#emc | Logs for 2010-10-23

[00:00:16] <andypugh> I am reading the datasheet for the driver
[00:00:18] <andypugh> http://snebulos.mit.edu/projects/voila/docs/datasheets/Oxford_QuadUART.pdf
[00:00:51] <andypugh> And that seems to hint that the EPP ports will be the 8bit ones. The 32-bit one is the serial ports
[00:03:44] <andypugh> Was there any documentation with the card? A driver CD?
[00:03:59] <paragon-wkshop> Only windows drivers
[00:04:33] <theorb> theorb is now known as theorbtwo
[00:07:30] <andypugh> You could also try each of the addresses and address + 0x400 for the high address, in case the hogh address is not being found.
[00:08:15] <paragon-wkshop> ok I'll give it a go!
[00:08:54] <andypugh> What did 0xd400 0xd480 and epp_wide off give you?
[00:09:27] <paragon-wkshop> did'nt work andypugh
[00:09:42] <andypugh> In what way didn't it work, though?
[00:10:30] <andypugh> And don't the docs on the CD tell you anything?
[00:10:43] <paragon-wkshop> PARPORT: ERROR: request_region(0xd480) failed | PARPORT: ERROR: request_region(0x400) failed
[00:11:17] <andypugh> Err, you missed out the "d" again
[00:14:17] <paragon-wkshop> god dam it may have worked...
[00:14:21] <paragon-wkshop> halcmd: loadrt hm2_7i43 ioaddr=0xd480 ioaddr_hi=0xd400 epp_wide=0 config="firmware=hm2/7i43/SVST4_4B.BIT num_encoders=2 num_pwmgens=1 num_stepgens=4"
[00:14:50] <paragon-wkshop> Oct 23 01:13:32 emc kernel: [ 5203.338244] hm2/hm2_7i43.0: IO Pin 042 (P3-37): StepGen #3, pin Step (Output)
[00:14:51] <paragon-wkshop> Oct 23 01:13:32 emc kernel: [ 5203.338249] hm2/hm2_7i43.0: IO Pin 043 (P3-39): StepGen #3, pin Direction (Output)
[00:14:53] <paragon-wkshop> Oct 23 01:13:32 emc kernel: [ 5203.338252] hm2/hm2_7i43.0: IO Pin 044 (P3-41): IOPort
[00:14:55] <paragon-wkshop> Oct 23 01:13:32 emc kernel: [ 5203.338256] hm2/hm2_7i43.0: IO Pin 045 (P3-43): IOPort
[00:14:56] <paragon-wkshop> Oct 23 01:13:32 emc kernel: [ 5203.338259] hm2/hm2_7i43.0: IO Pin 046 (P3-45): IOPort
[00:14:58] <paragon-wkshop> Oct 23 01:13:32 emc kernel: [ 5203.338263] hm2/hm2_7i43.0: IO Pin 047 (P3-47): IOPort
[00:15:00] <paragon-wkshop> Oct 23 01:13:32 emc kernel: [ 5203.338308] hm2/hm2_7i43.0: registered
[00:15:01] <paragon-wkshop> Oct 23 01:13:32 emc kernel: [ 5203.338312] hm2_7i43.0: board at (ioaddr=0xD480, ioaddr_hi=0x0000, epp_wide OFF) found
[00:15:08] <andypugh> That seems insane though, the hi address is below the low address
[00:15:27] <andypugh> But that does look to have worked
[00:15:31] <paragon-wkshop> The init and done light have switched off
[00:15:51] <andypugh> Good sign
[00:16:35] <andypugh> Still in halcmd?
[00:16:49] <paragon-wkshop> yeah but strange... Yes with halcmd.
[00:17:04] <andypugh> tab-completion should help with tis
[00:17:20] <paragon-wkshop> ioaddr_hi=0x0000 but specified ioaddr_hi=0xd400 ?
[00:17:29] <paragon-wkshop> ok
[00:17:34] <andypugh> setp hm2_7i43.0.led.0. 1
[00:17:43] <andypugh> (or something like that)
[00:17:59] <andypugh> You should be able to control the on-card leds
[00:18:25] <andypugh> (assuming the led driver is included in the emc2 release you have)
[00:18:37] <andypugh> Actually, type show pin to see
[00:18:59] <paragon-wkshop> lots of output
[00:19:07] <andypugh> Good
[00:19:15] <andypugh> Any with "LED" in the name?
[00:20:25] <andypugh> hm2_7i43.0.led.CR01 for example?
[00:20:37] <paragon-wkshop> no, have encoder, gpio, pwmgen, stepgen and watchdog.
