#linuxcnc Logs

Jun 10 2019

#linuxcnc Calendar

01:19 AM Deejay: moin
04:57 AM Tom_L: morning
05:11 AM Tom_L: jthornton, http://tom-itx.no-ip.biz:81/~tom-itx/irc/logs/%23linuxcnc/2019-06-09.html
05:11 AM Tom_L: 09:26 PM
05:19 AM jthornton: morning
05:19 AM XXCoder: yos
06:05 AM elmo40: moo
08:48 AM cerna: Generally speaking the RT threads running in kernel space have lower latency than user space ones, right? On that was based the RTAI based PP support, right?
09:17 AM * jthornton is making headway with the J1900 and latency with Lubuntu 19.04!
09:27 AM mozmck: jthornton: run this in a terminal and it will tell you the isolcpus setting you need: lscpu -p | tac | awk -F, '{ if(FNR==1) {lastcore=$2; cpu=$1; cpu0=$1} else if ($2==lastcore && cpu0>1) {cpu = $1 "," cpu} } END {if(cpu0>0) {print "isolcpus=" cpu} }'
09:28 AM mozmck: another setting you need to prevent random lockups is intel_idle.max_cstate=1
09:28 AM mozmck: and for some graphics bugs: i915.enable_rc6=0
09:31 AM jthornton: ok, thanks
09:31 AM jthornton: I got the j1900 down from 250us to 30us
09:31 AM mozmck: rcu_nocbs= and nohz_full=1,2,3 should be set to the same core(s) as isolcpus. I've not used those settings but it seems that I tested with them a little a while back and did not see a whole lot of difference. It could depend on the kernel version though.
09:31 AM mozmck: Not bad! What all did you do?
09:31 AM jthornton: slightly older kernel and disable ondemand
09:32 AM jthornton: I'll write it up and add it to my uspace page
09:32 AM jthornton: bbl
09:32 AM mozmck: Ah, yes. I disable freq scaling in the kernel config altogether. I also had the better latency with 4.1.x or earlier than with any newer kernel so far.
09:34 AM pcw_home: In general I dont care for the IsolCPUs options as it throws away CPU power, though it does seem to be needed in some cases
09:34 AM pcw_home: (the RPI definitely needs it)
09:35 AM jthornton: I tried disabling the governor in the kernel config but it kept going back to powersaver
09:35 AM jthornton: time for my 9:30 post office run
09:36 AM mozmck: pcw_home: isolcpus really helps on J1800 and J1900. Probably not as much on some of the other systems we've tested. It doesn't make a noticeable difference in system response
09:45 AM pcw_home: mozmck: does it help with Ethernet latency or just latency-test latency?
09:46 AM mozmck: I think both - we were only testing latency, but especially with J1800 we were getting following errors on occasion without it.
09:47 AM mozmck: Some also on J1900. They went away completely with isolcpus
09:47 AM pcw_home: have you tried my hal file hack that skip-cycles late/lost packets?
09:47 AM mozmck: Yes, I'm using that now - thanks!
09:48 AM mozmck: I did the isolcpus before that though - so I haven't tested with the hal hack an no isolcpus
09:48 AM pcw_home: that should fix the following error issue for lost packets
09:48 AM mozmck: Yes. I *think* though that the problem was occasional high latency.
09:48 AM pcw_home: what isolCPUs setting are you using? I may try on my J3355
09:49 AM pcw_home: (thats a 2 core more modern equivalent of the J1800)
09:49 AM mozmck: The best way to figure it out is run that command I sent earlier. I think on the J1900 it is just isolcpus=3
09:49 AM mozmck: Oh, that would just be isolcpus=1 then
09:50 AM mozmck: On some cpus 2 cores will share cache, and both need to be isolated - that script figures all that out.
09:50 AM mozmck: lscpu -p | tac | awk -F, '{ if(FNR==1) {lastcore=$2; cpu=$1; cpu0=$1} else if ($2==lastcore && cpu0>1) {cpu = $1 "," cpu} } END {if(cpu0>0) {print "isolcpus=" cpu} }'
09:51 AM mozmck: On my 8 core desktop that returns: isolcpus=6,7
09:52 AM pcw_home: I'm curious to try a Ryzen, looks like the newest may really give Intel a run-for-the-money performance wise
09:53 AM mozmck: Yes, even the first Ryzen's did. I'm still running an 8350 and it is not bad. If it would just crash I might justify buying a Ryzen - but it just keeps running!
