#linuxcnc Logs

Jun 19 2018

#linuxcnc Calendar

12:03 AM RyanS: called 3 local supplies looking for dowel pins "no stock -2 weeks" and variations. pity the poor Australian here
12:52 AM Lcvette: Anyone on here running step direction servos?
12:56 AM Lcvette: I'm starting to turn up the velocities a bit now that I have the limits set and tested and working and noticing that the servos aren't real happy at higher speeds
12:56 AM Lcvette: They get a bit jerky
12:57 AM Lcvette: Wondering if I need to change some settings but not finding allot of information on pulse type servos and Linuxcnc/Mesa settings
12:57 AM MrHindsight: have you read JT's guide to servo tuning?
12:57 AM Lcvette: I was looking it over but it appeared it was for analog
12:58 AM MrHindsight: make sure your pulse widths for Step are long enough..
12:58 AM MrHindsight: but no too long that they overlap each other at high step rates
12:58 AM MrHindsight: it's similar
12:59 AM MrHindsight: Step and Dir vs analog are handled by HAL
12:59 AM MrHindsight: the PID just gets output as Step and Dir vs an analog out
01:00 AM MrHindsight: PID are tuned the same
01:00 AM Lcvette: Ok I'll go back and read it now that I know it pertains to pulse type drives as well
01:01 AM Lcvette: They certainly need some tuning
01:02 AM MrHindsight: what servo mode are you using? Position is a good start
01:02 AM MrHindsight: what are these servos running?
01:02 AM Lcvette: Servo mode?
01:02 AM MrHindsight: I assumed a machine axis on a mill
01:03 AM Lcvette: Vmc
01:03 AM MrHindsight: a VMC for a spindle?
01:03 AM Lcvette: https://youtu.be/-iJX7SuJqT0
01:04 AM Lcvette: Vertical machining center
01:04 AM Lcvette: Xyz Axis's
01:04 AM MrHindsight: ok
01:06 AM MrHindsight: possible servo modes: position, velocity and torque
01:07 AM MrHindsight: position is the easiest for your application
01:07 AM MrHindsight: speed mode would be for a spindle or a motor that needs to run at some constant speed
01:08 AM MrHindsight: which servo drives are you using?
01:09 AM Lcvette: Yes I'm running in position
01:09 AM Lcvette: Hence pulse
01:10 AM Lcvette: Machine is up and running and scaled just is a bit jerky at higher speeds
01:10 AM MrHindsight: other modes are possible with Step/Dir but thats another discussion
01:10 AM Lcvette: Over 500ipm
01:10 AM Lcvette: Mac sould be 708ipm on x and y
01:11 AM Lcvette: Mac*
01:11 AM Lcvette: Max*
01:11 AM Lcvette: But it shudders a bit at those speeds when stopping and starting
01:20 AM IchGucksLive: morning from germany
01:21 AM hazzy-lab: Morning ichs
01:21 AM hazzy-lab: gn8
01:21 AM MrHindsight: Lcvette: post your configs for someone to see
01:22 AM IchGucksLive: Lcvette, servos shoudent be handled by step dir
01:23 AM IchGucksLive: Lcvette, the mesa encoder read cycle can handle up to 4 Milion pulses per second
01:23 AM MrHindsight: IchGucksLive: he has Step/Dir dervo drives
01:23 AM IchGucksLive: so the analog driveing mode is better and faster
01:24 AM MrHindsight: dervo/servo
01:24 AM IchGucksLive: MrHindsight, the servo drives have usaly more then that mode
01:24 AM MrHindsight: he said his are only Step and Dir
01:25 AM IchGucksLive: cand belive
01:25 AM MrHindsight: holy Jesus
01:25 AM MrHindsight: Lcvette: did you lie to me? :p
01:25 AM IchGucksLive: MrHindsight, they then m,aybe limited to 3000rpm
01:27 AM IchGucksLive: MrHindsight, how do i come to that conclusion
01:27 AM IchGucksLive: servo drives are international 200khz
01:28 AM IchGucksLive: 3000rpm = 50rp/sec
01:28 AM IchGucksLive: most strandard encoder are 1000ppr
01:28 AM MrHindsight: IchGucksLive: did you just come from the beer garden?
01:28 AM MrHindsight: why are you going on like this?
01:28 AM MrHindsight: Lcvette: has the questions
01:28 AM IchGucksLive: 50000pps internaly Quad is 200khz
01:29 AM Lcvette: They are only pulse
01:29 AM Lcvette: No analog
01:29 AM MrHindsight: SEE
01:29 AM MrHindsight: I give up
01:29 AM IchGucksLive: MrHindsight, of 2 beer garden for the next 10hr
01:29 AM IchGucksLive: beer
01:30 AM IchGucksLive: 8am here
01:30 AM Lcvette: Lol
01:32 AM Lcvette: https://pastebin.com/ksWW4SDf
01:32 AM Lcvette: Hal file
01:32 AM Lcvette: https://pastebin.com/wx03DHav
01:32 AM Lcvette: Ini file
01:41 AM MrHindsight: P = 3000
01:41 AM MrHindsight: I = 0
01:41 AM MrHindsight: D = 0
01:41 AM MrHindsight: I didn't look over all your numbers to see if the timings are all within range but...
01:42 AM Lcvette: Is that your recommendation?