[00:21:07] <andypugh> OK, I wasn't sure if that module had made it into the 2.4 branch yet
[00:21:17] <paragon-wkshop> I ma using the 10.04 2.4 release
[00:21:29] <paragon-wkshop> Oh ok
[00:21:41] <paragon-wkshop> can I get it through git?
[00:21:48] <andypugh> You have a multimeter?
[00:22:23] <paragon-wkshop> It's in the yard! What could I do with it?
[00:22:43] <andypugh> You could turn io pins on and off
[00:22:54] <andypugh> And see if the multimeter sees them
[00:23:11] <andypugh> Though you need the realtime threads running for that
[00:23:14] <paragon-wkshop> oh right and measure it with the multimeter
[00:23:42] <andypugh> I think you need to copy that config line into the INI file, and try starting EMC2 normally
[00:23:50] <paragon-wkshop> I may have a logic probe here
[00:24:40] <paragon-wkshop> what will the epp_wide=0 do to functionality the default=1 ?
[00:25:26] <andypugh> It will make it a little slower than it would be. You can try it with it on to see if it works that way too.
[00:26:52] <andypugh> You would need to unloadrt the 7i43 first, I think
[00:27:23] <paragon-wkshop> It's a no go with epp_wide=1
[00:27:57] <andypugh> I got that impression from Googling
[00:29:22] <andypugh> try again with "enable_raw" in the config file.
[00:29:53] <andypugh> (config string, I mean, and with epp_wide off)
[00:30:07] <andypugh> Actually, scratch that
[00:30:46] <andypugh> add "enable_raw" to the config string in the iNI file, and then start emc2 in the normal way
[00:31:06] <paragon-wkshop> andypugh: I just started emc with the updated config and I got a message Parport: linux parport parport1 does not support mode 4... continuing anyway
[00:31:31] <andypugh> Odd
[00:32:19] <paragon-wkshop> andypugh: so enable_raw with default epp_wide=1?
[00:32:43] <andypugh> No, with epp_wide off
[00:32:55] <paragon-wkshop> ok
[00:36:05] <paragon-wkshop> does this look of? loadrt [HOSTMOT2](DRIVER) ioaddr=0xd480 ioaddr_hi=0xd400 epp_wide=0 config="firmware=hm2/7i43/SVST4_4B.BIT enable_raw num_encoders=2 num_pwmgens=1 num_stepgens=4"
[00:38:00] <paragon-wkshop> that did not work. BTW the done and init leds remain off do I need to reset the board?
[00:39:01] <andypugh> That is in the HAL file?
[00:39:30] <paragon-wkshop> yes
[00:39:37] <andypugh> The LEDs stay off once the board has a valid firmware
[00:39:51] <andypugh> In what way did it not work?
[00:39:57] <andypugh> what does dmesg say?
[00:40:12] <paragon-wkshop> Oh I see... not like the pluto that reprograms every time it is started.
[00:41:32] <paragon-wkshop> hm2: loading Mesa HostMot2 driver version 0.15
[00:41:35] <paragon-wkshop> Oct 23 01:40:50 emc kernel: [ 6841.942035] hm2: unloading
[00:41:36] <paragon-wkshop> .....
[00:41:38] <paragon-wkshop> .....
[00:41:40] <paragon-wkshop> RTAI[hal]: unmounted
[00:42:30] <andypugh> pastebin it? Or was that really all it said?
[00:42:58] <paragon-wkshop> There was more... pasting now.
[00:43:27] <paragon-wkshop> http://pastebin.ca/1970450
[00:45:05] <andypugh> That's very odd
[00:45:43] <andypugh> Start emc from the comand line, maybe the error went to stdio not dmesg
[00:45:50] <andypugh> (just type "emc")
[00:46:30] <NTU> bash -x emc might give you more output too
[00:46:37] <paragon-wkshop> andypugh: I put the enable_raw at the end of the config line and emc has started but with ->
[00:46:39] <paragon-wkshop> PARPORT: linux parport parport1 does not support mode 4.
[00:46:40] <paragon-wkshop> PARPORT: continuing anyway.
[00:47:05] <paragon-wkshop> Axis is up.
[00:47:14] <andypugh> w00t!
[00:47:26] <andypugh> machine -> show hal configuration
[00:47:51] <paragon-wkshop> ok
[00:48:20] <andypugh> Then resize the window until the textbox and message box at the bottom are visible.
[00:48:54] <paragon-wkshop> done
[00:49:29] <andypugh> expand pins->hm2_7i43.0.raw
[00:49:48] <paragon-wkshop> done
[00:50:25] <andypugh> In the box at the bottom type "setp hm2_7i43.0.raw_write_address 0x0200
[00:50:43] <andypugh> (checking that that is spelled corectly)
[00:51:04] <andypugh> setp hm2_7i43.0.raw_write_data 0xFFFFFFFF
[00:51:24] <andypugh> setp hm2_7i43.0.raw_write_strobe 1
[00:51:26] <paragon-wkshop> and execute?