09:55 AM pcw_home: Ill have to try some Ethernet statistics (with the histogram component) with isolCPUs
09:56 AM mozmck: How are you doing the ethernet staticstics now?
09:56 AM cerna: pcw_homepcw_home: The J3355 does not have dual L2 cache, so you cannot isolate whole cache to RT.
09:56 AM pcw_home: just lookin at the max times
09:56 AM mozmck: I've been doing that but not with histogram - how do you do that?
09:57 AM mozmck: cerna: hmm. You can still use isolcpus but it probably won't be as helpful I bet.
09:57 AM pcw_home: D Garret expanded hist latency histogram to a general purpose histogram component
09:58 AM pcw_home: s/hist/his/
09:58 AM mozmck: Ah - I should figure out how to use that.
09:59 AM pcw_home: yeah its pretty neat
10:13 AM cerna: mozmck: Hmm, BTW do you by chance know where to reliably find such info like which cores are on which cache and such? It is always such pain.
10:30 AM mozmck: cerna: not offhand
10:38 AM mozmck: cerna: maybe from the datasheets?
10:40 AM cerna: mozmck: Yeah, sure, datasheets are the first think I look at. (It's not really important, I just though out loud.)
10:40 AM cerna: mozmck: Thanks.
10:42 AM mozmck: No problem. I haven't looked real hard for that - just pretty much check on the systems I'm testing on. The stuff I'm doing needs to be pretty generic - so I use that script in setup to figure it out dynamically
10:59 AM CaptHindsight: pcw_home: recent Ryzen experience https://github.com/NTULINUX/aggressive_kernel_patches
11:00 AM CaptHindsight: turn your 4+ core into a 2 core for the cache reasons to lower latency
11:01 AM CaptHindsight: 2 core Celeron 784 boards are ~22,000
11:02 AM CaptHindsight: as long as you aren't compiling much they seem to run LCNC just fine
11:03 AM CaptHindsight: ^^ with RTAI
11:47 AM cerna: When running the parallel port driver for step/dir control and encoder readout, the base thread of 25us (I assume update frequency) is common or do people change this?
11:49 AM pcw_home: Its standard to tweak the base thread rate depending on the host latency
11:50 AM pcw_home: CaptHindsight, not sure why some CPUs need no tweaks: http://freeby.mesanet.com/h97-g3258-preemt-rt.png
11:53 AM cerna: pcw_home: Even down, or only upwards? And I guess the complete step is then 2 x thread frequency, right?
11:53 AM CaptHindsight: pcw_home: it would help if you could trust BIOS
11:53 AM pcw_home: cerna up or down depending...
11:54 AM CaptHindsight: but it is such a pile of crap that you can't, and the tools they give the vendors to make ROM images is also broken
11:54 AM pcw_home: CaptHindsight: yeah especially if they can tweak stuff that not accessible by kernel options
11:55 AM CaptHindsight: you have no way of knowing if an option is ON or OFF unless you notice some obvious change in system behavior
11:56 AM CaptHindsight: yes, there are all sorts of legacy register settings in the CPU and chipset that BIOS has access to
11:57 AM pcw_home: yeah it may be just chance doe to BIOS settings (or Intel ME crap)
11:58 AM CaptHindsight: this all could have been better and cleaner but Intel didn't care and the Linux kernel devs 20 years ago WANTED to rely on BIOS
11:58 AM CaptHindsight: "don't take my BIOS away"
11:59 AM pcw_home: (was able to increase Ethernet thread rate from 1 KHz to 4 KHz on a HP by turning off AMT)
12:00 PM pcw_home: (AMT looking over your shoulder on all Ethernet access... nice)
12:00 PM CaptHindsight: it's for your protection
12:00 PM CaptHindsight: you from "us"
12:01 PM CaptHindsight: they wanted completely out of band functionality but can share the source with you since "think of what would happen if this gets into the wrong hands"
12:01 PM CaptHindsight: can/can't
12:03 PM CaptHindsight: so we went from spaghetti in assembly to EFI spaghetti
12:31 PM CaptHindsight: https://www.automationtechnologiesinc.com/products-page/nema-23/voltage-control-stepper-controller-n-driver/
12:31 PM CaptHindsight: microprocessor embedded, voltage control, miniature stepper motor controller. It can be adjusted through an analog voltage.
