#emc | Logs for 2009-01-27

Back
[00:02:58] <ds3> guess the real question is does the temperature input have to be provided in "real time" or can it go through a non RT subsystem like I2C
[00:05:44] <PCW> dmess: easiest is probably to use a TC--> V-F converter so HAL can just count frequency
[00:06:16] <PCW> (unless you need high speed)
[00:08:27] <ds3> Nope, just want to keep a little pot of platic at the right temperature
[00:09:09] <ds3> I assume HAL or some component can map things with just a mere table?
[00:09:48] <ds3> in that case, I could use a thermistor
[00:10:22] <PCW> Yep avoids cold-junction comp
[00:10:48] <ds3> exactly
[00:11:24] <ds3> does HAL care if the mapping from freq to temperature is nonlinear?
[00:11:45] <ds3> a 555 seems to be a simple way
[00:12:04] <PCW> Yep, that would probably do
[00:32:02] <jmkasunich> ds3: you could write a simple hal comp to linearize it
[00:32:07] <jmkasunich> hal doesn't care, its just a signa;
[00:32:10] <jmkasunich> signal
[00:32:21] <jmkasunich> but if you want the answer in degrees, you'll need to convert it
[00:32:39] <ds3> hal comp == kernel module or some userspace stuff?
[00:33:02] <jmkasunich> how often do you want to sample? do you care about delays in the 1/100th to 1/2 second range?
[00:33:14] <ds3> it would be so nice to not have put together yet another microcontroller to handle temperaturecontrol and another one to handle openloop motor speed
[00:33:27] <ds3> if it can update once a second, it should be fine, I think
[00:33:46] <jmkasunich> one possibility: use two software pwmgens
[00:33:50] <jmkasunich> one for the motor
[00:34:12] <jmkasunich> the other one you filter heavily and then compare it to the temp signal, bring one bit back in
[00:34:13] <ds3> trying to build a 3D printer loosely modeled after the Reprap design but using EMC to control
[00:34:40] <jmkasunich> write a simple comp (I'd do it in RT, but you don't have to) that changes the pwm out so it "hunts" around the actual value
[00:34:40] <ds3> Hmmm that sounds complicated, I might have come back and bug people on that
[00:35:19] <jmkasunich> ds3: the approach I was talking about is (not very) complex software that lets use use the simplest possible hardware
[00:35:30] <jmkasunich> no A/D converter or anything
[00:35:55] <ds3> jmkasunich: I don't have a full dev system setup so i am trying to avoid redoing kernel modules
[00:36:06] <jmkasunich> V/F converter + software counter will also work, might be more (or less) accurate depending on the V/F
[00:36:09] <jmkasunich> ah
[00:36:21] <jmkasunich> so you want a completely "c-less" solution
[00:36:40] <jmkasunich> wuss ;-)
[00:36:48] <ds3> just a solution that doesn't require all the headers
[00:36:53] <ds3> I can compile C just fine
[00:37:29] <jmkasunich> I've never tried to compile a user space hal component on a system that didn't have the proper HAL headers
[00:37:42] <jmkasunich> it works for sim, but that is a special case
[00:37:46] <ds3> for what you were proposing, what hardware do I need?
[00:38:10] <jmkasunich> something that gives a reasonable analog voltage range over the temperature range of interest, and a comparator
[00:38:47] <jmkasunich> might be as simple as a thermistor, a fixed resistor, and a $0.50 8-pin dip
[00:38:55] <ds3> something like a thermistor wired as a voltage divider feeding into a comparator along with an output from EMC integrated and feeding into the same comparator?
[00:39:01] <jmkasunich> yes
[00:39:20] <ds3> the integration would be my worry
[00:39:34] <jmkasunich> tracking A/D - if the EMC output is too low, add 1, if too high, subtract 1, you wind up hunting around the actual value
[00:39:43] <jmkasunich> not true integration
[00:39:45] <jmkasunich> just filter
[00:39:48] <jmkasunich> passive RC would work
[00:40:09] <ds3> not very good at figuring out required RC constants
[00:40:51] <jmkasunich> 1uF and 100K
[00:41:12] <jmkasunich> plus or minus a factor of 10 would probably work, this isn't rocket science
[00:41:30] <ds3> isn't there a min. value or odd things start happening?
[00:41:50] <jmkasunich> yes, you want the time constant to be 10-100 times the PWM frequency
[00:41:53] <jmkasunich> actually, you
[00:41:59] <jmkasunich> you'd use PDM, not PWM
[00:42:03] <jmkasunich> emc's pwmgen can do both
[00:42:30] <ds3> hmmm I can deal with this...just got to confirm the thermistor I have in mind won't melt at the temps I want to run it in
[00:42:37] <jmkasunich> how hot?
[00:42:46] <jmkasunich> many thermistors top out at 150C or so
[00:43:03] <jmkasunich> thermocouples go much higher, but they give tiny signals and need amplification
[00:43:10] <ds3> the heater tops out at about 210C... but I am dealing with ABS so it'll probally be around 120C
[00:43:27] <ds3> I know... thermocouple requires more work and cold junction compensation
[00:45:21] <ds3> another possiblity is I can put a glass packaged 1N914 on there as a sensor
[00:46:18] <jmkasunich> that will also be a millivolt level signal
[00:46:39] <jmkasunich> nothing an op-amp can't solve, but it is more complex
[00:48:40] <jmkasunich> what are you actually measuring (physically)? do you want an axial part, probe, SMT, etc?