[00:51:35] <andypugh> execute or enter
[00:51:45] <andypugh> you should see it update on screen I think
[00:52:06] <paragon-wkshop> Error popup .. Invalid Argument
[00:52:23] <andypugh> You need to spell it right
[00:52:59] <andypugh> I am doing all this from memory
[00:53:15] <paragon-wkshop> It's very much appreciated :-)
[00:54:22] <paragon-wkshop> looking at hostmot2 manpage..
[00:54:25] <andypugh> Any luck?
[00:55:05] <andypugh> You should see the pin names on the screen, there should be no need for a manpage
[00:55:35] <paragon-wkshop> setp hm2_7i43.0.raw.write_data 0xFFFFFFFF
[00:55:54] <andypugh> Ah, yes, I got a _ for a .
[00:55:56] <paragon-wkshop> You were dam close! . not _
[00:56:17] <andypugh> And the address too.
[00:56:45] <andypugh> then the strobe should turn on the onboard LEDs
[00:58:04] <andypugh> Assuming that the LED register doesn't wander about from firmware to firmware
[00:58:33] <paragon-wkshop> no did not work | write strobe = 8 bit IN FALSE hm2_7i43.0.raw.write_strobe
[00:58:42] <paragon-wkshop> No leds are on
[00:59:38] <andypugh> Are the address and data values set? (you might have to switch to the "watch" tab to get a live view
[00:59:38] <paragon-wkshop> The box at the bottom says setting parameter 'hm2_7i43.0.raw.write_strobe' to '1' Pin 'hm2_7i43.0.raw.write_strobe' set to 1
[00:59:52] <pcw_home> Even if the register wanders, the driver should get its address from the ModuleID...
[01:00:22] <andypugh> He is using raw mode, as the Led driver is not in the 2.4 branch
[01:00:45] <andypugh> So we have assumed 0x200
[01:01:30] <pcw_home> Oh OK. Register locs are coinstan but will change in the future to pack them more densely
[01:01:42] <pcw_home> constant now
[01:02:13] <andypugh> paragon-wkshop: Do the address and data values match what you set them to?
[01:02:42] <andypugh> The strobe should immediately return to zero, though
[01:03:48] <andypugh> pcw_home: So it is imperative to use register_stride and instance_stride and not assume?
[01:04:50] <paragon-wkshop> andypugh: setp hm2_7i43.0.raw.write_data 0xFFFFFFFF changes 8 u32 IN 0x0FFFFFFF hm2_7i43.0.raw.write_data
[01:05:08] <paragon-wkshop> But the strobe stays False
[01:05:12] <andypugh> It will
[01:05:18] <paragon-wkshop> register_stride ?
[01:05:25] <pcw_home> Yes but there are some issues with more complicated modules like sserial and resmod
[01:05:49] <andypugh> paragon-wkshop: don't worry about register_stride, that was a different conversation
[01:06:16] <andypugh> what's the raw.write_address?
[01:06:22] <pcw_home> (and UART and twiddler)
[01:06:34] <paragon-wkshop> 0x00000000
[01:06:45] <andypugh> I told you to set it to 0x200!
[01:07:04] <andypugh> (Bah! you can't get the staff)
[01:07:26] <paragon-wkshop> When?
[01:08:08] <paragon-wkshop> you mentioned 'So we have assumed 0x200' but I thought that was for pcw_home
[01:09:24] <paragon-wkshop> Yep you did .... sorry ... it's a lot to take in and I missed it! setp hm2_7i43.0.raw_write_address 0x0200
[01:09:50] <andypugh> now set the strobe to 1
[01:10:27] <andypugh> But note the typo, its raw.write_address
[01:10:36] <paragon-wkshop> andypugh: WE have light :-)
[01:10:57] <andypugh> In that case we can definitively say that you have a working 7i43
[01:11:17] <paragon-wkshop> andypugh: what does the setp hm2_7i43.0.raw_write_address 0x0200 do?
[01:12:23] <paragon-wkshop> I have 4 leds fron cr5 - cr8 is that as expected?
[01:12:30] <andypugh> When the strobe is set high the data in raw.write_data is copied to the register in the 7i43 at address raw.write_address
[01:13:45] <andypugh> Yes, as you miss-typed the raw.write_data. Should be 0xFFFFFFFF (eight Fs)
[01:13:47] <pcw_home> Yes because you wrote 7Fs instead of 8
[01:14:08] <andypugh> Actually FFFF0000 would work too
[01:14:15] <paragon-wkshop> OK, but why specifically 0x0200
[01:14:33] <andypugh> Ah. well, that happens to be the register that stores LED sate
[01:15:54] <paragon-wkshop> Oh I see. How did you know that? From a spec sheet?