01:42 AM Lcvette: 3000 p?
01:42 AM MrHindsight: try adjust D to see if it dampens the oscillations at start/stop
01:42 AM MrHindsight: NO
01:42 AM Lcvette: Ok
01:42 AM MrHindsight: that what I saw in your INI file
01:43 AM MrHindsight: ftp://ftp.pmccorp.com/pub/support/docs/tutorials/Servo_tuning_tutorial.pdf
01:43 AM MrHindsight: might also be of help
01:43 AM MrHindsight: in general for servos
01:44 AM MrHindsight: 1am here
01:44 AM MrHindsight: this should not be too big a deal to tune
01:46 AM renesis: ziegler-nichols it, if you can get it to oscillate and measure the periods
01:46 AM Lcvette: Cool I'll read it over
01:46 AM Lcvette: 2am here
01:46 AM Lcvette: Lol
01:46 AM renesis: https://en.wikipedia.org/wiki/Ziegler%E2%80%93Nichols_method
01:47 AM MrHindsight: is it oscillating at start stop or?
01:47 AM MrHindsight: just making loud noises?
01:48 AM Lcvette: It's hard to explain
01:48 AM Lcvette: Like it overshoots I guess and rewinds
01:48 AM Lcvette: Makes the machine shudder
01:48 AM MrHindsight: a few times
01:48 AM Lcvette: Kind of violent
01:48 AM MrHindsight: yeah sounds like it's oscillating
01:49 AM renesis: if oscillates periodically can tune w/ formulas
01:49 AM renesis: shit works, is magic
01:49 AM MrHindsight: http://linuxcnc.org/docs/2.5/html/motion/pid_theory.html
01:49 AM Lcvette: I remember Peter saying something about the p value being really high
01:50 AM Lcvette: I sould now I'm not closing the loop
01:50 AM Lcvette: Note
01:52 AM renesis: how do you do a pid open loop? just assume errors based on some diff equation?
01:53 AM MrHindsight: just leave the encoders disconnected and hope for the best?
01:53 AM MrHindsight: changing from beer to scotch
01:54 AM MrHindsight: adios
02:00 AM Lcvette: That's why I'm not sure how timing works
02:00 AM Lcvette: Tuning
02:12 AM MrHindsight: can't tune without feedback
02:12 AM MrHindsight: all you can do now is configure for open mode Step and Dir
02:13 AM MrHindsight: without feedback the PID is seeing no response after sending a move
02:42 AM Deejay: moin
03:17 AM rmu: hmm
03:18 AM rmu: using servos with step/dir and then again doing stepgen with PID with the same feedback that is used by the servo can't work
03:18 AM rmu: one control loop will fight the other
03:23 AM RyanS: i remember reading about counterfit Australian copper pennies (1930 worth a fortune) it sounds like they use die sinker EDM to copy the original. I wonder if copper coins could be used as the electrode to make a tool steel miniting die? I imagine the original would get destroyed so I can't see who would do that to a $50k coin
03:23 AM Loetmichel: ahem... hint please: is there a Gcode to rotate or flip the coordinate system 180°? just noticed that my workpiece mount and my program are 180° out of phase... :-(
03:24 AM Wolf__: flip machine over
03:24 AM RyanS: not that I advocate rare coin counterfeiting :-)
03:24 AM Loetmichel: RyanS: the original wouldnt be destroyed completely. just a bit eroded
03:25 AM rmu: Loetmichel: G10 L2 R180 ?
03:25 AM Loetmichel: you dont have to do it all with the coin. you can make a slightly smaller hole in the steel with a blank copper disk and only to the last few 1/10mm with the actual coin
03:25 AM Loetmichel: rmu: does that rotate the whole "machine?
03:26 AM RyanS: yeah and I guess a coin counterfiter doesn't care about the original they just want to dupe collectors
03:28 AM RyanS: surely the copy could be detected under a microscope?
03:28 AM rmu: Loetmichel: it does rotate one of your coordinate systems, see http://linuxcnc.org/docs/html/gcode/g-code.html#gcode:g10-l2
03:28 AM Loetmichel: ah... found it in the reference... wouldnt that be a "G10 L2 0 R180"?
03:28 AM rmu: strange there is no G92 R...
03:29 AM rmu: Loetmichel: G10 L2 P0 R180, but that change is persistent!
03:29 AM Loetmichel: a G10 L2 P0 R-180 after the job should correct it though?
03:30 AM rmu: no, G10 parameters should be absolute offsets that replace existing onse
03:30 AM rmu: so G10 L2 P0 R0
03:30 AM rmu: assuming you don't change coordinate systems in your program
03:31 AM Loetmichel: ah i see
03:31 AM Loetmichel: will do that and report afterwards ;)
03:35 AM MrHindsight: rmu: did Lcvette mention which servo drive he was using before?