[00:49:07] <ds3> let me find a picture that approximately shows it
[00:50:24] <ds3> something like this:
[00:50:25] <ds3> http://exmrclean.blogspot.com/2009/01/power-resistor-heater-experiment.html
[00:50:40] <ds3> I have a slightly different design but same idea
[00:53:48] <jmkasunich> glass DO-35 NTC thermistors rated to 300C are just over $1 at digikey
[00:54:10] <ds3> Hmmm
[00:54:24] <ds3> just need to find $24 more things to buy
[00:54:56] <jmkasunich> they also have bead style, same temp range, ~$5
[00:55:09] <jmkasunich> a bit handier since both leads come out the same end
[00:55:35] <ds3> I am tempted to try this thing I salved from a BBQ thermometer
[00:55:46] <ds3> salvaged
[00:55:52] <jmkasunich> that sounds like a plan
[00:56:20] <jmkasunich> although if it is the kind that gets stuck into a lump of food, it probably isn't rated much over 100C
[00:56:33] <jmkasunich> at 100C internal temp, food becomes rather crunchy
[00:56:49] <ds3> oh, doh
[00:57:00] <ds3> well, it was a dollar store item... not going to worry if it frys
[00:57:27] <ds3> it is in a metal package unlike the epoxy resin ones I have before
[01:02:53] <jmk-robot> hello from mini-ITX land
[01:17:46] <eric_unterhausen> one of my neighbors has an open router that has 100% signal strength, mine has barely over 50%
[01:21:36] <jmk-robot> yours in the basement or something?
[01:22:26] <eric_unterhausen> I had the cable modem put in a downstairs room at the end of the house, I'm upstairs, other end of the house
[01:22:36] <skunkworks> jmk-robot: must have gotten your os installed :)
[01:22:48] <jmk-robot> yeah, running hardy at the moment
[01:22:58] <jmk-robot> emc2 is installed, but I didn't boot tho the rt kernel yet
[01:23:04] <jmk-robot> installing build-deps for emc2
[01:23:14] <skunkworks> going to use a lot of hal?
[01:23:20] <jmk-robot> a bit
[01:23:27] <jmk-robot> pwmgen, encoder, pid
[01:23:29] <skunkworks> neat
[01:24:37] <SWPadnos> jmk-robot, how does the performance look compared to your desktop?
[01:24:54] <jmk-robot> hard to quantify yet
[01:25:18] <SWPadnos> have you tried your image processing stuff on it yet?
[01:25:23] <jmk-robot> it is thrashing the disk quite a bit (I'm using a laptop drive), but that is to be expected when installing packages
[01:25:26] <jmk-robot> not yet
[01:25:31] <SWPadnos> ah, ok
[01:25:45] <SWPadnos> what's your desktop?
[01:26:05] <jmk-robot> I'm using the full blown hardy desktop (installed from the alternate CD tho)
[01:26:21] <SWPadnos> err - what's the CPU in your desktop? :)
[01:26:26] <jmk-robot> oh
[01:26:32] <jmk-robot> core 2 duo, E6600
[01:26:36] <SWPadnos> ok
[01:26:38] <jmk-robot> 4G ram
[01:26:54] <SWPadnos> just so I have a reference when you tell us how much slower the atom is :)
[01:26:56] <jmk-robot> compiling EMC2 should be a nice speed test
[01:27:09] <jmk-robot> after I finish with packages, and reboot into RT
[01:27:38] <jmk-robot> I'm afraid that might run into the "that driver isn't part of the RT kernel" problem tho
[01:27:41] <SWPadnos> that's what I keep forgetting to do
[01:28:13] <SWPadnos> I downloaed the Jaunty Beta 3 ISO, and I keep forgetting to get a CDR to write it to
[01:28:16] <eric_unterhausen> is there a kernel config with the rt-kernel?
[01:28:25] <SWPadnos> alpha 3 I guess
[01:30:08] <jmk-robot> jmkasunich@robot:/boot$ ls *-rtai
[01:30:08] <jmk-robot> config-2.6.24-16-rtai System.map-2.6.24-16-rtai
[01:30:08] <jmk-robot> initrd.img-2.6.24-16-rtai vmlinuz-2.6.24-16-rtai
[01:30:23] <jmk-robot> eric_unterhausen ^^^
[01:31:56] <eric_unterhausen> I figured there was, that should make adding the modules you need fairly easy
[01:32:58] <eric_unterhausen> although last time I had the missing module problem I had to dig rather far into the kernel build menus to fix it
[01:40:04] <jmk-robot> cvs checkouts are starting to take a long time
[01:40:28] <eric_unterhausen> it's that BigJohnT guy, you just can't stop the gcode buddies
[01:40:46] <jmk-robot> whats funny is that he is on dial-up
[01:41:10] <eric_unterhausen> he's writing python all the time, no more tv for him, just python
[01:42:13] <BigJohnT> with a string from my house to the hwy through the woods
[01:43:06] <jmk-robot> geez, still downloading the docs dir
[01:43:22] <eric_unterhausen> I forgot about that, he's been going nuts in there too
[01:44:11] <eric_unterhausen> jmk-robot: is it snowing in Cleveland?
[01:44:34] <jmkasunich> I haven't looked in the last hour or so
[01:44:57] <jmkasunich> it hasn't the last couple of days - just a flurry sunday, 1/4" or so
[01:45:22] <eric_unterhausen> skunkworks lives in Wi, correct?
[01:45:27] <jmkasunich> yeah
[01:45:52] <eric_unterhausen> you are too close to be interesting as far as weather predictions go
[01:46:12] <eric_unterhausen> need to get Roland on here
[01:47:05] <eric_unterhausen> jmkasunich: are you building an autonomous robot?
[01:47:28] <jmkasunich> yeah
[01:47:39] <jmkasunich> gonna try anyway
[01:48:01] <eric_unterhausen> I'm helping to build a miniature segway type robot
[01:48:19] <jmk-robot> miniature? as in not for people to stand on?
