#linuxcnc-devel | Logs for 2015-06-25

[07:30:21] <skunkworks> zlog
[07:34:21] <skunkworks> I could just make a comp that reads a gcode file so that when I give it a Z position - I get an x - y position back. Easy peasy :)
[08:25:49] <jepler> ooh the "dragonboard 410c" is actually shipping -- 64-bit ARM for <$100. arrow says mine will ship tomorrow. http://www.arrow.com/campaigns-na/qualcomm/dragonboard-410c/
[08:32:49] <jepler> kernel is based on 4.0.0 so -rt should be possible
[08:32:53] <jepler> just what I needed, a new project!
[08:37:03] <jepler> .. 1.8v SPI on a 2mm-pitch connector .. sound familiar?
[08:39:41] <pcw_home> a purpose built FPGA card could just use 1.8V on one FPGA bank
[08:52:23] <pcw_home> And at least its built to some kind of standard (not just the hot dev board of the week)
[08:52:25] <pcw_home> and... the I/O pinout is in the 96 board standard Jolly good!
[08:59:32] <pcw_home> no provision for mounting daughterboards though :-(
[09:03:01] <pcw_home> Hmm Maybe there is (one picture shows soldered in spacers but others do show empty mounting holes)
[09:10:05] <Tom_itx> Seems like sale partner arrow is unable to provide the power supply connector
[09:10:12] <Tom_itx> hope you have one
[09:11:20] <pcw_home> I like the 7-18V input
[09:13:12] <skunkworks> better yet - give it the xy feedback and it also calculates the following error...
[09:14:23] <pcw_home> is there a peak detector comp?
[09:14:35] <skunkworks> yes
[09:14:53] <skunkworks> well - I assume so - that is how I have peak acc/vel...
[09:15:04] <skunkworks> just don't remember how I did it
[09:17:40] <pcw_home> next step is RMS error :-)
[09:22:13] <skunkworks> my nexus has a snapdragon.. I really like it.
[09:22:47] <skunkworks> 15.04 is out?
[09:23:01] <skunkworks> oh - that isn't lts
[09:25:07] <jepler> right
[09:26:15] <jepler> pcw_home: there's a component called "minmax"
[09:28:20] <skunkworks> hmm
[09:28:21] <skunkworks> https://groups.google.com/forum/#!topic/machinekit/FlCoP6Q2Tu0
[09:28:41] <skunkworks> that makes me think that they don't have their machine setup correctly...
[09:31:49] <pcw_home> Preemt-RT 4.05-rt4 seems at least as good as 3.18 on X86 (been up a week on my Core Duo at 3 KHz)
[09:32:49] <jepler> if I never read the word "m------k--" again I would be a happy man
[09:33:04] <archivist> skunkworks, the ability to change axis order was recent ish is that in the other codebase
[09:33:34] <skunkworks> jepler, sorry
[09:38:24] <jepler> skunkworks: I don't fault you for trying to be a helpful participant in all these different communities.
[09:39:29] <pcw_home> LTS maybe somewhat less important with Preempt-RT since its simple to build a suitable kernel
[09:39:31] <skunkworks> the preview if setup correctly should work though - right? THis looks right
[09:39:32] <skunkworks> https://youtu.be/R1DCXe9t3UE?t=1m22s
[09:40:37] <jepler> in axis in linuxcnc there are still limitations on the kinds of rotation that will be previewed correctly
[09:40:51] <pcw_home> linuxcnc/mint17.1 runs ok with Preemt-RT for example
[09:41:40] <jepler> but I believe that many 4- and 5-axis systems can be correctly previewed at least when the center of rotation is around 0,0 in the respective plane
[09:45:43] <skunkworks> heh - the tool doesn't quite look like it is pointing in the right direction in the preview
[09:48:31] <jepler> someone, maybe archivist, has a sane-sounding rotary axis setup that I failed (some time ago) to figure out the correct GEOMETRY string for
[09:49:44] <archivist> my 5 axis is still on an old emc2 so none of that nicety for me
[09:51:24] <jepler> I wonder who I'm thinking of then
[11:51:58] <seb_kuzminsky> jepler: hm2-encoder-dpll (357ce4229) looks really good to me, feel free to push to 2.7
[11:53:39] <mozmck> seb_kuzminsky: I think Curtis is probably talking about rebasing on master
[11:54:11] <jepler> seb_kuzminsky: thanks, will merge to 2.7 and push, hopefully this evening.
[11:54:17] <mozmck> Or maybe he doesn't realize that is how it is normally done?
[14:33:53] <mozmck> For some reason emails I send to emc-developers takes a long time to show up.
[14:39:16] <Tom_itx> maybe they are spam filtered
[14:40:11] <Tom_itx> all my emails to the Mrs go straight to her spam folder :(
[14:49:41] <mozmck> heh, I don't know. The list is on sourceforge, and they come through, but it takes hours or even a day or so sometimes.
[14:50:17] <Tom_itx> one reason i like irc.
[14:50:23] <mozmck> me too.
[15:11:07] <mozmck> jthornton: shouldn't it be "Hardware Abstraction Layer" instead of "Hardware Abstract Layer"? http://www.linuxcnc.org/docs/2.7/html/
[15:46:49] <mozmck> jepler: is your "[PATCH] WIP do iptables and sysctl configuration automatically" change one you are going to push?
[15:47:13] <jepler> mozmck: I'd like to. were you able to test with it?
[15:47:17] <mozmck> It seems to work with a 7i92 here.
[15:47:29] <mozmck> I may not know how to test as well as I should though.
[15:47:30] <jepler> ok
[15:48:25] <mozmck> I ran latency tests and things looked good - without doing the sysctl and iptables stuff manually.