12:33 PM CaptHindsight: Speed control through 0-5V voltage input 0.2-900RPM
12:33 PM CaptHindsight: Speed Range 8 Hz – 102K Hz, 0.15 - 1912.5 RPM (1.8° Motor)
12:34 PM CaptHindsight: Switch control start/stop, direction, enable
12:34 PM CaptHindsight: hmm so 5v control like a servo BUT you have to use a STOP/START to get it to fully stop
12:53 PM CaptHindsight: https://www.omc-stepperonline.com/new-arrivals.html/closed-loop-stepper-driver-0-30a-24-48vdc-for-nema-11-14-17-stepper-motor
12:55 PM miss0r: 'evening
01:14 PM cerna: <CaptHindsight "https://www.omc-stepperonline.co"> > encoders(Encoder resolution must be 1000-line)
01:14 PM cerna: That doesn't sound very promising.
01:15 PM CaptHindsight: looks like that is why their steppers have a 1000 line encoder as an option
01:16 PM CaptHindsight: 1000 line/4000 count or is it a 250 line/1000 count?
01:16 PM CaptHindsight: count/ppr, ppr=pulse per rev
01:20 PM * cerna sent a long message: < https://matrix.org/_matrix/media/v1/download/matrix.org/wYSUNFMmfiKoRHhlmSzkCRBh >
01:20 PM cerna: From https://www.omc-stepperonline.com/download/CL42T.pdf page 4
01:21 PM cerna: Bloody IRC with it's message limit.
01:21 PM cerna: It's 1000 PPR.
01:22 PM CaptHindsight: yeah :(
01:22 PM CaptHindsight: well it's from Nanjing :)
01:24 PM cerna: So?
01:27 PM CaptHindsight: so?
01:27 PM CaptHindsight: ever been to Nanjing?
01:30 PM cerna: Nope.
01:34 PM CaptHindsight: https://time.com/5603571/china-restricts-technology/
01:35 PM CaptHindsight: How will we ever be able to cut corners on our own?
01:42 PM cerna: Finally, they are starting to throw their weight around.
01:49 PM roycroft: we should not want to use china's technology
01:50 PM roycroft: to me it's a serious national security concern
01:51 PM roycroft: that said, i don't think the huawei hissy fit that the trump administration are throwing is based on national security concerns at all
01:51 PM roycroft: it's just another attempt at leverage in the trade dispute
01:52 PM roycroft: anyway, i think it's fine if china (or any other nation) decides to withhold technology from us
01:52 PM roycroft: we should be more self-reliant
01:54 PM gloops: 5G isnt just a matter of faster texting
01:54 PM gloops: huawei can bring the country to a standstill, literally, if they control our 5G
01:55 PM gloops: and with all this things, we need to be doing it ourselves to keep up in the tech race
01:55 PM rmu: 5G? whats that?
01:56 PM CaptHindsight: what I find interesting is how the US upholds US patents filed by the Chinese, but China tends to not uphold patents in China
01:57 PM CaptHindsight: it depends on how big you are in China and how well connected
01:59 PM gloops: rmu 5G is that new mobile phone thing
01:59 PM CaptHindsight: rmu: higher data rates, wireless
01:59 PM gloops: if we dont do our own 5G we will forget how to do this stuff, and then they really have got us by the balls
02:00 PM gloops: 5G is a revolution, the machine start to communicate with one another
02:00 PM CaptHindsight: stop letting idiots run co's that build things
02:00 PM gloops: like your washing machine and your car will be chatting with your kettle
02:00 PM rmu: but why
02:01 PM CaptHindsight: rmu: it's more for 24/7 connections to media
02:01 PM CaptHindsight: and spying on your car talking to your toaster
02:01 PM rmu: but why
02:02 PM rmu: why would i want that
02:02 PM rmu: what does it do that LTE and / or wlan doesnt
02:02 PM CaptHindsight: rmu: get everybody on the same hardware/software platforms...
02:02 PM CaptHindsight: control the cattle
02:03 PM CaptHindsight: rmu: oh, it's over a handful of providers that track you 24/7
02:03 PM CaptHindsight: wlan is just local
02:03 PM rmu: no
02:03 PM CaptHindsight: not controlled by those who know better than you :)
02:04 PM rmu: no difference to cell service now
02:04 PM CaptHindsight: the difference is higher speeds
02:04 PM rmu: this wholke 5g business and how it enables this and that is complete and utter BS
02:04 PM CaptHindsight: sure
02:05 PM rmu: higher speeds @ lower range
02:05 PM CaptHindsight: but you'll be forced to use it after 3g and 4g are obsolete
02:05 PM rmu: that won't become obsolete for a long time
02:05 PM rmu: 5g doesnt have range
02:06 PM rmu: microwave-stuff in the tens ghz
02:06 PM CaptHindsight: 900mhz cell seems like yesterday to me
02:06 PM rmu: 900mhz in us? whats that? GSM?