[01:48:31] <eric_unterhausen> no, the size of a large grapefruit
[01:48:58] <jmk-robot> dang, still downloading docs
[01:49:03] <jmk-robot> hasn't even gotten to the src dir
[01:51:33] <jmkasunich> yikes, my trunk CVS tree (which includes binaries, etc, since I've built in it) is 118M
[01:52:00] <jmkasunich> 15M in docs, 84M in src, the rest spread around
[01:52:10] <jepler> 102M here
[01:52:41] <jmkasunich> I probably have a lot of fpgs cruft down in there somewhere
[01:52:51] <jmkasunich> I'm keeping your wire busy
[01:53:44] <jmk-robot> this is gonna take a while - bbl
[02:20:39] <BigJohnT> say goodnight Gracie
[02:56:21] <jmk-robot> SWPadnos: you were curious about atom speed
[02:57:15] <jmk-robot> Atom: emc2 sim, make after make clean:
[02:57:15] <jmk-robot> real5m15.576s
[02:57:15] <jmk-robot> user4m17.940s
[02:57:15] <jmk-robot> sys0m27.082s
[02:57:42] <jmkasunich> core 2 duo, 2.4Ghz:
[02:57:46] <jmkasunich> real 2m26.835s
[02:57:46] <jmkasunich> user 1m29.798s
[02:57:46] <jmkasunich> sys 0m15.553s
[02:58:12] <SWPadnos> was the core2 make done with -j?
[02:58:26] <jmkasunich> not bad, for having 2/3 the clock speed, 1/8 the cache, and 1/8 the main memory
[02:58:36] <jmkasunich> no, but cpu load indicators showed it using both cores
[02:58:42] <SWPadnos> strange
[02:59:07] <jmkasunich> unless.... maybe one of jeff's commits triggered some farm activity at the same time
[02:59:11] <jmkasunich> lemme run the core one again
[02:59:13] <SWPadnos> heh
[03:14:28] <eric_unterhausen> i thought make was multithreaded?
[03:14:46] <SWPadnos> not by default
[03:15:19] <SWPadnos> and it's only multithreaded by file - ie, it runs multiple copies of the compiler
[03:40:34] <jmkasunich> I ran make again on the core 2 (with no other significant load, but still single threaded)
[03:40:53] <jmkasunich> Atom: 5:15 real, core2 2:01 real
[03:41:09] <jmkasunich> then I tried make -j, on the core2 it went down to about 1 minute
[03:41:23] <jmkasunich> on the atom it caused a massive swapfest that I still haven't recovered from
[04:01:55] <jmk-robot> but the network works!
[04:02:14] <eric_unterhausen> rt?
[04:02:17] <jmk-robot> yes
[04:02:25] <eric_unterhausen> rebuilt kernel?
[04:02:32] <jmk-robot> no, just our kernel
[04:02:42] <eric_unterhausen> so what seems to be the problem?
[04:02:49] <jmk-robot> no problem so far
[04:03:00] <eric_unterhausen> that people have with the live cd?
[04:03:05] <jmk-robot> no clue
[04:03:10] <jmk-robot> I didn't install from the live CD
[04:03:18] <eric_unterhausen> user has no clue, probably correct :)
[04:03:27] <jmk-robot> I did a stock hardy, then install.sh script to get the RT packages from our repo
[04:03:57] <jmk-robot> the only major weirdness right now is that I have to put the USB stick in to boot
[04:04:15] <eric_unterhausen> ok, that's a downside
[04:04:20] <jmk-robot> grub is using menu.lst from the hard disk, but apparently the MBR isn't right
[04:04:43] <eric_unterhausen> this is for all kernels?
[04:04:44] <jmk-robot> I think since the original boot was from usb, the installer wrote a grub MBR to the stick instead of the disk
[04:04:58] <jmk-robot> its a pre-kernel thing
[04:05:07] <jmk-robot> unique to the weird-ass way I did my install
[04:05:29] <eric_unterhausen> someone told me there are instructions in the ISO for that
[04:05:31] <jmk-robot> I booted from stick, fetched packages from my local server
[04:05:45] <jmk-robot> yeah, I'm sure I can fix it
[04:05:50] <jmk-robot> but not tonight - tired
[04:40:21] <eric_unterhausen> displayport -- proof that stupidity is not dead
[04:43:31] <eric_unterhausen> new dell computers ship with a displayport cable, but no displayport ports on the monitor, and I didn't see it on the comp either
[04:46:12] <eric_unterhausen> SWPadnos: did you buy a power supply?
[04:46:22] <SWPadnos> nope
[04:47:04] <eric_unterhausen> the one in your cart is $80 now?
[04:47:40] <SWPadnos> after rebate, yes
[04:49:54] <eric_unterhausen> http://www.newegg.com/Product/Product.aspx?Item=N82E16817341002
[04:51:55] <eric_unterhausen> they have some reasonable deals on Corsair/Silverstone, but those are top of the line and aren't likely to come down much
[04:52:00] <SWPadnos> 700W, not in the silencer series (though the PowerWhisper Technology may be the same thing)
[04:52:22] <SWPadnos> at least it
[04:52:25] <SWPadnos> it
[04:52:27] <SWPadnos> argh
[04:52:33] <SWPadnos> it's got a blue LED
[04:52:40] <eric_unterhausen> yah, important
[04:53:38] <eric_unterhausen> too bad hdmi cables cost $50 locally
[04:54:11] <SWPadnos> monoprice.com
[04:56:46] <DanielFalck> what's the trick to ignore homing altogether again? 'no home' in the axis ini? I am having a hell of a time finding info on this recent conversation on the mailing list and have a pure sim version and I don't care to home my virtual machine
[04:58:12] <SWPadnos> leave out all the HOME_*_VEL settings in the ini
[04:58:22] <SWPadnos> I think that does it anyway
[04:58:24] <DanielFalck> ok, thanks SWPadnos
[04:59:06] <DanielFalck> maybe I need some tips on better searching. I started with 'home' in my email client (thunderbird) and got a bunch of cruft
[04:59:19] <DanielFalck> I looked in the wiki
[04:59:30] <DanielFalck> found all kinds of info on home switches
[04:59:37] <SWPadnos> maybe it's in the manual?
