#linuxcnc Logs

Sep 05 2023

01:23 AM W1N9Zr1 is now known as W1N9Zr0
01:34 AM Deejay: moin
03:30 AM travis_farmer: Morning :-)
03:35 AM rigid: Has anyone ever made an effort to put threads on the separated isolated CPU cores?
03:36 AM rigid: afais, the isolcpus cmdline doesn't have a big effect when heavy slow stuff and lightweight quick stuff runs on the same core while other cores are idling
03:37 AM Scopeuk: you are usually better leaving that to the os for most things, it is aware of which core share cache etc to optimise general case performance. if you are running 4 threads on a 4 core and they are all bursty you might be able to beat it
03:39 AM rigid: Scopeuk: well, taskset shows otherwise. "leaving that to the os" is generally not a good idea for a dedicated system. linux defaults are usually set to maximum stability/compatibility
03:40 AM Scopeuk: the linux scheduler is very configurable, although the "completely fair scheduler" which became the default somewhere late 4 series early 5 series if I remember correctly probably hurts the performance of a linux cnc system
03:40 AM travis_farmer: my machine runs just fine without any isolcpus...
03:42 AM rigid: Scopeuk: scheduling is only an issue with a lot of processes. but even if there are just 4 processes running, latency can be improved by putting every process on a separate core
03:42 AM rigid: travis_farmer: let me guess: fast box + MESA card
03:43 AM travis_farmer: maybe
03:43 AM rigid: i'm trying to get that software stepping going >50khz
03:43 AM rigid: should be easily possible in theory. but i'm not sure if there are code changes needed
03:44 AM rigid: i'm trying my luck with cgroups
03:45 AM CloudEvil: :)
03:45 AM travis_farmer: sounds like when you get a honda civic to run 1000HP. can be done, but there are easier ways. why not just run a Mesa board?
03:46 AM robert1: rigid: look at busywait8 branch
03:55 AM robert1 is now known as rmu
03:55 AM rmu: rigid: look at busywait8 branch
03:55 AM rigid: travis_farmer: yeah, with a more hefty CNC i'd definetly need more I/Os but for now: 1. those are not exactly cheap and 2. I gut a 4-core 1.8GHz beast here. it's _easily_ capable of doing tens of Mhz on the GPIO port - in theory. In practice with a realtime OS running, at least a MHz should be possible.
03:55 AM rmu: https://github.com/LinuxCNC/linuxcnc/tree/busywait8
03:55 AM rigid: does that turn off multitasking? :)
03:55 AM rmu: isolcpus turns off multitasking on one core, busywait does what it says, it spins the cpu and doesn't use syscall to wait for deadline
03:55 AM rmu: AFAIK
03:55 AM rmu: so for software stepping, you would want to isolcpu two cores, run base thread on one core and servo thread on the other isolated one
03:55 AM rmu: will probably need some code changes but for first experiments taskset should suffice
03:58 AM rigid: when I run a halrun script, currently I get for "taskset --all-tasks -cp $(pgrep rtapi_app)": task running on 0-3, first child process=0, 2nd and 3rd childs=3
03:59 AM rigid: while on CPU 0 should just be used for linux stuff and childs should run on separate cores (that receive no - or almost none - interrupts)
04:02 AM rmu: you can pin interrupts to cores.
04:02 AM rmu: taskset --all-tasks -cp 7695
04:02 AM rmu: pid 7695's current affinity list: 0-5
04:02 AM rmu: pid 7696's current affinity list: 0-4
04:02 AM rmu: pid 7706's current affinity list: 5
04:02 AM rmu: sorry for paste. 7706 is servo thread. this system has no base thread.
04:03 AM rmu: everything else is running on core 0-4.
04:03 AM rigid: yeah, just the timer interrupt seems to go to all cores, at least I don't know otherwise
04:03 AM rmu: this is with isolcpus=6 on a 6 core amd something something
04:04 AM rmu: sorry isolcpus=5 of course
04:04 AM rigid: ok, if you only have one thread, that should be fine. but if you had a 2nd (time critical) thread, it would also run on 5 (the highest core actually, no matter if it's isolated or not)
04:05 AM rigid: your system is probably beefy enough to easily handle other stuff happening on core 5 like interrupts
04:05 AM rmu: what i proposed was 1) isolcpus two cores 2) use taskset to pin one thread to one of isolated cores each
04:05 AM rigid: i mean, +1GHz can do _A LOT_ of stuff in a few us
04:06 AM rigid: yep, taskset is probably easier than cgroups
04:06 AM rmu: that system is a 2010 vintage amd phenom, wouldn't call it particularly beefy in today's standards
04:07 AM rmu: for 50kHz step rate you will want latency to be in single digit µs range at most
04:07 AM CloudEvil: It's depressing how not completely 100% true that statement is.
04:08 AM rmu: CloudEvil: which one
04:08 AM CloudEvil: (single core performance basically stopped going up in 200?)
04:09 AM CloudEvil: 1997 computers are comedically slower than 2010 ones, 2010 to 2023 really not the same jump alas.
04:10 AM rigid: we can all agree that a 2010 x86 system is more than enough for a CNC controller :)
04:10 AM rigid: (depending on the actual CNC of course)
04:10 AM Scopeuk: most of the performance bump 2010 to 2023 has been on the io side
04:10 AM Scopeuk: well and performance per wat
04:11 AM CloudEvil: And multicore.
04:11 AM CloudEvil: And inserting more vulnerabilities due to speculative execution in every functional unit.
04:12 AM rigid: rmu: i don't grasp the concept of max. latency and problems introduced quite, yet. If I understand correctly, the problem is that linuxcnc does not more than one pulse per cycle (in my case of step/dir) and uses the kernel for timing pulse length/space
04:14 AM rigid: while - in theory at least - it should be able to send more than one pulse and use some highres timer for timing in one cycle
04:14 AM rigid: but i'm still struggling with linuxcnc actual architecture. so... :)
04:14 AM Tom_L: morning
04:14 AM rigid: moinmoin
04:15 AM Tom_L: the arch is a patchwork of bandaids
04:15 AM rigid: my first impression was actually of a solid arch. solid enough to work with a lot of bandaids :)
04:16 AM Tom_L: there _have_ been major improvements over the past 10 yrs
04:16 AM rigid: feature wise? or system architecture wise?
04:22 AM Tom_L: some of both
04:22 AM Tom_L: path planning improvements etc
04:23 AM Tom_L: not sure how much core improvements have been done.. i'm not a software engineer
04:24 AM rmu: rigid: "sending more than one pulse" means busywaiting between pulses
04:26 AM rmu: step rate of 50kHz means 100k signal transitions per second, ideally spaced 10µs, so if your uncertainty of wakeup is larger than 5µs there will be problems. even if it is around 1µs step timing will not be very uniform
04:28 AM rmu: next thing is, how do you get 100kHz out of the PC, parallel ports are a dying species.
