#emc | Logs for 2009-05-31

[00:14:56] <jmkasunich> if you have no electronics knowledge, you are not going to be able to do what you want
[00:17:59] <Danimal> so i should scrap my machine, rather than try to learn it?
[00:19:21] <jmkasunich> no
[00:19:32] <jmkasunich> you should try to learn enough to do what you need
[00:20:17] <jmkasunich> you wrote "someone mentioned that I could use an optocoupler driven off my 5i20 to avoid having to use a DAC"
[00:20:26] <jmkasunich> that tells us very little about what you need
[00:20:32] <jmkasunich> and nothing about what you know
[00:21:02] <jmkasunich> what input voltage does the VFD need?
[00:21:05] <Danimal> i'm trying to configure my hal
[00:21:47] <jmkasunich> that much I knew
[00:22:38] <Danimal> the analog side of the vfd is 10v
[00:23:12] <Danimal> it has it's own 10v supply
[00:23:17] <jmkasunich> does the VFD provide a 10V reference output? many do, so that you can connect a pot and get a 0-10V signal to go back into the vfd
[00:23:20] <jmkasunich> ok, good
[00:23:27] <Danimal> yep
[00:23:44] <jmkasunich> so, you need a plain transistor output optocoupler
[00:24:20] <jmkasunich> it needs to be connected so that when the light is on 100% of the time, you get 10V, 0% of the time gives 0V, and 50% of the time gives 5V
[00:24:25] <Danimal> plain transistor..ok i'll grab one... all i have is darlington and schmitt
[00:25:13] <Danimal> yea, i kinda grasp all that stuff, it was more of the software part that i'm having the trouble
[00:25:35] <jmkasunich> well, that's not the impression you gave earlier
[00:26:05] <jmkasunich> I wrote "if you have no electronics knowledge...." but it sounds like you _do_ have some
[00:26:38] <jmkasunich> it gets frustrating playing 20 questions on IRC to help people - the more specific you can be, the more likely we can help
[00:26:53] <Danimal> i didnt say i had no electronics knowledge, i said i had no background in this stuff, meaning the software side
[00:26:59] <Danimal> sorry for the confusion
[00:27:06] <jmkasunich> ok
[00:27:34] <jmkasunich> even though I'm sure I could do this config, I still need to read the man page for the details - nobody has it all memorized
[00:27:38] <jmkasunich> so, man pwmgen
[00:28:12] <Danimal> yep, i got it open... been reading it all day trying to make sense of it
[00:28:43] <jmkasunich> ok - do you understand the threads/functions part?
[00:28:50] <jmkasunich> wait
[00:28:58] <jmkasunich> you said you have a mesa card?
[00:29:17] <Danimal> yes to the mesa card, not entirely to the first question
[00:29:28] <jmkasunich> the pwmgen I was thinking about is software pwm
[00:29:35] <jmkasunich> mesa pwm is different
[00:30:35] <jmkasunich> do you have the rest of your machine working with the mesa, and just need to get the spindle going? or are you just starting?
[00:30:47] <Danimal> so i should be looking in the pwmgen section of the hostmot2 man, right?
[00:31:11] <jmkasunich> yes, probably
[00:31:17] <Danimal> yes, my machine works, spindle turns on, now i'm to the vfd part
[00:31:33] <jmkasunich> the axis motors are servo? or stepper?
[00:31:37] <Danimal> stepper
[00:32:33] <jmkasunich> so you've been thru the basics of hostmot2 - you turned on the proper number of stepgens, and you saw the pin listing in dmesg after loading the driver
[00:33:16] <Danimal> yep, i got that part, and i got a little grasp of the gpio stuff yesterday when i did my spindle on
[00:33:25] <jmkasunich> I don't know if the stepgens and the pwmgens compete for the same pins
[00:33:41] <jmkasunich> your first step will to be to get a pwmgen, and find what pins it is using
[00:34:06] <jmkasunich> that may already be done by default, it looks like the default for num_pwmgens is -1, which means "use all of them"
[00:34:47] <Danimal> when i set up the firmware, i chose 3 pwmgens
[00:34:57] <jmkasunich> ok
[00:35:08] <jmkasunich> (any particular reason for choosing three?)
[00:35:26] <Danimal> figured i may use them someday
[00:35:34] <jmkasunich> ok
[00:35:55] <jmkasunich> it looks like the pwmgen has three hardware outputs
[00:36:22] <jmkasunich> Not_Enable is strangely named, but I guess that means "goes low when enabled"
[00:36:48] <jmkasunich> you could probably use that instead of GPIO to start and stop your VFD, but if you already have that working, don't worry about it
[00:37:04] <Danimal> oh cool
[00:37:06] <jmkasunich> one of the other two outputs will drive your LED
[00:37:27] <jmkasunich> on the HAL side, there are only two pins: enable and value, plus a few parameters
[00:37:39] <jmkasunich> I believe EMC outputs the desired spindle speed in RPM
[00:38:00] <jmkasunich> you want to connect that with a hal signal to the "value" pin of the pwmgen
[00:38:24] <jmkasunich> the manual page says "value" is in arbitrary units
[00:38:34] <jmkasunich> what that means is "you get to decide what the units are"
[00:38:42] <jmkasunich> in your case, you want the units to be RPM
[00:39:01] <jmkasunich> how fast does your spindle go with +10V into the VFD?