[04:59:50] <SWPadnos> (I'm not sure really)
[04:59:53] <DanielFalck> maybe I should start there :)
[05:00:05] <SWPadnos> using sim/axis you can just hit home all anyway - there are simulated home switches in that config :)
[05:00:08] <DanielFalck> seems like it was in the emails just a couple weeks ago
[05:00:15] <eric_unterhausen> having to read the manual seems so unfriendly
[05:00:28] <DanielFalck> the manual is nice though
[05:01:10] <DanielFalck> I'm so used to google, I forget that we have a good manual
[14:13:34] <ChanServ> [#emc] "This is the #emc channel - talk related to the Enhanced Machine Controller and general machining. Website: http://www.linuxcnc.org/, wiki at http://wiki.linuxcnc.org/"
[14:40:21] <Odd_Socks> Who is logger_emc ?
[14:40:32] <archivist> a bot
[14:40:50] <SWPadnos> Topic Welcome! EMC (Enhanced Machine Controller) is a linux-based opensource CNC control. | Latest release: EMC 2.2.8 | http://www.linuxcnc.org | http://wiki.linuxcnc.org | Channel logged by logger_emc
[14:40:57] <SWPadnos> note the last bit
[14:41:01] <archivist> logger_emc: bookmark
[14:41:01] <archivist> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-01-27.txt
[14:42:04] <Odd_Socks> Maybe I should read all that stuf at the start ;-}
[14:53:49] <eric_unterhausen> freenode didn't manage to kick IRSeekBot this time
[14:57:17] <eric_unterhausen> IRSeek is a company that produces IRC logs
[15:00:54] <archivist> iirc IRseek has to get permission these days (from a channel owner)
[15:01:25] <eric_unterhausen> until yesterday they were just being kicked from here
[15:04:40] <eric_unterhausen> is emc organized enough to opt in?
[15:05:16] <SWPadnos> yes we are, thank you very much
[15:05:25] <SWPadnos> but we didn't opt in, as far as I know
[15:05:36] <SWPadnos> since we have our own (google-indexed) loggers
[15:06:01] <archivist> In #mysql I know who did the opt-ing
[16:55:23] <eric_unterhausen> so other than Pete, has anyone built an EMC controlled motor drive with a Mesa board?
[16:55:46] <eric_unterhausen> I want to build a drive for my 4 phase brushless spindle
[17:01:27] <skunkworks_> I don't know if anyone has done it.. but jmkasunich would be the one to get ideas from.. He wants to do a 3 phase drive run from mesa
[17:06:55] <SWPadnos> I suspect you'd need to mess with SoftDMC (from Mesa) - I'm not sure that a 1-5 kHz PWM update rate would be sufficient for a vector drive
[17:07:29] <SWPadnos> (or something else that can rotate the drove vector(s) in the FPGA rather than in the PC CPU)
[17:09:46] <jepler> or use jon elson's pwm brushless servo amplifier: http://jelinux.pico-systems.com/acservo.html
[17:09:59] <jepler> well, I guess when you say "build" you mean "build"
[17:10:48] <SWPadnos> bbl
[17:27:19] <eric_unterhausen> jepler: 4 phase
[17:28:26] <eric_unterhausen> I suspect a 1khz update rate is fine, we use a 10khz rate now, but that's for data acquisition as much as anything
[17:29:46] <eric_unterhausen> the joys of finding really nice surplus parts: life is never easy
[17:29:46] <jepler> oh I should have read what you said
[17:32:19] <jepler> without recompiling the fpga firmwares I think you only get 4 PWM signals per connector
[17:33:52] <eric_unterhausen> I only need 4, and their inverse
[17:34:11] <eric_unterhausen> I guess I could invert in logic
[17:40:33] <jepler> looking at the vhdl source, I don't see a mode that gives pwm and pwm-not
[17:41:31] <eric_unterhausen> bummer
[17:41:33] <jepler> the 4 output modes available are: pwm,dir; dir,pwm; pwm,0 or 0,pwm according to dir; pdm,dir
[17:45:24] <eric_unterhausen> so I could make it work with an inverter or vhdl
[17:45:25] <eric_unterhausen> bbl
[18:53:49] <jensor> In the discussion of base period set in the ini file, it gives the example for a base period of 50,000 you get a maximum step rate of 20,000 steps per second. This implies to me that if one uses quadrature drive where you get 4 steps per cycle, that you would end up with a maximum rate of 4 x 20,000 = 80,000 steps per second. Does this sound right?
[18:54:50] <jensor> Or is it only .25 steps per cycle for quacdrature?
[19:02:39] <cradek> with both quadrature and the new doublestep step/dir, you can get one step per base period. previously (before doublestep), it took two base periods to get to get a step with step/dir.
[19:02:58] <LawrenceG> jensor, each base period, you can advance or retard the quadrature state by one count
[19:03:07] <LawrenceG> :}
[19:03:15] <LawrenceG> hi chris
[19:03:18] <cradek> quadrature is still superior though. there are no timing problems with it, like step/dir has
[19:03:21] <cradek> hi
[19:05:24] <LawrenceG> cradek, I went to visit a local foundry yesterday... looking to see what they did, looking for casting supplies, and maybe a job!