04:46 AM rigid: rmu: yeah, but busywaiting just in one thread (on one core).
04:47 AM JT-Cave: morning
04:47 AM rigid: don't modern parallel ports support much higher frequencies nowadays with those EPP2.x modes? it's not a dying species like the ZIP-drive, it's a dying species like RS232
04:48 AM CloudEvil: EPP type modes are generally clocked in some manner, not GPIO
04:48 AM rigid: also, I'm using GPIOs but any kind of soft-pulse-generation should profit if that succeeds
04:49 AM rigid: CloudEvil: clocked doesn't necessarily mean not usable as GPIO. like e.g. an SPI clock makes a perfect 50% duty cycle pulse generator
04:49 AM rigid: but I haven't looked into them tbf
04:51 AM rmu: rigid: what kind of GPIO? you sure that is not internally connected via some evil i²c or similar thing?
04:51 AM rigid: rmu: rpi4 ... it is connected via some busses but still capable to do a few MHz when bitbanged
04:52 AM rigid: even more with DMA
04:52 AM rmu: rigid: in that case that is not a PC, not even a related animal, but it's own kingdom
04:53 AM rigid: yeah well. PCs also come with pinctl and GPIOs nowadays
04:53 AM rigid: certainly not the same animal tho
04:53 AM rmu: rigid: as you will need a break out board anyways, bite the bullet and get one with a USD1 microcontroller on it to generate predictable stepper timings
04:53 AM rmu: talk SPI to it
04:53 AM rmu: remora
04:55 AM rigid: what is that $1 MCU that controls multiple motors that'll work with linuxcnc without lots of coding?
04:56 AM rigid: I get the point, but still can't ignore that I got 2 idle cores at 1.8GHz and a hefty GPIO at service. The rest is "just" a software problem.
04:57 AM CloudEvil: In principle you can even lock those cores down so as to only be running from local cache.
04:57 AM CloudEvil: So you only face IO contention to cause delays, not memory.
04:57 AM * rigid will get a 7C81 anyways, to get encoder inputs
04:58 AM rmu: rigid: choose one https://github.com/scottalford75/Remora/
04:58 AM CloudEvil: That may be going a bit far, but still.
04:58 AM rigid: CloudEvil: exactly. i guess in practice it's not as easy but one should be able to let a single process run without the scheduler touching it a lot
04:58 AM rmu: full self driving cars are "just a software problem" according to some pundits, nevertheless, it's not there yet
04:59 AM CloudEvil: There is about eleven orders of magnitude difference in those problems.
04:59 AM rigid: i'd say it's not even the same problem class
05:00 AM rmu: the difference is that interfacing steppers to a rpi is a solved problem
05:00 AM rigid: remora looks cool
05:01 AM rigid: rmu: interfacing stuff with low latency directly from the rpi is aswell. but the combination of those... not really
05:02 AM rmu: remora seems to have code for a stm32g0 board, those things can be had for around usd1 https://www.lcsc.com/product-detail/Microcontroller-Units-MCUs-MPUs-SOCs_STMicroelectronics-STM32G070RBT6_C529340.html
05:03 AM rigid: i think i have those MCUs around still.
05:03 AM rmu: so as you will need an interface board for stuff anyways just put a cheap STM32 on it, those can easily generate accurate step trains up to tens of kHz
05:03 AM rigid: well, maybe I should focus on getting that millturn thing working before doing premature optimization :-P
05:05 AM rigid: rmu: tens of kHz? i can generate accurate steps with stepgen + GPIO at 50kHz with a max. drift of +/-2uS
05:06 AM rmu: rigid: so what is your problem then
05:06 AM rigid: 200kHz-ish would be nice
05:11 AM rmu: just curious, what for? even with 100-fold microstepping that would amount to 100 rotations per second and a stepper motor won't do that.
05:18 AM rigid: rmu: i'm not sure about the steppers. someone else is buying them. But if I had to guess, it's more for encoders & inputs.
05:20 AM rigid: and I guess 600-1800 RPM are in the possible range with transmission/gears
05:21 AM rigid: rmu: but yeah, i will focus on getting the current open-loop thing running with linuxcnc first. That's hard enough. https://linuxcnc.org/docs/devel/html/man/man9/millturn.9.html is basically the only documentation around for my use case :)
05:37 AM rmu: stm32 have timers that can directly count quadrature inputs, frequency usually up to half/quarter of µc frequency, so easily into the MHz
05:40 AM rmu: 600rpm is stepper territory, no problem there, but 6000rpm is not. typical stepper has 200 full steps per rotation. seems i messed up some zeros.
05:46 AM fdarling: rmu: I have written GD32 (STM32 equivalent) firmware that can do 125kHz stepgen with very stable frequency linearity. It's hard to get right, but it's possible! FPGAs make it so much easier. The reason you'd want such high frequency stepping is to both have high resolution and high speeds on the same machine.
05:51 AM fdarling: rmu: for example, a retrofit I did on a CNC router with Yaskawa servos has 10,000 counts per rev on the motor, with a 10:1 gear reducer. The pinion gear travels 150mm per rotation, or 15mm per servo rotation. So 10,000/15 = 666.66 counts/mm. If you want to go 800 IPM, 800/60*25.4*666.66 = ~225,775 Hz if you have 1 step per encoder count
05:56 AM fdarling: Yaskawa Sigma 7 servo drives support up to 4MHz stepping, though personally I wouldn't try to push it past 1MHz or so. Their serial encoders are 24-bit (~16M counts per rev), but the encoder feedback to the controls is divided down to something more reasonable for quadrature
05:58 AM rmu: yes interfacing servos with step/dir i can understand, but you would probably do it with ethercat or something like that if possible
05:59 AM fdarling: rmu: guess what -- the LinuxCNC EtherCAT driver doesn't currently support Yaskawa X-D
05:59 AM rmu: my machines live in the analog times of the 80ies
06:00 AM rmu: so no idea unfortunately
06:00 AM fdarling: I just spent 3 weeks trying to get a software developer to fix it, and then I eventually broke down and did it myself. I haven't gotten it mainlined yet, because I essentially commented out a bunch of device specific code to get it working (I need to rework them for my changes)
06:00 AM fdarling: I have always used analog velocity or torque setups. For retrofits, sometimes I deal with drives that only have step/direction despite being servos
06:01 AM fdarling: typically the old controls didn't use the encoder feedback and were running open loop with just alarm signals going back, but I switch it to have LinuxCNC getting the encoders
06:02 AM fdarling: I don't know about Leadshine (I have yet to test one), but JMC EtherCAT servo drives seem to have synchronization issues with their implementation of EtherCAT
06:02 AM fdarling: Yaskawa is very very good though
06:02 AM rmu: i have analog servo systems using a resolver for servo speed control and the CNC used an additional encoder
06:02 AM fdarling: https://forum.linuxcnc.org/ethercat/49457-ethercat-system-implementation-of-jmc-ethercat-servos?start=10#279330
06:03 AM fdarling: rmu: sounds like a Fadal ;-)
06:03 AM fdarling: they actually used a tachometer for speed control and resolver for position PID
06:03 AM rmu: GEC alsthom parvex
06:03 AM rmu: manual says "edition nov 90"
06:04 AM rmu: in french
06:05 AM travis_farmer: my old machine originally had DC brushed servos, but as it would have costed more to get them working, i changed to stepper, open-loop. may go back to servos at some point
06:07 AM travis_farmer: just because with a 4' x 4' router table, rapids around the machine take a while... ;-)
06:09 AM travis_farmer: my Y and X top out at 5340mm/minute right now, as configured
06:09 AM travis_farmer: if i calculated right...