03:42 AM RyanS: Loetmichel: https://www.ngccoin.com/news/article/842/ sounds like they don't do it to any really rare coins
03:51 AM rmu: MrHindsight: how should I know ;-), I didn't read all the backlog (I'm in UTC+02), but from first principles, doing two control loops using the same feedback and in endeffect controlling the same variable won't work
03:55 AM rmu: IMHO he could either use the servo drives in "position" mode, then there is no need to close the position loop a second time in linuxcnc, or he could use the drives in velocity mode, close position loop in linuxcnc, where he would set FF to 1 and everything else to O(0) ;-)
03:58 AM MrHindsight: I'm not a good mind reader is why I asked
03:59 AM MrHindsight: and I often forget that sometime people spend a lot of money on parts and don't know how to put them all together
03:59 AM MrHindsight: buy first. plan later
04:00 AM rmu: MrHindsight: i took a glance at the hal file
04:01 AM rmu: MrHindsight: it seems he is using hostmot2 stepgen with pid to generate step/dir signals that go into servo drive
04:02 AM * Wolf__ resembles that remark
04:02 AM RyanS: and at all costs never read the instruction manual
04:02 AM Wolf__: what instruction manual? lol
04:04 AM rmu: MrHindsight: step/dir amounts to position command, so servo drive closes position loop, no wonder that doing a second position loop around that leads to oscillations
04:05 AM MrHindsight: is that what he is doing now, I didn't look at his HAL file
04:06 AM rmu: yes it looks like that
04:06 AM MrHindsight: I thought if he got as far as he did and spoke with Mesa about the cards that he had a working plan
04:06 AM MrHindsight: wow
04:06 AM Loetmichel: rmu: Report: works like a charm... have to rework 2 dozen Monitor frames to fit the new "front soft keys" instead of the old side ones because Boss forgot that he bought the new HP TFTs to mount in there.... and we had old frames in storage... -> http://www.cyrom.org/palbum/main.php?g2_itemId=17127&g2_imageViewsIndex=1 (complete Thin client with the new TFTs underneath)
04:06 AM Loetmichel: ... and switching back works well, too
04:07 AM RyanS: damn i want a swiss lathe
04:08 AM Loetmichel: and the program i had (did that some time back when the same problem arised) was mader for fixing the Frame the other way around on the CNC ;)
04:08 AM rmu: MrHindsight: i think getting rid of the PID in linuxcnc should do it.
04:08 AM MrHindsight: Monday was pretty bad, is Tuesday any better so far?
05:52 AM jthornton: morning
05:53 AM XXCoder: 84f. nice and cool compared to what it was at work lol
05:56 AM Tom_L: morning
05:59 AM Deejay: howdy
06:40 AM jthornton: Tom_L: put the shower door frame in yesterday... getting close
06:41 AM XXCoder: oh yeah
06:41 AM XXCoder: what happened with pan? I remember something about it popping up
06:42 AM jthornton: the second time I tried to set it I had put some plywood down to take some space up and it swelled up and lifted the pan
07:33 AM Beachbumpete1: Good morning LinuxCNC
08:07 AM Deejay_ is now known as Deejay
08:34 AM JT-Shop: morning Pete
08:47 AM Tom_L: 76F, hi 91
08:47 AM Tom_L: supposed to get some rain but i don't see it
08:49 AM Tom_L: later..
09:28 AM pcw_home: Lcvette: should you read this later, if you have a closed loop system with encoder feedback, you must use a much smaller P term
09:30 AM pcw_home: for a "open loop" stepgen a P term of 1/servo thread period is appropriate so 1000 for a normal 1 KHz servo thread
09:31 AM pcw_home: this means that any position deviations are corrected by the next thread time
09:32 AM pcw_home: This high P term only works because the stepgen is a nearly perfect 1st order system, the servo drive is not...
09:51 AM Roguish: pcw_home: good morning Peter. do you have a 7I40 in stock (dual 400w driver) ?
10:21 AM pcw_home: probably though maybe only one flavor
10:40 AM Roguish: pcw_home: mesa_store lists 17 HV and 20 LV. I have 36v power supplies.. Which would be best? obviously the LV is ok, but would the HV work at 36v? how well?
10:51 AM IchGucksLive: hi all
10:54 AM IchGucksLive: nice day here in germany
10:54 AM IchGucksLive: getting summer tomorrow one day
10:57 AM -!- #linuxcnc mode set to +v by ChanServ
11:00 AM IchGucksLive: gloops, how is it after the glorious win
11:00 AM gloops: yes not a bad performance Ichs
11:00 AM gloops: always feel a bit safer with an opening win, bit of pressure off
11:00 AM IchGucksLive: 100% more then owers
11:00 AM gloops: i didnt see th Germany game, i know you lost - did they play ok though?