[19:05:51] <cradek> how did it go?
[19:06:17] <LawrenceG> I forgot how what a black hole they are.... made me rethink my furnace project....
[19:06:43] <LawrenceG> s/how/xxx
[19:07:26] <cradek> what do you mean black hole? (I assume they are very dirty and hot...)
[19:07:35] <jensor> It almost seems to me that when one goes through the sequence of states when the unit is outputting motion there are 4 unique conditions before the sequence repeats.
[19:08:50] <LawrenceG> had a very good talk with the owner... yes black gritty dust everywhere, and when they are pouring, its hot and smokey
[19:09:13] <cradek> jensor: in quadrature, each transition ("edge") represents a motion
[19:09:33] <archivist> LawrenceG, take a short term job there just to learn the ropes
[19:09:57] <jensor> Wait, I see every time the input changes the output changes and therefore you would get a step for every input change
[19:10:09] <LawrenceG> the good news is, that I can probably send them any casting projects I want done...
[19:10:13] <cradek> jensor: so to take a single step, you always just change one bit. with step/dir, to take a single step, you have to change one bit and then put it back
[19:10:32] <jensor> I see
[19:10:59] <cradek> it's a superior interface to a stepper driver, and sadly nobody seems to use it
[19:11:34] <PCW> eric_u: Ill have a multiphase PWM module soon. Do you need deadzone between PWM /PWM
[19:12:56] <jensor> I am using it and right now I am fighting an rtaip error. I have streched the base period out to 200,000 and haven't seen the error pop up yet. I was running with it set to 50,000.
[19:13:44] <PCW> Rats I guess I missed him
[19:14:23] <cradek> jensor: you have stepper drives that take quadrature?
[19:14:32] <jensor> yes
[19:14:40] <cradek> interesting - what are they?
[19:15:12] <jensor> I converted a Bandit CNC addon to a Brigeport
[19:15:45] <cradek> oh are they the old BOSS 5ish drives with the exploding transistors?
[19:16:16] <jensor> Probably- I've had t replace quite a few in the conversion process
[19:17:19] <jensor> Bandit was an antique CNC and was supported by Allen Bradley in later years
[19:18:06] <jensor> It had no internal memory and was normally setup with a tape input
[19:19:04] <jensor> I acquired the unit - it had no tape input - I programmed it manually. If power was lost it had to be reprogrammed manually
[19:19:53] <cradek> I bet you are happier now then.
[19:20:48] <jensor> Very much so - it can now do much more than before simpler,faster and probably more accurately
[19:21:45] <jensor> The original g code that it used was was very limited
[19:22:23] <jensor> the ngc doesn 't have a mirror command that they had available
[19:23:07] <jensor> And I found that the mirror command was handy
[19:24:22] <cradek> yes it would be nice if we had mirroring/scaling/rotation...
[19:26:38] <jensor> By the way, I am checking for rtaip errors by running Axis.ngc with a loop that goes iterates for 100 times. I turn the display power off and then come back later to see if the error has shown itself.
[19:27:42] <jensor> If I see it has appeared I stretch out the base period further and rerun
[19:28:47] <cradek> it sounds like you are not finding/fixing the actual problem
[19:29:40] <cradek> what kind of machine is it? what kind of video card?
[19:29:55] <jensor> Well wh;en I run the latency check, I've not seen any thing over 70 uS
[19:30:16] <jensor> It is an intel board, with nvidea
[19:30:21] <cradek> perhaps it's the use of opengl (AXIS) that triggers it. sometimes bad video hardware will do that
[19:30:31] <cradek> ah, nvidia. which driver are you using for the nvidia?
[19:30:45] <jensor> not sure
[19:31:06] <cradek> for sure, try the vesa driver if you're not using that
[19:31:15] <cradek> nvidia cards are trouble
[19:32:02] <jensor> Is there a list of compatible cards and would they still be available?
[19:32:45] <cradek> I don't know of a list. I use old matrox stuff. Many are fine if you just use the vesa driver.
[19:33:18] <jensor> I"ll look into that
[19:34:06] <jensor> You mentioned the opening screen, Iv'e never seen the error pop up at that time
[19:35:03] <jensor> I forget the command to list the drivers
[19:35:17] <jensor> I'll have to look it up
[19:45:22] <justa> 'lo
[19:46:28] <justa> Right; here goes my question; I'm sure I'll trigger a 'N00b-alarm' , but here goes.
[19:47:29] <justa> I'm new to the CNC scene (let that be said), but I'm looking to replace the software used for an X/Y (parport-connected) laser-engraver with EMC-2 on an rt-linux installation.
[19:47:35] <cradek> HONK HONK HONK
[19:47:43] <cradek> oh, sorry, not a question yet
[19:47:44] <justa> Reason being: the software is half-chinese, comes with USB-dongle, etc.
[19:48:09] <justa> cradek: I'm sure you'll time will come. Keep your hand on the button ;)
[19:48:15] <cradek> sounds good so far
[19:48:35] <justa> The thing is, it's a 'cheap chinese laser', one of many out there.. in this case the Rabbitlaser HX40A.
[19:49:02] <justa> Now, it doesnt seem to be a 'passive Par-port-with-a-ring-counter-and-fets' type of setup inside.
[19:49:47] <cradek> uh-oh, hope it isn't too "smart"
[19:49:48] <justa> There's altogether too much 40-pin DIP 'CPU-style stuff' (including an Eprom with 'Version 3.33' firmware-revision sticker on it) inside of the mahcine.
[19:49:58] <justa> yes; well.. that's what i'm worried about.
[19:50:00] <cradek> ouch
[19:50:10] <cradek> the old software works still?
