#linuxcnc Logs

Jul 10 2022

12:20 AM Thorhian[m]: That looks beautiful Tom. I wish I knew it existed. Are there any guides or docs on how to use it with LinuxCNC?
12:29 AM XXCoder: interesting
12:29 AM XXCoder: breakout board with rj45 jack type
12:29 AM Thorhian[m]: And it already works with hostmot2.
01:37 AM Deejay: moin
04:11 AM Loetmichel: mornin'
04:11 AM Loetmichel: *GIGGLE* just read a nice comment on reddit. "... saltpeter, an ingredient for making gunpowder was made from feces (human or otherwise) for a long time, until someone finally figured out how to get nitrogen directly from the air." Answer from some other commenter: "You're taking the piss.... and making it explode!" :)
04:25 AM pere: If I am not mistaken, that someone was Kristian Birkeland, who discovered the process as a by-product of his experiments with an electrical canon, and the curious powder discovered on the floor was the functionat for Norsk Hydro, currently Yara, one of the main produsers of furtilizer in the world.
04:28 AM pere: I remembered correctly, <URL: https://en.wikipedia.org/wiki/Kristian_Birkeland >
04:30 AM XXCoder: cannon?
04:30 AM pere: yes, typo.
04:31 AM XXCoder: ok :)
04:47 AM Loetmichel: pere: that urine can produce nitrates was known for a loooong time. I think the chinese used that for their fireworks.
04:56 AM JT-Cave: moning
05:01 AM JT-Cave: wow I actually slept deep enough to get to REM state last night, first time since accidentally poisoning myself with Bacillus Thuringiensis
05:13 AM Tom_L: morning
05:18 AM XXCoder: jt awesome :) hope next day will be even better
05:31 AM JT-Cave: I actually spent 5 hours in the machine shop yesterday and almost finished a couple of little parts
05:31 AM JT-Cave: went to put the valve on the slide and I realized I forget a couple of clearance holes lol
05:35 AM XXCoder: doh
05:36 AM JT-Cave: it's a weird mount for a loctite ca glue applicator
05:37 AM JT-Cave: well I just disabled translations for the Mesa Configuration Tool after waiting over 4 months for something to be translated, adds extra work for me with no return
05:41 AM Tom_L: the tool seems pretty solid now though
05:44 AM JT-Cave: thanks, still toying with being able to use GPIO instead of daughter card
05:56 AM Tom_L: no doubt you could go on forever with it. there are so many combinations possible
06:00 AM JT-Cave: yeah there is a point where the return is not worth the effort
06:01 AM XXCoder: lst 10% is 90% of work
06:10 AM JT-Cave: take translations after spending hundreds of hours getting it to work any time I change anything in the ui file I have regenerate the translation file, if no one is going to translate the extra effort has no payback
06:22 AM pere: JT-Cave: are you using weblate to handle translations?
06:22 AM JT-Cave: yea
06:22 AM JT-Cave: but I"m fixing to delete it
06:23 AM pere: then I will not disturb. :)
06:23 AM JT-Cave: weblate has not be updated in quite some while
06:24 AM JT-Cave: weblate seems to attract folks to larger projects not my little tool
06:24 AM pere: translators are needed for linuxcnc and the documentation too, <URL: https://hosted.weblate.org/projects/linuxcnc/ >, for those that do not want their mother tongue to die. :)
06:25 AM pere: JT-Cave: Personally I like to translate small projects, as it is easier to bring them to completion.
06:28 AM pere: so, my X and Y is homing just fine, but Z is not. X and Y home by moving in the negative direction until hitting the home and limit switch, then slowly back off until the switch releases. Z on the other hand...
06:31 AM pere: Sorry, X and Y move in the positive direction. Z will move in the positive direction too until hitting the home and limit switch, and then continue in the positive direction past the switch and stop when it come to the other (aka wrong) side of the switch. The documentation claim this should be controlled by the sign of HOME_SEARCH_VEL, HOME_LATCH_VEL and HOME_FINAL_VEL, but I am unable to spot the difference in in <URL: https://github.com/SebKuzminsky/Maz
06:31 AM pere: akVQC1540/blob/bon-dev/MAZAK-VQC1540-20170501/MAZAK-VQC1540.ini >
06:33 AM pere: btw, I got at_pid limping along, and managed to get some P, I and D values that seem to not overload the axis controller voltages too often, at least. Still a large following error.