[00:39:35] <Danimal> 3800
[00:40:05] <jmkasunich> ok - when "value" is 3800 RPM, you want duty cycle to be 1.0 (100%)
[00:40:20] <jmkasunich> the man page says that "dc = value / scale"
[00:40:45] <jmkasunich> so if you set "scale" to 3800, then when value is 3800, duty cycle will be 3800/3800 = 1.0 = 100%
[00:40:57] <Danimal> ahh ok so thats what that means
[00:41:10] <jmkasunich> when value is 1900 RPM, duty cycle will be 1900/3800 = 0.5 = 50%
[00:41:12] <jmkasunich> etc etc
[00:41:17] <Danimal> gotcha
[00:43:51] <Danimal> it's easier to understand in layman's terms
[00:44:10] <jmkasunich> so, it sounds like you already had a pretty good understanding of things, just led astray by one or two details
[00:44:20] <Danimal> i'm a machinist, not i'm basically onlu educated in that
[00:44:45] <Danimal> -not =so
[00:44:49] <Danimal> +so
[00:45:15] <Danimal> well i've been at it for a few weeks now.. trying to make sense of it all
[00:45:19] <jmkasunich> I pretend to be a machinist ;-)
[00:45:48] <Danimal> i wish i could only pretend lol
[00:46:46] <Danimal> i got my first cnc machine last month, a shizuoka with a bandit control, so i decided to try the conversion
[00:47:25] <jmkasunich> you own your own shop, or work at one and got a machine on the side?
[00:47:49] <Danimal> i'm starting my own shop... been in business since april
[00:48:22] <jmkasunich> tough time to be getting started - good luck to you
[00:48:52] <Danimal> it's going ok believe it or not
[00:49:32] <jmkasunich> all manual at the moment?
[00:49:45] <Danimal> aside from the shizuoka, yep
[00:49:57] <Danimal> i do more tooling than production
[00:50:13] <jmkasunich> ah, that makes sense
[00:51:00] <jmkasunich> I've done a few paying jobs on the side - enough over 10 years to pay for my tools, etc
[00:51:12] <jmkasunich> but if I was trying to make a living at it I'd have starved by now
[00:51:57] <Danimal> haha yea... i'm not making as much as i used to, but it've only been doing work for a month
[00:53:08] <Danimal> actually, i take that back, i'm making about the same, but i'm spending everything i make on shop stuff
[00:53:11] <KimK_> Danimal: A Shizuoka? Have you seen Kirk Wallace's machine shop URL? (it's in the sig of all his mailing list posts) He has a Shizuoka too. Photos on the web of his EMC2 retrofit.
[00:53:39] <Danimal> yes, his is part of the reason i decided to use emc2
[00:54:22] <Danimal> i was unsure of the whole toolchanger integration, but he assured me it could be done
[00:55:31] <KimK_> "jmkasunich wrote: I pretend to be a machinist ;-)" That was a pretty fine router table I saw in Wichita from a pretend machinist. Got any pictures up yet?
[00:58:21] <jmkasunich> just the one I took there
[00:58:52] <cradek> you sure got a lot done in a few days.
[00:58:53] <jmkasunich> http://jmkasunich.com/pics/pcbmill-table-2937.jpg
[00:59:30] <jmkasunich> since then I've (temporarily) mounted the linear rails and X servomotor, I should take another pic
[01:00:03] <KimK_> Danimal: Yes, I'd like to talk to Kirk about his tool changer too. On the Kasuga I'm doing, we kept the Dana/Summit stuff in the TC intact, we didn't try to take over all TC operations. We plan to keep the M10 trigger/response, or whatever it has already.
[01:01:18] <KimK_> Maybe it was M10/M11, I don't have a list in front of me
[01:01:45] <Danimal> me too, do you have the schematics for it? i havent read into it too much, but it looks like the m function cable should be fairly easy to understand
[01:02:08] <Danimal> i have the whole manual/schematics for it if you want it
[01:02:17] <Danimal> in pdf
[01:02:33] <KimK_> Yes, we have what I think is a complete set (paper), but not scanned in
[01:03:01] <Danimal> lemme know if you want a pdf version, i can send it.
[01:03:14] <KimK_> Ah, excellent. Sure if you send it, I'll compare and report any differences
[01:03:27] <Danimal> you have a quickdraw, correct?
[01:03:43] <Danimal> i am unaware if there are different versions
[01:03:58] <KimK_> Are you keeping the M10/M11 (or whatever) or are you stripping it out and taking over?