[19:50:44] <justa> I've trawled a couple of forums (sawmill creek, cnczone) and more people have asked about any experiences with replacing the software with Mach3/etc
[19:51:36] <justa> The software works, perhaps.. we havent unpacked it for the reason that Windows XP and software-dongles and all that already is a hoop for us, let alone the horrendous reviews the software gets.
[19:51:46] <cradek> seems like you have a few choices. you could try to decipher the interface, or throw away the smarts and keep only what you can use
[19:52:07] <justa> I'm sure we'll end trying it out, but i've seen some really impressive results of machines (retrofitted to be ) using EMC-2
[19:52:08] <cradek> even if you keep nothing but motors and home/limit switches (assuming it has them) your job is not too hard
[19:52:25] <cradek> what kind of motors are they?
[19:52:51] <justa> *nod*. The latter option has been in my thoughts; however it's ofcourse a lot more work and might be a hard decision to squeeze buy some of the people here.
[19:53:28] <justa> Being able to 'hot-drop' EMC-2 in there with little hardware destruction is prefferred, but I get the point.
[19:53:28] <cradek> I'm not sure how I can help you
[19:54:03] <justa> *nod*. Really, I was just wondering if anyone actually had some practical idea about what the kind of protocol could be that these things typically use.
[19:54:09] <cradek> EMC wants to control the motors in realtime - if it's not set up for that, well, not much you can do from outside its box to fix it
[19:54:42] <cradek> hm, good question. but I would be only guessing, sorry
[19:54:45] <justa> Interestingly enough; some of the pins on one of the 40pin chips have been cleanly dremmeled off. The brand/type/markings on the CPU have been removed too.
[19:54:47] <cradek> have a logic analyzer?
[19:55:08] <cradek> typical obfuscation from the evil side of hardware-land
[19:55:26] <justa> No. Though depending on the speed it operates on, i might be able to rig together something USB-connected to monitor some traffic.
[19:55:46] <cradek> sounds fun
[19:55:58] <cradek> it's probably somewhat unlikely that EMC will be a good match for whatever you find, though
[19:56:06] <justa> I guess I'll be checking the newlydraw software suite to see what's in some of the 'Advanced Device Settings' that I see in some screenshots.
[19:56:24] <cradek> I would expect that it takes vectors in (maybe even in hpgl or some other old/common format)
[19:56:53] <justa> Hmm. So, EMC-2 is really not equipped to handle anything that releases motor-control to 'another entity than itself' ?
[19:57:01] <cradek> EMC wants a much "dumber" (directly controllable) device
[19:57:35] <cradek> even servo drives that work with EMC are fairly dumb - velocity or torque command out to the drive every ms or so, position feedback back into EMC
[19:57:52] <cradek> for steppers, it outputs steps in various configurable ways
[19:57:52] <justa> *nod*. Right. I reckoned perhaps the hardware.conf file (i forget what the file is called) perhaps had options allowing 'smarter' machines or even some bit of protocol transcription via an outside process
[19:58:13] <cradek> nope, there is really no provision for that currently.
[19:58:22] <justa> Yeah; I saw some of that; the stepper-config etc.
[19:58:23] <cradek> (to some extent, that way lies madness)
[19:58:46] <justa> I'm actually (as a coincidence) also working on a collection of old printers and scanners for my own little DIY cnc
[19:59:00] <justa> which I aim to rig up with EMC2 into a working system
[19:59:13] <justa> A nice score , so far.
[19:59:23] <cradek> fun. some old deskjets etc have neat little servo motors with encoders
[19:59:55] <justa> Hmmm, I actually have two laserjet 4's.. but they don't seem to have any steppers.
[20:00:04] <cradek> I have a little lathe that uses a single L298 (dual H bridge meant for a bipolar stepper) to run two servos
[20:00:37] <justa> An old Matrix 'Star LC24' does. And a a combi scanner/printer from XEROX has two aswell aswell as nice linear guides, belts, gears, etc.
[20:00:42] <cradek> encoder position feedback to the PC, PWM out to the H bridges. simplest/cheapest servo design ever.
[20:01:00] <justa> Hmm; servo-controlled, fun :)
[20:01:19] <justa> reason to go for servo instead of steppers ?
[20:01:39] <justa> (I realize that with encoder-feedback you at least 'know' where your axis is)
[20:01:41] <cradek> they are faster, and it was partly a test of the concept (successful)
[20:01:59] <cradek> yes, it's neat. you can disable the amps and drive it around by hand, the readout on the screen follows.
[20:02:16] <cradek> if it bumps (crashes) into something, you get a following error and it shuts off
[20:02:25] <justa> That is pretty fun indeed
[20:02:47] <cradek> closed loop servo systems are superior to steppers in some ways, but they are more complex.
[20:03:05] <justa> optical encoders on a stepper-controlled setup would provide that too, but less 'need' to, I guess.
[20:03:26] <cradek> encoders on steppers could definitely detect a following error
[20:03:53] <cradek> but you've lost all your simplicity though as soon as you do that - no benefit to steppers anymore
[20:04:16] <justa> yeah.. hence the lack of 'need' for it.. Hmm, still fun to have.
[20:08:29] <justa> Right; I guess I'll be giving Newlydraw a spin and *hurk* use windows for it. At least it seems to accept PNG, TIFF and GIF (no SVG.. *sigh*)
[20:27:35] <jepler> oh boy, a new hard drive is to have a 2TB capacity.
[20:31:07] <justa> and only consume between 5 and 7 watt when ACTIVE
[20:33:07] <fenn> does that mean I can seal it in a watertight/airtight case?