02:07 PM CaptHindsight: pre GSM
02:07 PM CaptHindsight: analog cellular
02:07 PM rmu: extremely wasteful use of spectrum IIRC
02:07 PM rmu: even GSM was not very optimal
02:08 PM rmu: and had "artificial" problems with range and speed of handsets
02:08 PM CaptHindsight: https://images-wixmp-ed30a86b8c4ca887773594c2.wixmp.com/f/03116b88-73ee-4ebf-999b-320bd6f8f4a8/d9wdows-a7274f04-796a-4a21-b073-c7d47e513391.jpg/v1/fill/w_1032,h_774,q_70,strp/motorola_brick_cellular_phones_by_redfield_1982_d9wdows-pre.jpg?token=eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJzdWIiOiJ1cm46YXBwOjdlMGQxODg5ODIyNjQzNzNhNWYwZDQxNWVhMGQyNmUwIiwiaXNzIjoidXJuOmFwcDo3ZTBkMTg4OTgyMjY0MzczYTVmMGQ0MTVlYTBkMjZlMCIsIm9iaiI6W1t7ImhlaWdodC
02:08 PM CaptHindsight: I6Ijw9MTEyNSIsInBhdGgiOiJcL2ZcLzAzMTE2Yjg4LTczZWUtNGViZi05OTliLTMyMGJkNmY4ZjRhOFwvZDl3ZG93cy1hNzI3NGYwNC03OTZhLTRhMjEtYjA3My1jN2Q0N2U1MTMzOTEuanBnIiwid2lkdGgiOiI8PTE1MDAifV1dLCJhdWQiOlsidXJuOnNlcnZpY2U6aW1hZ2Uub3BlcmF0aW9ucyJdfQ.7xVXdL7MwpUkfyPQqtMYiuf20dpx-cqtfcC195_cnic
02:08 PM CaptHindsight: holy bejeus
02:08 PM CaptHindsight: sorry
02:09 PM CaptHindsight: https://www.wired.com/wp-content/uploads/archive/images/slideshow/2009/02/gallery_cell_phone_history/Motorola_DynaTAC_8000X.jpg this was you hand held cell phone
02:09 PM rmu: i know those motorola bricks
02:09 PM rmu: miami vice-era mobile phones
02:10 PM rmu: standby-time on battery measured in hours
02:16 PM CaptHindsight: https://en.wikipedia.org/wiki/5G#New_radio_frequencies
02:18 PM CaptHindsight: so yeah it's for very close with any sort of barrier e.g. walls, trees, dirty laundry, , not through walls, etc
02:19 PM CaptHindsight: rmu: the big thing is the femto, pico, micro cells everywhere
02:23 PM CaptHindsight: my bet is that they will also use mesh networking but keep tracking a #1 priority
02:27 PM andypugh: Anyone fancy making a spindle speeder? I reckon this would work, but don’t have the time to make it to test.
02:27 PM andypugh: https://a360.co/2Iya5k3
02:27 PM andypugh: No gears, friction drive. Only 7 machined parts.
02:28 PM andypugh: It’s based on a supercharger design that was used in production cars, so should work pretty well.
02:29 PM CaptHindsight: what surfaces provide the friction?
02:30 PM CaptHindsight: ^^ link doesn't load for me
02:30 PM andypugh: How long dod you give the link?
02:31 PM andypugh: Works with Chrome and Safari, so what are you using?
02:31 PM CaptHindsight: 2 minutes so far
02:31 PM CaptHindsight: FF
02:31 PM andypugh: Ah. Do you have Chrome?
02:31 PM CaptHindsight: no
02:32 PM andypugh: It’s a 3D model viewer by Autodesk, I doubt that it is in any way W3 compliant.
02:32 PM mozmck: I can view it fine here in FF
02:32 PM CaptHindsight: Konqueror no luck either
02:32 PM andypugh: But if you can’t see the pictures then there seems little chance of explaining it.
02:33 PM mozmck: Does the outer ring have to be held from turning?
02:33 PM DaViruz: pretty much a planetary gear set with friction drive instead of tooth drive?
02:33 PM andypugh: Yes, That’s the reactin ring, and also the spring that holds the bearings in cotact with the high-speed spindle shaft
02:34 PM andypugh: Yes, it is entirely a planetary gear set with no teeth.
