I'm sure not gonna use them for the intended purpose
now for the 50 pin ribbon to screw terminal board that you dig out of the office dumpster 8-)
I have at least one of those ;-)
and somewhere I have at least one of those Opto22 racks...
I wonder how much ribbon cable is on these rolls
inside diameter 3.5", outside diameter 10.5"
about 100 feet I think
jmkasunich: I'm in for one card .. if 50-pin is an old SCSI standard, I bet I can scavenge some of the ribbon cable at the office
I can find an infinite supply of old scsi cables
imagine now easy it would be if cables were 40 pin like IDE... 8-)
should be easy to make a breakout board if you can find suitably cheap screw terminals...
it's 24 or 25 signals per connector, with lots of GND?
if so, it's an easy singlesided board with ground plane in between everything
someone earlier said there's a GND per signal
24 signals, 24 grounds, plus power feed, I think
the power feed is either 3.3 or 5, jumper selectable
400mA total (for all three cables)
I have one 50-pin header to screws and 2 50-pin D-shell to screws breakout boards
no reasonable 50 position screw terminals on mouser
that is actually not enough if you are running all outputs to Opto22 boards
72 opto-22 modules at 10mA each or something like that?
that assumes all of the modules are outputs
if you have a mix its not so bad
plus 5i20 allows external power to Opto22 cards, plenty of grounds so no problem
I wish digikey would let you search on price
[01:05:49] <steve_stallings> http://www.winfordeng.com/products/brk2x25.php
sometimes available on eBay for less
$0.80 per pin
you like that, try National Instruments
the cheapest screw terminal blocks I've found so far are $0.32 per pin for 10 pinners ($0.29 per pin if you buy 10 pcs = 100 pins)
given that there are only 24 actual signals on each cable, the breakout board doesn't really need 50 screws
only if it is general purpose and you don't know which wires are ground
but if I was gonna build a breakout board (what cradek mentioned) I'd put some signal conditioning on it anyway, whether pullups, or optos, or whatever... so its not general purpose anymore
I'm sure I could not make and sell a general purpose version for less than $40
do like everyone else does, sell the PCB and let them find their own components
pullups could just be resistor packs that you solder in, or not
I prefer to address the mid and upper range of function/price in products I produce
even at $0.29 a pin, you have $15 just in terminal blocks
jmkasunich what gauge wire?
another $5 for the ribbon cable connector
Jymmm: for what? the cheapest screw terminal blocks?
jmkasunich I can chek the local surplus
Jymmm: this is just speculation - most of us either already have what we need, or will scrounge it
$5 for PCB, $25 in parts, fails 3X rule for parts cost to retail price
by a lot
jmkasunich but if the surplus has 10,000 of them and are willing to sell for $25
you might not need 3X for a product like that that doesn't need any testing and has no semiconductors or other sensitive parts, but this isn't even 2X
that is great when surplus offers what you need, and I acknowledge this by not trying to compete with surplus
Hi All...this is wb9mjn...
no its not, its ubuntu
this is ubuntu 5 Live EMC2 CD...
The ubuntu 6 is definately messed up...will not boot...
Question - How do I get this to install on the hardrive ?
the silence is deafening.......
Yea....somebody who knows something will come along eventually...
I get a pass on this since I am a windows weenie
I like the look and feel of this ubuntu....
what won't boot?
I pass because I installed both ubuntu "5" (5.10, aka breezy) and ubuntu "6" (6.06, dapper) from home-burnt CDs with no troubles
Has been tamed down of its Linux non-ergonomic-ness...
so I have no idea what is causing your problems
Well, that is what this is....5.?? from a home burnt CD...
most often it's an incorrectly burnt CD, or just a CD error
So, how do I move on and get this installed on the Hard Drive ?
ubuntu: as far as I know there was no EMC2 live CD using ubuntu 5.10 (or 5.anything)
I think its 5.05 ...
jmkasunich: yes there was a breezy live cd
and the "normal" 5.10 CDs are not live CDs either
shows how much I know
It came right up....the 6.06 multiple burnings, and trials, and it always would
stop on the ISOLINUX thing...
ubuntu: did you try installing a STOCK ubuntu 6.06?
or an EMC live one?
ubuntu: did you check the md5sum of your iso before you burnt it?
I tried the ubuntu 6.06 Live EMC2 CD download from linuxcnc.org, but NA and EU sites..
I think it has a bug...
LOTS of people have installed it
did you check the md5sum against the sum posted on the website where you downloaded it?
I used it ok (6.06 Live EMC2)
This is my only Linux machine...so could do it now, that I have 5.05 up,
it's your iso, your CD burn, or your hardware
but no, not when I was trying to get something to work....
Lots of people swear at that John Arvin guy too...do a google on the "disk error 80" or 32....
I had hardware issues....on this computer, which I resolved, but
ubuntu 6.06 has problems with motherboards using Intel chipset that has SATA only, but add on chip for parallel ATA (IDE)
I tried the CD's on another and RC_46, and this worked...but none of the
ubuntu 6.06 's i burnt....I agree there is a bit shifting error, but think
again, did you check the md5sum of the iso before you burned it?
its in the source of the ISO, not the download...both EU and NA copies were
the exact same size...have not done a compare on em yet..
were people looking for 50pin brkout boards? http://www.winfordeng.com/products/brk2x25.php
No SATA here...
don't compare them. CHECK THE MD5SUM
Only two EIDE controllers...
length means nothing
A compare will tell me if they are identical....which would imply the
downloading is working...
the md5sum tells you with 100% certianty.
But then again, cannot do the md5sum until I can get this thing on the hard-
which brings me back to my original question...hi...
what did you burn the CD with? do the md5sum there
Yea, I am familiar with CRC's...
I burned the CD with my XP system....
That I normally type to you guys on....its in the other rooom....
you can do md5sum on the downloaded image while on the xp system
you likely can do it inside the program that burned the cd
... looking for the md5sum program... fo xp
Roxio EasyCD Creator ?
... ill be back with an answer
google search for winxp md5sum
fourth hit or so
Download it here: http://www.etree.org/cgi-bin/counter.cgi/software/md5sum.exe
Yep...lots of shareware programs there...that is a start...
Have to go turn the other system back on...
But still want to install this one...Even if I find the CD's are bad, how can I
get a good one ? The downloading is just not working, then...
I got this image from cncgear....hey wait- this is ubuntu 5.10 ...check out http://www.cncgear.com/EMC/
I don't think you can install from the 5.10 live cd, but I might be wrong
(I don't recommend 5.10 for a new system anyway)
5.10 had separate live and install CDs
Welp, cncgear does not have an install version...unless maybe the coolcnc is that?
coolcnc is something else entirely
I recommend you get the one from linuxcnc.org or the EU mirror listed there
how can I check the ubuntu version ?
actually running on here...
the cncgear site says it is 5.10 and is live and can be installed "You have the option of installing if you wish" but doesnt have the md5sum...
I tried those, both, no go...
in a terminal: cat /etc/lsb-release
that's surprising, the file /etc/lsb-release should be readable by all
just forget the cd that wont install.. becuz: it wont instal: then goto http://www.linuxcnc.org/index.php?option=com_content&task=view&id=21&Itemid=4&lang=en
get the newer ubuntu & the md5sum
then we'll know how to advise you
using the file browser, i can get to it .... 5.10 breezy badger...
wonder if the new CD(r,rw) will work off here....
that is where I originally got the ISO's for 6.06 tomp...
nope...cannot eject the cd here...!
network install ?
then just get the md5sum, it's very smalll txt file, use xp to verify the image you downloaded... then we know that parts ok
yep....going to switch rooms here, and bring up the other system...
md5sum of ubunto 6.06 live cd 6ee5048eb9cb424aa030dfedccc5386b
'ello 'ello 'ello, that you ubuntu?
I guess I should change that link on cncgear to the 6.06 liveCD version
leave the 5.10 on there !
I can link to both
ok....no help in this md5sum program...
How do I get it to do the whole disk in the CD ?
are you using it on the .iso image or the burned CD?
want to do both, to see where the problem is...
you downloaded a .iso file, test that file
i think the md5 is going to check out fine....and the problem is with the original file that the MD5 was generated on..
first things first - you burned the .iso image as an image, right, not as a file?
( he gets to the isolinux text when he tries to boot )
ah - ok
the command should be something like `dd /dev/cdrom ifs=2048 | md5sum`
( he's on an xp system )
then it won't work ;)
md5sum is available for Windows, but you can't check the disc, only the .iso
tomp: there are several people giving him conflicting advice (and he's on two systems at once)
ok, md5sum checks out...
wb9mjn: EXACTLY what did you check?
the CD, the .iso file that is still sitting on your XP box?
the download from the NA mirror...
Now, how do I do the CD's ?
so you compared the md5sum of the .iso file on your XP box, against the md5sum posted on the website where you downloaded it?
I don't think you can check the CD with MD5Sum, but your burning software may have a verify command
both the NA mirror and EU mirror files have the same md5sum, matching the sum on the web page...
ok, so the download (both of them) worked correctly
well...they downloaded identically...
(you only needed to do one, and then verify the m5dsum, but thats water under the bridge)
wb9mjn: What speed did you burn the ISO's at?
neither would boot in the EMC computer...
wb9mjn try at 4x
what speed is your burner?
but, it could be the old CD's...but then again, the 5.10 burned on the same old CD media...
the burner is up to 32 x...I believe...
At least that is what Roxio permits...
And I believe that is based on hardware lookup...
wb9mjn: I've come to notice than MANY ISO's will have issues if not burned on slower speed. the 4x always seems to work.
Roxio does not permit anything slower than 8x ...
disable av, turn off everything, blah blah blah
I could try the newer media I got today ...which is supposed to be good to 52 x...
[02:03:44] <Jymmmmmm> http://iso-burner.qarchive.org/
wb9mjn: instead of trying this and that, you need to figure out how to VERIFY a burned disk
but, I will try this and that eventually, too...
back in 90
I thought the Ubuntu CD installs were capable of self testing the CD from the first menu upon boot
Maybe they are, and maybe that is why the boot is failing ?
no, it shows up as a menu item on the stock Ubuntu images, don't know about EMC ones
the menu also includes a memory test option
steve_stallings: I'm not sure he's getting to the menu
yes - I believe those options are there on the LiveCD
he says it fails at the "isolinux" stage
its been too long since I last installed, I don't remember if that is before or after the menu
and I'm not about to reboot just to try it
it's part of the ment
the emc2 livecd is 99% the same as the dapper cd -- it's just dapper with the updates and emc2 packages added
err - sorry. isolinux is the boot system, before the menu
wb9mjn: tell us again what you see happen... anything after "isolinux" ?
menus, disk check, memtest, installer - all the same
ok, thats where he says its crashing, so if he never gets to the menu, he can't use the menu cd tester
I just read someting about padding...
then it likely that Ubuntu 6.06 is not compatible with his motherboard for some reason, this is exactly what I hit with my new motherboard
you know - there have been several people who seem to think that they can boot the normal Ubuntu 6.06 disc and not the liveCD - I wonder if ther eare unintended differences ...
This ubuntu 5.10 has EMC2...
wb9mjn: that is irrelevant
The isolinux version is newer on the 6.06 ...