06:36 AM JT-Cave: there are 4 possible homing sequences as defined by the sign of HOME_SEARCH_VEL and HOME_LATCH_VEL
06:36 AM JT-Cave: sounds like you want type 2 search for switch , back off, then slow search back?
06:37 AM JT-Cave: the other variable is are you using USE_INDEX?
06:38 AM pere: all as on HOME_USE_INDEX = NO. currently running ini file is in the git repo above.
06:39 AM pere: what confuses me is that as far as I can tell, all the axis have the same homing sequence defined, but only two of them work.
06:39 AM JT-Cave: ok, is the sequence above what you want
06:39 AM pere: I want quick positive until switch, slow negative until switch is released and them quick movement to the offset.
06:40 AM pere: I got quick positive until switch, slow positive until past switch, then confused.
06:40 AM JT-Cave: ok then search and latch must be a different sign
06:40 AM JT-Cave: I can't open your link so maybe pastebin it
06:41 AM JT-Cave: http://linuxcnc.org/docs/stable/html/config/ini-homing.html#_homing_sequence
06:41 AM pere: because of the stupid line split? Perhaps <URL: https://github.com/SebKuzminsky/MazakVQC1540/blob/bon-dev/MAZAK-VQC1540-20170501/MAZAK-VQC1540.ini > work better?
06:41 AM JT-Cave: ah there we go
06:41 AM pere: JT-Cave: already read the doc, followed it and got confusing setup.
06:43 AM JT-Cave: this may be your issue HOME_FINAL_VEL = -42 perhaps should not be negative
06:43 AM pere: ah. testing. back in a minute.
06:46 AM pere: nope, did not help
06:46 AM JT-Cave: did it change anything
06:46 AM pere: I think it is in the latch step speed it go wrong. nothing changed.
06:47 AM pere: I've tried to change the latch sign, did not help.
06:48 AM JT-Cave: is Z moving off the switch during the search?
06:49 AM JT-Cave: or moving past it?
06:49 AM JT-Cave: if so that's sounds like an electrical issue
06:51 AM JT-Cave: the ini is very straight forward and if you don't get the correct behavior then it's a machine issue like the switch or something
06:52 AM pere: Z is on top of the switch at the end of the search.
06:52 AM pere: so it need to back off the switch to find home. the switch is 3-4 cm wide, or so.
06:52 AM JT-Cave: then does it back off the switch?
06:53 AM pere: nope, it continue past the switch.
06:53 AM pere: could well be a machine issue, the Z axis is not well tuned yet.
06:53 AM JT-Cave: that's odd for sure because x and y are configured exactly the same
06:54 AM JT-Cave: try slowing down the search
06:58 AM * JT-Cave goes to get his chicken chores done before it gets hotter outside
06:59 AM _unreal_: ya I didnt get too far designing the pendent. I need to test a bunch of features on the controller. the documentation is a tiny big lacking in details. so I'm not sure if a pin= function button. or connecting to the sensor device.. and so on...
06:59 AM _unreal_: sigh
06:59 AM _unreal_: oh well time to clean the pool
07:00 AM _unreal_: Or at least make food then clean the pol
07:01 AM CloudEvil: JT-Cave: It's annoyingly hot at the moment. (Admittedly here, that is 23C)
07:07 AM pere: reducing search speed seemed to help.
07:09 AM pere: <URL: https://github.com/SebKuzminsky/MazakVQC1540/commit/08a9371f88e6948d1d4e8fc24d9f1fddb207697c > got it limping along.
07:28 AM JT-Cave: CloudEvil, same here but it feels cool compared to last week
07:32 AM ZincBoy[CAON][m]: I use one of these in my system. It works great and makes building a control panel quite easy. Andy created a HAL component to decode the keypad scan codes to pins (matrix_kb) so that part is much easier now.
07:44 AM Tom_L: pere, have you done any debounce on the switches?
07:45 AM Tom_L: the z switch coule be excessively noisy
07:50 AM tcurdt: on startup I am getting "X Error of failed request: BadWindow (invalid Window parameter)" and then it shuts down ... is that the axis UI telling me it cannot start for some reason?