[01:16:06] <andypugh> Yes
[01:16:29] <paragon-wkshop> pcw_home: Yep I had 7F ;-) but now I have 8 :-)
[01:16:46] <andypugh> I am in the middle of writing a driver for the Mesa cards, so I have the regmap open on screen
[01:17:31] <pcw_home> regmap is somewhere in the distribution I think
[01:18:28] <andypugh> So I know that 0x5b00 is the smart-serial command register
[01:18:35] <andypugh> for example
[01:18:35] <paragon-wkshop> Did you download that from xilinx? The reason I ask is that I am keen to get into vhdl programming and have been reading a couple of tutorials.
[01:19:01] <andypugh> No, that is specific to the Mesa firmware
[01:19:48] <andypugh> You don't ned to know the register addresses, the hm2 drivers handle all that stuff.
[01:20:33] <paragon-wkshop> ofcourse... I got my microchip pic head on where the spec sheets have all the reg info, but of course fpga are just a bunch gates / logic.
[01:20:41] <andypugh> The only reason I had you using raw mode was so that we could see for sure that the card really was working, without having to attach extra stuff
[01:21:30] <andypugh> Right, with that minor triumph, I am off to bed.
[01:23:00] <paragon-wkshop> Thanks very much for help andypugh :-)
[01:23:13] <andypugh> No problem.
[01:48:21] <NTU> NTU is now known as NTU_brb
[06:06:02] <Jymmm> Jymmm is now known as Red70sShow
[06:06:19] <Red70sShow> Red70sShow is now known as Jymmm
[08:42:01] <MARKUSBEC> MARKUSBEC is now known as m4rku583C
[13:09:21] <Paragon39> Has anyone used eagle cad? I am trying to find the library that has a 50 way idc straight pcb connector!
[13:11:00] <cradek> plain row of 50 pins with .1" spacing? I think those kind are in 'pinhead' iirc
[13:12:12] <Paragon39> cradek: Looking for the same connectors that are on the 7i43 card it's for a breakout.
[13:17:02] <Paragon39> cradek: .1 = 2.5mm, I have found 50way 2mm in the library you sugested but not a 2.5mm
[13:18:48] <DaViruz> pinhd-2x25
[13:18:59] <archivist> .1 != 2.5mm
[13:19:21] <Paragon39> archivist: No?
[13:19:25] <DaViruz> exists in three versions, 0.1" angled, 0.1" straight, 2.0mm straight
[13:19:46] <archivist> its .1 which is 2.54mm
[13:20:31] <Paragon39> I was working to 1 decimal space ;-)
[13:20:58] <archivist> well having dome many pcb's that other place matters :)
[13:21:00] <archivist> done
[13:21:32] <DaViruz> if it's like 4way it might not matter, but 0.04mm error 25 times becomes alot
[13:21:44] <Paragon39> DaViruz: Got it thanks. Now I have to find the lib for these ... lol http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=360304241849&ssPageName=STRK:MEWNX:IT
[13:34:02] <Paragon39> Found them con-wago-500 w237-103 :-)
[13:41:38] <ries_> ries_ is now known as ries
[14:10:15] <lilalinux> is there a simple and free cam for linux that takes a dxf and lets you do radius correction (or even G41/G42)?
[14:10:27] <lilalinux> dxf2gcode isn't yet able to do that
[14:10:47] <pcncx> lilalinux must be why Linux ships with a compiler
[14:10:55] <lilalinux> hm?
[14:11:03] <cradek> heekscnc is pretty good at path offsetting and pocketing
[14:11:18] <cradek> whether it's simple I won't venture
[14:11:20] <pcncx> cradek hey good morning yo uda man!
[14:11:23] <lilalinux> cradek: thx
[14:11:41] <pcncx> cradek do you know who maintains the pdf files on linuxcnc?
[14:12:31] <cradek> cncx: depends which. some are periodically autogenerated from the docs that are in git, and if you mean those, the answer is in a roundabout way jthornton
[14:12:52] <cradek> if you mean other pdf files, the answer is probably nobody
[14:13:08] <pcncx> cradek does anybody ever look at them? Because i just did and the table of contents in them is sort of funny
[14:13:24] <pcncx> well in the EMC2_Integrator_Manual.pdf
[14:13:24] <cradek> you're asking me questions I can't answer...