11:00 AM gloops: sometimes unlucky
11:00 AM IchGucksLive: badest performence ever says the newspaper
11:00 AM gloops: hmm, not like the germans, theyre usually up there with the best sides
11:01 AM gloops: England could get to last 8 maybe, but very early days
11:02 AM IchGucksLive: so tomorrow finals day until holiday i hope i can mount all mashines to full motion
11:02 AM IchGucksLive: i hope i got a team of electronics stand bye
11:03 AM IchGucksLive: with mechanics and Red wire it is a miss
11:04 AM IchGucksLive: gloops, and as a old man it is hard to get moving around under over side of mashine
11:04 AM IchGucksLive: hi makerman1
11:04 AM IchGucksLive: gloops, YT channel is now down to 800hits per day
11:04 AM gloops: yes it doesnt get any easier Ichs
11:04 AM IchGucksLive: gloops, while coming from 5k
11:05 AM gloops: Ichs hmm, strange
11:05 AM IchGucksLive: arduinocnc takes its toll
11:06 AM IchGucksLive: on a serial command you can make benefit of full BT even Network or radio
11:06 AM gloops: everybody wants USB
11:06 AM IchGucksLive: usb is the badest thing you can go
11:06 AM gloops: main complaint on facebook - linuxcnc ok - but cant use laptop
11:07 AM IchGucksLive: i got left hand side a pile of 12 notbooks running cnc
11:07 AM IchGucksLive: at <20k
11:07 AM pcw_home: USB is a bad idea even for buffered CNC controllers
11:07 AM IchGucksLive: you need to reed what Motherboard works
11:08 AM IchGucksLive: pcw_home, problem is at precision of 0,05mm even 0,1 at 3D printers it works perfect
11:08 AM pcw_home: Its not a precision issue
11:08 AM IchGucksLive: Noewone woudt even blame the USB its always the mechanics
11:08 AM pcw_home: ist a bad interface reliability wise
11:09 AM JT-Shop: dang anti-static bubble rolls have gotten expensive
11:09 AM pcw_home: so not for serious machines
11:09 AM IchGucksLive: pcw_home, i'am with you but they dont here US
11:09 AM IchGucksLive: pcw_home, there are only a view retrofitsa
11:10 AM IchGucksLive: and the hobby people even the scalebuilders are with the arduino people
11:11 AM IchGucksLive: all they say is IT WORKES for ME
11:11 AM gloops: have seen a lot struggling with arduino on facebook though
11:11 AM IchGucksLive: gloops, step in promote lcnc
11:11 AM gloops: seems prone to noise like problems
11:11 AM gloops: i do, very few join here though
11:11 AM gloops: never heard of irc
11:12 AM IchGucksLive: agree
11:12 AM IchGucksLive: im ade e help line video and only 200hits
11:12 AM IchGucksLive: i maybe take the headline wantai Rattmotors 24/7 helpline
11:13 AM IchGucksLive: as mot of us here work with this items
11:13 AM IchGucksLive: most
11:14 AM IchGucksLive: gloops, i think they sold items at 20k euros to me only
11:15 AM gloops: seems to be more sellers now for the stepper/driver bundles, trying to sell them by auction
11:15 AM gloops: coming from northern europe
11:16 AM JT-Shop: dang last time I got anti static bubble roll it was $25 for 125' x 12" now I can only find 250' rolls for $70 and up
11:18 AM IchGucksLive: gloops, the 3d printer boost
11:19 AM IchGucksLive: gloops, nema17 a mass
11:20 AM IchGucksLive: gloops, al we miss is precise redy made worm gearing afortable low price
11:20 AM IchGucksLive: gloops, 27:1
11:20 AM IchGucksLive: best to go
11:21 AM IchGucksLive: 27*400=10800 / 360 =30 steps per deg
11:22 AM IchGucksLive: gloops, gives you a 35mm Diameter at 0,01 Mantel line
11:22 AM gloops: well, i doubt ill be building anything soon, too much to do
11:23 AM fragalot: 'sup
11:23 AM IchGucksLive: gloops, on a standard 70mm diameter part its 0,02mm spacing at full torge and full speed
11:23 AM IchGucksLive: if you go up to 800steps/rev Stepper
11:24 AM IchGucksLive: you are in 135diameter will break your mashine
11:24 AM IchGucksLive: so all in on 27:1
11:24 AM IchGucksLive: preciscen wormgear module 2.5 at A 50mm
11:25 AM IchGucksLive: lets see some costs on homemakiong
11:25 AM gloops: coming from Switzerland now https://www.ebay.co.uk/itm/UK-SHIP-3Axis-Nema-23-Stepper-Motor-425oz-in-Driver-4-2A-CNC-Router-DM542A/163096404205?
11:26 AM IchGucksLive: gloops, expensive
11:27 AM IchGucksLive: gloops, https://www.ebay.de/itm/EU-SHIP-3-axis-NEMA23-stepper-motor-425oz-in-Dual-Shaft-CW5045-Driver-CNC-kit-/253646464197
11:27 AM IchGucksLive: gloops, with board and newer drivers
11:28 AM gloops: nearly £60 more though
11:28 AM fragalot: "UK ship".. item location switzerland
11:28 AM fragalot: is nothing sacred any more
11:29 AM IchGucksLive: gloops, no shipcost and more items better drivers
11:30 AM gloops: their social media dredging has revealed USA and UK are the top 2 buyers
11:30 AM IchGucksLive: gloops, why is there a parportcable but no BOB
11:31 AM IchGucksLive: wheard seller
11:31 AM IchGucksLive: im off need to do somthing
11:31 AM IchGucksLive: till tomorrow
11:31 AM IchGucksLive: Gn8 the old german is leving ROFL
11:31 AM gloops: ok have a good day
11:32 AM gloops: the chinese will continue to cut prices, even until theyre making £1 per ship
11:40 AM gloops: well, suppose i better have half an hour trying to get more old junk working again
11:45 AM Lcvette: MrHindsight: "did Lcvette mention which servo drive he was using before?" K Series Chinese Servo Drives they are Position Drives only
11:49 AM Lcvette: MrHindsight: "MrHindsight: and I often forget that sometime people spend a lot of money on parts and don't know how to put them all together" when my buddy and i placed our order we requested analog drives but the order was sent wrong. because it was a machine build and not a direct send to us, the drives were sent to the machine mfg to be installed and then it took several weeks before the machines were completed and then severa
11:51 AM Lcvette: MrHindsight: by the time the machines l;anded, were rigged and brought back unloaded wired and powered and tested and discovered that they were not analog, almost 6 monthsd had passed since the purchase date and the drives were in the US from china. the manufacturer was insistant it wasn't there fault and it became a mess. it was less expensive to purchase a 7i76 then to replace the drives at that point
11:51 AM Lcvette: now you have the back story
11:53 AM MrHindsight: https://doc.diytrade.com/docdvr/2394064/47325192/1506481821.pdf is this the manual for the drives?