07:51 AM JT-Cave: sounds like a linux issue
07:52 AM JT-Cave: what os
07:52 AM CloudEvil: It's a generic xwindows issue
07:52 AM CloudEvil: (yes, linux)
07:53 AM tcurdt: debian 10, based off the official iso ... ok, maybe because I switched to another vnc then
07:54 AM tcurdt: yup!
07:55 AM pere: Tom_L: nope
07:55 AM pere: miss0r: sad to report that the magic smoke escaped from the spindle controller just now. :(
07:56 AM CloudEvil: :/
08:00 AM tcurdt: when I am in the component view dialog ... wasn't there a way to have a visual way to see the state of the pins? ... watch is empty
08:00 AM JT-Cave: axis?
08:00 AM tcurdt: yes
08:01 AM JT-Cave: Machine > Show LinuxCNC Status
08:01 AM JT-Cave: or for hal pins you have to add them to the watch window one at a time
08:03 AM tcurdt: I think I've seen the latter ... that could these dots in the watch screen, right?
08:03 AM * tcurdt searches the docs on how add the watches
08:04 AM JT-Cave: do you know where the watch window is?
08:05 AM JT-Cave: Machine >
08:05 AM tcurdt: This watch tab is what I am looking at ... but it's still empty http://linuxcnc.org/docs/html/hal/halshow.html#_footnote_1
08:05 AM JT-Cave: Machine > Show
08:06 AM JT-Cave: Machine > Show Hal Configuration
08:06 AM tcurdt: ah ... got it
08:06 AM JT-Cave: good
08:06 AM tcurdt: within the tab I have to click the name
08:07 AM JT-Cave: with the watch window open you have to select each pin or parameter you wish to monitor
08:07 AM tcurdt: don't see how to remove them though
08:07 AM JT-Cave: the menu
08:09 AM * JT-Cave heads for the shower
08:41 AM pere: why is it that a spindle motor can not be run directly from three phase main power?
08:43 AM ZincBoy[CAON][m]: Spindle motors are not designed to run directly from mains power. You can try but it will very likely result in the smoke escaping from the motor as well.
08:44 AM pere: I have seen that elsewhere, but I hoped someone could explain why.
08:44 AM JT-Cave: depends on the motor, in the case of my VMC it's a high powered servo
08:46 AM ZincBoy[CAON][m]: The volts/hz are very different than a standard 3 phase motor. Means that if you run mains power voltage and hz it will be wrong for the motor and overload it.
08:47 AM pere: My motor is 10kW. Is it the volts/hz relationship that is different? My motor seem to run on 230V already.
08:49 AM rmu: pere: what speed (rpm) is it rated?
08:49 AM pere: Not sure.
08:49 AM pere: vague memories of 3000 or 6000 rpm
08:50 AM rmu: pere: AC induction motors with 1 pole do 60 rpm per Hz. if the motor is designed to run at up to 400Hz you should at least reduce voltage for lower speeds (or use some more advanced control scheme)
08:51 AM rmu: 3000rpm could be 1 pole motor at 50Hz (real rpm will be more like 2850)
08:52 AM pere: can a VFD do this?
08:52 AM rmu: a dumb VFD usually lowers voltage with frequency
08:52 AM pere: I got a <URL: https://www.se.com/us/en/product/ATV71HU30M3Z/variable-speed-drive-atv71-3kw240v-emc-filterw-o-graphic-terminal/ >. No idea how stupid it is.
08:53 AM rmu: why would you drive a 10kW spindle with a 3kW FU?
08:56 AM pere: rmu: because I got one at hand and would like to see if the motor can turn.
08:56 AM ZincBoy[CAON][m]: A VFD can do this but it must have an encoder input. A simple V/Hz drive will NOT work. You will need a fairly high end drive with a sensored vector mode.
08:58 AM rmu: ZincBoy[CAON][m]: depends on what you want to do. my spindles run "open loop" without encoder feedback on a "dumb" V/Hz VFD. but of course you can't rigid tap with those.
08:59 AM rmu: pere: i think you can give it a go
08:59 AM pere: what can go wrong?
09:00 AM rmu: pere: just program prober voltage / frequency from you motor, and a smaller power / current
09:00 AM rmu: the drive probably won't accept a 10kW motor
09:01 AM ZincBoy[CAON][m]: It really depends on the motor. The one in my VMC will NOT run on a standard V/Hz drive. Pere’s motor is very likely the same as it is the same era and style as mine.