[20:33:25] <alex_joni> justa: that's only for SSD's
[20:33:39] <alex_joni> I doubt a 2TB platter drive will use 7W ;)
[20:33:41] <fenn> no there are "green" hard drives that use that much
[20:33:52] <fenn> in 1TB at least
[20:34:00] <alex_joni> < 5400 RPM?
[20:34:06] <fenn> dunno
[20:34:24] <alex_joni> I think I read about those somewhere
[20:34:33] <fenn> swp was evangelizing them
[20:34:40] <archivist> hmm 2TB on raid 10...that would be nice
[20:34:42] <alex_joni> nothing like a 15k raptor :D
[20:34:48] <alex_joni> archivist: 50 ftw
[20:35:03] <fenn> 2TB drive spinning at half the speed is the same data rate as a 1TB?
[20:35:07] <archivist> raid 10 for databases
[20:35:26] <fenn> but anyway, might as well just use a ram disk, with ram so cheap now
[20:46:58] <jepler> fenn: 2TB of RAM is still a bit expensive
[20:49:29] <jepler> and probably uses more than 7W
[20:50:11] <jepler> this article says 7W active http://hothardware.com/News/WD-2TB-Caviar-Green-Monster-Drive-Preview/
[20:57:43] <jensor> cradek, I remember now, Several days ago someone told me to change a file that specifies the video driver to specify vesa, which I did and then when I rebooted it wouldn't let me login. Fortunately I was able to get to a terminal window and redit the file back and got going again. Don't remember the file name.
[20:58:20] <jensor> So vesa won't work for me
[21:01:12] <pjm_> was it xorg.conf?
[21:02:07] <jensor> I'll check
[21:08:05] <jensor> Why, yes that is the file and the driver that is listed for the nvidea card is nv and the problem occurred when I changed from "nv" to "vesa"
[21:11:36] <justa> vesa has no 3d accel
[21:12:16] <jensor> Is that bad?
[21:12:58] <jensor> Is that why it wouldn't let me log in?
[21:14:03] <justa> no. It's likely that VESA just won't work with your card. When you say 'doesnt let me log in', i'm guessing it doesnt show any login-screen
[21:14:18] <justa> or does it show you a graphic login-prompt but it doesnt work ?
[21:14:56] <jensor> No, a login screen comes up burt when I log in it would just cycle to the same log in screen again
[21:15:08] <cradek> jensor: maybe you just didn't get the change right. as far as I know, nvidia cards do all have vesa support built in.
[21:15:29] <justa> that's interesting. You can try to login with 'control-enter' instead of 'enter' when confirming the password.
[21:16:01] <justa> (or select 'Safe session' somewhere in the session-selection).. It could be that something in your login-startup config is broken (dunno why...beyond the scope)
[21:16:20] <jensor> I tried changing it a second time to vesa and got the same result
[21:17:00] <justa> please try the above.
[21:17:01] <jensor> Fortunaltly the are other options to select from and I was able to get to a terminal window
[21:17:16] <justa> it's likely not a graphics-card issue but a login-session issue
[21:17:30] <justa> perhaps it's trying to start compiz (and ofcourse, that won't ever work with vesa driver)
[21:17:42] <justa> Yes; so it's a session issue.
[21:17:49] <jensor> oh
[21:18:22] <justa> whatever your default desktop is doing, it requires features the vesa driver doesnt support. Typically, 3d-stuff/DRI-support..
[21:18:35] <justa> other option is that your disk might be full
[21:18:51] <justa> or the config-files in your homedir that it tries to read are not readable, corrupted, etc.
[21:19:00] <justa> What do you run ?.. KDE ?.. Gnome ? XFCE ?
[21:19:09] <SWPadnos> defaults, nost likely
[21:19:12] <SWPadnos> which means gnome
[21:19:18] <jensor> gnome I think
[21:19:28] <justa> (what you can do with gnome, for example, is startup with the terminal option and try 'gnome-session' on the prompt there.
[21:19:50] <justa> it'll start the gnome session-manager 'by hand' and you can watch what it tries to do; you'll see any errors it'll print.. etc
[21:20:12] <jensor> I don't really get an option directly to select a terminal mode
[21:20:34] <justa> it's often called 'safe login' or something along that line
[21:20:44] <SWPadnos> start up with the nv driver, go to System -> Preferences -> Desktop Appearance, and turn off desktop effects
[21:20:48] <SWPadnos> then try vesa again
[21:21:29] <justa> OR try pressing 'CTRL+ENTER" instead of "enter" after the password. That is a/the way to start 'a clean X with just a terminal'
[21:22:21] <SWPadnos> that's not as useful in this case though, since who knows where the "desktop preferences" option is stored? :)
[21:22:29] <SWPadnos> (if that's the problem)
[21:23:44] <justa> true. Anyhow.. there's just so much I can debug ;)
[21:24:49] <jensor> under system, preferences, I don't have "desktop appearance'" on the menu, just "desktop background"
[21:26:29] <SWPadnos> hmm
[21:26:49] <SWPadnos> oh, it's just "Appearance" in the menu
[21:27:23] <SWPadnos> and then the "Visual Effects" tab
[21:27:47] <SWPadnos> tell us the setting, then set it to "None" if it's not there already
[21:30:10] <jensor> don't have appearance on the menu
[21:31:21] <SWPadnos> is this 6.06 or 8.04, or not Ubuntu at all?
[21:31:34] <jensor> I have pasted the contents of xorg.conf at http://pastebin.com/d2b7952fd
[21:31:38] <jensor> 6.06
[21:31:47] <SWPadnos> oh. then you don't have desktop effects :)
[21:31:54] <cradek> the line you would change is 93
[21:31:59] <cradek> to Driver "vesa"
[21:32:13] <cradek> and change nothing else
[21:32:27] <jensor> yep thats what i did when it gave me the login problem
[21:32:36] <SWPadnos> in lower case letters?