[14:13:36] <pcncx> the numbers and the headings run together
[14:13:43] <cradek> ok, I think that one is autogenerated
[14:14:00] <pcncx> yeah well somebody should lok into why the stuff is overlapping
[14:14:07] <cradek> url please
[14:14:11] <pcncx> figured someone might want to know
[14:14:19] <pcncx> oh ah let me get it
[14:19:52] <pcncx> sorry about that
[14:22:11] <pcncx> hmm it only seems to do it when I choose fit width so I guess it is this dumb reader here
[14:23:35] <Jymmm> No, It does it down near 8.2.100
[14:24:00] <pcncx> yeah when the numbers get longer that is when it overlaps
[14:24:05] <Jymmm> actually ALL of chapt 8
[14:24:13] <pcncx> but when I wasn't in fit width I didn't see it
[14:24:34] <Jymmm> s/8/8.2/
[14:24:55] <Jymmm> 8.10
[14:25:26] <Jymmm> 10.2.10 is bad
[14:25:35] <pcncx> yeah that is where it is really bad for me too
[14:26:03] <Jymmm> 12 thru 13
[14:26:55] <Jymmm> actually chap 12 thru 36
[14:26:57] <pcncx> this machine has so little software on it i have to mail an xwd to my other machine to convert it
[14:28:15] <Jymmm> cradek: http://linuxcnc.org/docs/EMC2_Integrator_Manual.pdf
[14:30:21] <pcncx> http://img408.imageshack.us/img408/8923/contentso.png
[14:30:34] <pcncx> that is what i see
[14:31:01] <pcncx> just looks funky figured someone might like to know
[14:31:26] <pcncx> ignore the xterm in the top left
[14:32:14] <pcncx> on the brighter side I got my add on parallel port card installed into this box this morning
[14:32:26] <pcncx> and did an lspci -v on it too
[14:34:18] <pcncx> now i have to clear some junk away so I can setup my scope and see whos what
[14:34:56] <pcncx> unless someone here knows a better way of determining which port is which on the list
[14:35:50] <pcncx> I think that is how I figured it out the last time
[14:36:06] <Jymmm> Usually the slot closest to the CPU is the lower address.
[14:36:41] <pcncx> this card only fit in one of two slots because of how the machine is made
[14:36:49] <pcncx> its an oddball
[14:37:04] <Jymmm> so one onboard, one pci?
[14:37:25] <pcncx> well yes it has an onboard parallel port but I put in a PCI card too
[14:37:43] <pcncx> I want to run off the add in PCI card leave the integrated port alone
[14:38:12] <pcncx> but the way the PCI slots are in this machine the card would only fit into specific PCI slots
[14:38:31] <pcncx> they're not normal connectors it is one molded thing
[14:38:36] <Jymmm> 0378 or ?bc3 iirc is onboard commonly
[14:38:53] <pcncx> yeah but lspci gives me 5 addresses
[14:39:07] <pcncx> and only tw really coorspond to actual ports on the card
[14:39:10] <Jymmm> which are?
[14:39:10] <pcncx> two even
[14:39:29] <pcncx> and out of those two each only is to one port
[14:39:43] <pcncx> so I could just hook up my drivers and try each one out
[14:40:02] <pcncx> but that introduces other variables into the mess
[14:40:12] <pcncx> like do I have my driver timings setup right
[14:40:23] <pcncx> with the scope i could just see a pulse
[14:40:32] <pcncx> and be like hey this is it
[14:41:15] <pcncx> though if I recall correctly I'm going to need a port pulser
[14:41:34] <pcncx> I think i used this thing called K74.c
[14:42:01] <pcncx> so I am up for hearing about an easier method
[14:42:33] <Jymmm> how about answering my question?
[14:42:53] <pcncx> oh yeah 0378 is the onboard
[14:43:01] <pcncx> least I'm pretty sure
[14:43:11] <pcncx> think I've seen that somplace in this machine
[14:43:17] <pcncx> sounds familiar
[14:43:51] <pcncx> yeah it is in heinfo
[14:43:56] <pcncx> hwinfo even
[14:44:06] <pcncx> E-3400_hwinfo.txt:i/o:1 0x0378 - 0x037a (0x03) "parport0"
[14:44:28] <pcncx> I wonder what hwinfo would tell me with the card in it?
[14:52:38] <elmo42> one way to find out
[14:52:54] <pcncx> elmo42 oh I'm there already
[14:53:02] <elmo42> ;)
[14:53:14] <pcncx> it is a bit different a little more helpful
[14:53:27] <pcncx> it seems to marrow things down to two
[14:53:31] <pcncx> narrow even
[15:55:04] <Jymmm> Hi LawrenceG
[16:41:10] <LawrenceG> Jymmm, greatings from the north
[16:53:58] <Jymmm> Nanook of the north!
[16:55:19] <LawrenceG> no nook here!
[16:58:22] <Jymmm> heh
[17:01:12] <LawrenceG> Jymmm, pse send some s.cal nook!