06:34 AM JT-Cave: rooster just crowed
08:27 AM rigid: hm, can anyone reproduce "linuxcnc /usr/share/doc/linuxcnc/examples/sample-configs/sim/axis/vismach/millturn/millturn.ini" triggering the dialog "This version of LinuxCNC separates the concepts of Axes and Joints which necessitates changes to the INI and HAL files..." ?
08:27 AM rigid: neither "Yes" or "No" seem to work
08:28 AM rigid: "Yes" gives "The supplied INI file is already at version 1.1 and should not need updating"
08:29 AM rigid: "No" fails for reasons I'm trying to find out
08:31 AM rigid: I think, this could be the relevant log output: https://gist.github.com/heeplr/6cc5e7d7ad2f2642aae815d93a53b532 but what does it even mean? :)
08:36 AM Scopeuk: looks like it tried to load a kernel module and it said no
08:36 AM Scopeuk: could be permissions, could be an invalid configuration
08:37 AM travis_farmer: older versions of LinuxCNC didn't use joints AND Axes. if you are using a newer (2.9) LinuxCNC, and an older config, it will cause that, IIRC
08:38 AM travis_farmer: maybe millturn uses an older config, so you may need to update it to be like the new configs
08:38 AM travis_farmer: http://linuxcnc.org/docs/2.9/html/config/ini-config.html#sub:ini:sec:axis-letter
08:41 AM rigid: it seems "MIN_LIMIT_TURN" is gone
08:46 AM travis_farmer: There! finally my father's accountant got the Taxes done!!! :-) :-)
08:47 AM travis_farmer: so later today i can find out what i am getting for a tax return :-)
08:48 AM rigid: hm, the config seems to work on the rpi4 ... so it must be some gentoo-specific problem
08:49 AM travis_farmer: that could be too...
08:54 AM rigid: oh no... the x-axis of millturn moves the turning brackets not the milling toolhead :-/
08:54 AM rigid: I hoped I could just use that :-(
08:55 AM rmu: why start with something easy and common
08:56 AM rigid: you think first getting the XYZ mill to work without A axis would help?
08:58 AM rmu: i have no idea. to me a millturn looks something like that https://www.wfl.at/maschinen/millturn/m20-m20-g which seems to be on the upper end of complexity
08:59 AM rigid: that's one kind of millturn (not the upper end tho compared i guess)
09:00 AM rigid: rmu: mine is a simple XYZ mill (H-Frame) with an additional A axis. two lathe-like brackets that hold the piece to mill and rotatable with a stepper motor
09:01 AM rigid: the lathe-turn-thingy is just placed inside the XYZs cube
09:01 AM rmu: rigid: i would not call it millturn if you can't do lathe operations
09:01 AM travis_farmer: i was thinking this https://www.grizzly.com/products/grizzly-8-x-16-variable-speed-combination-lathe-mill/g0769
09:01 AM rmu: which you can't if your lathe spindle is driven by a stepper
09:02 AM travis_farmer: it would be a mill with rotary axis then
09:02 AM rigid: well, you could do lathe operations. just from the top. the A axis could turn quickly. But it's not used for that. It was used to turn the piece upside down to process the backside. recently experiments were made with some proprietary software and mach-3 (i'm not involved there). Now I wanted to give linuxcnc a shot.
09:03 AM rigid: travis_farmer: more like this https://www.youtube.com/shorts/-O9pJfWHei8
09:05 AM travis_farmer: yeah, mill with rotary axis, or CNC router with rotary axis, for sure
09:06 AM rigid: i guess the millturn example is the closest example I get :-/
09:10 AM rmu: your millturn vismach example has switchable kinematics, you sure you tested the right one
09:10 AM rigid: quite, i'm using mill mode. The other one is turn mode, where I guess it just behaves like a lathe with the A axis constantly rotating at fixed RPM
09:11 AM rigid: in both modes, the workpiece moves X and Y, not the the toolhead
09:13 AM travis_farmer: i thought the toolhead was generally Z... i may be wrong
09:14 AM rigid: is it? i didn't check other vismach stuff
09:14 AM rmu: Z usually is in direction of spindle
09:15 AM rigid: but it's not always the table/workpiece that's moving in X and Y, is it?
09:15 AM rmu: don't know what you mean
09:15 AM rigid: iirc I saw the toolhead moving in vismach
09:15 AM rmu: so what
09:15 AM rigid: rmu: i mean toolhead fixed in X/Y plane vs. workpiece/table fixed in X/Y plane
09:16 AM rigid: rmu: so when you got a rotating table or a turny thingy like in my case, the visualization is wrong
09:16 AM rmu: don't understand, only relevant thing is relative position tool / workpiece
09:16 AM travis_farmer: http://linuxcnc.org/docs/2.9/html/user/user-concepts.html#sec:machine-configurations
09:16 AM skunkworks: lol - travis_farmer was quicker..
09:17 AM travis_farmer: :-)
09:17 AM * skunkworks was going to find that.
09:17 AM rmu: if your vismach example is wrong tough luck, you will have to adapt it ;)
09:17 AM rigid: rmu: yeah, i'll try that. trying to find the millturn code :)
09:18 AM rigid: ah, got it
09:20 AM rigid: "turnmill" would probably be an appropriate name for those kind of 4-axis mills :-P
12:49 PM rigid: "I have no idea what this does, but it seems to be important for tool tip visualization."
12:49 PM rigid: props to the unsung heroes writing documentation...
12:49 PM travis_farmer: lol
12:59 PM rigid: are vismach python scripts supposed to run standalone?
01:00 PM rigid: the docs recommend to add a shebang so it can be started as script but a minimal example gives strange errors
01:33 PM CaptHindsight: https://www.youtube.com/watch?v=02LDnhiZFTo how many Z axes does this lathe have?