11:54 AM Lcvette: yes
11:56 AM MrHindsight: did you read the comments by rmu ?
11:57 AM Lcvette: yes but im not following
12:00 PM Lcvette: rmu: using servos with step/dir and then again doing stepgen with PID with the same feedback that is used by the servo can't work
12:00 PM Lcvette: rmu: one control loop will fight the other
12:01 PM Lcvette: we were only wanting to try and use the encoder feedback to try and use the encoders for homing
12:02 PM Lcvette: however I connected all of the encoder wires not just the index because i was not certain at the time if they would or would not be needed for it
12:02 PM Lcvette: i can disconnect them if need be
12:03 PM pcw_home: You can use encoder feedback with position mode drives but the P term must be small
12:04 PM pcw_home: you had 3000 for P which is close to 100 times too high
12:04 PM Lcvette: I remember you saying that!
12:04 PM Lcvette: i remember you said 1000
12:04 PM Lcvette: oh
12:04 PM Lcvette: maybe not then
12:04 PM pcw_home: yes 1/tservo but only if the feedback is from the stepgen
12:08 PM Lcvette: i read through the docs last night trying to find something to tell me what to do but did not find much in the way of this
12:08 PM Lcvette: i found the stepper examples
12:08 PM pcw_home: try P of 20 or so
12:08 PM Lcvette: ok
12:08 PM Lcvette: switching to machine computer
12:09 PM pcw_home: And try open loop ( just to make sure the motion issues are not drive/something-else related )
12:11 PM Loetmichel: *GNHIHI* my wife does not only "money launering" but also "data cleansing"... just pulled not only 2*2 eur, 1*1 eur, 1*50ct, 2*20ct und 1*2ct out of the door seal of the washing machine while unloading it but also 2 of my USB thumbdrives ;)
12:13 PM lcvette_: i changed the x axis and it gave an x axis folloing error code during homing sequence
12:14 PM pcw_home: is it set up to home to index?
12:14 PM lcvette_: no
12:15 PM lcvette_: it didn't even make it to the switch
12:15 PM lcvette_: i reset and tried jogging and it goes an inch or tewo and throws same code
12:15 PM lcvette_: Axis X following error
12:16 PM pcw_home: what are your following error settings?
12:16 PM pcw_home: or maybe just post you current hal/ini files
12:17 PM MrHindsight: he did 10 hours ago https://pastebin.com/ksWW4SDf
12:17 PM MrHindsight: https://pastebin.com/wx03DHav
12:17 PM Tom_itx: JT-Shop, how wide is your bubble wrap?
12:18 PM lcvette_: https://pastebin.com/9W49L8T6
12:18 PM lcvette_: ini
12:19 PM pcw_home: lcvette_: set all pid_maxerror values to 0
12:19 PM lcvette_: ok
12:20 PM lcvette_: https://pastebin.com/gksMTDj9
12:20 PM lcvette_: hal
12:21 PM JT-Shop: 12"
12:22 PM JT-Shop: last few rolls were 125' x 12" x 1/2"
12:22 PM lcvette_: i don't see pid_maxerror
12:23 PM pcw_home: its in the hal file
12:23 PM lcvette_: min_ferror?
12:23 PM lcvette_: oh
12:23 PM lcvette_: setp pid.y.maxerror .0005
12:23 PM lcvette_: ?
12:23 PM lcvette_: .0000
12:24 PM lcvette_: for all
12:24 PM pcw_home: yes (0 means don't bound)
12:25 PM lcvette_: setp pid.x.maxerror 0
12:26 PM Tom_itx: JT-Shop, https://www.uline.com/BL_7657/Anti-Static-UPSable-Rolls
12:27 PM lcvette_: ok changed them all to 0
12:27 PM pcw_home: yes for all (all using encoder feedback)
12:29 PM lcvette_: ok changed them all and changed all P to 20
12:29 PM JT-Shop: Tom_itx: thanks, I missed that
12:29 PM lcvette_: no getting following error in z axis
12:29 PM lcvette_: now
12:29 PM lcvette_: immediately
12:33 PM pcw_home: I would expect this since you have somewhat insane velocity and acceleration settings
12:34 PM pcw_home: 280 IPS and 5Gs is pretty scary
12:34 PM lcvette_: metric
12:34 PM JT-Shop: how do you calculate the Gs?
12:35 PM cpresser: 9.81m/s² is one g
12:36 PM lcvette_: oh you probably didn't know that because i was tinkering last night trying to figure how to display the step increments diferently
12:36 PM lcvette_: but the ini is setup in metric
12:38 PM pcw_home: OK metric I had the idea is was inch
12:38 PM lcvette_: my bad
12:38 PM pcw_home: you may need to increase the ferrors for tuning
12:40 PM pcw_home: JT-Shop or 384 I/s^2
12:42 PM JT-Shop: cool, thanks
12:42 PM lcvette_: ok so from here do i follow JT's tuning guide?