[01:05:09] <Danimal> KimK_, i gutted the whole bandit control, but i'm leaving the dana/summit stuff intact
[01:05:37] <Danimal> i may use the board with all the reed switches, but not sure
[01:05:46] <KimK_> We have one (Quickdraw? I'd have to look, it looks like Kirk's, though) but ours is without the stepper motor controlled vari-drive, we have a 5-step pulley. So ours is missing the speed up/down buttons and related circuitry
[01:06:14] <Danimal> ah ok
[01:06:43] <Danimal> mine has it, but it wont be used... i'm going direct drive and using the vfd
[01:06:43] <KimK_> Dana/Summit offered it both ways, apparently
[01:07:33] <jmkasunich> if you rely _only_ on a VFD for varying speed, you might be dissapointed
[01:08:00] <jmkasunich> if you have a 2HP motor, and use step pulleys or vari-speed sheaves to vary speed, you get the whole 2 HP over the entire speed range
[01:08:28] <jmkasunich> if you use a VFD, you only get the whole 2HP at top speed - at half speed you only have 1HP, at 1/4 speed you only have 1/2 HP
[01:08:39] <Danimal> i'm gunna ditch the whole salsbury drive system
[01:08:55] <KimK_> It's probably a good average compromise. You can always hand crank it to the bottom for big-tapping or to the top for high speed light routing and have a PyVCP input for a "gear change"
[01:09:20] <Danimal> hmmm
[01:10:13] <KimK_> We plan to have at least a PyVCP "gear change" input, maybe some real switch input bits, not sure yet
[01:10:25] <Danimal> i was under the impression that you keep the torque with the vfd
[01:10:32] <Danimal> but i've been wrong alot
[01:10:48] <jmkasunich> torque remains constant as you slow down the VFD
[01:10:59] <jmkasunich> but with gears, belts, etc, torque increases as you slow down
[01:11:08] <Danimal> ahh ok
[01:11:30] <jmkasunich> * jmkasunich designs vfd's for a living ;-)
[01:11:43] <Danimal> i'm keeping the hi/lo gear
[01:12:00] <Danimal> my motor is 3hp though
[01:12:13] <Danimal> not that it's much of an improvement over 2hp
[01:12:35] <jmkasunich> I don't know how often you actually take a 2HP cut
[01:12:49] <KimK_> Ha, we machine hackers want all the power we can get, LOL
[01:13:06] <Danimal> i think my steppers will croak before i use 2 hp lol
[01:13:28] <Danimal> KimK_, how should i send over the manual?
[01:14:24] <KimK_> Our Bandit had the servo/resolver option, so our motors are a little stronger. And we have a 5HP spindle motor.
[01:14:56] <KimK_> Up to you. File transfer right here, or email, either is good, which would you like?
[01:16:13] <Danimal> i can do a file transfer, but it's been years since i've done it on here so i may need a walk through
[01:16:47] <KimK_> What's your client? (I'm on Ubuntu Xchat)
[01:17:18] <Jymmm> Just use http://filebin.ca/ up to 50MB
[01:17:37] <KimK_> There you go
[01:18:04] <Danimal> ok we'll do that
[01:18:09] <KimK_> And observers get to take home a copy too
[01:19:03] <Danimal> yep!
[01:19:08] <Danimal> it's uploading
[01:20:40] <Danimal> i need to get a pendant for my machine eventually
[01:20:44] <KimK_> After you get that going... What did you do with your empty Bandit, get rid of it, or put stuff in it? What stuff? Dave is trying to put a small CRT in the front half, with a keyboard tray in front, but that might have to go, I need front panel switch space, etc.
[01:21:23] <Danimal> i put the stepper drivers in it, but the front half is empty
[01:21:37] <Danimal> i just put the new drivers in place of the old ones
[01:21:44] <KimK_> We're not sure about the PC either, inside would be great, but we might have to cable next to it for for stuff inside
[01:21:52] <Danimal> i'm using the power supply
[01:22:27] <KimK_> OK
[01:23:03] <Danimal> i'm planning on getting rid of the bandit's box, and putting the stepper drivers and the power supply in a box on the side. i have some good sized cabinets that i took off, so i may use one of those
[01:23:06] <KimK_> We traded our resolvers for encoders, we'll see how that goes.
[01:23:19] <Danimal> sounds like a good way to go
[01:23:33] <Danimal> i'm going to switch to servos eventually
[01:23:39] <cradek> I'm really happy with the pico resolver to quadrature conversion board
[01:24:11] <cradek> they are expensive but considering what a new encoder costs (and the pain involved in mounting it), it's probably a wash
[01:25:10] <Danimal> as for the pc, i'm going to make something on a arm off to the side of the machine that will house my pc, 19"lcd (might swap it out for a 15"), and have a heyboard/mouse tray on it
[01:27:00] <Danimal> http://filebin.ca/jfbmqh
[01:27:07] <Danimal> there's the pdf
[01:28:47] <KimK_> cradek: We thought about keeping our resolvers, but the USDigital E6 encoder kit won us over on price, resolution, and speed. http://www.usdigital.com/products/encoders/incremental/rotary/kit/e6/
[01:28:58] <KimK_> Danimal: Thanks, they're coming in now
[01:29:09] <Danimal> no prob
[01:29:40] <Danimal> i have some large steppers
[01:30:23] <Danimal> somewhere in the ballpark of 1300oz/in
[01:30:37] <Danimal> i thought they were large anyways
[01:30:39] <cradek> KimK_: 10k ppr, wow
[01:30:42] <KimK_> The E6's ended up costing about $80 each with the index and differential options, plus another $40 or so for a 35' cable (encoder connector to bare wires)
[01:31:09] <Danimal> more torque than my poor ford ranger
[01:32:08] <cradek> whoah, pico systems got a web store thingy
[01:32:12] <KimK_> Yeah, we went with the 2500 line ones. And I think we'll be OK on speed too, so, good. (he said, crossing his fingers)
[01:32:27] <cradek> I ended up with 204800 counts/inch
[01:33:29] <KimK_> Wow, that's quite a bit! Gear/belt reduction?