01:36 PM CaptHindsight: https://www.youtube.com/watch?v=x5r9AtzuWwU
01:41 PM markgonz: Hello friends
01:44 PM markgonz: Im really struggling to get the spindle running from linuxcnc
01:45 PM markgonz: I wonder if there is some linuxcnc expert that I can hire by the hour or something
01:45 PM CaptHindsight: markgonz: which spindle?
01:45 PM CaptHindsight: which control/drive?
01:46 PM markgonz: is a mechatron 800w ER11 high frequency spindle
01:46 PM markgonz: and omron mx2 vfd
01:46 PM CaptHindsight: spindles are usually pretty easy to get spinning
01:46 PM markgonz: I have set it up and can make it run from the vfd
01:46 PM CaptHindsight: how is the Omron connected to LCNC?
01:46 PM markgonz: however from commands on linuxcnc is another story
01:47 PM markgonz: now using rs485 usb dongle
01:47 PM markgonz: my setup is intel minipc and mesa 7i96s
01:47 PM CaptHindsight: are any commands responding or is there any response from the VFD?
01:47 PM markgonz: I tried all the ways there are, analog 1-10v from mesa, and modbus
01:48 PM CaptHindsight: is LCNC actually connecting to the VFD?
01:48 PM markgonz: no response from the vfd
01:48 PM markgonz: here is where i get lost
01:48 PM CaptHindsight: so which is your preferred method to connect?
01:48 PM markgonz: checking modbus communication I cant understand how to do it
01:48 PM markgonz: I want to try vfdmod
01:49 PM markgonz: https://github.com/aekhv/vfdmod/wiki
01:49 PM CaptHindsight: what has happened with all your different tries 0-10V, rs485?
01:50 PM CaptHindsight: well lets get you crawling before you try to run
01:50 PM markgonz: so 0-10v I read the manual and I can not understand the wiring
01:50 PM CaptHindsight: sounds like config issues, but lets see
01:50 PM markgonz: i have to say I am a noob
01:50 PM CaptHindsight: it's ok
01:50 PM markgonz: i have cnc machining exprience and precision parts but this is another world
01:51 PM markgonz: thanks!
01:51 PM CaptHindsight: yeah, no problem, no need to apologize
01:51 PM CaptHindsight: so right now you have a USB -->rs485 dongle?
01:52 PM markgonz: yes
01:52 PM CaptHindsight: does the Linux kernel know that the dongle is there?
01:52 PM markgonz: mm
01:52 PM markgonz: I am not sure
01:53 PM CaptHindsight: open a terminal and type: lsusb
01:53 PM markgonz: btw I have the problem I am using the minipc of the machine now, so if I conect to the mesa card to check I lose wifi
01:53 PM markgonz: yes lsusb i can see it
01:54 PM CaptHindsight: well the USB dongle doesn't need the Mesa card to operate
01:54 PM markgonz: if I unplug the dongle and repeat lsusb is not there
01:54 PM CaptHindsight: https://paste.debian.net/
01:54 PM markgonz: ok
01:54 PM CaptHindsight: paste the results here and then post the link it creates
01:55 PM CaptHindsight: for example mine looks like this https://paste.debian.net/1291046/ in the first 3 lines
01:55 PM markgonz: https://paste.debian.net/1291047
01:56 PM markgonz: is device 005
01:56 PM CaptHindsight: is this the USB -->485 dongle? 4 Bus 001 Device 005: ID 1a86:7523 QinHeng Electronics CH340 serial converter
01:56 PM markgonz: yes
01:56 PM CaptHindsight: ok great
01:58 PM CaptHindsight: I have to go in a minute but you should be able to connect to the VFD and see if it responds ....
01:58 PM CaptHindsight: section B2 of https://assets.omron.eu/downloads/manual/en/v3/i570_mx2_users_manual_en.pdf
01:59 PM markgonz: ok no worries
01:59 PM markgonz: I will read more
01:59 PM CaptHindsight: does the VFD drive the spindle if you use the manual buttons on the VFD?
01:59 PM markgonz: yes
01:59 PM markgonz: and i have set up the vfd to recieve modbus commands now
02:00 PM CaptHindsight: ok, that is one hurdle that you have already made it past
02:00 PM markgonz: yes luckly
02:01 PM markgonz: I think my problem is undestanding HAL and pins and the custom hal stuff
02:02 PM markgonz: seems so complicated
02:02 PM markgonz: also finding the adresses in hex on the manual was hard for me
02:10 PM CaptHindsight: ah, i think you were chatting with solarwind about this on Saturday
02:10 PM markgonz: yes I think so
02:11 PM markgonz: told me not to dissapoint him getting the right hex adress from the manual
02:12 PM CaptHindsight: https://forum.linuxcnc.org/49-basic-configuration/38139-control-of-an-omron-mx2-vfd-via-rs485-usb-adaptor
02:12 PM markgonz: LOL thanks to him I think I got that part right
02:12 PM CaptHindsight: look at this ^^
02:13 PM markgonz: ah yes I found that thread the other way
02:13 PM markgonz: day*
02:14 PM markgonz: seems that the wj200 driver that should work is almost imposible to get to work on rs485
02:14 PM markgonz: I also tried that route
02:14 PM CaptHindsight: https://wiki.printnc.info/en/controllers/linuxcnc/vfd
02:15 PM markgonz: yes also that one I read
02:15 PM markgonz: and I did ad user to dialout group permission
02:15 PM markgonz: to USB0
02:15 PM markgonz: but is another vfd and custom hal is not the same
02:16 PM markgonz: is another driver
02:16 PM markgonz: I tried checking communication with this but doesnt work because is another driver
02:17 PM markgonz: sudo chmod 666 /dev/ttyUSB0
02:17 PM markgonz: $ halrun
02:17 PM markgonz: halcmd: loadusr -W hy_vfd -d /dev/ttyUSB0
02:17 PM markgonz: maybe I should try but replacing hy_vfd with wj200
02:17 PM CaptHindsight: yeah the next thing is being sure that serial is getting to and from the VFD
02:18 PM markgonz: true
02:18 PM markgonz: thats where I am stuck
02:18 PM markgonz: I can try things without knowing exactly what Im doing but can take forever
02:19 PM markgonz: also I read that these dongles are really buggy and also not helping my confidence
02:20 PM CaptHindsight: yeah that as well
02:20 PM CaptHindsight: do you have 2 dongles and 2 PC's? then you can test them and connections
02:21 PM Tom_L: what happened when you tried analog 0-10v in to the vfd?
02:21 PM Tom_L: imo that would be the easiest to set up
02:22 PM markgonz: yes I bought two dongles
02:22 PM markgonz: analog 0-10v the manual didnt make any sense to me
02:22 PM CaptHindsight: how about 0-10V, what did or did not happen?