SWPadnos: different kernel (in particular no ACPI)
yep, and realtime added
though wb9mjn isn't getting there, so that's probably not it
disk error 80 (or 32) AX=4280, drive 9F ...
same to you!
was that on XP or booting or in Linux ... ?
on the linux machine and on the xp machine both...
the disk error?
actually, the disk error number was different on the xp machine,,,,its a scsi system...
the "80 (or 32)" was while trying to boot with the cd? on 2 diff hdwr systems?
but errored at the same time...i think the disk error number was 02 on the XP machine...
ok. there's something wrond with the disc. did you "finalize" the burn or do disc-at-once?
wb9mjn: can you tell us EXACTY and precisely without skipping anything or assuming we know anything, what happens when you try to boot the live CD?
the 5.10 worked on both, very nicely...
many problems happen when the BIOS understands the disk configuration, but the Linux installer does not
that's a different disc, so it doesn't help for 6.06
i did track-at-once....I seem to remember that that has always worked...but one of the three 6.06's disks was
(though it does tell us that the PC can boot from CD)
done disk-at-once...still did not work...
the 5.10 was done track at once...
lets wait for what he says happens when he tries to boot
yes,,,the disks were all finilized...
again: wb9mjn: can you tell us EXACTY and precisely without skipping anything or assuming we know anything, what happens when you try to boot the live CD?
When booting the live CD (6.06) the isolinux id line comes up...the CD stays on, and then eventually it gives:
disk error 80 AX=4280, drive 9F
The error numbers can vary, sometimes 32....different on the xp machine 02 ?
google isolinux "disk error 80"
[02:16:22] <SWPadnos> http://www.fedoraforum.org/forum/archive/index.php/t-47526.html
other people have had this problem, with other linuxes, its completely unrelated to EMC
I did some looking around for this problem by googling...
And tried the SBM, but all that is for is to get it to the isolinux in the first place, which it was doing already...
That is if the CD cannot be booted from , by BIOS setting, which was not the case in my system...its been used
Results 1 - 10 of about 675 for isolinux "disk error 80"
for multiple BDI's, and RC_46....
The only answer I found was that the SBM program was needed, which I tried, and it was just to get the CD to boot...
add "dapper" to the google search
Which is a problem I never had...
[02:19:03] <SWPadnos> http://www.linuxformat.co.uk/modules.php?op=modload&name=PNphpBB2&file=viewtopic&t=4525
are you booting from a DVD ROM drive, or a CD burner, or a plain old CD reader?
SWPadnos: he been there & tried the SmartBootManager too
no, the original was a straight CD, then replaced it with a CD Burner when found out there was a problem with it...
though they do mention "isolinux incompatible with the BIOS" as well
does a stock Ubuntu 6.06 disc work?
Found out the problem, by booting off the DVD/CDburner combo on the XP system here...
SWPadnos: isolinux is identical
(I suspect it won't, unless the burn is bad, and the new burn is good)
Then I had similar results with the XP system and the EMC system....
wb9mjn: is your system a Core 2 Duo?
no...the XP is a Xeon 2.4 GHz, dual processor, with one processor installed, and 10 K RPM SCSI....
that was my problem, SATA only chip set, but he said no SATA on his
I'm asking about the one that won't boot, not the XP one
Are track-at-once/disk-at-once valid options when burning an ISO?
The EMC computer is a 800 MHz PIII with dual EIDE controllers (one is on the CD, other on the hard drive)...
finalize may be, but the trac/disc-at-once may not be
is the CD an RW or plain old CDR?
wb9mjn: thank you for that info - we need facts. unfortunately I'm even more confused now, since a good old P3 system should be easy to install on
I beleive both CD's are RW...the new one certainly is....both act the same to all CD's now...
the new CD is in the EMC system...
I'd burn on non-rewritable media
ok, there are three steps in this process..... lets be logical about this
1) downloading .iso file from the net
2) burning .iso file to CD
Yes, the XP system is a RW...and I have been buring on non-rewritable media, for all the disks...
3) hardware issues on the target system (the one you are installing to)
wb9mjn: you're not speaking clearly
Downloading is apparently good...
RW is rewritable media
ok - that was my question - I want to know the media type, not the drive type
are you talking about the _disk_ or the _drive_
we can't read your mind ;-)
Ok...I was refering to the drive itself...
the media is one-time...
they're 700 meg CD-R, right, not 650?
step 1 can be verified by doing an md5sum of the iso file while its on the windows machine's hard disk, and comparing to the website
you did that, so we know step 1 went correctly
step 2 can be verified by doing an md5sum of the burnt CD and comparing that to the website
you haven't done that yet
yep...md5sum OK on .iso files resident on windows system...
or, on Windows, using the "verify" feature of your burning software ...
nope...have not found windows program for that by googling...there are some Linux programs...
one way or another, you NEED to verify that the disk burned correctly
Roxio should be able to do it
Let me take a look at the Roxio Easy CD Creator 5 help file...
maybe I should be using the disk copier, rather than the general utility ?
I don't think so
I can't help you about burning CDs on windows
I've never in my life burned an iso onto a CD in doze
[02:29:40] <tomp> http://twiki.org/cgi-bin/view/Wikilearn/CdromMd5sumsAfterBurning#In_Windows
how to check cd md5sum in widoze
soory... it goes on to say "i dont know of a way..." so use tools on the good ubuntu 5 isntead
I haven't been able to do it after the fact, but there is a "verify after burning" option in my CD writing software (NTI CD-Maker 2000)
there should also be an option to verify a CD against an image or another CD, but I think Windows changes the way the device is accessed depending on whether there's a filesystem on the disc or not
do not see anything on MD5SUM, Check Sum...
you won't see that - the software will only (maybe) let you compare the CD to the original image, to verify that it burned correctly
they don't give a rat's as sabout the md5sum
use the good ubuntu 5 system to chk the cd... SWP said the command should be something like `dd /dev/cdrom ifs=2048 | md5sum`
tomp: the ubuntu 5 system is running as a live cd, so unless there is a second CD drive, he can't take the live CD out and put the suspect one in
wb9mjn: are there two cd drives on the EMC system?
not seeing anything....
Nope..only one drive...
ok, forget that
going to look into tomp's link...
the .iso files are still on the XP system, right?
maybe the simplest thing to do is to burn another CD with "verify" or whatever turned on
wb9wjm... NO the link said it didnt know how
instead of trying to figure out how to verify an already burnt CD
jmk : doh! you're right
Oops...its says he does not know of a way to do it in Windows...
yes, roxio CAN verify the burned cd that was just burned , so make a new cd (http://mail-index.netbsd.org/netbsd-help/2006/10/29/0000.html)
does not say how tomp....I did do a "test" when I made one disk, but that just tested the media, not the burn...
The burn was after the test....
a proper verify reads every single byte of the freshly burned CD _after_ burning, and compares it to the original .iso file
a "test" is not a verify
nope...it just tests if the computer will not buffer underun, I think...when the files are accessed...
need a third system !
from the bsd web page "so you should have Roxio verify it after burning (that is an
Put RC_46 on it, and then md5sum the CD's...
verify should be a checkbox on the burning options page - in the same kind of place where you see options like "finalize disc"
nope...its not there....maybe a newer release of Roxio ?
Yes...I see the finalize disk, and used it...
could be. I was just looking though version 4 menus, and didn't see the option :(
wb9mjn: which particular Roxio software are you using?
My version had "create cd from image" as an option.
or something like that...
SWPadnos: not on menu, on popup dialog after chhosing the action
I didn't create a disc, so I didn't get that far ;)
uh, he had a cdrw and a cd, can both be put into the linux box? boot from cd, write & verify with cdrw?
SCSI and IDE, IIRC
or "Small Computer System Interface and Integrated Drive Electronics, If I Remember Correctly" :)
acronyms are so much shorter :)
I'm looking at Roxio support - searched on "create image", got a result saying to burn the disk image using Disk Copier.
both drives are CD RW ....one is also a DVD Rom...
The the one used to create the diskes was the DVD Rom Combo CD RW ...
Easy CD Creator 5 ...
can 2 cdrw go into the linux box... boot from one, burn to other?
you need access to the ISO for that (wither the HD or via network)
I replaced the EMC computer CD with a CD RW ....that resulted in the matching of failures/successes with XP system..
got a usb drive that'll hold the .sio image? (AND 2 cdrw devices)
disk copier will not take an .iso file source...has to be a disk...
yes...have a USB drive that would hold it,....
but, it has other stuff I cannot take off it...
how would I make the USB bootable anyway from the .iso ?
well, as jmk pointed out, step 2 is only half done - a disc has been burned with something, but it needs to be verified
do you have any PCs that have Linux installed on the HD?
yep....and the only linux system I have up now is a live one, which cannot have the CD removed to do the md5sum...
burn the 'other stuff' to a cd :-)
Could put the RC_46 drive back in....
(dont try to boot from usb, just holds the file)
reboot and see...although with the new CD it might be a problem...
well, it's long and torturous, but you could install the 5.10 CD, then copy the iso over and (a) verify the 6.06 disc and/or (b) burn a new one ;)
ie, install 5.10 to HD
The Disk Copier program only accepts Disk Drives for the source...
Hi hi...Yes, SWP, that is why I was asking how to get the 5.10 live CD to install to disk...
there should be an install option, but I could be thinking of dapper ...
I might try the new media I got today first...
I could not find it, but the intall option might still be hiding somewhere...
maybe not for 5.10
New media is next thing...
the LiveCD thing was new for Ubuntu in the 5.10 era
I'd burn another disc, at the slowest speed you're allowed (which should be 1x or 2x), and look very hard for a "verify after burning" checkbox
well, you can probably burn at 4x
I usually use 8x, on my 12x RW drive, and have no problems. I have lots of problems at 12x though
i'm shutting down, booting live ubuntu 5.10 and looking
Ok, burned a new CD...going to shut down the EMC system and see what happens...
was there a verify option?
No Verify option..
But, it did work !
ok. I'd say the other CD was bad ;)
Came right up....
err - cool
now run the verify on the menu
The old media I had was apparently not burning fast enough...
tomp, it was the CD itself
It will take a while to install on the EMC omputer...
the CD media was not fast enough...
verify first - it may save you some trouble
Maybe because its 70 F in here ?
should be the second-to-last menu option, I think
And maybe in the summer when its 90 F, it wrights faster ?
no luck: my ubuntu 5.1 cd is the 'install' version, not the 'live' version
Thanks anyway Tomp...I think I am on my way...I got new 52 X CD Media when I was at the store today...
oh! good news ( of a sort )
Just tried it, and it booted!
how did you >know< it was the cd
he burned another one ;)
Its booted now, but there is no IRC client so cannot talk to you from it...
And there is a big "INSTALL" Icon right there on the desktop...
gaim is supposedly an IRC client, but I think it sucks
well, now have some fun, learn about apt-get (which you do have)
you are now in charge of 'i-cant-boot-my-cd' department :)
I even burned it at 32 X....
oh, i see, the discussion talked about cd drive speed & i thought about media, which is usually the limter ( for cheap ass me buying the 100 for 12$ stuff )
The old media speed was 8X, but the XP Combo drive is allot faster, but I set Roxio to 8x ...