[01:33:47] <cradek> the resolver is geared to the ballscrew
[01:34:58] <jmkasunich> what is the pico website anyway
[01:35:04] <cradek> pico-systems.com
[01:35:10] <cradek> not picosystems.com, I see
[01:38:19] <Danimal> well i think i'm gunna eat then try to finish the grout in my bathroom
[01:38:24] <KimK_> Dave's should be 10,000 with either 5mm/rev or .200"/rev (haven't measured), so 50,000/in or 50,800/in
[01:38:43] <Danimal> thanks for the help jmkasunich
[01:39:07] <Danimal> and good chatting with you again KimK_
[01:39:23] <cradek> KimK_: sounds like a good choice of resolution
[01:39:24] <KimK_> You bet, keep us posted on how you're doing
[01:39:35] <Danimal> will do
[01:39:42] <Danimal> slow but steady
[01:40:04] <Danimal> with lots of headaches
[01:40:09] <jmkasunich> Danimal: glad I could help - sorry I was a bit grouchy at the start ;-/
[01:41:17] <KimK_> cradek: A good choice? Do you find that your 204800/in is too much? Why isn't that even better?
[01:41:42] <Danimal> dont blame ya jmkasunich, i'm not very good at explaining stuff..sorry for the confusion
[01:42:28] <Danimal> adios!
[01:42:29] <cradek> no, the 204800 is fine - the only way to have too high a resolution is if you overrun your encoder counter's max speed
[01:43:21] <KimK_> Well, I know you do some pretty fine work over there, so high-res is the order of the day, I'd guess?
[01:43:48] <cradek> super high res on X is nice. the lathe is very good.
[01:43:52] <jmkasunich> I suspect that the benefit going from 50K per inch to 200K per inch is very very small
[01:44:00] <cradek> yes that's surely true
[01:44:17] <cradek> http://timeguy.com/cradek-files/emc/no-backlash.png
[01:44:28] <jmkasunich> my mill will most likely have 63K per inch, and that seems like plenty to me
[01:44:37] <cradek> red is resolver->quadrature count, cyan is a glass scale I attached temporarily to X
[01:46:35] <cradek> I thought I would generate a screw comp table, but I think the screw is as good as, or better than, the scale
[01:47:01] <KimK_> That is a nice pic, what was the glass scale resolution? (since the steps are easy to see) Or, what's the vertical scale in inches?
[01:47:51] <jmkasunich> 0.001 per div
[01:49:11] <jmkasunich> I bet the scale is 2.54 counts per thou, or 0.01mm (10 microns) per count
[01:49:18] <KimK_> If the scale is .001"/div, then was the scale metric? (it doesn't quite repeat)
[01:49:26] <cradek> yes I think so
[01:49:37] <KimK_> oops, that was confusing
[01:49:49] <cradek> somehow the scale's readout box (all TTL) reads in inches though. baffling.
[01:50:06] <KimK_> scale1 = halscope , scale2 = the glass scale
[01:51:03] <KimK_> maybe the human display rounds to the nearest value?
[01:51:12] <KimK_> nearest metric value
[01:51:18] <KimK_> whatever
[01:51:30] <cradek> yes the input scale of the scale (haha) was 2540 when I had it hooked up
[01:51:59] <KimK_> OK. So 25.4 strikes again.
[01:51:59] <cradek> KimK_: it has so many segments burnt out (numitrons) that I can't really tell how it works, I can only barely tell it's in inches
[01:53:44] <KimK_> Really? You need Numitrons? I've been hanging on to a few for no good reason that I know. (Numitrons aren't as "retro cool" as Nixies, in my opinion.)
[01:53:46] <jmkasunich> oh, that is the scale from the workshop last year?
[01:54:00] <cradek> jmkasunich: yes, or two years ago, I forget
[01:54:04] <cradek> KimK_: nope, thanks :-)
[01:54:29] <KimK_> OK
[02:05:35] <John_F> Hi, I just got my computer running with the 5I20 board. I thought I would play around with Hal to make the I/O work. Questions: I should "loadrt hm2_pci" not hal_m5i20 right?
[02:06:12] <cradek> you can use either driver but hm2_pci is the new and currently-maintained one
[02:06:50] <John_F> also, what is the param hm2_5i20.0.pet.watchdog.tmax ?
[02:07:35] <cradek> tmax values tell you the longest time a function has taken to execute so far
[02:07:46] <cradek> lots of functs have tmax parameters you can read
[02:09:39] <John_F> OK so if I pet the dog at a certain interval tmax will tell me the maximum time it has taken so far?
[02:10:05] <cradek> no, not the interval, the time it took to execute
[02:10:27] <John_F> execute petting the dog?
[02:10:38] <cradek> yes how long did it take to pet
[02:10:50] <cradek> tmax of the motion controller, or pid, etc etc is more interesting
[02:11:08] <John_F> got it thanks
[02:11:13] <cradek> these are really for debugging or optimizing only
[02:11:26] <cradek> not generally needed
[02:14:32] <John_F> When does the watchdog actually start watching?