[17:01:37] <Jymmm> Say what? lol
[17:02:12] <LawrenceG> nevermind... wife might object
[17:02:20] <Jymmm> PM it =)
[17:23:01] <pcw_home> http://www.ww.com/dilliecam
[17:24:45] <cradek> well that's unusual
[17:30:53] <pcw_home> House broken but man, ruminants have bad breath, being portable fermenters
[17:42:56] <archivist> there was a tv prog over here a few days ago about a south african household with a tame hippo
[17:43:34] <archivist> locking it out was hard word
[17:48:24] <pcw_home> Don't argue with a hippo
[17:48:52] <archivist> http://hubpages.com/hub/There_is_a_Hippo_in_our_house
[17:55:34] <pcw_home> Amazing
[18:25:27] <willburrrr2003> I am wiring my e-stop into my control panel today. It controls a relay in my panel will signal emc2 from one of its contacts, and the second contact will break power to the control relay that turns the spindle on/off stopping the spindle. I have one more contact left and was thinking of breaking the power supply feeding my hobbycncpro stepper driver board. Is it ok to just kill power to the
[18:25:28] <willburrrr2003> board while it is running the steppers in this way?
[18:27:54] <_AR_> so i was just working on my cnc frame
[18:27:59] <_AR_> started cutting a piece of metal
[18:28:05] <_AR_> the was like 'fuck this' and i quit
[18:28:08] <_AR_> then*
[18:29:18] <willburrrr2003> soo...you were cutting a piece of metal and your machine up and quit on you?
[18:32:11] <_AR_> no i quit
[18:43:56] <theorbtwo> You were cutting a piece of metal by hand for your frame?
[18:49:27] <_AR_> theorbtwo, well yeah
[18:49:43] <theorbtwo> (machine tools)++
[18:52:02] <_AR_> theorbtwo, ya i wish
[18:52:52] <Jymmm> willburrrr2003: Is cutting power to the driver board also disconnecting the filter caps that handle the BEMF?
[19:31:06] <_AR_> _AR_ is now known as AR_
[21:26:31] <tom3p> any examples of hal pid loop that are not velocity? ( like temperature or water conductivity ?)
[21:27:00] <andypugh> They all look the same
[21:30:06] <tom3p> hmm, the output might be a solenoid for a valve, so its a but more bitbangy than a velocity control using pwm
[21:30:48] <andypugh> PID doesn't really work with a digital control signal.
[21:31:14] <andypugh> You might as well use the window comparator.
[21:31:52] <tom3p> analog in ( from a ac volateg across carnbons in water, yields a voltage across a shunt, ) to a threshold, to a valve
[21:32:05] <andypugh> Though perhaps a PID with a lot of D would let you close the valve before it was too late :-)
[21:32:37] <andypugh> Is the valve full-open or full-closed?
[21:32:56] <tom3p> :)
[21:33:20] <tom3p> usually a valve is one of theose two when connected to a solenoid
[21:33:41] <tom3p> ah i didnt say solenoid the 2nd time
[21:34:14] <andypugh> You might find that a PID with a threshold in the PID output switching the valve might work. D would make it close a bit before necessary, then I might open it again as the undershoot went on.
[21:35:18] <tom3p> thx i can try that and just a hal window compartor ( i have to control water conductivity for a test monday )
[21:35:42] <andypugh> Can't you use mercury?
[21:37:22] <tom3p> uh a bit too conductive, not sure what you're asking
[21:37:48] <andypugh> I was being flippantly facetious.
[21:38:14] <tom3p> oh the question was suppository ;)
[21:39:00] <andypugh> (and backing up a bit, I could imagine a valve operated by solenoids and ratchets, that opened and closed in discrete steps, and that would be a challenge to code in HAL)
[21:39:06] <tom3p> i was surprised how many circuits and calibrations for DIY were on the net
[21:39:47] <tom3p> maybe use a water pump on a vfd ( thats possible with the hdwr i have at hand )
[21:40:32] <tom3p> oh a stepper based valve? used them, good thought
[21:42:21] <tom3p> i thought there was an example of temp control using pid in the reprap/emc stuff but cant track it down
[21:43:20] <tom3p> i think it may be here http://geo01005-ideas.blogspot.com/
[21:45:34] <tom3p> "trying to fully implement PID heater control for my EMC based repstrap"
[21:47:06] <andypugh> Heater via PWM is definitely sensible.
[21:47:40] <DaViruz> can emc run at all on non-realtime linux?