02:22 PM Tom_L: go as far as you can with what you're doing before jumping ship
02:22 PM CaptHindsight: ah
02:22 PM markgonz: the manual said something like 0-5v or something and had to make calculations to make it work 0-10v
02:23 PM Tom_L: what board do you have that has analog out?
02:23 PM Tom_L: mesa
02:23 PM markgonz: 7i96s
02:23 PM Tom_L: ok
02:23 PM markgonz: I found some info on german how to make it work
02:23 PM markgonz: and some other peple have told me to go 0-10v
02:24 PM Tom_L: i'm not sure on that one but my 7i47s i had to supply a 10v source to the analog
02:24 PM Tom_L: it's alot easier than modbus imo
02:24 PM CaptHindsight: page 52 in the MX2 manual
02:25 PM markgonz: https://www.youtube.com/watch?v=IkHypYjUlz8&feature=youtu.be
02:25 PM markgonz: however is not the same vfd as me
02:25 PM markgonz: ok im gonna look at it now
02:26 PM CaptHindsight: https://i.imgur.com/nt6yWdB.png
02:26 PM CaptHindsight: only 3 connections, H, O and L
02:27 PM markgonz: https://assets.omron.eu/downloads/manual/en/v3/i570_mx2_users_manual_en.pdf
02:27 PM markgonz: i am looking the manual here, maybe is the complete manual but is not page 52 here
02:28 PM CaptHindsight: yes there is :)
02:28 PM markgonz: ok I see the pic now
02:30 PM CaptHindsight: the 7I96S also has one RS-422/RS-485 interface available
02:31 PM CaptHindsight: 7I96S page 12 , ANALOG SPINDLE INTERFACE
02:32 PM Tom_L: capn does that one have onboard 10v source?
02:32 PM Tom_L: i haven't tried analog on that yet
02:32 PM CaptHindsight: I don't know
02:32 PM markgonz: https://postimg.cc/Dmt5WjxZ
02:32 PM CaptHindsight: I'm just reading the doc
02:32 PM markgonz: this is my vfd wired to the spindle now
02:32 PM Tom_L: if it's like the rest, you need to supply the 10v reference
02:33 PM markgonz: so I need a 10v power supply to use analog?
02:33 PM CaptHindsight: and his VFD has a 10V source/reference
02:34 PM CaptHindsight: so Spindle+ to H
02:35 PM CaptHindsight: Spindle Out to O
02:35 PM CaptHindsight: Spindle- to L
02:35 PM rigid: i suppose there is no way to have the A axis rotate the world in the preview instead of rotating the tool?
02:36 PM markgonz: but my L is taken
02:36 PM markgonz: https://postimg.cc/Dmt5WjxZ pic of my vfd
02:36 PM Tom_L: are AL0-2 the analog?
02:37 PM CaptHindsight: where does your L wire go to?
02:38 PM markgonz: L wire is some singal from spindle
02:38 PM markgonz: I dont remember exactly
02:38 PM markgonz: but I followed the spindle manual for the vfd
02:38 PM CaptHindsight: L is the 0-10V common or Ground
02:38 PM CaptHindsight: UVW goes to the spindle
02:39 PM Tom_L: yeah that's how mine is iirc
02:39 PM markgonz: yes uvw goes to spindle
02:39 PM CaptHindsight: H is the 10V analog reference voltage
02:39 PM markgonz: so you think my H wire to spindle is wrong?
02:40 PM markgonz: I will recheck my spindle manual but I think is what it said
02:40 PM CaptHindsight: O is the 0-10V that gets divided by the potentiometer in the 7i96s
02:40 PM CaptHindsight: L is the common ground for analog voltage
02:40 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Mill_Steel/Spindle/BLDC_Wiring2.jpg
02:40 PM Tom_L: not that it will help, but there is mine
02:41 PM CaptHindsight: anyway I gota go again
02:42 PM markgonz: no worries
02:42 PM markgonz: Im curious how did you all learn all this stuff
02:43 PM markgonz: is by trade or are all of you electrical automation engineers
02:43 PM Tom_L: same way you are
02:43 PM CaptHindsight: 50+ years of abuse, study, tinkering, classes, etc etc
02:43 PM Tom_L: gawd you must be an antique
02:44 PM markgonz: awesome
02:44 PM CaptHindsight: a few of us are engineers, the rest are software devs and some just hobbyists
02:44 PM markgonz: Im only 30 so I need more time
02:44 PM markgonz: ok interesting
02:45 PM Tom_L: you're probably in the right place to get it goin though
02:46 PM markgonz: yes true!
02:46 PM markgonz: IRC seems so cool to me
02:46 PM markgonz: my frist time thanks to lcnc
02:46 PM markgonz: otherwise I would not have discovered
02:47 PM markgonz: https://postimg.cc/VdwwqR6f
02:47 PM markgonz: my mill build coming along
02:47 PM skunkworks: I used to be 30...
02:48 PM markgonz: made from granite surface plate
02:48 PM markgonz: and custom granite columns
02:48 PM markgonz: mill is for making watch parts
02:48 PM CaptHindsight: me too for a whole year
02:48 PM markgonz: cocacola for scale
02:49 PM Tom_L: you do watch repair?
02:49 PM Tom_L: another guy here does that as well
02:50 PM markgonz: I make watches
02:50 PM markgonz: started doing repair but not anymore
02:50 PM markgonz: machining is more fun
02:50 PM Tom_L: he makes parts for them
02:50 PM markgonz: ah cool
02:50 PM markgonz: whats his name?
02:50 PM skunkworks: there is a guy who re-winds accutron coils..
02:50 PM Tom_L: here.. cradek
02:51 PM Tom_L: timeguy dot com iirc
02:51 PM markgonz: wow impressive
02:52 PM skunkworks: cute machine! looks solid.
02:52 PM markgonz: thanks!
02:52 PM * Tom_L feels a late afternoon nap coming on
02:53 PM markgonz: https://postimg.cc/PNVXjZrT
02:54 PM markgonz: another angle
02:55 PM markgonz: my idea at first was getting servos with ethercat
02:55 PM skunkworks: Nice - looks like fine enough screws to do some precise machining..
02:55 PM markgonz: but if I can not wire a spindle ethercat would be insane for me
02:55 PM markgonz: yes my idea is all my parts for the watch are smaller than 30mm
02:56 PM markgonz: so I go really fine pitch, 2mm
02:56 PM markgonz: maybe is too small but I dont need high speed machining
02:56 PM markgonz: I wanted 4mm pitch but there was an offer on 2mm
02:57 PM markgonz: they are preloaded nut and c3 quality
02:57 PM markgonz: hopefully work well because I spend all my savings on this LOL
02:57 PM skunkworks: Nice!
02:58 PM markgonz: I will do a forum post once is finished
02:58 PM markgonz: maybe youtube video
03:06 PM markgonz: so anyone can tell me cool IRC stories
03:07 PM markgonz: I guess some of you were here on the good era
03:07 PM markgonz: any other channels I should check out now?