[02:15:18] <c0ldfront> hi
[02:48:36] <John_F> cradek: OK I just tried a simple experiment. http://pastebin.ca/1441820 used halmeter to look at watch dog bit. It goes true as soon as I start. what is missing?
[02:50:40] <cradek> hm, I'm not sure
[02:51:20] <cradek> maybe you should add read to the thread too?
[02:51:38] <cradek> what is hm2_5i20.0.watchdog.timeout_ns?
[02:52:14] <John_F> the timout is default value of 3B9ACA00 ns
[02:52:35] <John_F> shoud be mor that 1000000 right?
[02:54:11] <cradek> looks like more, to me
[02:54:15] <John_F> add read to the thread? read what?
[02:54:43] <cradek> hm2_5i20.0.read
[02:55:03] <cradek> on my system I have read, then write, then pet watchdog
[02:55:27] <John_F> OK I will try that
[03:41:56] <John_F> cradek: ? I think it works now. The start command needs to happen right away, needs to be in the .hal file. Does that make sense?
[04:42:17] <bglackin> I am trying to load EMC2.3 sim
[04:42:33] <bglackin> Just ran through everything upto
[04:42:43] <bglackin> sudo apt-get install emc2-dev
[04:43:12] <bglackin> got ---
[04:43:22] <bglackin> The following packages have unmet dependencies:
[04:43:49] <bglackin> emc2-dev: Depends: emc2-sim (= 1:2.3.0~beta1) but 1:2.3.1 is to be installed
[04:44:02] <bglackin> E: Broken packages
[04:44:27] <bglackin> Any ideas?
[04:44:59] <bglackin> I will take my answer while away should anyone read this - thanks and gnite.
[07:08:33] <alex_joni> bglackin: still there?
[07:09:13] <alex_joni> it would help if you could say what version you are trying (6.06 or 8.04, 32 or 64 bit, etc)
[07:12:35] <alex_joni> bglackin: the package names should be emc2-sim_2.3.1 and emc2-sim-dev_2.3.1
[10:42:43] <JanVanGilsen> Can somebody explain why the do_homing function in homing.c keeps getting executed after my axis has been homed? If I add reportError("dummy"); between line 202 and 203, then I home one joint, the message keeps getting printed after the joint was homed. seems a bit strange to me ...
[10:46:09] <micges> do_homing does nothing when no joints are homing
[10:46:39] <micges> see line 233:
[10:46:42] <micges> case HOME_IDLE:
[10:46:42] <micges> /* nothing to do */
[10:46:42] <micges> break;
[10:51:00] <JanVanGilsen> is it because there are joints that aren't homed, and thus are in the HOME_IDLE state?
[10:52:42] <JanVanGilsen> oh i see after it reached the HOME_FINISHED state it goes into the HOME_IDLE state ...
[10:53:04] <JanVanGilsen> thx
[10:55:09] <micges> exactly :)
[11:42:20] <cheiron> i need some help with a remote setup (using nml and emcsvr)
[11:42:47] <cheiron> i always get strange errors
[12:26:24] <BigJohnT> to upgrade emc without a network connection is this the only file I need http://linuxcnc.org/emc2/dists/dapper/emc2.3/binary-i386/emc2_2.3.1_i386.deb ?
[13:06:52] <JanVanGilsen> I'm having problems reading the hal pin i created in the homing.c file
[13:11:02] <JanVanGilsen> I can see the pin with the hal meter, but don't know what i'm doing wrong while tryning
[13:11:07] <JanVanGilsen> to read it
[13:14:33] <BigJohnT> what kind of error do you get?
[13:15:12] <JanVanGilsen> none, it just reads 0.00000000
[13:17:09] <JanVanGilsen> but on the hal meter it shows an other value
[13:20:09] <JanVanGilsen> I assume there's something wrong with the c code I wrote, ...
[13:20:28] <BigJohnT> where are you trying to read it?
[13:21:05] <JanVanGilsen> in do_homing in homing.c
[13:21:36] <JanVanGilsen> i'll get an usb drive and pastebin the files =)
[13:29:01] <JanVanGilsen> motion.c http://pastebin.ca/1442268
[13:29:28] <JanVanGilsen> homing.c http://pastebin.ca/1442271
[13:29:59] <JanVanGilsen> mot_priv.h http://pastebin.ca/1442275
[13:30:42] <JanVanGilsen> I marked the things I added with comments containing "JVG"
[13:34:04] <BigJohnT> mmm breakfast
[13:37:12] <JanVanGilsen> I'll wait until jmkasunich is here, he did instruct me on how to do it.
[15:01:04] <JanVanGilsen> aaaaarg... stil no success on reading the halpin
[15:07:53] <tom3p> what does 'test' return (re: reportError("Joint%d: %f", joint_num, test); ) ? you seem to have been looking to see if anything was working.