Its very very old media (2003 ?) ...
I have only used half the spindle...
I imagine its slower the colder it gets...
media degrades over time, burning is a chemical/optical process
The 5.10 disk was burned when the sun was streaming through the windows late morning...
The other disks were late friday night....
what color sox were you wearing?
white both times....
So, the media was probably marginal, and the 5.1 disk just got lucky....
if the CDs are 3 years old you should probably just throw them away
no, burn windows boot cds on 'em :)
Or at least not use them for computer stuff anymore....
There are still about 25 of em left...
you can set drinks on them
or tile your bathroom
how much are you worth per hour? ( you already spent that much!)
There was a project making kids tops out em...just turn a spindle....
I'm in the same boat - I have about a third of a stack of 100 that I got several years ago
Yea, and they are Systemax....
"free after rebate", so they weren't the best even then
see, I told you guys I am cheap...
as I've gotten more and more into linux, I find myself burning fewer and fewer CDs
What is wierd is the the paper sleeves for the disks cost half as much as the disks!
more...$16 for the disks, and $9 for the 100 sleeves...
I copy things around on the network instead of burning and sneakernetting them, for example
jmk: how do you find stuff?
just put them in the few directories that you allow to be shared...
Like CNC and CAM ....
if all the stuff that used to require cds, now is 'on the net... how do you find stuff ( large volume of files, need 1 )
especially with VMWare, where you just point the virtual drive at the ISO ...
finding on a cd is no different than finding it on a hard disk
i use loads of cds and i use gtktalog
I don't have that much stuff
there's no need to burn, since under Linux (and with the right utilities under Windows), you can mount and look at files in isos
I keep most stuff in the home directory of my main machine
and I have a directory called isos ;-)
mount -o loop -t iso9660 /mnt/myimage /path/to/my.iso
uh, that still requires hard drive space, i like to offload stuff, but mounting iso is fast
too bad there isn't an option to mount -o loop -t iso9660 ~/isos/* ;)
like 'virtual drive' on m$
to get a subdir for each image there ;)
for iso in isos/* ; do ;-)
well, hard disc space is less expensive than CDRs these days
unless you get freebies all the time
300+ GB drives are ~$90
jmkasunich@ke-main-1006:~/blog/shoptask$ ls ~/isos/*.iso
looks like BDI is almost twice as popular as all other OSes combined ;)
no, just 4 times older
i'm so cheap i havent looked for a long time ( I'm a web hoarder tho, store every interesting site I cross... migawd 300G <100$!
even 500G is in the 30cents/gig range now
i was going to buy some of those 50cd jukeboxes for my over 500 cds
good night all...thanks...
tomp: I wonder how many of your 500 CDs are still 100% readable?
bet its less than you would like
hmmm - 500 CDs * 700M each (worst case) = ~350G, or about $100 ;)
wow - 320G, $77.99
i gtktalog-ged them earlier this year... ok
tiger direct? or where
it's on PriceWatch - I haven't checked where (or brand) yet
jmk: 100% yeah, doubtful
but 400G are $128, and 500G are $168
so the puny 320's can't be too bad ;)
puny... my 256 byte rca 1802 system is crying
interesting - even the 750G drives are below 50 cents/gig
just barely though
I remember paying for hard drives at ~$90,000 per gig ;)
I paid $400 for a 40meg once.
we could only get a 0.01G drive though
funny to put 10M in that perpective L(
tomp: did you make an Elf?
cosmac elf: made it animate a face on a b&w tv, had it under my arm at a halloween party... like headless horseman
heh - read the log, did you?
Actually no, I was talking about "Can't burn slower than 8x boy"
[03:42:26] <Jymmmm> http://www.terabyteunlimited.com/downloads/burncdcc.zip
right - he couldn't, but the new media worked for him (he burned another CD which worked)
steve_stallings is now known as steves_logging
sometimes I wonder about the junk I collect
what the heck am I gonna do with 4 5/8" eyebolts
jmkasunich: We STOPPED windering about the junk you collect LONG ago.
you can bolt down a small milling machine, and/or hoist it up to the ceiling
1,000,000 self sealing stem bolts.
I already have 4 3/8 eyescrews in the joists above the shoptask
5/8" eyebolts are probably good for 4000 lbs each
3/8 are too small - you collected these to replace those
except these are bolts - they won't go into wood joises
no - wait, you're planning on getting a *real* lathe, and those weight 3000 pounds ... ;)
yeah, I've always wanted a Monarch 10EE
of course, theres no place on that lathe for an eyebolt
and I don't have a 3000 lb capacity crane to lift it with anyway
I just know that a couple months after I get rid of these I'll actually have a need for them
our cousin Aaahnold would be very disappointed if we couldn't lift a 3000-lb lathe
that's why you keep stuff
jmkasunich how many do you have?
You mean the govenator?
take 'em to the fest, there may be some bigger hdwr next year
I think I have 1 5/8 eyebolt - that's the size of the lift hole on a Bridgeport
put 'em on top of your speaker... makes people think you have some real good stuff :)
in fact, mine may be in the BP right now (or is that the lamp?)
SWPadnos: If you do it right, there is no difference.
well, I found a place to stash em, so they escaped the trash for now
It's raining hard, and I can't figure our what to eat tonight. Something comforting sounds good though.
hey I have a question about pid and deadband
delivery, so you don't have to go out in the rain
I'm kinda burned out on pizza.
go to that all-you-can-eat Korean buffet place
Korean Palace or some such
SWPadnos: I'm at work
argh.,.. maybe cause I'm asian, and get sick of asian food... K-buf is okay once in a while... but argh
oh, in that case, a cheeseburger
ok if the machine moves within the deadband should the oid output go to zero?
sorry pid output
the I term can be non-zero even with zero error
usually when the axis is stopped the I term will be very small though
but wont the machine continue to try to move?
if it does, then the PID will correct
suppose you have a vertical axis (knee or spindle) and it takes an amp or so to hold the weight
the PID shouldn't integrate any error less than DEADBAND in magnitude, no?
so you need a non-zero output to hold it still
SWPadnos: correct, it won't integrate more if you are in the deadband
but whatever is already in the integrator will still be there
I see that makes sense
so the machine wont continue to move but the servo will exert some effort
SWPadnos: Pizza just SUCKS here. I haven't had a good pizza in years.
as cradek once described, the servo should move freely within DEADBAND, but should be a brick wall once you hit that limit
Jymmmm, then I wouldn't order pizza. my suggestion was a cheeseburger
though you can get good mexican food, so that's a good option as well
what is the bias for
on the amplifier?
that should work like an offset - if 0 isn't "no motion" for your hardware (?)
ChrisMorley: its rarely used, but in that "heavy vertical axis" case, you could use bias to compensate for the current needed to hold it still
SWPadnos: Yeah, finding some pretty good holes int he wall lately, but I'm in foreighn territory here.... almost the entire community is asian.
Jymmmm, then I'd get asian food
Jymmmm, find the most crowded place, and go there
SWPadnosL Nuh, Uh.... they have those fish you have never seen in your life before.... life the Chinese Pharmacys
snake fangs, etc
ok - I don't go for too many eyeballs in my food either
usually bias/offset is used when the control puts out 0 (dont move) yet the device moves... the bias will 'null' it
SWPadnos the problem is that THEY DO! lol
tomp, that's true in the quill/knee case as well - if you output 0, the knee will fall
no the position loop will hold it inplace, increasing current to keep no motion
ok .I'm actually working on the pid of a large lathe.
PID will also compensate for an electrical bias
it proves to be difficult
what do you see happen?
but I don't know wha i am doing! learning!
can you move at all?
often the machine oscillates in the beginning of the move
ok, and how does it move by hand? (no electricity)
this is a cnc lathe no handles.but using a socket wrench -it is smooth
I found on my lathe that too much, or not enough, D will give oscillations (too much P will too)
what are regular p, i and d numbers?
but finding the right P is easy
yes d is very sensitive
the actual numbers vary to much for us to give you specific suggestions
d ranges in the .01 area
but usually D is quite small
but something like: I is half of P, D is like 1/100th of P
wow, thats _very_ small
p is around 1. I thought it would be higher
how do you have things scaled?
feedback is encoder, you should be scaling that from counts to inches or mm
did you find what value of P barely gives sustained oscillation?
ChrisMorley: something about your scaling is strange
with 0 everything else
can you go over that with us please!
yes, no I no D
yeah your scale is off
scale of voltage /velocity?
can you post your ini and hal files to pastebin.ca?
yes, DAC voltage to velocity
no i'm sorry that computer is not on net
(if your amps are velocity mode)
Chris are you using emc? or is this a general PID question?
if you send 1 volt out of the DAC, how fast will the axis move?
the scale is 1
I never checked
no idea at all what the relationship between DAC volts and axis speed is?
do you know how fast it will move when maxed out?
I have to max at 18 inches a minute. the encoder modual cannot keep up
(thats one of the first things I would try to figure out)
18ipm? thats slow
(at least for most "large servo lathes"
yes the machine can go 200
you have dacs but no hardware quadrature counters?
ok, two minutes ago I asked you how fast it would move when maxed out and you said "no idea"
now you say 200 ipm ;-)
jmkasunich: that was in response to a different question
the machine can go 200 ipm when using old control
the old control is broken
and you are using the same amps as vefore?
ok, then it will still do 200ipm\
just emc can not count the encoder fast enough
what are you using for DACs on your EMC machine? m5i20? motenc? ppmc? servo-to-go?
pwm through parallel port
are you directly driving H-bridges somehow?
(I was assuming that the old amps had +/-10V analog inputs)
yes they do use +-10
using the h-bridge to power the amps
so are you driving a +/-10 input, or are you driving the H-bridges directly? cant be both
sorry . computer runs h-bridge. h-bridge powers amps
I'm still confused
usually an H bridge is part of the amp
yes but I don't have an dac card in computer
so I was using h-bridge and pwm to make one
"servo amp" usually consists of a +/-10V input that possibley goes thru a speed loop, then goes to a PWM generator, and finally drives an H bridge. all of that is one box
to drive the H bridge directly you have to go hacking around inside it
yes I know. I have two
one I made
one in the amp
oh, you are using EMC's PWM to run a LOW POWER H bridge that makes a +/-10V signal, and that runs to the regular amp?
the one I made produces the +- 10 volts for the other
I think I understand now
you gotta explain these things to us, we can't read your mind
thought I did sorry
and does it look like +/-10V with a little integration cap on the output? ( like with a scope? )
so, my first step would be to disconnect the motors from the big amps, put a meter on the +/-10V signal, and make sure that works
make sure that you can get 10V, 5V, 0V, -5V, -10V
I have done that
(by setting values on pins using halcmd, and measuring with a meter)
and what did you see on the physical meter/scope
It is out a little max voltage is about 11 volts
so with the pwmgen scale set to 1.0, you probably get 10V out when you command 1.0 to the pwmgen
pretty close yes
or 11V in this case
if you set maxdc to about 0.91 or so, that should limit the output to 10V
next step, connect the motor to the amp, and if possible, decouple the motor from the screw (remove a belt or whatever)
now that i think of it I am usig PWD
pulse width demons
pulse width density
that shouldn't matter
you are filtering the signal before you send it on to the amp, right?