03:07 PM travis_farmer: i remember old IRC, but NNTP was almost better, IMO
03:08 PM markgonz: never heard of NNTP
03:08 PM travis_farmer: like email, but more like an early forum. email with threads
03:08 PM travis_farmer: and a PITA to serve ;-)
03:09 PM markgonz: oh cool
03:09 PM travis_farmer: stands for Network News Transfer Protocol
03:09 PM markgonz: are some of you on hackernews?
03:09 PM travis_farmer: never heard of it...
03:09 PM markgonz: interesting
03:10 PM markgonz: https://news.ycombinator.com/
03:10 PM markgonz: is like a forum that popular threads go up, really active but old school feeling
03:10 PM markgonz: very interesting reads there if you are into computer/software
03:11 PM markgonz: but also many more topics
03:13 PM markgonz: also you can search for most popular threads in the past day, month etc
03:13 PM markgonz: Im sure software devs here probably now it
03:14 PM skunkworks: I have a lot of random videos... https://www.youtube.com/@MakersEase/videos
03:14 PM markgonz: cool Im gonna subscribe now
03:14 PM skunkworks: Thanks!
03:14 PM markgonz: I like this kinds of vids
03:15 PM skunkworks: poor production value.. But I make up for it in painfulness...
03:16 PM markgonz: LOL
03:16 PM markgonz: I love straight to the point homemade videos
03:16 PM markgonz: are usually the most useaful
03:19 PM markgonz: https://www.youtube.com/shorts/Po_1_zV2v4g
03:19 PM markgonz: my channel
03:19 PM markgonz: I tried to edit some clips on my phone but I hate this format
03:19 PM markgonz: I will try to make better videos and no music
03:19 PM markgonz: and horizontal
03:20 PM skunkworks: lol - how do you like fusion?
03:20 PM markgonz: fusion is great I think
03:21 PM skunkworks: the closed captioning says 'foreign music'..
03:21 PM markgonz: but I dont know anything else
03:21 PM markgonz: so you cant hear music?
03:24 PM markgonz: your matsuura is so cool
03:44 PM rigid: hm, a mode that comments gcode would be quite helpful
03:48 PM JT-Cave: new ac tomorrow 'I hope'
04:06 PM * roycroft wonders why windows is so insane
04:06 PM roycroft: i went to build a vm by cloning an old windows 10 install
04:06 PM roycroft: windows decided it was time to do updates - about 2 years' worth
04:07 PM roycroft: that's fine, but what it did was download all the updates, and then apply them in REVERSE chronological order
04:07 PM roycroft: and updates that depended on previous updates failed
04:07 PM roycroft: it went all the way down the list until it got to the oldest update
04:08 PM roycroft: then i got to reboot, run windows update again, where it downloaded again all the updated that had failed, and proceeded to apply them in reverse chronological order
04:08 PM roycroft: i'm on the fourth iteration of this now
04:09 PM roycroft: it seems that installing the oldest ones first, even if they are superceeded by newer ones, would be more efficient in just about every respect
04:13 PM Tom_L: JT-Cave, thats taken a while
04:17 PM travis_farmer: just got my Taxes back! and i am getting enough back to get Vetric Vcarve Pro :-D
04:17 PM JT-Shop: yep
04:17 PM JT-Shop: does your router work?
04:17 PM Tom_L: my router?
04:17 PM JT-Shop: his
04:17 PM Tom_L: oh
04:17 PM travis_farmer: more or less. i need to learn the software too
04:17 PM Tom_L: was afk a sec
04:18 PM roycroft: the vm i'm trying to build is for vcarve pro
04:18 PM travis_farmer: it works enough to make things...
04:18 PM JT-Shop: start making money
04:18 PM travis_farmer: i will, when i get the software :-D
04:19 PM Tom_L: you need some ink...
04:19 PM travis_farmer: lol
04:22 PM roycroft: "you need some ink" sounds like you're sending him to a tattoo parlor
04:22 PM JT-Shop: lol
04:23 PM Tom_L: they are both permanent...
04:23 PM travis_farmer: heck, i got a hundred grand ;-) https://www.tjfhome.net/community/index.php?media/20230905_171816-jpg.18/
04:24 PM JT-Shop: don't spend it all in one place
04:25 PM travis_farmer: lol, i will try not to
04:26 PM Tom_L: reminds me of the movie "Gone in 60 sec"
04:26 PM travis_farmer: lol
04:27 PM JT-Shop: never saw that one
04:27 PM Tom_L: nicholas cage
04:27 PM travis_farmer: no... though i did get maybe enough to get "some" shop material to build with, i think
04:28 PM Loetmichel: roycroft: the "new" fad isnt ink but sticking your arm/other appendage into a laser engraver and get a branding. For the clinically insane that is. Kids, dont try that at home.
04:29 PM Tom_L: Loetmichel, that's used for removing birthmarks etc
04:29 PM Tom_L: it _is_ somewhat of a thing... doesn't fall under home remedies though
04:34 PM Loetmichel: Tom_L: i am talking cheap chinese lasercutter meant for wood/acrylics.
04:34 PM Loetmichel: when those were "new" that was a fad.
04:34 PM Loetmichel: to use it to get a custom branding.
04:34 PM Loetmichel: usually mandated by peer pressure. "i double dare you"
04:41 PM CaptHindsight: laser tattoo printer
04:42 PM travis_farmer: https://cdn.soft112.com/laser-cut-finger-prank/00/00/0G/1U/00000G1U3A/pad_screenshot.jpg
04:44 PM rigid: do I need to change the kinematic to have the preview correctly represent my A axis turning the workpiece?
05:23 PM CaptHindsight: rigid: https://www.youtube.com/watch?v=De5uZbKe8wI
05:40 PM rigid: CaptHindsight: hm, that's not exactly my use case but i'll look into the custom kinematics stuff
05:40 PM rigid: this is the kind of thing I'd like to control https://www.youtube.com/shorts/VA8zebWiH6Q
05:47 PM CaptHindsight: rigid: I'm sorry, what is the problem?
05:51 PM rigid: CaptHindsight: i adapted the millturn vismach UI to my use case (so that the A axis isn't a moving table but still and the tool moves in XYZ). But the normal axis visualization still rotates the tool for A axis change, not the work (or the world for that matter)
05:52 PM CaptHindsight: vismach is pretty much DIY, I think it's great but hardly anyone uses it
05:53 PM rigid: that way, if you repeat the same XYZ path where your piece rotates after each pass one step further around the X axis, the path in the visualization gets just drawn over by a different tilting tool. Instead of drawing multiple similar paths around a common center
05:54 PM rigid: vismach works fine. i think it's my kinematics that's wrong. lcnc seems to think my workpiece is immovable and my toolhead tilts
05:54 PM rigid: not sure if that's a bigger problem but all the path visualization is incorrect for sure
06:51 PM roycroft: i'm on iteration six of the windows updates now
06:51 PM roycroft: this time for sure!