[15:12:52] <JanVanGilsen> 0.0000
[15:13:55] <JanVanGilsen> I can see the halpin change, but it doesn't affect test
[15:16:34] <tom3p> all the halpins were created? ("retval = hal_pin_float_new(buf, HAL_IN, &(addr->home_offset), mot_comp_id);")
[15:17:27] <JanVanGilsen> Yes the halpins did get created
[15:22:55] <tom3p> ok, so 'test' and the tool that looks at test, are not looking at the same thing (eg addr vs content) or, looking at same thing different ways ( eg char vs float )
[15:23:38] <tom3p> that comment is result driven ( not guided by reading the code you wrote )
[15:24:19] <tom3p> what tool is used to examine 'test' ( besides the debug print you wrote )
[15:24:38] <JanVanGilsen> none :)
[15:25:37] <tom3p> " what i sthis halpin of whicj you speak? "I can see the halpin change, but it doesn't affect test"
[15:26:09] <tom3p> what is this halpin of which you speak ;)
[15:26:31] <tom3p> oh and how is it related to 'test'
[15:27:03] <tom3p> laundry duty, bbl
[15:29:03] <JanVanGilsen> I connected the halpin (axis.0.home-offset) with a pyvcp slider
[15:29:35] <JanVanGilsen> i can see the cange of the halpin with the hal meter.
[15:30:11] <JanVanGilsen> but there is no effect on 'test'
[15:31:26] <JanVanGilsen> I have to go now, will be back in a couple of hrs ..
[15:41:19] <tom3p> 'reducing' your code yields double test = *( &(emcmot_hal_data->joint[joint_num]) )->home-offset ); try the same de-referencing on some known value.
[15:41:20] <tom3p> ( maybe make sure making the struct larger is ok )
[15:41:38] <tom3p> and, this is really stuff for #emc-devel
[15:46:12] <tom3p> haha checking on some pointer stuff, stanford talks about pointers & 'pointees' :)
[15:47:40] <tom3p> even better, the pointee in examples is '42' ( for all the douglas adams fans )
[16:11:43] <tom3p> maybe joint_data = &(emcmot_hal_data->joint[joint_num]); test = joint_data.home_offset; ?
[18:12:31] <roh> hi there
[18:13:22] <roh> is http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?A_New_Approach_For_Using_Joypads_With_EMC2 too new to have joyhandle.ko already built in 2.3.1 or is there a 'better way'
[18:14:14] <roh> i now have the axis and buttons of my joypad in hal, but it seems i am missing the joyhandle.ko to tweak the axis to nonlinear for better handling
[18:26:21] <kirk_wallace> Hello. I need an electronics opinion. I am hooking up a brushed DC motor to turn the tool carousel. I have a multi voltage transformer that was on the original machine. I have a 12 DC supply connected to this transformer and it only runs the carousel motor. Using a 12 volt battery, I measured 7 Amps. With the 12 Volt supply I measure 8 Volts, so I am dropping 4 Volts at something not much less than 8 Amps. Is my transformer up to the task?
[18:27:36] <archivist> any capacitors after the rectifiers?
[18:28:02] <archivist> else sounds like no it isnt
[18:28:16] <kirk_wallace> 50k uF
[18:30:03] <archivist> then definitely struggling
[18:30:47] <archivist> what happened to the original motor
[18:31:33] <kirk_wallace> Okay, I thought so. I'll have to dig up another supply.
[18:31:56] <kirk_wallace> The original was an AC induction motor.
[18:32:24] <archivist> add a vfd to drive the original
[18:35:24] <kirk_wallace> It was connected to a Geneva drive, so I got a start-stop for each to position change. Really anoying. I am going to use the DC motor to accel/decel each way to the tool position. An encoder is in the works too. I also need a PWM driver.
[18:35:44] <kirk_wallace> rm "to"
[18:37:24] <kirk_wallace> For now I'll use Jon's pico drive, but it's a bit of overkill. I'd rather use it where its really needed.
[18:37:36] <kirk_wallace> it's
[18:48:50] <tom3p> an of the shelf desktop suited to emc according to the wiki latency test page http://www.tigerdirect.com/applications/SearchTools/item-details.asp?sku=N52-1070+P (needs more mem & chuck the os)
[18:58:21] <kirk_wallace> 15 GB hard drive? Maybe flash instead?
[19:00:31] <roh> tom3p thats quite a lot of money for that old junk
[19:01:00] <tom3p> haha go look for new junk that works with emc, get it home & have it fail the test
[19:01:13] <tom3p> (bad attitude from bad experience)
[19:02:20] <roh> tom3p *g* .. i see where you are heading.. i meant that it could make more sense to build a machine from old parts or 'known good' parts which one already has in use and 'are the right age' now.. just buy something nicer e.g. for the workstation
[19:03:15] <roh> i just took one of the 'old boxes' with a via based msi board and a duron out of the corner.. since notebooks nowadays are like 4 times as fast it was just not used anymore
[19:03:36] <roh> still it 'seems to be too slow' sometimes
[19:03:57] <roh> we use 18usec now i think, 16 was 'too sharp'
[19:06:01] <tom3p> example: i'm in cincinnatti now for maybe 3 or 4 months, got a great new sable2015 stepper machine in from taipei. what can i drive it with? not my 2 laptops, so what can i buy?
[19:08:25] <roh> is it a problem getting good, not dead, used p3 class or socket A (amd) based boards with intel, via or amd chipset?