difficult to disconnect motor
dangerous not to
tomp: not to what? disconnect, or filter?
you gotta do what you gotta do
how would I filter it?
ok, before we even get that far
well, the filtering thing is a complication no matter how you look at iut
filtering is easy, use a RC network
the problem is deciding how much
too much, and it adds phase lag to the system making it hard to tune
like a charge pump?
no, just an RC lowpass filter
R from input to output, C from output to ground
how are you driving the (probably single ended) input of your existing amp with the differential and unbalanced output of an H bridge?
sorry not sure what U mean
its pretty simple to turn the output of pwmgen into a +/- 10V signal with an opamp and some passives
doesn't the servoamp input have one terminal that is supposed to go from -10 to +10
and another terminal that is ground?
ok, that is single ended
a single signal, relative to ground
an H bridge has TWO output signals, and neither one is grounded
so how did you get them to play nice together?
square peg, round hole I think
just wired them together. ignoance is bliss!lol
? shorted the hbrdg output?
ignorance is not bliss
or cnx''d 1 side of hbrdg to gnd?
then no longer +/-
ChrisMorley: it is not terribly difficult to turn the output of pwmgen into a +/-10V signal, if you know some electronics
it can be done with an opamp, some resistors, a cap or three, and a +/- 12 or 15V power supply
how much electronics do you know? I thought you said you _built_ the H-bridge?
yes I did build it . I found the plans on the net then played around till it worked for me
ok, do you have a URL for those plans? paste it here so we can look
usually H-bridges are used to directly drive a motor, NOT to drive some other stage
[04:37:29] <tomp> http://controls.ame.nd.edu/microcontroller/main/node40.html
pwm to analog, single opamp
tomp: that circult is unipolar, but its a step in the right direction
[04:38:26] <tomp> http://www.allaboutcircuits.com/vol_6/chpt_6/9.html
I am finding it hold on
ChrisMorley: take your time
yeh, those are unipolar
everybody is using pics to do this
tomp: and the second one will blow the transistor unless it is heavily derated, or the load and wiring has very low inductance
I use AVRs ;)
you use PICs or AVRs to _make_ PWM maybe
you gotta use good old analog to do anything fun with the PWM
not to directly (via transistors, of course) drive a motor, unless it's a tiny unipolar stepper from DigiKey
interesting drawing... they don't even show transistor symbols, just the packages
if use search around it does
"leads-up dead bug style" ;)
I used different transisters and played with the resister
at least they oriented all the transistors the same way ...
how did you decide that it was "working"?
were you spinning a small motor with it?
the lathe moved
actually I tried a small moter and leds first
"moved at a rate that seemed proportional to the output", or "moved"?
it seemed to work fine. other then tuning
precisely controlling the motors is much different from just making them turn
so when you sent 0.1 to pwmgen, the lathe moved slowly, when you sent 0.5 to pwmgen it moved 5 times faster, and when you sent 1.0 to pemgen it moved at 200 ipm?
then it didn't work
I never tested it that way.
I'n not trying to be a prick here, so please don't get offended... but there is a lot more to this than what you've done so far
I used the etch-servo in emc2 and set the max speed down to 18 ipm
in that tilden bridge.. the 2 "-" inputs are not pwm inputs... they're supposed to be active 1 at a time
stepgen can do that type of output - type 2, I think
tomp: that circuit is really unsuitable for what he's trying to do
if he had a suitably small motor on it, pwmgen could certainly drive it
yes type 2 is what I am using I did cahne the etch-servo to suit
but its not right for making +/-10V to drive another amp
ChrisMorley: are the existing lathe servoamps commercial ones?
yes it is an Okuma lathe
yeah - to get the scaling right you'd need ~-2V for ground, and ~12 for + (or thereabouts)
well, there would be an offset, among other things
that circuit simply can't make a single ended negative output
in the 1st circuit i saw, the opamp could be connected to +/- 10V (or 15) instead of the suggested +12 & 0
not with a ground reference the same as the control inputs
ok guys if I take a battery and wire it to the amp the motor moves
and if i change the leads around it moves the other way
battery to the input of the Okuma amp you mean?
what ever voltage I used a 9 volt
did it move pretty darned quick?
yes very quick
I would expect that 10V would give you 200ipm
so 9V gives 180 ipm
(you are brave - I would do the battery test with a much lower voltage)
a 1.5V battery should give you 30 ipm
so why would an h-bridge not be the same ?
for one thing, a battery is DC, not PWM
for another, the battery is a floating source, and the H bridge is not
floating with respect to ground
ok but if I did'nt use pwm It would work ?just reeally fast?
ya sorry I am way over my head I see
connecting the H bridge circuit you showed us to the inputs of the Okuma amp will do something, but I have no idea what. I'm sure it won't do the right thing
it might work in one direction but not the other, for example
this is an op-amp differential amplifier
[04:55:40] <jmkasunich> http://en.wikipedia.org/wiki/Image:Opamp-differential.svg
I dont understand the "floating source"
the voltage on the battery isn't connected to ground on the PC or on the servo drive
how much electronics education do you have?
it "floats" relative to them
we aren't going to be able to teach you from square one
I don't want to insult your inteligence by assuming you know nothing, and I don't want to confuse you by assuming you know more than you do
I read books - self taught. I really don't want to waste your time either
I happy that you are helping . thanks
as long as we are communicating clearly, its not a waste of time
do you know anything about op amp circuits?
the basic principles
you understand ground, and that most voltage measurements are made with respect to ground?
could you build and understand this circuit: http://en.wikipedia.org/wiki/Image:Opamp-differential.svg
the power supplies for the op-amp aren't shown, for this application they would be +/- 12 or 15 volts
Rf give neg feedback-yes?
if V2 is grounded, it is a standard inverting amplifier
like this: http://en.wikipedia.org/wiki/Image:Opampinverting.svg
Assume R1 (or Rin) is 5K, and Rf is 10K
if you put +5 volts on V1 (or Vin) you will get -10V out
if you put zero volts on V1 you will get zero volts out
now lets put some voltage on V2 instead of V1 (leave V1 at zero)
make R2 = R1 = 5K
and Rg = Rf = 10K
if you put +5 Volts on V2, with zero on V1, you will get +10V at Vout
just a thought before I go to bed - the best way to run this lathe, which will be simpler and give full speed, is to buy an interface with some dacs and quadrature counters - since you have amps already you might get by with spending just a couple hundred bucks
after a lot of work, 18ipm is going to be very ... unsatisfying
cradek is right
if you have a really really tight budget, the approach you are taking can be made to work
but it is more effort, and the results will be dissapointing
unless you need to cut something before Monday ...
lol no this is all about learning
we all like a challenge but sometimes the easy way is the good way
thank you for your help
hardware encoder counters will be able to track 200 ipm
and regular DACs will give better results than converting the PWM from EMC into +/-10V
yes I knew that .I just wanted to play with it now!
(I'll be happy to continue explaining how to do that conversion)
yes If you don't mind
well, there would be a lot of learning involved of you build a circuit like jmkasunich is describing, then at some point change over to a Mesa 5i20 FPGA board, and make an FPGA configuration to replace stepgen type 2 outputs at higher speed ;)
still looking at the differnetial amp picture
R1 = R2 = 5K
Rf = Rg = 10K
if V1 = 5V and V2 = 0, Vout = -10
if V1 = 0V and V2 = 5V, Vout = +10
if V1 and V2 both = 0, Vout = 0
with you so far
so, drive V1 with the "down" output of pwmgen, and drive V2 with the "up" output
the opamp takes the two 0 to +5V signals from pwmgen, and turns them into a single +/-10V signal
jnk is the mcgiver of emc
the only problem is that Vout is still PWM - its not a nice steady value like the battery was
jmk is an electrical engineer ;-)
MacGyver - mcgiver sounds wo McDonalds like ;)
back to the filtering problem
we want to filter the signal before the opamp
which can be overcome with rc circuit
so, instead of using 5K for R1 and R2, split them into two resistors that total 5K
split R1 and R2 into two resistors in series
then connect a capacitor from the midpoint of R1 to the midpoint of R2
the size of the capacitor determines the amount of filtering
* jmkasunich digs out the calculator
ChrisMorley, what BASE_PERIOD setting are you using?
same as etch 20000 for fast I think
the smaller the base period, the higher the PWM frequency and the easier it is to filter
20000 is good
I think the capacitor should be between 0.5uF and 1.0uF
it might be hard to find a non-polarized cap that large
you could make all the resistors 10x bigger and use 0.1uF
wouldn't two polarized caps to ground give similar filtering?
ie, each leg separately
in this case, yes
thats probably a better approach anyway
(for a generic diff amp, you'd need tightly matched caps to retain common mode rejection, but that doesn't matter here)
ok - also we know the input range is unipolar, which helps
because we only ever drive one side at a time, the other is zero
if I was more ambitious, I'd draw up the complete circuit (with R1 and R2 split, and with the capacitors in there) and post it on the web somewhere
but not tonight
so a polarized cap from 'center' of r1 to gnd, and another polarized cap from 'center' of r2 to gnd
tomp: yes, exactly
same value for each cap (unless you want attack and decay to be non-symmetric, which you don't)
Chris? got it?
ogot it thank you -busy drawing circuit
10% tolerance is fine, but don't make one of them 0.1uF and the other one 1uF
actually, if R1&R2 is 5K (2 x 2.5K in series) and Rf&Rg is 10K, then you want 1.0uF
for 500uS time constant (unless I need to think about it more ;) )
what kind of amperage will this supply ?
that depends on the op-amp, but it should be unimportant ...
its just a voltage (a couple of mA)
this isn't meant to drive a motor
the output of this circuit goes to the +/-10V input of the big Okuma amps
the Okuma amps will want very little current input, look at the wires they used
mA's is enuf
these are "signal" circuits, not "power" circuits
well I have some building to do!. I will buy a card eventually. what do you recomend?
what does '500 uS time constant' mean... how long it takes to go min to max?
if you build this circuit and get it working with EMC2's pwmgen and software encoder counting, then later you could probably use Jeff's "pluto-p" circuit (about $60)
in RC circuits, the time constant (R * C) is the amount of time it takes to charge the cap to 1/e of the applied voltage
err - 1-1/e, or about 69%
or a Mesa card (at some point), for $199
ChrisMorley: Jeff's thing is stull experimental
actually, that can be done with PWM/DIR as well, with some changes to the op-amp circuit
oh, jeffs is pwm/dir, not up/down?
didn't know that
I think there's an option on the Mesa to invert the PWM signal when in reverse DIR
dunno about Jeff's
it is up/down
the opamp circuit is up/down, and that is what I prefer
if PWM can be inverted by the FPGA, then you can still use that
gives you a nice accurate zero point, and simple analog circuitry
attach DIR to - and PWM to +
or the other way around
oh, so with DIR=0, out is proportional to PWM
since PWM=!PWM when in reverse, you just flip the two inputs from 0 to + at the same time, then gradually lower the PWM output average to go firther negative
I think so
and with DIR=5V, out is 10V - PWM
or something like that
something like that
seems icky to me
I thikn that's there, and if it isn't, then it can be done in the driver
or change the FPGA ;)
If I was stuck with PWM/DIR, I'd add logic to the opamp circuit to convert it to up/down
or add an invert pin to the op-amp circuit
anyway, lets not confuse Chris
the circuit as described works with pwmgen in up/down mode (I think thats mode 2)
you should test the circuit before you hook it to the Okuma amps
put a meter on the output, and send 0.1, 0.5, 1.0, -0.1, -0.5, and -1.0 into the pwmgen
(using halcmd - disconnect the signal from EMC)
don't forget 0 ;)
if you get the appropriate voltages out of the circuit, then you can connect to the amp
yes, test zero too
how did it get to be 12:30?