02:34 PM JT-Shop: andypugh: I ordered that M30 x 3.5 tap
02:34 PM mozmck: Why would that be better than a planetary gear set then? Isn't slippage possible?
02:34 PM andypugh: The main objective was to use as many off-the-shelf parts as possible.
02:34 PM DaViruz: ease of manufacture i imagine
02:35 PM mozmck: I guess it would be easier to make
02:35 PM andypugh: Yes, it’s meant to be easy to make. Also it should be smoother than a set of gears running at 10,000 rpm.
02:35 PM andypugh: It’s a 10x speeder.
02:35 PM CaptHindsight: bigger rollers turn smaller roller
02:36 PM mozmck: It looks neat for sure!
02:36 PM CaptHindsight: what is the ratio?
02:36 PM CaptHindsight: in:out
02:36 PM andypugh: I am not sure :-)
02:36 PM CaptHindsight: ~2-3x
02:36 PM andypugh: I think it is about 10:1
02:37 PM CaptHindsight: ah
02:38 PM andypugh: Yes, 100mm (and a bit) outer ring and 10mm central shaft.
02:38 PM andypugh: So 10:1 to go with my 1000 rpm main spindle.
02:39 PM andypugh: That viewer has an explode-tool and slider that might make the construction more obvious.
02:40 PM CaptHindsight: rollers ned to be strong enough to handle the RPM and also not expand enough to make them lock up
02:40 PM CaptHindsight: ned/need
02:41 PM andypugh: The rollers are only doing twice spindle speed.
02:52 PM Tom_L: when i add axisui.notifications-clear-info to the following line
02:53 PM Tom_L: net tool-changed axisui.notifications-clear-info iocontrol.0.tool-changed <= hal_manualtoolchange.changed
02:53 PM Tom_L: i get an error: http://tom-itx.no-ip.biz:81/~webpage/cnc/configs/error/axisui_error.png
02:53 PM Tom_L: but hal configuration shows: http://tom-itx.no-ip.biz:81/~webpage/cnc/configs/error/axisui_error1.png
02:53 PM Tom_L: any ideas?
02:54 PM Tom_L: setp also generates the same error on that pin in the main.hal file
02:54 PM Tom_L: however if i put it in the postgui.hal file it does not generate an error but setp has no visible effect on the pin
02:55 PM JT-Shop: your not loading it in the postgui.hal file
02:56 PM Tom_L: no but i thought i'd try it there to see
02:56 PM Tom_L: seems to be the order things load in
02:58 PM JT-Shop: no axis pins are there until after all the regular hal files have ran
02:58 PM JT-Shop: that pin is not there yet in the main.hal file
02:58 PM Tom_L: so how can you assign it to a net without getting an error?
02:59 PM JT-Shop: it has to be in the postgui file
02:59 PM perry_j1987: just got a notification that a threadmill has been delivered
02:59 PM JT-Shop: put this net tool-changed axisui.notifications-clear-info in your postgui file
02:59 PM Tom_L: so i need to move the manualtoolchange pin to the postgui.hal file....
03:00 PM JT-Shop: no
03:00 PM JT-Shop: just any lines that have axis pins
03:00 PM Tom_L: that's where the signal comes from
03:00 PM JT-Shop: perry_j1987: I looked in the mailbox but didn't see a threadmill
03:00 PM perry_j1987: heh
03:00 PM JT-Shop: that's?
03:01 PM perry_j1987: terrified im going to break the thing
03:01 PM Tom_L: hal_manualtoolchange.changed
03:01 PM JT-Shop: are you sure they delivered to the correct address lol
03:01 PM JT-Shop: paste.ubuntu.com your postgui hal file
03:01 PM perry_j1987: i got a parallel breakout board enroute now to get this 6040 converted to linuxcnc as well
03:01 PM Tom_L: the notificatons come from a tool change so i figured i'd use that to trigger the clear
03:02 PM JT-Shop: perry_j1987: good idea
03:02 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/configs/NEW_MILL_2.7/
03:02 PM Tom_L: there's the whole thing
03:03 PM Deejay: gn8
03:03 PM Tom_L: that was before i added the axisui.notification-clear
03:03 PM Tom_L: which is in the main hal file
03:03 PM Tom_L: right now
03:06 PM Tom_L: JT-Shop, there's nothing about axisui in the postgui.hal file right now
03:07 PM JT-Shop: just testing here
03:07 PM Tom_L: i can put it there but it will be with the hal_manualtoolchange.changed pin
03:07 PM Tom_L: i was reading something else about loading
03:07 PM Tom_L: something about a 2 pass but i didn't read it all, it was late
03:08 PM JT-Shop: yea you don't need to worry about 2 pass
03:08 PM JT-Shop: I put this line in the postgui.hal file for the axis sim and it does not error net tool-changed axisui.notifications-clear-info
03:09 PM Tom_L: assigned to a trigger pin does it change state?