[19:14:05] <tom3p> the salvage houses around Ky & S. OH are reclaiming everything. i cant find squat used. maybe this is local, maybe a trend. on the web, the wiki listed stuff is disappearing.
[19:15:56] <Jymmm> http://weirdstuff.com/
[19:17:12] <tom3p> one of the links on wiki leads not to the jetway mobo, but to a p/s, other leads to 'discontinued' item http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Hardware_Requirements
[19:17:18] <tom3p> Jymmm: will look thx
[19:18:18] <tom3p> do they have any of the listed units?
[19:18:45] <roh> tom3p well.. stuff listed there is only 'some' ... actually there are heaps of hardware which shouldnt make any trouble out there
[19:18:53] <Jymmm> Avoid the Tyan Thunder it has a BIOS settings issue
[19:19:46] <tom3p> ahah i was advised to ge the tyan thunder and got 2 of 'em, and mem and ... one of many attempts to get something that passes the tests
[19:19:55] <Jymmm> http://www.weirdstuff.com/cgi-bin/item/15061
[19:19:59] <roh> i would for example guess that a ibm thinkpad T40p or so should work great. fast enough, can hold decent amounts of ram, decent graphics, should work out of the box with 8.04 and also still has a real parport
[19:20:15] <Jymmm> roh: Nope, no laptops
[19:21:21] <Jymmm> That intel mobo has the best chipset
[19:21:33] <Jymmm> though limited am
[19:21:38] <Jymmm> ram
[19:21:39] <roh> Jymmm any sane reason?
[19:22:53] <Jymmm> Yeah, 100,000+ laency
[19:22:56] <roh> basically half the stuff which is listed as 'doesnt work' out there for various reasons looks like 'stupid choice of hw' to me... e.g. using anything with either nvidia graphics or chipset is plain stupid and in the end will lead to 'strange' errors and bugs.
[19:22:56] <Jymmm> latency
[19:23:20] <roh> Jymmm did you test a t40p?
[19:23:38] <Jymmm> roh: I tested a LOT of laptops.
[19:24:17] <roh> laptop != thinkpad, and even there one has to be careful. (e.g. dont buy nvidia)
[19:24:27] <Jymmm> T23, T40, T60, 1500, Fujitsu,
[19:24:53] <Jymmm> roh: WTH are you alking about? All thinkpads are laptops. and always have been
[19:25:15] <Jymmm> Be they IBM or Lenovo
[19:25:20] <roh> Jymmm yes.. but not all laptop-makers have a clue how do not fuck it up completely.
[19:25:33] <roh> e.g. acer or fuji
[19:26:28] <roh> in the end a 40th class thinkpad is nothing else than a pentium-m class 'intel chipset solution' as close to the intel-recommendations as possible
[19:27:22] <Jymmm> roh: Look, you are adamant about using a laptop, you've been given warnings not to due to excessive latency, I wash my hands.
[19:27:26] <roh> and they are cheap and come in masses... about 300E for a fully equipped one used atm.. lots of formerly leased machines which now get replaced with new ones and are sold refurbished
[19:27:59] <roh> Jymmm i wanted a sane reason. you dont give one but fud around. thats what is pushing me
[19:28:37] <roh> especially since there is a tp600 listed as 'works'
[19:30:23] <roh> does not help people who have less clue about hardware than ourselves.
[19:33:15] <archivist> roh 99% of laptops have power management issues that cause high latency
[19:33:53] <roh> you can run a tp40 completely without any acpi and apm whatsoever.
[19:34:37] <roh> archivist am not recommending hw i think is sane without reason. after all one gets a bit of experience after >10 years of using linux
[19:36:19] <roh> i know people who use these devices as temporary servers. fits into 1U 19" easily and can hold 2 disks and has proper intel ethernet and stable cooling
[19:39:31] <tom3p> haha , so after finding this stuff ( mobo , a P3 850 slot1 cpu , and PC100 memory for it) , i see the max mem is 384meg, and ubu 8.04 emc live has minimum 384 meg. this is the typical end of many hours spent looking for suitable hardware for emc.
[19:40:25] <archivist> I have it on an AMD 800 meg and on an intel P4
[19:41:08] <Jymmm> tom3p: P3 can do 512MB, depending on mfg
[19:41:12] <roh> tom3p sounds good.. doesnt it work?
[19:42:17] <tom3p> the unit suggested has max 384 meg (3x128) ... sounds good? it barely makes the minmum requirement, not useful to pursue if its marginal
[19:42:56] <tom3p> http://www.intel.com/support/motherboards/desktop/se440bx2/
[19:44:42] <tom3p> archivist plz tell me the mobo, maybe i can find one of those
[19:45:20] <archivist> the P4 is an oddball, its integrated in the display
[19:46:10] <Jymmm> tom3p: Damn, intel bites in the ass again
[19:46:23] <roh> p4 is mostly 'hotter than p3 at ca same speed' ;)
[19:47:07] <Jymmm> tom3p: 512MB http://www.weirdstuff.com/cgi-bin/item/15007
[19:47:31] <Jymmm> tom3p: and comes with cpu and some ram
[19:47:50] <tom3p> thx, will look ( really Thx Jymmm)
[19:48:27] <Jymmm> tom3p: sorry about the previous one, I forgot that intel has a tendancy to limit ram on all their desktop mobo's
[19:49:00] <tom3p> np
[19:49:07] <Jymmm> tom3p: Now, I have a few server intel mobo's that exceed most... like P3 or 4 with 8GB ram =)
[19:49:25] <Jymmm> and mgmt via searial too
[19:49:44] <tom3p> if you've run the latency test & want to part with one, lemme know
[19:50:24] <Jymmm> tom3p: I guess I could bench it (if I can actualy FIND my bench)
[19:50:58] <tom3p> i gotta post pix of this minimill, its really solid
[19:51:07] <Jymmm> tom3p: HF???