12:43 PM lcvette_: JT-Shop: thanks for that b y the way read it last night very good read
12:44 PM JT-Shop: your welcome
12:46 PM fragalot: turns out if I grunt enough I can pick up the base & column of the rf45 and bolt them together alone :P
12:47 PM fragalot: also - bolting it down onto the factory supplied base warps the base by .2mm
12:49 PM Rab: It's all fun and games until your giblets escape through your abs.
12:49 PM lcvette_: mating surfaces not flat?
12:49 PM gloops: bend the knees not the back
12:50 PM gloops: fragalot - piece of cheap chinese tape measure - .2mm
12:50 PM fragalot: hehe
12:50 PM fragalot: lcvette_: kinda hard to expect them to be tbh, the base is made out of a 12mm steel plate weldment, and there's a 1mm sheet metal pan sandwiched in between
12:50 PM gloops: you cant rely on cig packets these days
12:51 PM fragalot: that said - it's very possible that it's the casting's base that isn't flat as I didn't re-machine that but just bolted it down onto that bigger mill's table
12:52 PM lcvette_: oh I read that to mean bolting the base and the column together
12:52 PM lcvette_: you mean the base and the stand
12:52 PM fragalot: ys
12:52 PM fragalot: :P
12:52 PM lcvette_: ok
12:52 PM lcvette_: makes more sense
12:52 PM lcvette_: shim it
12:52 PM fragalot: the column mates perfectly
12:53 PM fragalot: I did
12:53 PM lcvette_: ok
12:53 PM fragalot: 3 points were perfect, 4th needed several shims hovering around .2mm
12:54 PM fragalot: https://imgur.com/gallery/8I60hHW ha.
01:10 PM miss0r: Picture or it didn't happen
01:11 PM miss0r: fragalot: I've put dielectric fluid in the EDM. and I have added the hydraulic fluid to the z-axis... Still nothing works
01:11 PM * miss0r is starting to get worried :D
01:11 PM fragalot: there's nothing to see
01:12 PM fragalot: sounds like a sensor or wiring fault
01:12 PM miss0r: yeah.. I'm pouring over that now
01:13 PM miss0r: I just want to see it do a little machining, so I can rest easy knowing it works "overall". Then I can start finetuning stuff
01:25 PM miss0r: *cough* excuse me... *cough*
01:29 PM fragalot: what was it
01:29 PM lcvette_: miss0r: NICE!!!!!!!! CONGRATULATIONS
01:30 PM miss0r: heh
01:30 PM lcvette_: I kow the feeling of that victory!
01:30 PM miss0r: fragalot: Well, I still have an issue with a thermostat/level switch. But I can move the quill up and down now (error 40, apparently)
01:30 PM miss0r: So, now I should be able to run a spark test :D
01:31 PM fragalot: :P
01:33 PM miss0r: holy hell it can fill the work tank fast !
01:33 PM miss0r: 3 seconds or so, then it is about to come over the top
01:33 PM miss0r: I had to regulate the ballvalve down alot.. the drain isn't fast enoug :D
01:33 PM fragalot: xD
01:58 PM Lcvette: pcw_home: do I need to tune a pulse setup? Or is this something I've created by going the encoders feed back too the Mesa cards?
02:05 PM Lcvette: Meaning, it I eliminate the encoders for the acis and treat the drives as if they were stepper drives would things work better?
02:07 PM Lcvette: The tuning is over my head with the fact that everything i can find seems to to be written for analog and it makes an already daunting task even more confusing
02:12 PM miss0r: fragalot: i found the error.
02:13 PM miss0r: Apparently, when I set it up, I rotated a hose, that was fixed with a screw hoseclamp. The screwhead on said hoseclamp, grapped a wire, and ripped it with the terminal it saw inserted to from a level switch.
02:13 PM miss0r: Fixing it as we speak
02:13 PM fragalot: I know some of those words
02:13 PM miss0r: but imagine this: worst working position ever, combined with very inaccesable work
02:14 PM fragalot: not sure about the order of them
02:14 PM fragalot: hehehe
02:14 PM miss0r: lool
02:14 PM miss0r: I can barely get two fingers in there
02:14 PM fragalot: this is why I have an entire drawer dedicated to oddly shaped pliers :P
02:16 PM miss0r: hehe so do I :D but the space is very scarse. and I have to put a cable shoe on the end of what looks like a 0.34mm^2 wire that is too short to get it to a good position
02:16 PM fragalot: urk
02:16 PM miss0r: yeah
02:16 PM miss0r: I will get it just set, and try to solder it through the hole
02:17 PM fragalot: dirty
02:18 PM miss0r: I think this will work
02:19 PM miss0r: side note: I spilled alot of kerosine earlier... I have a shitty cleanup task ahead of me tonight
02:19 PM * fragalot hands miss0r some of those specialty soaking rags
02:22 PM miss0r: I wish that worked - the handing me stuff over the internet part
02:22 PM miss0r: Those would've been great here
02:23 PM fragalot: worth the investment
02:23 PM fragalot: you'll have to put off buying that chuck for a bit
02:25 PM miss0r: I am so retarted it hurts ! ! !