09:02 AM rmu: pere: any link to the motor or datasheet? is it async induction motor?
09:02 AM pere: rmu: how will it reject it? killing fuses, magic smoke, refuse to run, die on the spot?
09:02 AM pere: rmu: will see what I can find.
09:03 AM rmu: the vfd will indicate an error condition if it doesn't like something
09:03 AM rmu: DON'T disconnect cabling between vfd <-> motor while it is running, some VFDs don't like that and let out the magic smoke
09:03 AM ZincBoy[CAON][m]: It might run at low rpm but it will overload when trying to go to higher rpm.
09:04 AM rmu: pere: the motor should have a type label indicating type of motor, power, frequency, voltage, amperage, cos phi etc...
09:06 AM rmu: pere: vfd should have parameters for power / frequency voltage of motor, be sure to program those accordingly
09:06 AM pere: The motor is Mitsubishi SJ - 9 - A.
09:11 AM rmu: hmm. doesn't really say how many poles / frequency this beast runs at
09:13 AM rmu: i'm sorry i have to go. but IMO nothing too bad will happen on that VFD. just try @ 50 Hz, program it to 200V, and look if it turns, how it runs, if it makes strange noises, and the resulting RPM
09:15 AM pere: what are poles? it is a three phase motor. the volt and amp info is behind the encoder, so hard to read. the readable parts say this:
09:15 AM pere: RPM 1500 1500 5000
09:15 AM pere: CONT (KW) 5.5 5.5 4.0
09:15 AM pere: 1/2H (KW 9 9 6.5
09:22 AM JT-Shop: pere, got a photo of the spindle motor?
09:22 AM ZincBoy[CAON][m]: Pretty much the same as my motor. Try starting the vfd at less than 10hz. It will probably work at low power and rpm.
09:24 AM pere: JT-Shop: only of a fraction of the info plate.
09:26 AM pere: JT-Shop: <URL: https://www.werktuigen.com/m-mitsubishi+sj+-+9+-+a+spindle+motor/wt-724-5819 > seem to match it pretty well.
09:29 AM JT-Shop: it does say AC spindle motor
09:30 AM pere: yes. At the top of the plate it states AC SPINDLE MOTOR, THREE PHASE INDUCTION MOTOR.
09:30 AM JT-Shop: it is similar to my vmc spindle motor which is an AC servo motor but as I understand an AC servo is just a motor
09:30 AM JT-Shop: it should work with a vfd then
09:31 AM pere: can linuxcnc do the encoder, and give speed instructions to the vfd?
09:32 AM pere: <URL: https://controlrepair.com/shop/tamagawa-seiki-rfh1024-22-1m-68a-5v-dc-optical-shaft-encoder-52763#attr=384139 > is the encoder.
09:33 AM JT-Shop: yes, if you are controlling the vfd with linuxcnc
09:33 AM JT-Shop: and you can read in the encoder to linuxcnc
09:34 AM jpa-: ZincBoy[CAON][m]: not running with V/Hz control sounds like just having wrong parameters in the VFD (e.g. my spindle motor specs 400Hz @ 230V, so running with stock settings of 50Hz @ 230V would burn it out)
09:42 AM pere: note, it is just almost the encoder. look like it, but mine is marked OSE not RFH.
10:08 AM tcurdt: I have connected (just the DIR of) one motor/driver to see if I can control it via linuxcnc. Before I get out the scope - is there a way to see what the stepgen via linuxcnc?
10:08 AM tcurdt: as soon as I try to move the axis I am getting a "following error"
10:08 AM tcurdt: and I don't quite get how that could be
10:09 AM tcurdt: since linuxcnc has no idea about the FE of the driver
10:09 AM tcurdt: so that's a FE inside linuxcnc?
10:10 AM roguish[m]: JT-Shop: are you alive and well today????
10:11 AM pcw--home: tcurdt: default Mesa step/dir setups use a local PID loop to control the position. If there is a setup error of some kind, you can get a following error.
10:24 AM tcurdt: pcw--home I tried not to stray away too far from the generated config ... but that's what I have so far: https://gist.github.com/tcurdt/242330bc590b1b6d75d6ec9681c819d4