hey by the way if I saved a hal cmd list
time flew at SR-71A speeds tonight
how do you load it ?
if forget the comand now
halcmd -f <filename>
but it didn't work -f something
halcmd -h shows the options
It didn't load I will try again!
thank you so much for all the help!!!!
you are welcome
you may need to setup the shell environment. from the emc2-head directory, run ". scripts/emc-emvironment" (no quotes)
err - emc-environment
best of luck
hey , i had probs w loading too, will try that env script
dunno if it matters - it probably shouldn't for halcmd, but the loadrt / loadusr commands may need the help
those circuits weren't really svg, you had to click on another link to see the svg version ( then ffox blows up when you leave )
Obviously, we don't want our tools--including our programming languages--to be more complex than necessary. But one aim should be to make tools that will serve skilled professionals--not to lower the level of expressiveness to serve people who can hardly understand the problems, let alone express solutions. We can and do build tools that make simple tasks simple for more people, but let's...
man - Bjarne Stroustrup can really cut to the chase:
...not let most people loose on the infrastructure of our technical civilization or force the professionals to use only tools designed for amateurs.
sounds a bit like the EMC philosophy, actually :)
make it for pros, and try to make it accessible also
uh, software pros or machining pros ;-)
err - whichever ;)
ideally, only people who are both
oh, both of them?
yeah - I'm sure there are at least two
nite all, you guys really helped 2 people tonight
that's a rate of 700+ per year!
argh - two like that every night and I'd be insane well before a year ;-)
likely before the end of this year ;)
well - good night
Hi All...this is on the new Ubuntu 6.06 install....
Added in the packages I wanted...now ready to start learning Hal/EMC2 ...
wb9mjn: usually people are around to answer questions if you get stuck :)
I would advise starting with the manual(s)
there is an installed version (Applications->CNC->EMC2 User Manual)
Yep....I was on last night....
I was stuck on the Ubuntu install...turned out to be the old CD media I was using...
Taking an install break, to get this phase of stuff organized....
wb9mjn: I saw the logs .. good thing you figured it out:)
Well, its what I could do, when I could do it, so I tried it....and it worked...
03alex_joni 07v2_1_branch * 10emc2/debian/ (.cvsignore changelog control control.in emc2-dev.files rules): starting to backport dpkg-improved from HEAD. testing on .cvsignore
Had a touch of Up-per weather down here this week...
03alex_joni 07v2_1_branch * 10emc2/debian/ (configure drivers.files.in emc2.files.in rules.in changelog): more merging of improved-dpkg
It seems to be spreading around quite a bit these days.
rayh: what's up-per?
As in UP per ...
Youp per ...
There are several ways of spelling citizens of this little country.
Well, I have not been there in 35 years....
Hi Chris. Don
You need to come back. We have a great bed and breakfast here at crooks cove
and the price is just right.
was 50 degrees here yesterday - I'm glad spring is finally here
It was darn cold in that part of the country.
Worse I think than here.
Did anyone look at the "
We are going into a mini slush season this week...
10 inches of snow on the ground, and going to be 40 the next few days...
Nothing like Calumet, though....8 foot high snow banks melting in March and April...
I hit some real patches of ice on the ride back.
Almost trashed a gas pump.
The Icicles are falling off the building outside my window just now...sun is out...
That sounds like an improvement.
My brother had a Sciiroco, and it would get hung up on the crest of hills leading up
to the pumps, almost did the same thing once...
Been there several times traveling between the west coast and here during winter.
I had an issue with EMC/GUI the other day.
rayh: what issue?
Incremental jog 0.0001 was smaller that a single step -- stepper./
so press it several times to make one step.
But the move was always in the same direction.
You gouing to get to the mountains for vactation this year.
rayh: I tried but could not reproduce that problem
rayh: probably so
I set scope to show step and dir, and triggered on the step
I'll have to try it here when I get set back up.
the dir always showed the right direction
sorry for the flood
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-5.10/emc2.files: ubuntu 5.10 specifics backport
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-5.10/etc/udev/rules.d/emc2.rules: ubuntu 5.10 specifics backport
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-5.10/etc/modprobe.d/emc2: ubuntu 5.10 specifics backport
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-5.10/etc/xdg/menus/applications-merged/cnc.menu: ubuntu 5.10 specifics backport
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-5.10/usr/share/pixmaps/emc2icon.png: ubuntu 5.10 specifics backport
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-5.10/usr/share/applications/ (emc2-doc.desktop emc2.desktop): ubuntu 5.10 specifics backport
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-5.10/usr/share/desktop-directories/cnc.directory: ubuntu 5.10 specifics backport
the dir does always go back down to low a while after the step pulse (don't know why) but it was correct during the pulse
cradek: I think the dir always corresponds to the current direction of movement (and for 0 movement dir=0 by default)
ok I guess so
the other choice would have been to always reflect the last state
not sure what is better
probably doesn't matter at all does it
it probably does in some cases
because of setup and hold times
right, so you just have to predict the future correctly to avoid the setup times
sounds like branch prediction to me ;)
(or not use step/dir)
up/down would be so much better ;)
quadrature is better still - noise resistant
yeah, but harder to interpret
harder = more expensive
yes a little
another flood :)
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-6.06/etc/modprobe.d/emc2: ubuntu 6.10 specifics backport
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-6.06/etc/xdg/menus/applications-merged/cnc.menu: ubuntu 6.10 specifics backport
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-6.06/emc2.files: ubuntu 6.10 specifics backport
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-6.06/etc/udev/rules.d/emc2.rules: ubuntu 6.10 specifics backport
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-6.06/usr/share/applications/ (emc2-doc.desktop emc2.desktop): ubuntu 6.10 specifics backport
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-6.06/usr/share/pixmaps/emc2icon.png: ubuntu 6.10 specifics backport
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-6.06/usr/share/desktop-directories/cnc.directory: ubuntu 6.10 specifics backport
cradek: do you have some system around to test the deb building ?
I can get to one
let me get a branch checkout
a breezy would be best
I can probably do that later
no problem .. dapper is fine too
cd emc2_1/debian && ./configure -r
cd emc2_1 && debuild
is debuild the same as dpkg-buildpackage?
similar.. it calls dpkg-buildpackage
and fakeroot, and whatever there is still needed
I only use debuild when I don't want to mess with options passed
you probably need to apt-get it
or use dpkg-buildpackage :)
hmm, it's not running realtime so it won't work now I bet
configure -r tries to build packages for the running kernel
didn't look at sim much though
I'll try sim
you need to make a dir extras-sim-Ubuntu-6.06 and add the needed extra files in there
I didn't get an error about it not being there
sure you won't.. but the packages won't be complete
(no menus & such) ;)
before we give this to anyone else I want to figure out a way we won't need 1:2.1.0 to be newer
have to run... bbl
[15:32:57] <jepler> http://lwn.net/Articles/194664/
2.1.0-alpha0 it's called now
I think if we change it to 2.1.0~alpha0 it will be an older package than 2.1.0 when we release it
I don't know what "dak" is but note it says at the top that apt and dpkg have supported this for a long time
Debian's archive maintenance scripts
03alex_joni 07v2_1_branch * 10emc2/debian/changelog: change to 2.1.0~alpha0 so it'll be older than 2.1.0 as suggested by jepler
[16:01:32] <jepler> http://firstname.lastname@example.org/1994733.html
verified that on 5.10 and 6.06 the ~ works
2.1.0~alpha1 is less than 2.1.0 and 2.1.1. 2.1.1~alpha1 is less than 2.1.1 but greater than 2.1.0.
jepler: did you just the pluto powered up - and it was getting that hot?
sorry typing with one hand. (eating with the other).. Did you just have the pluto powered up and it was getting that hot? (nothing hooked to it)
skunkworks: after I upload the pluto_servo hardware it gets hot
it stays hot if I disconnect the parallel port without resetting it
it stays cool if it is connected to the parallel port but unprogrammed
wow - your program is hot ;)
I guess so
probably the hardest the pluto has had to work. Most people must just blink led's with it.
might have to glue a little heat sink on the top?
reseting it resets it to 192.168.1.1
going back to working on trim.
jepler: where you able to find a reset solution for the pluto? I see no one has replied to your post.
Oops,,,it crashed sitting here.....
might be the power saving stuff...although that worked fine on the 5.10 ...
The screen saver was locked up, and it did not respond to anything...
screen savers are bad
especially if they use 3D calcs.. it really stresses the processor
but no matter what, it shouldn't crash
I would advise either deactivating it, or uninstalling the 3d screensavers
jepler: I know.. :/
Yep...had the planetary gear saver running....Or Prius PSD as some people call it...
Might have overheated the video board ?
only the processor maybe
what kind of video board do you have?
Slip the box covers back on for now, to get the air flow back to the design..
Welp, looked at the processor temp with the Bios...31 C ....
if it's ati or nvidia, it won't be using the 3D of the board (the open source drivers don't have that)
Its a Radeon, older model, just has a short heat sinc, no fan, but there is no card nex
t to it ...
The board sits invervted in the slot though...so not optimum for convective flow...
PC fan is running, and I have an extra fan besides the PS fan...
I meant to say CPU fan...
let it run somemore,,,see what happens...
does anyone need a dapper vmdk ?
jmkasunich: How soon are you placing the order for the mesa stuff?
definitely before the end of the year
I'm thinking about going ahead and getting the m5i20 for the lathe spindle encoder
I guess we should get together a list of the 5i20 cards and peripheral cards people want
give me a few minutes
yeah - I'd been thinking the pluto might do well, but with the overheating thing...
overheating is possible with any FPGA - you can do that programmatically ;)
but the Pluto is definitely not an industrial design
Well, my Taig is being used as a hobby lathe, but i know a guy who'd like to do a plasma cutter...
it would be good to get experience with the card we'd use for that.
I like the Mesa card, though it's still not perfect
I'd like to see more protection on the I/Os (though you can get that with the add-on cards)
SWPadnos: imo, putting the protection outside is better anyway
you can mix and match to suit the need
true, but there isn't even any static protection on the I/Os, and there's no fuse on the 5V / 3.3V lines, which go directly to the PC power supply ...
oh, that.... I thought you meant isolation, etc
there's one PTC onthe card, near the main 5V/3.3V jumper
hmmm - red herring - the PTC happens to be near the P4 jumper
I suppose you tack solder a leaded pico-fuse to the voltage select jumper pins...
hmmm - interesting product idea, that is. fuse/jumpers ...