02:25 PM fragalot: grabbed the iron from the wrong end again?
02:26 PM miss0r: I just wanted to do a quick test before putting the covers back on. FUCK ! I forgot to reattach the secondary flooding hose. it shot 2-3 liters of kerosine over my worktable, the wall, behind the surface grinder ect, before I could turn it off !
02:26 PM miss0r: I'm just standing here looking at it in disbelieve
02:27 PM fragalot: whoops
02:27 PM miss0r: I'm still unsure if I should cry or laugh
02:27 PM miss0r: Tiping towards crying
02:28 PM miss0r: well - we will find out of my new laptop is a liquid reciliant as its pumped up to...
02:28 PM gloops: less haste, more speed
02:28 PM hazzy-lab: oh no miss0r!
02:28 PM fragalot: miss0r: at least it's a dieelectric?
02:28 PM hazzy-lab: right!
02:28 PM miss0r: indeed
02:29 PM miss0r: odor free my ass... When you have 4m^2 of surface covered with it, you can start to smell it
02:29 PM fragalot: lol
02:29 PM miss0r: wI blame the french
02:29 PM miss0r: I*
02:30 PM gloops: better learning these lessons in your own place miss0r, could have been someone elses
02:30 PM gloops: look on the bright side lol
02:34 PM miss0r: gloops: Not helping
02:34 PM miss0r: Well, that *IS* a good thing. But you need to keep in mind, my shop is so tightly packed, that a million things are now soaked in kerosine
02:35 PM skunkworks: miss0r: we had a couple oopses -- none that bad though ;)
02:36 PM miss0r: also, my walls are 'raw' porous concrete. It will stain for ever
02:37 PM miss0r: and the pressure is insane.. it flew ~2meters in the air, before hitting a drawersystem, exploding all over the place. comparable to a gardenhose at full blast.. inside my shop... with kerosine
02:37 PM gloops: beginning to look like a real workshop finally lol, its fine, takes a bit of getting up paraffin, doesnt evaporate very quick, but shit happens
02:37 PM fragalot: I can't quite find the exact image i'm looking for
02:38 PM fragalot: but this is close
02:38 PM fragalot: https://proxy.duckduckgo.com/iu/?u=https%3A%2F%2Fi.ytimg.com%2Fvi%2Ft7_b2Tljf9Y%2Fhqdefault.jpg&f=1
02:38 PM miss0r: gloops: yeah..
02:38 PM miss0r: fragalot: Had you only shown me that before the fact :D
02:39 PM fragalot: https://www.youtube.com/watch?v=t7_b2Tljf9Y full scene :P
02:39 PM miss0r: also, that little fix didn't fix the overall problem ! *sigh*
02:39 PM fragalot: :(
02:39 PM miss0r: alright. I have reached a point where I am ready to laugh at my own misery
02:39 PM Deejay: re
02:39 PM miss0r: wb
02:40 PM Deejay: tnx
02:40 PM miss0r: nta
02:40 PM miss0r: (Not This Again)
02:41 PM fragalot: miss0r: probably damaged the sensor more than just that loose wire
02:41 PM fragalot: what type of sensor is it
02:41 PM fragalot: communication or analog?
02:41 PM fragalot: or dead simple digital
02:41 PM miss0r: French model from 1982 - Take a wild guess? :D
02:41 PM fragalot: how many wires :P
02:42 PM miss0r: 2 (if the one I fixed is indeed the broken one)
02:42 PM miss0r: This error indicator apparently has 2 thermostats and 2 level switches
02:42 PM miss0r: Trying to look it up in the wiring diagram at the moment
02:42 PM miss0r: And this machine doesn't have a 'trust me, everything is alright'-switch to override it
02:43 PM pcw_mesa: Lcvette: the tuning is trivial, you basically just see how much P term you can get way with
02:44 PM fragalot: miss0r: can you override the switches individually?
02:44 PM miss0r: no electrical diagram either in the documentation
02:44 PM fragalot: hm
02:44 PM miss0r: I suppose I can. But did I mention how inaccessible they are ? :D
02:44 PM pcw_mesa: but you will likely have to increase your ferror limit to see what is going on
02:44 PM fragalot: miss0r: the switch is, but where does the wire go?
02:44 PM miss0r: further in the back :D
02:45 PM fragalot: xD
02:45 PM * miss0r gets out the multimeter, and tries to look like he knows what he is doing
02:45 PM fragalot: you may not want to hear this
02:45 PM fragalot: but can you move the machine forward
02:46 PM fragalot: to a position where it's actually accessible from the back
02:46 PM miss0r: I realy cant
02:46 PM fragalot: bleh
02:46 PM miss0r: its as far forward as the table allows. and it is 300kgs
02:46 PM fragalot: you (and I) need one of those little lift tables
02:46 PM fragalot: on wheels
02:47 PM pcw_mesa: Lcvette__: you may be better off with open loop depending on how much delay there is in the drives
02:47 PM miss0r: Yeah. I had on here to lift it up. But I returned it
02:48 PM miss0r: The thermostat right next to the switch I repaired doesn't change its contacts, regardless of what temperature I set it to.. I will take that apart and see what is up
02:50 PM pcw_mesa: Lcvette__: but open loop will not support homing to index (at least currently)
02:54 PM Isleape1 is now known as Isleape
02:55 PM lcvette__: pcw_mesa: I wouldn't mind trying open loop , currently the new home sensors seem to be repeating to about a +/-0.0001"
02:55 PM lcvette__: I can live with that
02:56 PM lcvette__: now that I have everything else finally working I'm not sure how to switch to open loop though..lol
02:56 PM miss0r: fragalot: That still didn't fix it. I replaced the microswitch in the thermostat, and it now works
03:00 PM fragalot: right, so both the level & thermostat are confirmed to work, yet you still get an alarm?