[21:47:55] <skinnypup> simulator
[21:49:10] <DaViruz> i see. i guess hal is too tied into rtai
[21:49:39] <pcw_home> EMCs software PWM at a few Hz would work with normal AC SSRs
[21:51:53] <andypugh> 1Hz PDM could work nicely for a solenoid valve
[21:52:13] <tom3p> pcw_home, thx! the dufuss who built used 230Vac coils on the solenoids, and i figger 1 or even .25hz is fast enif to control microsiemens in a 55 gallon tank
[21:52:44] <skinnypup> automotive isv <idle control> valve comes to mind
[21:53:22] <andypugh> Eh? Idle control is all electronical (and is immensely complex)
[21:54:04] <tom3p> thnx again, have to look back into PDM
[21:54:08] <cradek> I think some iac valves are solenoid run by pwm (but some are stepper motors too)
[21:55:33] <tom3p> really? displacement of solenoid by pwm? like voltage-magField versus return spring ?
[21:55:53] <cradek> yes I think so
[21:56:20] <tom3p> thats pretty cool, like a cheapo 'rc servo'
[21:56:34] <skinnypup> some work on solonoid duty cycle, and some are steppers varying the opening
[21:57:02] <tom3p> woof, well i best go try some stuff now, thx
[21:58:05] <andypugh> <off at a tangent> Idle control is an interesting example of PID. The final output is torque demand (which is internally converted to fuel demand and then open-time for the various injections, depending on the current injection pattern). It is basically a PID except that the P, I and D terms are pulled out of 3D maps depending on things like how far from setpoint you are, the coolant temperature and such factors. Then you ch
[21:58:05] <andypugh> whole different set of the same maps depending on which gear you are in. (engineers spend months idle-driving up and down hills and over bumps of various shapes in all gears to tune the resonances out and avoid stalls).
[21:58:31] <cradek> wow
[21:59:35] <cradek> recently I had a rental car that couldn't drive 5mph. it could either coast/idle or accelerate to a speed way over 5mph, based on .1mm of movement of the accelerator pedal.
[21:59:51] <cradek> (I was in manhattan where driving 5mph is an important thing for a car to be able to do)
[22:00:30] <cradek> ... obviously they didn't test that kind of condition
[22:01:32] <andypugh> You don't get to 2000 16x16 maps and 40,000 individual variables by avoiding complexity :-)
[22:01:53] <andypugh> What car?
[22:02:16] <cradek> it was a hyundai model mumblemumble
[22:02:21] <andypugh> I think it is unlikely to have been a Ford diesel, so I am probably off the hook for that.
[22:02:27] <cradek> :-)
[22:03:08] <andypugh> I have been driving the new Focus this week, blimey that car is full of toys.
[22:03:11] <cradek> I'm jaded though - often when I drive an automatic I wonder "didn't those morons try driving this thing before they started selling them!?"
[22:03:44] <andypugh> There is a rubber strip that pops out and wraps round the door edge when you open the door (all mechanical)
[22:03:57] <cradek> this thing was toyish too - lots of controls that did stuff I didn't care about - that must make it seem expensive and luxurious to suckers
[22:04:16] <cradek> andypugh: sounds like a mechanical nightmare
[22:04:27] <andypugh> it has a reversing camera that draws lines on the screen depending on steering input to show where you will end up
[22:04:37] <cradek> haha
[22:05:23] <cradek> this thing stuck itself in overdrive and stayed there even if you slowed down to 35, against all obvious signs that it wasn't going to work out
[22:05:25] <andypugh> For parallel parking there is no need for that. You tell it you want to park, it measures gaps using the distance sensors till you drive past a big enough space, then it steers you in while you work the forwards-reverse.
[22:05:40] <cradek> andypugh: you're shitting me?
[22:06:13] <andypugh> It knows the speed limits and shows you them on the dashboard (GPS database)
[22:06:32] <cradek> ah - nice way to sell a subscription service
[22:06:54] <cradek> "for only $39.99/month, it'll stop whining that you are speeding when you're not"
[22:06:59] <andypugh> http://www.motorator.com/videos/605
[22:07:21] <andypugh> It is a feature on my GPS, so seems to be free
[22:08:50] <andypugh> That video shows auto park it on a Lincoln, but the Focus is aiming at $20,000
[22:09:32] <cradek> "This is a great feature for our customers (... who are laughably unskilled drivers)"
[22:09:54] <tom3p> casafrog said he fixed GL hash in emc by loading libgl-mesa-swx11, when i try to do that synaptic says it's going to remove gnome desktop
[22:10:21] <cradek> do you mean ubuntu-desktop?
[22:10:34] <tom3p> lemme try again
[22:11:22] <tom3p> yes ubuntu-desktop
[22:11:31] <cradek> that one doesn't matter
[22:11:31] <tom3p> isnt that fatal?