03:09 PM Tom_L: i just tested it with setp there and nothing changed
03:09 PM Tom_L: i'll try a 'signal' in a little while
03:10 PM JT-Shop: yea setp may only work once, I assume you have to toggle the pin after the message to clear it
03:10 PM Tom_L: i didn't get an error with setp axisui.notification-clear in the postgui.hal either
03:11 PM Tom_L: i've got an idea what's going on now at least
03:11 PM JT-Shop: what messages do you get for a tool change?
03:11 PM Tom_L: i don't remember, i'd have to check again
03:11 PM Tom_L: i get 2 per tool change though
03:12 PM Tom_L: i'll take a screen shot of it once i get in the door here
03:12 PM JT-Shop: I don't get any, is your post generating that?
03:12 PM JT-Shop: ok
03:13 PM Tom_L: it's not an error message, just information
03:21 PM JT-Shop: putting net tool-changed axisui.notifications-clear-info in the postgui.hal file does clear any messages at tool change
03:22 PM Tom_L: i'll give it a try here in a bit
03:23 PM Tom_L: you don't have to assign it a value?
03:26 PM JT-Shop: it?
03:26 PM * JT-Shop sure was glad he screwed that 5" x 7" 14ga plate to a 1x6 before drilling the 22mm holes in it
03:29 PM Tom_L: you don't have to assign "net tool-changed axisui.notifications-clear-info" a value?
03:29 PM pcw_mesa: 22 mm holes in 14 ga can make a nice finger trimmer if it grabs...
03:29 PM Tom_L: net needs 2 parameters i thought
03:29 PM JT-Shop: no, it's a bit
03:30 PM JT-Shop: aye, didn't want my fingers trimmed with a small part like that
03:30 PM JT-Shop: net creates a net, it must have a signal name tool-changed and a pin axisui.notifications-clear-info
03:31 PM JT-Shop: a net being a connection between a signal and a pin
03:31 PM JT-Shop: didn't jmk come up with hal?
03:31 PM pcw_mesa: Yes
03:32 PM pcw_mesa: Its a very EE/Net-listly kind of thing
03:35 PM pcw_mesa: One of LInuxCNCs best features ( but that's from a EE point of view )
03:37 PM JT-Shop: yea hal made a huge change in LinuxCNC
03:38 PM JT-Shop: I never could get a BTi to install on anything then someone came out with the LiveCD and bam I was able to run EMC2
03:45 PM Tom_shop: ok i added: net message-clear axisui.notifications-clear into the postgui.hal file
03:45 PM Tom_shop: and i still get messages
03:47 PM Tom_shop: nevermind
03:48 PM Tom_shop: i see what i did now
03:50 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/configs/error/toolchange_message.png
03:50 PM Tom_L: that's the message i was getting fwiw
03:51 PM andypugh: That looks to be a message from a remap sub.
03:51 PM Tom_L: it's got remap in the config yes
03:51 PM Tom_L: but i'm not currently using it
03:51 PM andypugh: So the simpler solution would be to edit the remap routine if you don’t want the messages
03:51 PM Tom_L: i could just remove remap now
03:51 PM Tom_L: since i'm not using it currently
03:51 PM andypugh: I am pretty sure that you are using it.
03:52 PM Tom_L: i added it when i was testing spindle orient
03:52 PM andypugh: Those messages are almost certainly from that
03:53 PM Tom_L: axisui.notifications-clear clears them now
03:53 PM Tom_L: but it has to be in the postgui which i wasn't aware of yesterday
03:54 PM andypugh: Tom_L: Yes, there the messages are, in your “change.ngc” file (not the toolchange.ngc file)
03:54 PM Tom_L: there's a few things in the config i really should clean up
03:55 PM Tom_L: good to know where the messages come from though
04:39 PM jthornton: pcw_mesa: are you running uspace on a J1900?
04:42 PM cgipython: pastebin
04:42 PM cgipython: sorry wrong window
04:42 PM jthornton: lol
04:49 PM pcw_mesa: I have ( A Gigabyte brix BXBT 1900)
04:50 PM jthornton: what kernel are you running?