06:51 PM roycroft: it is rebooting
06:51 PM roycroft: i am crossing my fingers and awaiting with bated breath
06:52 PM roycroft: and it is fscking again
06:52 PM roycroft: or whatever the ntfs equivalent of fsck is
06:56 PM Tom_L: 10?
06:56 PM Tom_L: 11?
06:56 PM roycroft: 10
06:56 PM roycroft: my hardware won't support 11
06:56 PM roycroft: and that's fine
06:57 PM roycroft: it finished booting
06:57 PM roycroft: but the big question is: has it finished updating?
06:58 PM roycroft: and no
06:58 PM roycroft: five updates to go
06:58 PM Tom_L: other day i started a metric tool table for my cad cam and actually got a grbl post working with no mods
06:59 PM * roycroft wonders if he'll live long enough to see windows fully-updated
06:59 PM Tom_L: you mean linux?
06:59 PM Tom_L: take your pick
06:59 PM Tom_L: evils of different colors
07:00 PM roycroft: i mean windows
07:00 PM Tom_L: i know what you meant :)
07:00 PM roycroft: if it were linux it would have been done yesterday
07:00 PM Tom_L: then there's the dependency hell linux puts us in
07:01 PM roycroft: and it's rebooting again
07:01 PM Tom_L: or versions that won't work until versions are available
07:01 PM roycroft: so now we
07:01 PM roycroft: are finishing upgrade iteration 7
07:01 PM Tom_L: when was the last time you updated?
07:01 PM roycroft: and there is at least one more iteration to go
07:01 PM Tom_L: .... yesterday?
07:01 PM roycroft: a couple years ago
07:01 PM Tom_L: :)
07:02 PM roycroft: i don't mind that it's having to do it a bunch of times
07:02 PM roycroft: what i mind is that it downloads all the updates, then tries installing them in reverse chronological order, and the ones that depend on older updates fail
07:02 PM roycroft: then, when the oldest ones are applied, it does the exact same thing all over again
07:03 PM XXCoder: I find it weird that it does that
07:03 PM Tom_L: you shouldn't wait so long
07:03 PM Tom_L: you upset the ms gods
07:03 PM roycroft: there is this really big feature update - h22 or something - that is the newest one
07:03 PM roycroft: and it has downloaded 7 times now
07:03 PM Tom_L: i'll run 7 until it won't run any longer
07:03 PM roycroft: 22H2 is what that update is named
07:04 PM roycroft: it looks like there's only one update left besides that
07:04 PM roycroft: so maybe two iterations at most?
07:05 PM roycroft: at least the updates are going a little faster now, as there are fewer to download and fail each time
07:05 PM acer: I cant at this time but GOOD LORD. this .75KW brushless driver is killing me. I'd love to build a 5 axis machine if I had the space/time
07:05 PM acer: :(
07:06 PM acer is now known as _unreal_
07:06 PM roycroft: and it doesn't fsck every time it reboots - only every third time or so
07:10 PM roycroft: it's down to the last update, so hopefully this nightmare will end soon
08:05 PM _unreal_: so I'm working on building a server up. I got a free 1u server
08:05 PM _unreal_: being given memory and new cpu's for it
08:06 PM _unreal_: going for the max cpus which I beieve are e5-2470's
08:06 PM _unreal_: 8 core 3.1ghz 20mb cache
08:06 PM _unreal_: 2x
08:06 PM _unreal_: so 16 cores 32 HT cores
08:07 PM _unreal_: going to use it for CAD and graphics
08:08 PM _unreal_: a guy is giving me maybe the max memory it can take? if so that will be 192mb ram
08:08 PM _unreal_: granted not the fastest for CAD but the fact that it will have 96 gigs of RAM available for CAD will be nice
08:09 PM _unreal_: most CAD softwares only use one cpu/core for certain tasks. Rendering will be multi threaded but raw geometry tasks not so much
08:11 PM roycroft: well finally, after 10 update iterations and almost 24 hours, the windows vm is up-to-date
08:11 PM roycroft: until the next update, of course
08:11 PM roycroft: for cad, at least half the cost of the machine should go towards the display adapter
08:12 PM _unreal_: o,O
08:12 PM _unreal_: fun
08:12 PM _unreal_: roycroft, ahh no
08:12 PM _unreal_: CAD is CPU based
08:13 PM _unreal_: there are functions that can use the GPU. but in general CAD is cpu and single core based.
08:13 PM _unreal_: now rendering. can be GPU and MULTI THREADED core supported using every bit of power there is for # crunching
08:13 PM _unreal_: but CAD no faulse
08:16 PM roycroft: maybe the machines i use for cad already have sufficient cpu and memory
08:16 PM roycroft: i just know that a crappy frame buffer makes it a total dog
08:16 PM roycroft: a high end graphics card makes it run fast
08:17 PM Tom_L: The biggest reason why CPU is the standard in 3D rendering is simply that it has far greater overall quality than GPU. If you want your renders to be precise and your output quality to have the highest standards, then CPU rendering is the best choice.
08:17 PM Tom_L: Does CAD require more CPU or GPU?
08:17 PM Tom_L: AutoCAD scales primarily with CPU performance. It can run surprisingly well even on integrated graphics so long as you don't embark on any 3D-heavy work or modeling.
08:18 PM Tom_L: doen't say much about the rest
08:18 PM Tom_L: CATIA leverages both CPU and GPU to deliver a high quality user experience so there are several things that need to be considered in architecting your NVIDIA GRID Virtual Workstation (vWS) solution.
08:19 PM roycroft: yeah, i don't think autocad requires as much gpu as solidworks
08:19 PM Tom_L: so it depends on the cad
08:19 PM roycroft: when i upgraded the graphics card on my current cad machine, autocad ran a little faster, but solidworks ran like 10x faster
08:19 PM roycroft: and i'm not talking about renders
08:19 PM roycroft: just redraws, moves, etc
08:19 PM roycroft: the normal day-to-day stuff
08:19 PM Tom_L: i seldom do a rendering
08:20 PM roycroft: i do renders usually at the end of a project, if at all
08:20 PM roycroft: sometimes when i'm posting a milestone report on a project
08:20 PM Tom_L: you can do scenes etc with catia i know
08:20 PM roycroft: that cad machine is weird though
08:20 PM Tom_L: placing the model in a scene for astetics
08:21 PM roycroft: i remember when i upgraded from a 1080p display to a 4k display solidworks was a lot faster
08:21 PM Tom_L: i did learn that in class but never use it
08:21 PM roycroft: one would think it would run slower, all other factors being equal (which they were)
08:21 PM Tom_L: i can't see 4k anyway
08:21 PM Tom_L: :)
08:21 PM Tom_L: lines get too narrow
08:25 PM CaptHindsight: was it 10 years ago that displays had stalled out at 1080p?