[22:11:37] <tom3p> uh ok
[22:11:38] <cradek> nope
[22:11:48] <tom3p> thx
[22:11:55] <cradek> it's a package that does nothing but depend on packages that ubuntu thinks are important
[22:12:27] <tom3p> ok, it just sounded bad
[22:12:36] <cradek> I agree it does
[22:13:02] <andypugh> Oh, I forgot voice-control, adaptive cruise control (it slows down if there is a car in front) and the city-safety auto-braking that Volvo developed. That works really well: http://www.youtube.com/watch?v=aNi17YLnZpg&feature=related
[22:13:05] <cradek> you may have more-than-usual trouble if you do an OS upgrade to the next LTS in the future
[22:13:16] <cradek> so you might want to reinstall ubuntu-desktop before trying that.
[22:13:31] <cradek> auto braking is an awesome idea
[22:14:34] <cradek> andypugh: nice video - it did stop fast, that's for sure
[22:15:08] <cradek> "I would never trust a computer to do my breaking." </youtube commenting moron>
[22:15:28] <cradek> "I would never trust a computer to do my spelling." </me>
[22:16:41] <andypugh> When do the realtime threads start? When the HAL file finishes? At the first addf?
[22:16:52] <cradek> halcmd start
[22:17:17] <alex_joni> andypugh: you usually set up the whole HAL config, then do start
[22:17:21] <andypugh> I mean in the generic emc startup sequence?
[22:17:24] <alex_joni> so everything is up by then
[22:17:31] <alex_joni> after loading the HAL files
[22:17:46] <andypugh> Hmm.
[22:17:47] <alex_joni> (except for postgui stuff, that happens later)
[22:18:24] <alex_joni> andypugh: or is there a start in one of the files, I don't quite remember
[22:18:31] <andypugh> So, there is potentially quite a long wait between exiting the hostmot2 driver and the threads starting.
[22:18:43] <cradek> yes, up to forever
[22:19:46] <andypugh> I wonder how the watchdogs work?
[22:19:58] <alex_joni> andypugh: http://git.linuxcnc.org/gitweb?p=emc2.git;a=blob;f=scripts/emc.in;h=67fde33e8ffc9f7aa8cfa1d863f6700072f53655;hb=HEAD#l640
[22:20:03] <cradek> good question
[22:20:20] <alex_joni> I think there's a special thingie before the first petting
[22:21:03] <andypugh> That's what what worries me, as the sserial needs it's own petting.
[22:22:50] <andypugh> I think I know how to do it. It doesn't need "petting" until the first "do it", so as long as the last thing the driver does is a restart, it ought to be fine.
[23:33:06] <DaViruz> ouch, absolute encoders were not cheap even on ebay
[23:34:37] <andypugh> How are you interfacing them to EMC?
[23:35:32] <DaViruz> i'm not, this is for a different project
[23:35:41] <DaViruz> azimuth feedback for my radio telescope
[23:37:41] <andypugh> Unusual toy.
[23:38:00] <DaViruz> i'm an unusual kind of guy
[23:38:31] <andypugh> Looking for anything in particular?
[23:39:04] <DaViruz> aliens would be nice, bit i'll settle for quasars
[23:40:00] <andypugh> I wondered if it was a SETI style thing. Otherwise it seems a lot less immediate than optical.
[23:40:44] <andypugh> I don't expect you get the same "ooh! Pretty" with a radio telescope?
[23:40:53] <DaViruz> no, not really :)
[23:41:04] <DaViruz> especially not with my crappy angular resolution
[23:41:28] <DaViruz> to get any real detail you need crazy big dishes
[23:42:17] <andypugh> Anything from Gliese 581g?
[23:42:46] <andypugh> (They reckon there is a planet with liquid water there)
[23:43:09] <DaViruz> 20.5ly away :/
[23:44:09] <andypugh> Going there would be something of a commitment, 'tis true. But time dilation would make the decades fly by.
[23:44:50] <DaViruz> not much help for us who get left behind :/
[23:46:42] <andypugh> Should I get back to my FTL drive project then?
[23:47:19] <DaViruz> please do
[23:47:42] <DaViruz> never mind einstein crying in his grave
[23:48:01] <DaViruz> don't mind, rather
[23:49:13] <DaViruz> i can't reach gliese 581g from here :/
[23:50:26] <DaViruz> much of the sky is obscured by the earth at this latitude
[23:50:55] <andypugh> Build a tower.
[23:51:17] <DaViruz> i'd rather move closer to the equator
[23:51:24] <DaViruz> this has a few other pleasant side effects
[23:51:55] <DaViruz> no ice on the windshield in the mornings, and no snow up to the waistline..
[23:52:17] <andypugh> Sweat, mosquitos, sunburn. No thanks.
[23:52:58] <DaViruz> that's true, i dislike those more
[23:55:35] <andypugh> I like to visit snow, I confess it is not much fun when you only have bikes.
[23:59:07] <DaViruz> maybe i should get a bike