04:51 PM pcw_mesa: This was a while ago so probably 4.1 something (I can check)
04:52 PM jthornton: I'm running 4.19.1-rt3 at the moment and I'm in performance mode but still get ~165us latency
04:53 PM jthornton: if I reset it stays around ±36us for a while
05:23 PM tubes4branes: hi
05:41 PM abrock: hi
05:44 PM abrock: I have a cnc mill and I abuse it for taking fotos. I'd like to automate the process in the following way: Every time the CNC machine hits M01 it stops, executes an external program which triggers the camera and resumes as soon as the external program terminates
05:44 PM JT-Shop: M100-199 do that not M1
05:47 PM Tom_L: JT-Shop, thanks for helping clear that up for me..
05:49 PM _unreal_: got to love south florida, I just went up into the attic to try and find something. oh and I JUST took a shower. I'm now drenched, I'm going to have to rinse off later on...
05:49 PM _unreal_: oh and sweet. just got my limit switches in the mail
05:49 PM abrock: JT-Shop: this is perfect, thank you
05:54 PM abrock: taking 81 fotos with 40 seconds between each is a horrible way to spend time, can't do anything serious, always waiting for the movement to finish
05:55 PM * jthornton waves from down in the beer cave
05:56 PM jthornton: no problem Tom_L
05:56 PM abrock: also "Did I take a foto from this position already? O shit, I moved twice and didn't take a foto in between, need to restart the program"
05:56 PM Tom_L: i do need to go thru my config and clean a few things up
05:57 PM jthornton: mine always need some cleanup lol
05:57 PM andypugh: abrock: Are you filming machining, or using the mill as a camera platform?
05:57 PM Tom_L: i was testing a few things for a while
05:57 PM Tom_L: rigid tapping, spindle orient etc
05:58 PM Tom_L: but i'm not really using either one
05:58 PM abrock: andypugh: using the mill as a camera platform
05:58 PM jthornton: I like to just create a new configuration for things like that, that's why my qtpyvcp mill has 6 versions lol
05:58 PM andypugh: If the latter, then you could set up the camera shutter on the mist-coolant pin, for example.
05:59 PM abrock: the camera is controlled via usb
05:59 PM Tom_L: jthornton, are you gonna try to threadmill that M30 thread?
06:00 PM andypugh: Though an M100-M199 might be better in some ways, as you can send a pulse-release sequence rather than M7 / G4 P1 / M9
06:00 PM jthornton: no, too expensive... I'm going with a $15 chinlee tap that andypugh found on fleabay
06:01 PM Tom_L: ahh well yeah for 15 bux probably would too
06:02 PM abrock: actually, two cameras. One is a system camera and I get the fotos via gphoto2 and the other is a very cheap usb "webcam" with no autofocus or similar. The second is only for verifying that the first one's image stabilizer etc. doesn't ruin the orientation
06:02 PM andypugh: Chinese small print “guaranteed for threads in bread dough and weaker pastries. Also for mercury (temporarily) and sodium on a good day. Lead is right out”
06:02 PM Tom_L: at least he has somebody to blame now
06:02 PM abrock: hehe
06:03 PM abrock: lead is a horrile material, I recommend not using it
06:03 PM andypugh: abrock: OK, so M100 for one camera, M101 for another, and a little bash script to setp the parport pin, pause, and reset.
06:04 PM andypugh: (you can probably send a message to USB from the bash script too. I leave the details as an excercise :-)
06:04 PM jthornton: lead wine glasses didn't do well
06:05 PM jthornton: andypugh: could you use a digital out and in as well?
06:06 PM andypugh: jthornton: lead crystal has been around for centuries, and AFAIK is still considered fine to use?
06:07 PM abrock: nah, I just write one program which triggers both cameras at the same time. I already have that, I just need to put it on the PC and name it accordingly. Currently it's on my laptop and I trigger it by hand
06:07 PM andypugh: digital IO sounds like it might be a bad starting point for a USB camera
06:07 PM jthornton: yea glass, I was talking about pure lead glasses that change the taste of wine but also kill you
06:07 PM andypugh: Google is suggesting : streamer -c /dev/video0 -b 16 -o outfile.jpeg
06:08 PM andypugh: I think of glasses being made of glass. Do you mean lead goblets?
06:08 PM abrock: I'm using OpenCV for the webcam, very easy
06:08 PM andypugh: I have a bit of a sideline in white-metal drinking vessels.
06:09 PM andypugh: But hopefully only tin, zing, antimony and bismuth are to be found.
06:10 PM abrock: Earlier the mill cut my USB cable and I'm very cross with it
06:11 PM jthornton: that the ones you repair the handles and stuff?