[19:52:29] <tom3p> no, sable 2015 from taiwan, all alum frame, real 1.5mm pitch ball screws ( dbl nut), real linear guides. tight ( i build machines too )
[19:53:27] <Jymmm> tom3p: Well, were be the pics already?! Eeeeeeeesh
[19:53:41] <Jymmm> come on! Com on! HURRY UP DAMNIT!!!!
[19:53:55] <Jymmm> We're Waaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaiting =)
[19:54:04] <Jymmm> tap tap tap
[19:54:06] <tom3p> doh, finding stuff on multiboot, multi vm systems is hell. ( sorry i think i gotta reboot to linux )
[19:54:32] <roh> *g*
[20:02:00] <colin_> is the latency test mostly for stepper drives and par ports ?
[20:02:17] <archivist> no
[20:02:38] <colin_> can say iv every really bothered to do the test
[20:02:56] <colin_> cant
[20:03:52] <BigJohnT_> BigJohnT_ is now known as BigJohnT
[20:04:19] <colin_> is it less of an issue if you use fpga boards like mesa ?
[20:05:16] <Jymmm> tom3 http://www.intel.com/support/motherboards/server/l440gx/ 2GB and support for 1GHz cpu's
[20:07:45] <roh> colin_ ive ran the test some time, then used 16000, and after some weird errors, but nothing serious or bad happening ive upped it to 18000 and havent seen the warnings more than once since
[20:08:31] <colin_> iv only seen an rtai error when i put the live cd in my lappy
[20:09:29] <tomp> tomp is now known as tom3p
[20:09:29] <tom3p> http://imagebin.ca/view/wcYWz_v.html
[20:09:48] <tom3p> i had to gimp it down to sendable size
[20:09:58] <Jymmm> tom3 http://www.intel.com/support/motherboards/server/l440gx/ 2GB and support for 1GHz cpu's
[20:10:20] <tom3p> oh the hi rez went thru, cool
[20:10:56] <jmkasunich> what does it use for a spindle? I see a drill chuck?
[20:11:42] <tom3p> die grinder
[20:12:08] <tom3p> feels tight too, (push push bend)
[20:12:29] <tom3p> didnt pull chuck off yet
[20:12:49] <tom3p> the 2 plates are for mounting it ( a collar )
[20:13:13] <Jymmm> did that driver board come with it?
[20:13:47] <jmkasunich> JanVanGilsen: check other invocations of "reportError", I don't think that kernel code can print floats
[20:14:04] <jmkasunich> I could be wrong - if some other place is using it for floats then I probably am
[20:14:23] <jmkasunich> to be safe, try multiplying it by 1000 or something, then convert to int and print
[20:42:17] <tom3p_> Jymmm: yes, the driver board came with it
[20:47:41] <Jymmm> tom3p wheredid you buy it from?
[21:02:55] <tom3p_> ebay ( he didnt have one ready b4 i left taiwan, but no big diff in price)
[21:04:47] <tom3p_> that mobo has to have the chips on the memory set up a certain way and the inet sellers dont specify ( eg 32m X 8 on frontside of ds sdram and 32m x8 on rear side and only 2 pcs allowed )
[21:05:10] <tom3p_> per the intel app sheet
[21:05:40] <Jymmm> tom3p_: url?
[21:05:56] <tom3p_> uh minimill or app sheet?
[21:07:03] <tom3p_> app sheet http://download.intel.com/support/motherboards/desktop/d815eea/A1696401.pdf minimill just put sable 2015 into finder
[21:07:43] <tom3p_> table 5 in app sheet specs the mem cgfs allowed
[21:07:49] <tom3p_> cfgs
[21:13:56] <tom3p_> now this newegg kingston 512 pc133 mem shows 8 chips on 2 sides so must (?) be 32mx8, 32mx8 http://www.newegg.com/Product/Product.aspx?Item=N82E16820153101 ( got 1 solution :)
[21:14:35] <JanVanGilsen> *back
[21:25:01] <JanVanGilsen> jmkasunich, you're correct "%d",(int)(test*1000) works =)
[21:32:36] <tom3p_> sqr peg in round hole? just cast it & scale it a few times ;)
[21:40:19] <JanVanGilsen> I've got the homing with index only and variable offset working :)
[21:42:19] <JanVanGilsen> now I need to write a hal component that rounds down the position from te potentiometer to the nearest index :)
[22:04:45] <JanVanGilsen> well, i'm off to bed, thx for all the help
[22:06:52] <tom3p_> good work ( still looking at how to get n like modulus without seeking it in a loop)