.ss LR shells!!!
but only old copper ones
Nah, the shells work better, teaches the dumbass not to use a substitue fuse
.22s are for 1/4 x 1-1/4 fuses, pennys are for edison (house) fuses
everybody knows that
[19:30:14] <jmkasunich> http://jmkasunich.dyndns.org/cgi-bin/blosxom
* Jymmmmmm has never lived in a house that uses fuses.
well, I'm a young'un - I grew up with circuit breakers, not those funny screw-in glass things
I'm gonna add the normal and 5-piece prices for each board, and a short description of the various 7i ones
or just link to their site (though ti isn't laid out all that well)
man - I'm bummed that I'm not wealthy
there are several new 9MP monitors being sold on eBay - the good ones
but they're $3500-4500 each or thereabouts
Though you know what.... As popular as blade fuses are today, it's very hard to find project fuses holders for them - in blocks, inline are plentiful.
9mp are the big LCDs?
only 22 inches, but 204PPI
oh screw that.... I want to be able to ring the persons neck live and in person!
[19:33:41] <SWPadnos> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=330056858530
you know - people don't really think about it much, but a DVI-D monitor actually has 4 or 7 channels that run at up to 1.65Gbit/second
(single or dual-link, for the 4 vs 7)
so a dual-link DVI cable is almost as fast as 10GigE
how fast is SATA ?
but it can tolerate a bit error rate several orders of magnitude higher
3 Gbits/sec for SATA-2
jmkasunich: you have a dapper vmdk ..right?
no, I'm running the dapper farm "slot" on this machine directly
vmdk = virtual machine for vmware server
DVI is also limited to several meters, instead of 100M
its becoming clear that multiple VMs running at the same time is not pretty
it would be good to be able to detect changes on the host, then make the VMs run in sequence somehow
to much work
I;m not sure if you can just suspend/resume by asking VMWare nicely
so SWPadnos: which I/O board(s) were you interested in?
I'm looking at the mesa site now
I've seen resume and suspend take 30 seconds each, so they're not a cure-all
it's still faster (and less wear and tear) to avoid thrashing
I have to many other projects on my plate right now to mess with the farm as long as its working
steves_logging: you around?
if there are no memory/disk issues, then running in parallel should be about the same time as running in sequence. if there are, then a sequence will finish faster
jmkasunich, I noticed you're cleaning ;)
that's at least two jobs, maybe three (or 4, for me)
does anyone run the xilink dev tools on linux? (8.2i Design evaluation dvd )
hmmm - I think I'd like a 7i37, and I'd love to have one of the RS-422/485 boards for differential inputs (though I'm not sure they match the pinout of the FPGA config)
jepler does (maybe not that version)
and skunkworks and cradek, I think :)
oh wait - I'm wrong
the Pluto is an Altera chip
tomp: I've downloaded the webpak (I think its the same as the DVD), haven't had time to install it yet
the webpack is identical, except that it doesn't support very new, very old, or very large chips
and I think it's missing some optimization options
i bought the dvd,, and ditto on the install issue jmk
does the 5io20 having pwm instead of analog bother you?
it's got a very fast timebase, so filtering should be pretty easy ...
like last nite opamp circuit?
steves_logging is now known as steve_stallings
yes, only smaller values ;)
can be, if you want
so you'd use pwm & filters rather than chg the xilink code
either way would work
hi JMK, was out enjoying the sun, 15 degrees warmer today than Saturday, plus bright sun, down right plesant outside
you'd need hardware and firmware to get real analog
and the 8 encoder channels, what was the original design idea ( with 4 drive ouputs ), was it dual encoder joints? or just spares
either. there;s also an 8-axis version for the FPGA
same 200k gate chip?
nice, somebody there can pack stuff in tight
200k gates is plenty to put a DSP in there (one of their configs has that), plus all the I/O and quadrature/PWM functions
have you looked at the G-Rex from Geckodrive?
heard a bit, didnt look
hi steve_stallings, pretty nice here too
that's got a custom processor, 8 channels of step/dir and quadrature, a bunch of I/Os, and it fits into a 50kgate chip (as far as I can see - I can't find my magnifying glass ;) )
regarding compile farm, couldn't the scripts just be hard coded with delays to stagger the compiles?
lots of things _could_ be done
very few that I'm willing to mess with at the moment
just figured that delays would be simple, says he that has not run VMWare yet
there are already delays - making them *controlled* delays isn't as simple ;)
right now all the VMs are running
sleeping most of the time
they wake up every 5 mins and check for a commit
if there is one they do a cvs up and build
when all are building at once, it gets slow
I thought the VMs were running a script, and the script could determine that a compile was needed, but pause a fixed time bases on slot number
there are loads of ways of doing it (none of which jmk has the time to try right now, I gather :) )
what he said
so steve_stallings, which (if any) breakouts do you want
there are shared folders where the host can place files that the guests look for, and the host can control who compiles that way (place file farm_1_go in the sahred dir, and farm slot 1 starts ...)
SWPadnos: grex impresive ( 400$), 5i20 more impressive (199-150$), how do i sign up to buy one? (my stg isa slot is a hassle )
we may be able to get to 25 of the 7I31's ;)
you can put me down for a 7i33
jmk - add tomp ;)
jmk: please add the pwm3analog 7i33 to my 5i20 and lemme know how to pay
do we have to have multiples of 5 to get the deal?
5 or more
good, i see more already on 5i20
the deal is for each kind of board - we have the 5 needed to get the good price on the 5i20 already
oh - the 7I33 will do fine for me as well - that gives me the differential encoder inputs I want
we need that red thermometer jerry uses on his tc marathon
"ok folks we got 4 now, call us soon"
SWPadnos: So, are you ditching the 7i34?
I'll let a friend of mine know about this - we may be able to get a couple more
jtr, I'm not sure yet
I won't need it for the servos, but I may want to try it out for other projects
as far as how to pay, etc... I'll want an email from each person, with full name, shipping address, and exactly what you want
then you can mail a check
I'll place the order with my credit card as soon as we are all agreed, and ship stuff out as the checks arrive
I'd like to add $10 per person to cover re-shipping
I do think the LED pass-thru boards would be handy, as in Yep, there's the coolant-on signal.
that means the 5i20 is $169 instead of $199
I can't find any info about that board (the LED one) on the website (except for the price)
me either - I looked through the pdf directories and got nada.
maybe a phone call or email to mesa to make sure its not vaporware....
want to see a picture of the 7i31 ?
got 1 in my hands.
I guess its not vaporware ;-)
not much to say about a pass-through with LEDs, I guess ;)
does it have a test header as well (so you can connect a scope)?
Roguish: that pic would be nice
I'll link to it from the group buy page
working on it. low res phone camera.
i'll have a file in a few.... where to put it?
if you have a website, just post it
if not, mail it to me
jmkasunich at att dot net
imagebin.net works also
thats easier probably
50 pin ribbon to screw breakout boards: http://www.winfordeng.com/products/brk2x25.php
they have a 10+ discount, but its only $3, probably not worth it
jmk: depending on how you ship, maybe people oughta include a phone #
and I just realized that people shouldn't send checks until I place the order
any shipping charge from mesa to me needs to be divvied up
I think their pricing actually included ground shipping (I seem to remember not getting charged for shipping when I ordered my 5i20)
file commin' at ya.
SWPadnos: good, that simplifies things
on the winford, i needed other than 50pin hdr, and just inside the hdr is 1 more set of holes just like the 50 pin, i removed the 50, added my 2, cheap
those breakout boards are needed at 3:1 to the 5i20's, since there are 3 headers per card
yeah, unless you are doing something different
well - up to 3:1
some of the boards, like the stepper and H bridge, have dedicated connectors, motors connect direct
sure - the 7i33 changes that, as well as some of the others, but moany of the interface cards also end in 50-pin headers
err - the 7i33 also has a 50-pin output header :)
looks like the 7i31 has buffers to drive the LEDs, thats nice
no loading on the I/O signals
regarding the isolator card, the 7i37 does match the current 5i20 headers j3 and j4. with the way peterv wrote the emc code the first header, j2, does mate up exactly.
hmmm - I'd like to charge my stuff - can you take a credit card (or should I place the order)?
I can't take credit cards
there's simliar breakout now that you have volume, from niu mechatronics lab (was for 3x50 on stg)
ok. I can, via PayPal (though there's a fee)
and i just found one from clemson,, with artwork http://www.ece.clemson.edu/crb/ece496/stg_breakout_inst.htm
i got the other url downstairs..
SWPadnos: should I keep the list and let you handle the rest? or should I mail you the webpage and let you take over completely?
certainly the final emailed orders need to go to the guy whose credit card is getting charged
umm - maybe mail me when it's all settled. I'll be travelling for a week, starting in a couple of days
I'll put in some info about the fact that you will place the order, and can take paypal
ok. I'll check the rate for CC charges,I think it's ~3.5 or 4%
what is the paypal premium? 3%, 5%, something like that?
something like that
I think the fee is 2.9% + $0.30
[20:30:35] <tomp> http://www.mech.northwestern.edu/courses/433/ServoToGo/BreakoutBoard/ad/
I have a business account
7i31 pic linked
cool - it looks like the headers on the right are suitable for scope use
which? nwu or clemson
I think that is specific to the STG pinout tho
brb.. gotta pack (again)
yeh, but the board layout is simple to change
I thikn I also get some small discount on FedEx and UPS (I have accounts anyway)
no free shipping for that amount of boards/money?
to me maybe, from me - no! ;)
If this was happening closer to cabin fever, some could be "delivered" there.
SWPadnos: should we try to put a time table down?
wonder if thalx is going this year.
everybody in by such-and-uich a date?
some kind of timetable would be good
if your are gonna be gone for a week starting in a few days, that means you get back around what, 12/20 maybe?
for you budget minded: http://www.daqstuff.com/50_pin_breakout_500013.htm
the screw terminals are not as convenient as the Winford though.
but half the price.
the first row of screw terminals would be a bear to use
and murphy's law says that the row with the signals - the other row will be the grounds ;-(
yes they are. front accessible row is odd numbered, back bad row is even numbered.
rotate the cable
daqstuff is here in chicago ( naperville) they dont have a building, only web orders :( (the ni breakout is close)
back bad row is even - then murphy loses
the 5i20 pinout is even grounded, odd signal
dunno about the pinouts of the various interface boards
unless you come out of the 7i33. it's mixed.
gotta go, bye
well, my cheapskate friend doesn't want any Mesa stuff right now :(
SWPadnos: I don't want to sign you up for any more work, so I'm asking first... should we announce this on the emc-users list? or better to keep it smaller and more managable?
when it was me and checks, I wanted to keep it to people I know
I was thinking about an announcement, actually.
with paypal, the people can pay before you place the order
to minimize your risk
I'm gonna add links to each interface board and a few other details to that page
I don't mind reshipping to a larger number of people
and I'm willing to accept emails from interested parties, and add them to the table
I may give Mesa a call on Monday and see how much splitting up they're willing to do
those rates may be for "EAU", so they could be used to doing multiple shipments
estimated annual usage
if we were talking eau of hundreds, and monthly ships of 10s, then yes
but for total of 5-10, they're not gonna want to ship onesies
especially if the price includes shipping, that's probably one reason for the nice 5+ discount, because the ship in one box
no, though there will be ~25 or more total units
who wants to play guess?
you have to guess
guess how many users exist at linuxcnc.org
heh, I'm closer than SWP!