06:12 PM andypugh: Well, replacing the glass bottoms is the big job.
06:13 PM jthornton: is the bottom held in place by peening the metal over?
06:13 PM andypugh: Currenlty I have one waiting for attention made in 1903.
06:14 PM jthornton: wow
06:14 PM Tom_L: andypugh so if i'm not using remap, 'run from here' will work ok in 2.8?
06:14 PM andypugh: No, there is a ring soldered in that clamps up the glass and the rubber / silicone seal
06:14 PM andypugh: Tom_L: Within limits.
06:14 PM _unreal_: Oh I'm so close
06:14 PM Tom_L: no worse than in 2.7?
06:15 PM _unreal_: I'm only minus 2 things being fully setup
06:15 PM andypugh: Certainly no worse.
06:15 PM Tom_L: fair enough
06:15 PM _unreal_: my limit switches. my limit switch wire interface which Ineed to go out and drill all the holes on the PCB
06:15 PM _unreal_: AND a spindle motor PSU/setup for control
06:15 PM andypugh: I think you probably have to take care of spindle speed, coolant and the right tool yourself.
06:16 PM _unreal_: all motors are in place
06:16 PM _unreal_: andypugh, you have what that was built in 1903?
06:17 PM andypugh: A pewter tankard
06:17 PM Tom_L: andypugh i noticed a few sitting around on your living room centerpiece
06:17 PM andypugh: Yesm that’s one of those in the bicycle pics.
06:18 PM andypugh: _unreal_: https://youtu.be/bSOyh4JrluI if you want the whole story
06:19 PM andypugh: Hmm, that link seems to start half-way through.
06:19 PM andypugh: Though that could be just me
07:20 PM _unreal_: wow I just had an I am DUMB moment
07:20 PM _unreal_: was going nuts probing the wire hardness trying to make sure the switches worked
07:20 PM _unreal_: but I was testing the unpluged wire
07:20 PM _unreal_: and going WTF......
07:23 PM andypugh: Something a lecturer told me at college “The result of an experiment is never wrong. But perhaps you were not doing the experiment you thought you were doing?”
07:26 PM Tom_itx: :)
07:27 PM Tom_itx: i removed remap in the config and all is well with the world again
07:39 PM tiwake: https://tiwake.com/worn_out_screw.jpg
07:40 PM Tom_itx: just re'broach it the next size up
07:41 PM andypugh: tiwake buy a rotary broach the next size up and repair it. Shouldn’t coust much more than $100 for the tooling.
07:41 PM tiwake: heh
07:41 PM tiwake: pretty sure a good rotary broach is a lot more than $100
07:42 PM tiwake: I'm pretty happy with my new camera too
07:43 PM tiwake: spent $1,700 on camera, tripod, couple lenses and a couple other things
07:43 PM tiwake: it better be good
07:49 PM _unreal_: I must say that is the first timme I have ever seen someone turning a cup
07:49 PM _unreal_: :)
07:50 PM _unreal_: ok well I think I'm going to go get chines or subway not sure which its 8:45pm and my daughter needs to eat
08:44 PM CaptHindsight: anyone know why JT has a howto for Stretch + LCNC intsalls when there is a Stretch ISO here?
08:44 PM CaptHindsight: http://www.linuxcnc.org/testing-stretch-rtpreempt/
08:44 PM CaptHindsight: why the need for this? http://gnipsel.com/linuxcnc/uspace/debian9-rt.html
08:44 PM Tom_itx: so you can select your own desktop etc
08:45 PM Tom_itx: or just for fun
08:45 PM CaptHindsight: thanks
08:45 PM CaptHindsight: so either work
08:45 PM Tom_itx: yes
08:45 PM CaptHindsight: good
08:45 PM Tom_itx: afik
08:45 PM Tom_itx: i think i did both at one point
08:46 PM CaptHindsight: was wondering if he found some gotchas with the LCNC ISO
08:46 PM Tom_itx: he's been doing several howto builds
08:47 PM CaptHindsight: have been using Sid
08:48 PM CaptHindsight: want to try Stretch
08:52 PM CaptHindsight: Tom_itx: know if there is a RTAI kernel in the Linuxcnc Stretch ISO
08:55 PM CaptHindsight: looks like preempt_rt only
08:55 PM Tom_itx: sorry, phone. yes only preempt-rt
09:02 PM CaptHindsight: thanks
09:03 PM Tom_itx: np
10:00 PM Tom_itx is now known as Tom_L