08:25 PM roycroft: you can adjust for that
08:25 PM roycroft: 4k is nice
08:25 PM roycroft: and the whole point of 4k is that you're not supposed to see it :)
08:26 PM Tom_L: CaptHindsight, probably :)
08:26 PM Tom_L: i have a 4k display but that pc doesn't get alot of use
08:27 PM CaptHindsight: I wanted 4k on the late 90's for CAD and EDA
08:28 PM _unreal_: Tom_L, rendering and CAD-geometry are very different. GPU'S do not generally help when dealing with CAD.
08:28 PM Tom_L: _unreal_, that was somewhat my point
08:28 PM _unreal_: ahh
08:29 PM Tom_L: however i believe each cad is different in that regard
08:29 PM roycroft: well most modern cad packages do some of both, and not in isolation
08:29 PM Tom_L: to lump them all together would be unfair
08:29 PM CaptHindsight: also depends on your settings
08:30 PM roycroft: autocad is still mostly 2d-oriented
08:30 PM roycroft: and that's where the cpu and memory are everything
08:30 PM CaptHindsight: how much rendering, shadows etc do you have turned on
08:30 PM Tom_L: acad has a place... just not sure where :)
08:30 PM roycroft: but 3d parametric software benefits from a good gpu quite a bit
08:30 PM _unreal_: well CAD in general is 2D making 2d shapes INTO 3d
08:30 PM CaptHindsight: acad is still #1 for architecture
08:30 PM Tom_L: architecthure
08:30 PM roycroft: i still use it almost daily
08:30 PM CaptHindsight: and its variants
08:30 PM Tom_L: wrong button :)
08:30 PM CaptHindsight: revit etc
08:31 PM roycroft: and i use solidworks almost daily
08:31 PM Tom_L: for those types i'm sure it's great
08:31 PM roycroft: but that may all change soon, as i start using vectric software
08:31 PM Tom_L: i did know a guy that did all his cad cam in acad though
08:31 PM Tom_L: 3d
08:31 PM roycroft: for what i mostly do these days, vcarve pro is probably the ideal software for me
08:31 PM Tom_L: used nc polaris on top of it
08:32 PM roycroft: if i ever get the mill conversion done i'll use solidworks/solidcam for the mill, and vcarve pro for the cnc router and shaper origin
08:32 PM Tom_L: doubt you can even find that now
08:33 PM Tom_L: i won't update mine since it's paid for and does full 3d with a cam that works seamlessly with linuxcnc
08:34 PM Tom_L: it has a few quirks but show me one that doesn't
08:38 PM roycroft: it's all about what you're used to
08:40 PM Tom_L: when i did the catia classes, i also did everything in solidworks but mainly focused on catia since i like it better
08:40 PM Tom_L: they are similar but different
08:40 PM CaptHindsight: it's best to not get used to any of them, just run all of them and keep forgetting which one does what way and how
08:41 PM Tom_L: my kids did catia, autocad, mastercam, solidworks i think
08:41 PM CaptHindsight: SW, NX, Catia, Creo
08:41 PM Tom_L: so at the time i got to see what each had
08:41 PM Tom_L: never messed with NX or Creo
08:41 PM CaptHindsight: I have run Inventor but never used Acad for much
08:42 PM Tom_L: years ago i did my property with acad
08:43 PM Tom_L: i don't use any of them as much as i should anymore
08:43 PM CaptHindsight: same with EDA, Cadence, Altium, etc
08:43 PM CaptHindsight: all do it the same and different
08:44 PM Tom_L: cadence owns the shop my kid left
08:44 PM Tom_L: parent company
08:44 PM CaptHindsight: why do the open source tools make everything so clumsy to use?
08:44 PM CaptHindsight: too many steps
08:45 PM Tom_L: they all start from scratch and give up before it's done
08:45 PM CaptHindsight: are the GUI's and left or right clicks copyright or patented?
08:46 PM CaptHindsight: Kicad added some new feature abut you had to use it by some multiple key press vs just a right click then scroll the dialog etc
08:50 PM CaptHindsight: and Freecad, ouch
08:54 PM _unreal_: CaptHindsight, ?
08:55 PM Tom_L: i like that i can pan zoom rotate etc in catia without lifting off the mouse
08:56 PM Tom_L: alot quicker
08:56 PM Tom_L: the cad room was full of those space balls but nobody ever used them
08:57 PM _unreal_: I always wanted one of those space balls
08:57 PM _unreal_: I believe there is an opensource 3D print version
09:04 PM _unreal_: Tom_L, https://www.youtube.com/watch?app=desktop&v=NqHIdklkzfw
09:05 PM _unreal_: https://github.com/FaqT0tum/Orbion_3D_Space_Mouse
09:06 PM Tom_L: i fiddled with them a bit but didn't like em
09:06 PM Tom_L: not the OS one
09:16 PM _unreal_: os one
09:16 PM _unreal_: ?
09:18 PM Tom_L: open source
09:22 PM roycroft: i have a 3d mouse
09:23 PM roycroft: it took a while to get used to it, but i love it
09:23 PM roycroft: it especially makes assemblies go much faster
09:25 PM _unreal_: what 3d mouse do you hyave
09:25 PM _unreal_: have?
09:26 PM _unreal_: everything I do currently is just with a standard mouse
09:26 PM _unreal_: well 3 button/wheel
09:27 PM roycroft: it's one of the 3dconnextion space mouses
09:28 PM roycroft: i've had it for years - it's an older model, but it works great
09:29 PM roycroft: https://www.ebay.com/itm/202248608818
09:29 PM roycroft: that one
09:31 PM _unreal_: ya thats not cheap
09:32 PM _unreal_: I may try my hand at building that thing I just linked to
09:32 PM _unreal_: who knows
09:32 PM roycroft: it's not, but it's really well made and works amazingly
09:32 PM roycroft: it weighs about a kilo
09:34 PM Tom_L: yeah the school had a room full of those but they had all the buttons etc to go with it
09:35 PM Tom_L: rather cumbersome
09:36 PM Tom_L: https://www.cdw.com/product/3dconnexion-spacemouse-pro-wired-3d-mouse/2565617?cm_ven=acquirgy&cm_cat=google&cm_pla=NA-NA-3D%20Connexion_MT&cm_ite=2565617
09:38 PM _unreal_: hum
09:38 PM _unreal_: dont really even know that i need a space mouse
09:38 PM _unreal_: but the opensource one is interesting to me
10:22 PM roycroft: the buttons never appealed to me
10:23 PM roycroft: but the 3d knob is great
10:24 PM roycroft: pan/zoom/tilt/rotate all with the one knob