[21:32:43] <jensor> yes
[21:32:48] <cradek> I wish I understood exactly which kind of login problem you had
[21:33:14] <SWPadnos> what resolution do you have your desktop set to?
[21:33:47] <jensor> The login screen comes up , itype in user name enter, password enter and it loops around to the same screen again
[21:33:58] <jensor> How do I tell
[21:34:06] <jensor> what the res is set to
[21:34:19] <SWPadnos> so the video driver is working, but as justa said, something in your session is not working with the VESA driver
[21:34:42] <SWPadnos> System -> Preferences -> Screen Resolution (I think)
[21:34:59] <SWPadnos> or just Resolution, I don't remember exactly what it's called on 6.06
[21:35:21] <jensor> 800 x 600
[21:35:26] <BigJohnT> it's the same on 6.06 SWPadnos
[21:35:30] <BigJohnT> I just looked
[21:35:30] <SWPadnos> ok
[21:35:45] <SWPadnos> jensor, can you pastebin the file /var/log/xorg.0.log ?
[21:36:15] <BigJohnT> jensor: is the problem that you are stuck at 800 x 600?
[21:36:28] <SWPadnos> oops - /var/log/Xorg.0.log
[21:38:13] <jensor> no I can change it with r
[21:38:39] <jensor> I can chage screen res with a cmd - I need to look it up in my notes
[21:39:37] <jensor> That log file wouldn't show the problem now because its been a week or so since I went through that boot problem
[21:40:08] <SWPadnos> I want to see the log from a successful bootup with the nv driver
[21:40:17] <jensor> ok
[21:40:33] <SWPadnos> looking at the web a little, it seems that several people have had various issues with that monitor
[21:40:46] <SWPadnos> it's a native 1280x1024 unit, so that's the best resolution to use
[21:41:00] <SWPadnos> but apparently there's a lot of flicker unless you put in a custom modeline
[21:44:18] <jensor> SWPadnos, the file is at http://pastebin.com/d34e935d0
[21:44:28] <SWPadnos> thanks
[21:44:42] <SWPadnos> so you set your desktop to 800x600 on purpose? is that where you want to leave it?
[21:45:16] <jensor> I set it there because the axis screen seemed to look best there
[21:45:28] <SWPadnos> oh, strange :)
[21:45:35] <SWPadnos> you can maximize AXIS :)
[21:46:15] <SWPadnos> you can also change the font size of the DRO (though I don't know exactly how), and there's a BIG DRO tab in the upcoming EMC2 2.3
[21:46:27] <SWPadnos> most things should look better at the native 1280x1024
[21:49:31] <jensor> I used xrandr to adjust the resolution
[21:49:44] <SWPadnos> oh
[21:49:59] <SWPadnos> I don't know that xrandr is usable with vesa
[21:50:17] <SWPadnos> (it could be, but that's an accelerated drawing interface, so I don't know for sure)
[21:50:38] <SWPadnos> you should use the screen resolution application in the system->preferences menu instead
[21:51:03] <jensor> so if I use the gui to set it, I may not have the boot problem?
[21:51:16] <SWPadnos> I don't know :)
[21:51:57] <SWPadnos> this could be relevant: http://osdir.com/ml/freedesktop.xorg.modular/2006-09/msg00001.html
[21:52:41] <SWPadnos> so the monitor is a Viewsonic VA721?
[21:53:21] <jensor> yes
[21:53:44] <jensor> no, ve 510s
[21:54:08] <SWPadnos> ok, the string is wrong in the xorg.conf file :)
[21:54:09] <jensor> Viewsonic VE 510S
[21:54:51] <jensor> If I remember I couldn't find the model 510s as an option in the setup
[21:56:13] <SWPadnos> ah, ok. so it's 1024x768 native. that makes more sense with the Xorg.0.log contents
[21:56:18] <jensor> So I selected the VA721 and that seemed to work ok
[21:56:33] <SWPadnos> except that the VA721 is a 1280x1024 panel
[21:56:40] <jensor> oh
[21:57:35] <jensor> should I change the designation in xorg.conf to VE 510S
[21:57:47] <SWPadnos> no, that won't help
[21:57:58] <SWPadnos> hold on a sec, I'll have a couple of links for you in a minute
[21:58:05] <SWPadnos> or hold on a min - you know :)
[22:02:17] <SWPadnos> ok, try putting the following into your xorg.conf file, inserted between line 99 and 100 in your pastebin:
[22:02:24] <SWPadnos> Modeline "1024x768@60" 64.56 1024 1056 1296 1328 768 783 791 807
[22:03:04] <SWPadnos> I used information from here: http://www.viewsoniceurope.com/UK/Support/NonCurrentLCDs/LCDE2/VE510s.htm
[22:03:34] <SWPadnos> (plus a guessed 65 Mz maximum dot clock, which is as close as I wanted to calculate 1024x768x85 in my head)
[22:03:39] <SWPadnos> MHz that is
[22:03:53] <SWPadnos> and I entered the information here: http://xtiming.sourceforge.net/cgi-bin/xtiming.pl
[22:04:31] <SWPadnos> You will only notice a difference if you switch to 1024x768 mode though (which is what the login screen uses)
[22:04:58] <SWPadnos> you may have to disable xrandr as well. I don't know how to do that
[22:06:35] <SWPadnos> actually, disable xrandr first, then mess with modelines
[22:06:43] <jensor> ok, will try these -will take a while
[22:07:02] <SWPadnos> ok, no problem
[22:07:04] <SWPadnos> good luck
[22:07:10] <jensor> have to first figure disabling xrandr
[22:07:19] <SWPadnos> yeah, google should be your friend there
[22:07:27] <jensor> ok