[15:48:33] <jepler> OK
[15:48:38] <jepler> thank you for testing
[15:48:58] <mozmck> I did list the iptables rules and it shows the hm2_eth rule, and it gets populated while linuxcnc is running.
[15:50:11] <jepler> that's good testing too
[15:52:26] <jepler> I'll dust it off and ask seb for review before pushing to 2.7.
[15:53:51] <mozmck> ok, thanks! It will be helpful.
[16:19:00] <KGB-linuxcnc> 03Sebastian Kuzminsky 052.6 14a41db 06linuxcnc 10docs/src/gui/gladevcp.txt docs: fix a markup typo in gladevcp docs * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=14a41db
[17:06:53] <KGB-linuxcnc> 03Jeff Epler 052.7 ffc13ed 06linuxcnc Merge branch 'jepler/hm2-encoder-dpll' into 2.7 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ffc13ed
[17:26:31] <KGB-linuxcnc> 03Jeff Epler 05jepler/hm2-eth-simplify 66f0669 06linuxcnc 10docs/man/man9/hm2_eth.9 10src/hal/drivers/mesa-hostmot2/hm2_eth.c hm2_eth: do iptables and sysctl configuration automatically * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=66f0669
[17:27:12] <jepler> mozmck: I updated that patch, do you want to test again? I didn't change anything substantial.
[17:46:55] <PCW> I think I have a relatively simple way to run multiple hm2_eth cards efficiently (when I ever get some free time)
[17:48:56] <jepler> on one ethernet interface?
[17:49:06] <PCW> yeah
[17:49:40] <jepler> there are a few places the driver will have to change to support >1 7i80, whether that's on 1 or more ethernet interfaces
[17:49:44] <PCW> with a gige switch
[17:50:15] <jepler> now you have to worry about latency in the switch too?
[17:50:57] <PCW> yeah theres store+forward time but thats about it on good switches
[17:52:32] <PCW> I somehow had the mistaken impression that GigE didn't allow small packets but that's only true for half duplex
[17:53:58] <jepler> hm, and even the latency wouldn't matter terribly if the driver could send all its read packets and then receive all the responses
[17:54:07] <jepler> rather than sending one read packet and waiting for response, etc
[17:54:20] <jepler> not sure right offhand if that's a big change to the driver structure..
[17:55:16] <PCW> yeah it would be best if the driver could send all packets in one call not sure if that;s in the network API
[17:56:11] <jepler> that would be a step beyond what I was thinking of
[17:56:16] <PCW> on GigE the time between the ~100 byte packets is very small
[17:56:56] <PCW> the switch is basically a concentrator
[17:58:03] <PCW> the 10/1 ratio in speeds means the store+forward delays are minor
[18:57:43] <linuxcnc-build> build #3223 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/3223 blamelist: John Thornton <bjt128@gmail.com>
[19:14:26] <Tom_itx> PCW you around?
[19:23:17] <linuxcnc-build> build #3214 of 1200.rip-lucid-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1200.rip-lucid-i386/builds/3214 blamelist: Sebastian Kuzminsky <seb@highlab.com>
[19:36:13] <linuxcnc-build> build #3224 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/3224 blamelist: Sebastian Kuzminsky <seb@highlab.com>
[20:29:51] <PCW> Tom_itx: Yeah
[20:37:52] <Tom_itx> wondering about the 7i47s
[20:38:10] <Tom_itx> if you have any
[20:39:00] <Tom_itx> i haven't checked this one complete yet but at least the PWM isn't pwming on it and possibly the spindle
[20:39:21] <Tom_itx> i probably should have gotten one with the other board
[20:39:42] <Tom_itx> but not all the 5v stuff was blown out
[20:41:44] <Tom_itx> i could put the 7i47 back on but i don't really wanna rewire it yet again
[20:46:50] <PCW> fixed the store...
[20:47:39] <Tom_itx> i did test those questionable IO on the 7i90 and they're working fine
[20:48:35] <PCW> yeah I guess just the HV on the bus switch power done-em in
[20:49:14] <Tom_itx> do you know off hand what would go on the 7i47s if the 5v got too high?
[20:50:04] <PCW> well unfortunately almost every chip uses 5V
[20:50:10] <Tom_itx> figures
[20:50:23] <Tom_itx> i may not bother with it right now
[20:50:36] <PCW> (unlike the 7I90 thats almost all 3.3V)
[20:50:53] <Tom_itx> yeah, i didn't see anything that popped
[20:51:06] <Tom_itx> but it's not outputting anything
[20:53:03] <PCW> I can send you a untested one without connectors if my Wife doesn't find out
[20:53:04] <Tom_itx> does the analog part use 5v as well? or just from it's own input?
[20:53:19] <Tom_itx> i don't mind
[20:53:38] <Tom_itx> i've got plenty of connectors by now :D
[20:54:23] <PCW> the analog part is driven by the ACT04 inverter (which is likely the first to go from overvoltage)
[20:55:09] <PCW> (that is the analog section OPTOs are driven by the ACT04)
[20:55:11] <Tom_itx> when i enabled the spindle it went full on so it's probably bad
[20:55:45] <PCW> probably the analog section and OPTOS are OK
[20:56:14] <Tom_itx> i'll do some more poking around with it when i get a chance
[20:56:24] <PCW> (since the OPTOS have series resistors on the 5V side and no 5V connections on the other)
[20:56:28] <Tom_itx> i think i'll go ahead and get one from the store
[20:56:43] <PCW> bbl dinner time
[20:56:53] <Tom_itx> or if you have a used one etc
[20:57:59] <Tom_itx> i'll just be glad when i can get back to feature testing instead of fixing boards
[21:05:06] <Tom_itx> ordered