I wonder if I ever signed up ;)
SWPadnos: if you post the group buy, use this link: http://jmkasunich.dyndns.org/cgi-bin/blosxom/shoptask/m5i20-group-buy.html
ok - that's the permalink?
the 7i33 manual mentions two different versions - one with a female connector on the bottom of the board,
there's only one on the price list
the other with a male connector on the top. Guess the idea is to mount the 7i33 on the 5i20.
jtr: that's probably for use with their PC-104 stuff
i noticed that. may be the same price. I'd rather use the ribbon cable.
hadn't thought about the PC-104
you still get a ribbon on the output connector
that ain't bad of a price...
even at $200.
mind you a G101 is $400
err... that 5i20
the breakout board version is consdierably less
of the G-Rex
yep. I like the single-wide PCI card. gonna take enough room with the ribbons hanging off it.
ribbons take little room... if you handle them correctly. i.e. FLAT.
the G-Rex G101 is $208 in singles, $191 at qty. 10
the G100 is $399 in any quantity
wasn't it that G100 then that was around 400?
there are two versions of the G10x
5i20 with breakout boards is gonna be in the $300+ range, depending on the breakouts you use
[21:23:24] <A-L-P-H-A> http://geckodrive.com/product.cfm?pid=19
the main difference is if you want USB vs. PCI
and of course if you want EMC support ;-)
[21:23:43] <SWPadnos> http://geckodrive.com/product.cfm?pid=17
and less lag for servo update cycles ...
for me the clincher is the open architecture and ability to reprogram the FPGA
yeah - that's possible for the G-Rex, but not as easy
for me, the clincher is the USB interface, or the need to make my own processor board that can do RTNet
I tried going to pid=18
skunkworks, you need some MEsa cards
you want a 5i20 don't you?
I read about the deal :)
How much is it down to?
[21:25:43] <jmkasunich> http://jmkasunich.dyndns.org/cgi-bin/blosxom/shoptask/m5i20-group-buy.html
for you, $499
skunkworks: , for you... $2000000
A-L-P-H-A, ID 18 may have been the G102, which is the carrier board for development using the G101
oh... I forgot about that.
SWPadnos: should I put 12/20 as a deadline?
I won't be back in the office until then, so that may be a bit premature
don't want to crash into christmas....
let me find out what their holiday schedule is
Q... so this 5i20, is like a G100 in functionality?
I'll put 12/20 for now, we can always extend it
somewhat, but it ahs more I/O and it's relatively easy to reprogram
also, there's an EMC driver ;)
it doesn't do step/dir at the moment, only PWM for servos
oh. I have no servos... so no use to me atm.
PeteW was going to do a step/dir version "when he has the time", but that hasn't happened yet
couldn't we steal the oss code from gecko?
well, making a step generator isn't all that hard, and the FPGA is reprogrammable, so ...
that is a good deal
there isn't any OSS code for the FPGA
or at least the logic, and port it?
I thought there was...
SWPadnos: I intend to work on step/dir FPGA code too
step generator logic: count+= adder; if count > LIMIT step()
so even if peteV doesn't, EMC _will_ have step/dir for the 5i20 within a few months
so... if that's a promise, where do I sign up?
the emc hal as is has pwm+direction - correct?
[21:31:17] <SWPadnos> http://jmkasunich.dyndns.org/cgi-bin/blosxom/shoptask/m5i20-group-buy.html
how do I pay?
I can take credit cards via PayPal
there will be a small surcharge though
SWPadnos: tell me the total to ship it to canuckia as well...
that would depend on what you want, I imagine
it's a PITA to do that though, with the customs delcarations and all ;)
I'm "not" looking at the $45 breakout board...
you'll have to buy at Rodney's to get me to do that for you ;)
but I don't know what these other boards do.
I'm assuming this will give me better then the 26000 period time I got now, right?
or is that still dependant on my machine.
how about 33 or so for an effective BASE_PERIOD?
A-L-P-H-A: you eliminate the "BASE_PERIOD", and let the FPGA (which has its own clock) generate the step pulses
A-L-P-H-A: this is hardware step generation
ie, WAAAAAAAAAYYYYY faster ;)
but you must understand that step/dir support is a few months away, not as soon as you get your board
Jmk yes. I understand that.
hey - why isn't cradek on the list?
ok, if you can live with that, and swp is willing to ship to .ca, then I can add you right now
okay... I'm up for it... what other recommendations are there to get at this same time?
I/d suggest one 7i37 as well, for just about everyone getting a 5i20
for people who want/may want servos, the 7i33
for step-dir, you just need a breakout board (order yourself, from winford or the other supplier, see the bottom of the page)
and if you want to be able to debug easily, a 7i31 or two
if you want limit switches etc, with optical isolation, get a 7i37
7i31 is a luxury item, not needed for the base machine
is the 7i37 a 50-pin ribbon cable on its isolated side too?
the aux IO header would be best with a 7i37, remember that there are 3 I/OO headers on the card
A-L-P-H-A: I just added you, only the 5i20 so far, lemme know if you want any of the interface boards
I could fairly easily make these board myself at significantly cheaper costs... but at the cost of my own time.
jmkasunich, can you add 3x 7i31 and a 7i37 for me?
A-L-P-H-A: we were looking a digikey last night - for the plain breakout boards (the d-500013), you probably can't buy the parts for the selling price
(unless you have a surplus source)
note that we're not doing a buy of those, you can order direct
one more 31 and one more 33, and we get the 5+ price
$7 off the 31 and $17 off the 33
well, we have the 5+ price, I'll just keep the fifth one ;)
20% discound meanswe get a freebie for buying 4
SWPadnos: would you be comfortable mailing it at surplus made in USA electronics, valued at $20?
jmkasunich: please put me down as definite for 1 5i20, 1 7i31, 1 7i33, and 1 7i37
it's near Christmas, I can send you a gift ;)
but you have to pay me for it ;)
in hugs and kisses right?
err - no
SWPadnos: you just lost your free 31
not pizza either
* Jymmmmmm secretly hands SWPadnos the lump of coal to put in A-L-P-H-A's stocking
add me in on a 7i33 as well.
just incase the future I go servo.
but since you don't know what he wants for Christmas, you'll have to send him a gift certificate.
Jymmmmmm, - where's your participation in all this, hmmmm?
jtr, money order is better... I can cash it and spend it anywhere, as opposed to just whereever the issuing place is.
SWPadnos: I know you guys are looking at mesa's, but nfc what for
I'll take a 7i37 too
servo, or (in a few months) hardware step/dir
fast / smooth step generation, servo control, quadrature input, and lots of I/O
SWPadnos: I'm stepper
makes for fast and accurate machines ...
mesa is sure gonna be happy for christmas :)
SWPadnos but jmkasunich just said a few months
well, limited by the mechanics instead of the computer
buy now, use later!
hurry operators are standing by!
but by the time they invent 1/1280 microstepping it'll be too slow
be the first 20 callers to order, and receive a free gift!
I will let you guys know this week.
ok, we have about a week and a half to settle on the final list
there's 1/10000 microstepping already ...
I think some drives can do ~500k steps/rev
stupid, but true
Is this an alternative to Jon's card?
SWPadnos: uh... does it really move?
SWPadnos: huh, and here I thought I was joking
it will be
yes, but significantly more capable than jon's stuff
Jymmmmmm, the FPGA is reprogrammable, so it can be a replacement
the stepgen stuff will be very similar
it's also faster to communicate with it, since it's over the PCI bus instead of a parport
But still SW (emc) dependant or can it still be used as "pass-thru" a la TurboCNC?
it's just I/O, if you want
but he has 8 i/o, this thing can easily do 24 or more I/O with 4 axis of steppers
Jymmmmmm: this is all for EMC
so you can do somewhat faster step generation with it, in software (but not TCNC - no drivers)
it can be used with other stuff, but you are on your own - need to write drivers
okay... that 7i37 is way too expensive in my tastes... I could wire some opto-iso stuff inline on the scsi cable, and have it work. not as clean, but way cheaper.
ok, so no way to make it pass thru (generic programming of FPGA as a paraport I/O) sorta thing?
A-L-P-H-A: you can also use opto-22 stuff (from ebay) for generic I/O
Jymmmmmm, yes, it can be that way, but not as a parport
the ribbon cable pinout is compatible with opto-22 boards
jmk, cool. then why are you buying that? just wondering.
that being the 7i37
Jymmmmmm, you can use all 72 I/Os as generic I/O if you want, and emc can generate steps on them just like a parport
because for $55 its convenient
SWPadnos: Hmmmm, I just use TurboCNC to cya, nuttin agianest emc
but that is a rather expensive way to get 72 I/O
oh - I thought you had upgraded
jmkasunich: fair enough. put me down on one of those 7i37... I don't have to build it, or anything fancy.
I guess you have a week to change your mind
SWPadnos: I did (sorta), Hunted down and bought a new machien with no on-board crap and got 18000ns out of it, but that was JSUT before all this home shit started and haven't had a chance to test it.
well... if I consider my time to be worth $20/hr... that's an hour buying the parts, hour + to make it in eagle, 2hrs to etch, drill, solder the whole thing for a one off... doesn't make sense.
damn - no free 7i37 :(
still less then $300... and I'll have something as useful as a G100, for lots less.
so, we are at the 5 piece price for the 31, 33, and 37
(and the 5i20 of course)
SWPadnos: ? FREE?
and close to the 25 piece price if they combine SKUs
SWPadnos: and if the Xylotex still keeps stalling on me, I need to buy geckos (if the issues isn't my steppers)
A-L-P-H-A, the price for 4 is the same as the price for 5, so if there are 4orders, I can have the freebie ;)
jmkasunich: how's your spam filter? :)
SWPadnos: hahaha. punk.
thought you were going to be the guy that orgasnizes weekend trips... so you can go free.
which you were.
alex_joni: you mean with my email on the sidebar?
I thought about obfuscating it, but I noticed that jepler and cradek publish theirs on their blogs
alex_joni: the spam filter in thunderbird is pretty good actually.
I haven't got any in my inbox for a few days now... I had to retrain it.
I've been pretty amazed with the new filters my ISP just enabled
my isp does a lot of filtering too
yeah.. most do
only 2-3 spam/day get through, and maybe one false positive per week
textdrive.com is who I'm using... they use something, which I don't think I've turned on.
down from ~75/day before
alex_joni: you joining the group buy?
axis ( the gui) interfaces to hal pins, is that thru python? ( can I write pygtk/pyqt that can talk with hal pins? )
Ok, does anyone have a workning prototype of the mesa in use already?
I have one that's not in use
I think Roguish has one, don't know if he's using it yet
petev has one (or works on a machine with one)
I think Ray may have one, but I'm not sure
A-L-P-H-A: no, too much shipping :)
cheap at any price ;)
What's Chris using on his machine paraport?
shouldn't be more than ordering direct ...
yeah, just plain parport
or the pluto-p
I think he's getting a Pluto
I have the x and y on a test machine, with the z motor clamped to the bench. 3 yaskawa drives and motors, tuned and running.
alex_joni: .ro was it?
Roguish: with the Mesa?
alex_joni, you can pick up your stuff at Fest ;)
yes, mesa 5i20.
yaskawa are nice, i have 3 400watters
gotta run - later
where's the fest going to be this year?
someone say NY... or Detroit.
same as last year
7i33 & 7i37
how much is the cheapest and have many I/O ?
maybe 7 hrs from detroit
maybe I'll go this year... show my ugly mug.
* alex_joni heads to bed.. going on a trip tomorrow
see you Alex
5-6hrs to Detroit for me.
you're about 5-8 hours closer than me
I'm in Toronto.
I'm in Vermont
30 minutes east of downtown TO.
15 minutes east of Lake Champlain ;)
what's the IL location? going to map it.
last year, it was about a 22-hour drive each way
[22:02:42] <SWPadnos> http://www.cnc-workshop.com/
cameron, il really but mapquest wont find it
tells me 13hrs 29 minutes.
next door to sunny Galesburg
ok - that's not bad. it's about 1180 miles for me :)
I may fly this time
who lives in Lansing? or Near Langing?
I'll car pool with them there?
I sure as heck am not carrying as much junk^high quality product with me this time :)
hmmm - I don't think I can take enough computers on a plane though. damn
jepler: axis ( the gui) interfaces to hal pins, is that thru python? ( can I write pygtk/pyqt that can talk with hal pins? )
SWPadnos: you flying?
tomp: in emc2.1, there's a Python module called "hal" which any python program can import and use.
I thought I might, but I probably will want to take more stuff than flights would allow
jepler: outrageous! thank you
tomp: are you using the CVS version or are you using 2.0?
cvs, but 2.1 doesnt ring a bell... need to check the downstairs system ( it's only a cpl weeks old checkout of head)
tomp: OK, HEAD a couple weeks ago will do
tomp: Take a look at src/emc/usr_intf/axis/scripts/hal_manualtoolchange.py and you'll see one simple use of the python 'hal' module
tomp: it creates a "manual tool change" component which pops up a window when it's time for the user to change tools
jepler: even better, thanks, will leave this evening for exploring that
h = hal.component("hal_manualtoolchange")
h.newpin("number", hal.HAL_S32, hal.HAL_IN)
h.newpin("change", hal.HAL_BIT, hal.HAL_IN)
h.newpin("changed", hal.HAL_BIT, hal.HAL_OUT)
creating a component can be as easy as this ^^^
then refer to h.number, h.change, etc -- read or write them, as appropriate
so an analog dashboard with a few knobs and sliders, that can be on the desktop with emc2, is possible
(after i get range widgets working with real io)
tomp: yes, if you've already found that halvcp won't meet your needs then this might
halvcp has no floats except spin_button, while gtk & qt have sliders and pots, or i can try to put them inside hal_vcp
i just thought a scriptde language would be easier for me
tomp: I tend to agree with you on that
i gotta copy down your example & get to work :)
tomp: that code i pasted above is from hal_manualtoolchange.py
so just write down the filename
dont know if this is handy , but i used a cfg to describe the type of toolchanger on opther systems,
and if user had none, he could cfg the system to pause on a toolchange
and wait for him to do it
either no action, manual action or several types of toolchangers could be defined behing the tool change command
that's what hal_manualtoolchange does
cool, will dumop it to lpt now
since emc will pause until the "changed" pin becomes true, if you do anything other than connect change (out) -> changed (in), it'll wait
(from the motion controller ...)
this hal stuff is neat, so the toolchanger can have a dummy loopback :)
yep - that's the default, I think (for the sample configs)
bye again, thanks again
or you can run it onto CL for hardware toolchangers, like on the mazak
is there an emc mailing list I should be aware of?
I thought everybody knew about the mailing lists
infin in a sec
mailing list info available from here: http://linuxcnc.org/index.php?option=com_content&task=view&id=4&Itemid=8&lang=en
there's another online archive of the lists here: http://dir.gmane.org/index.php?prefix=gmane.linux.distributions.emc
I'm hungry... but definately not pizza today
I finished that whole pizza yesterday
dinners almost done here
N'Awlins style red beans and rice - been simmering all afternoon
I've just started to smell the beef stew/soup my wife is making
SWPadnos: Don't take this personally, but.... STFU!
I just "tested" the red beans... nice and spicy
rice is cooking now
Has anyone actually talked to mesa folks yet?
the company I mean
why do you ask?
SWPadnos what did they say? Or did you just talk about ordering thier products?
I'll be callingthem about htis group buy tomorrow as well
I talked to them a long time ago, not about this stuff yet
Well, if the emc devleopers are willing to support their product, there's a good chance of getting a running discount as it would promote their produtcs.
EMC would add a small percentage to their sales - I'm not sure they'll give us a running discount
With as many hobbiest getting into this, it wouldn't be a bad pitch to sell to them.
I'll ask about that kind of thing, but I'm pretty sure I already know the answer ;)
Hell, they could sell to all the sherline base as well.
and that has to be at least 20,000
not really - their FPGA card is ~10% of the cost of the complete sherline CNC mill kit
I think the sherlines are $2500 complete
I mean for those that want to upgrade (so to speak)
They may want indistial sales, but hobbiest is getting BIG these days.
well, could be, but I'm not optimistic :)
I'm 40 minutes from them.
hobbyists are also a PITA - lots of questions and almost no repeat sales ...
lots of oopsies without admiting it was them
Sure, but they aren't providing support for it either
for an industrial controls company, hobbyists are often a losing proposition
eh, sales are sales in these times.
support is also support ...
and not necessarily included
there is PAID support you know.
it is for most companies, which is why they like bulk customers who are educated instead of one-off customers who need lots of hand-holding
SWPadnos: how farmiliar are you with the mesa code?
(not meaning to be too harsh here, but there are economic realities to deal with)
Right, but discount could mean no support (other than RMA)
I've looked it over, but haven't recompiled it
I remember you saying you had a 8 axis version of it at one time?
was that you?
I need to install the Xilinx toolset (I have a license), but haven't decided exactly where to do it
yes, I have an 8 axis version, in addition to some bugfixes and pin changes
but I can't verify it, so I haven't checked it in
(they're from Pete Wallace of Mesa, via Pete V)
my question is - how hard would it be to change the code to be up/down instead
not too hard
(not that it is a big deal)
since there are so many peripherals, they use an internal data bus, but jepler's code could be used with a little massaging
or a full stepgen, complete with multiple step types, since there are zillions of gates
it it an acex fpga?
no, but VHDL is a generic FPGA language
actually, it's a generic logic language
the pluto could be downgraded to a lathe :)
with 72 pins - the 8 axis ver would be nice I bet. 8 encoder inputs would come in handy when you start thinking about spindle and table rotaion
there are 8 encoder inputs with the 4 axis version
time to eat - bbiab
oh - never mind then
the pluto's chip is "10,000 gates" and the 5i20 is "200,000 gates"
so a lot more stuff should fit
jepler: I hope so
it would be nice to have GPL'ed "chunks" so you could mix and match
jepler: your not too impressed with the pluto?
one guy wants 3 step/dir for the axis, and two encoders for spindle and jogwheel
another wants PWM and more encoders
skunkworks: it's under-engineered
jmkasunich: yeah, if you can fit in all the functions you want (e.g., 8x PWM, 8x stepgen, 8x quadrature counter) and do the I/O configuration flexibly .. otherwise, you're stuck using their software to "re-link", or trying to guess at the common combinations and include a bunch of firmwares
jepler: I'm not communicating clearly
I was specifically thinking of _not_ being able to fit everything in there, so you'd want to mix and match
this would NOT be joe user stuff, it would require doing an FPGA config
that is not nearly as tough as what I was contemplating
but if there was a sort of "bus" in the FPGA, and a standardized interface between the chunks and the bus, you could avoid doing everything from scratch just because you need one more encoder and one less PWM
even nicer if the HAL driver could read the bus and discover what you have, but I'm not sure I'm willing to dedicate much FPGA resources to that
maybe tonight I should see about installing the xilinx software
I upgraded to EMC2 2.05 and am still using 2.6.15-magma and now I am getting real time errors
can you paste the message you are seeing?
sure one sec
(here if its short, pastebin.ca if its long)
[ 497.738217] In recent history there were
[ 497.738220] 1698308, 1694840, 1693890, 1701662, and 1692920
[ 497.738233] elapsed clocks between calls to the motion controller.
[ 497.738236] This time, there were 2752224 which is so anomolously
[ 497.738239] large that it probably signifies a problem with your
[ 497.738242] realtime configuration. For the rest of this run of
[ 497.738245] EMC, this message will be suppressed.
rebooted and same thing
* skunkworks raises hand :)
I think the version you are running now does a real time latency test. have you ever run a latency test on your computer?
yes, I had problems with a computer with built in video and ran the test on the ccomputer I am now using
And the test was ok?
yep, and I've had no problems until now
Do you have a usb drive connected now?
nope, scsi only
I mean like a key chain drive (thumb drive) or the like
onboard video, or separate video drive?
separate video card I mean
I understood, it is agp
were you doing anything wehn that message popped up?
dragging around windows, stuff like that?
I should run the real time test again, If i can find the link with the details
machine is dedicated to use as cnc controller
[23:42:29] <skunkworks> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TroubleShooting
jmkasunich: I don't know a lot about either one, but your idea sounds a little like opencores.org and the wishbone interconnect .. http://www.opencores.org/browse.cgi/by_category http://www.opencores.org/projects.cgi/web/wishbone/wishbone
lat max is bouncing some, but it is never more than 60 % of BASE_PERIOD
try beating on the machine while the test runs - drag windows around, browse the web, thrash the disk, etc
* Jymmmmmm grabs the baseball bat
Heh, run glgears =)
jepler: those projects go well beyond what I had in mind
not doing anything just watching, twice lat max has spit out huge numbers, like 30 * base period
thats not good
doesn't correspond to anything you can see happening?
kdibble: How many on-bord things does this mobo have?
not doing anything now, I'm ssh-d into the box
laptop fans turning on and off, for example (don't think this is a laptop...)
Intellistation Xeon 1.7, on board scsi, AGP video
kdibble you using the SCSI ?
all disks are scsi
cd's are ide
What *I* foudn is that mobo's w/o on-board do much better - if you can find em.
all was well until today
when I upgraded
what version did you upgrade from?
I'm not sure what version got the latency detector
you might have been having the problems before, but they weren't being detected
I thought the detector was in 2.0.3 as well
the wiki will say
no, I was wrong
"Starting in EMC2 version 2.0.4, the number of CPU cycles between invocations of the real-time motion thread is tracked."
great, so I guess I've had a hardware problem for a while and it is just now getting caught
talk about mixed feelings... :-(
could it be gnome, polling for a cd or usb insertion?
I wouldn't expect that, but who knows
the things that traditionally cause trouble are shared video memory, DMA, and power management
no power management, Xeons dont support it
you might want to try disabling some power management stuff in the BIOS
I'm only using scsi so no dma
AGP video card
no auto-reduction of clock rates, etc
thats the three biggiies
no, Xeons only run wide open as far as I know
I don't think any of us have xeons
you might want to report the latency test results to rtai.org (they have a mailing list)
the EMC specific stuff won't be of interest to them
but they have a wider range of experience with differnet hardware, maybe they know something about Xeons