03:01 PM lcvette__: pcw_mesa: I guess I could just make another PNCConfig without encoder feedback and replace the axis information in the Hal and ini?
03:01 PM lcvette__: try that out?
03:02 PM pcw_mesa: Did you try changing the P value (after widening the ferror limits)?
03:05 PM lcvette__: pcw_mesa: I'd be shooting in the dark not knowing what I'm doing
03:06 PM lcvette__: pcw_mesa: I'm smart enough to admit when I'm not smart enough to figure something out with the resources available to me..
03:06 PM pcw_mesa: Follow the tutorial, you just start low and increase P until it becomes unstable than back off about 20/30%
03:07 PM lcvette__: increase P?
03:08 PM lcvette__: like I said I think I'm an open loop guy
03:08 PM lcvette__: just not something I will be able to get a grasp on
03:09 PM pcw_mesa: widen ferror limits, start low with P, halscope following error, increas P with calibrate menu
03:10 PM lcvette__: widen limits
03:11 PM pcw_mesa: yeah so you dont get an instant following error
03:11 PM lcvette__: any number
03:11 PM lcvette__: just big so it doesn't error
03:12 PM pcw_mesa: say to 1 and 10 mm (not too huge for safety)
03:13 PM pcw_mesa: enough to differentiate oscillation from too much P gain from lagging motion from too little
03:14 PM pcw_mesa: if the ferror limits are too low you can't tell what is going on since you will just get an immediate fault
03:16 PM lcvette_: FF leave at 1?
03:16 PM lcvette_: FF1 = 1
03:17 PM lcvette_: pcw_mesa: https://pastebin.com/Vukd1M9K
03:17 PM lcvette_: thats the x axis setting from the ini
03:18 PM pcw_mesa: yes FF1 is always 1 if the scaling is correct
03:18 PM lcvette_: ok
03:18 PM lcvette_: it is
03:25 PM lcvette_: in the tutorial for channel 3 he is using joint-vel-cmd
03:25 PM lcvette_: is that for velocity mode
03:25 PM lcvette_: ?
03:25 PM lcvette_: http://gnipsel.com/linuxcnc/tuning/servo.html
03:27 PM pcw_mesa: We are treating your servos as if they are running in velocity mode (which will be a close approximation if they do not have significant delay)
03:29 PM lcvette_: what speed should i be testing at?
03:29 PM lcvette_: max?
03:30 PM pcw_mesa: usually errors are worst during accell/deccel
03:35 PM lcvette_: ok im not sure im using halscope right
03:36 PM lcvette_: i have it looking like JT's screen but when i jog i just get horizontal lines
03:37 PM JT-Shop: what trigger are you using?
03:37 PM lcvette_: normal
03:37 PM JT-Shop: no, what source?
03:38 PM JT-Shop: look at the lower right corner of halscope
03:38 PM lcvette_: ahhhh
03:38 PM lcvette_: channel 4
03:38 PM lcvette_: yours says channel 3
03:38 PM JT-Shop: which is?
03:38 PM lcvette_: f-error
03:38 PM JT-Shop: change that to vel-cmd
03:38 PM lcvette_: i did it exactly the same as your tutotial was setup
03:39 PM lcvette_: same channel numbers
03:39 PM JT-Shop: so as soon as the velocity increased the trigger is tripped
03:39 PM lcvette_: just missed the trigger channer at the bottom
03:40 PM lcvette_: changed it
03:40 PM lcvette_: i still just have horizontal lines
03:43 PM lcvette_: no squiggly lines or anything
03:43 PM Tecan: https://www.youtube.com/watch?v=st8dkGzJWtg penetrating liquid showdown.
03:44 PM tjb1: So if I were to gang proximity sensors into 1 input, should I be placing diodes on each before the input so they cant backfeed each other?
03:44 PM JT-Shop: does the black line up top next to the text TRIGGER? move when you jog?
03:46 PM JT-Shop: tjb1: 3 wire prox or 2 wire?
03:46 PM tjb1: 3 wire
03:47 PM lcvette_: it does on channel 4 but not on channel 3
03:47 PM lcvette_: ooh i got some squiggly line
03:47 PM lcvette_: f-error
03:48 PM JT-Shop: did set the trigger level?
03:49 PM JT-Shop: move the sliders and see the lines move and put the level just above 0
03:49 PM JT-Shop: for joint vel cmd
03:52 PM tjb1: JT-Shop: pretty sure they are npn
03:52 PM lcvette_: think i screwed it up
03:53 PM lcvette_: when i try to move joint velocity command to near level all the lines go flat across again
04:05 PM Deejay: gn8
04:10 PM lcvette_: open loop is looking like the right course of action here
04:10 PM lcvette_: this sucks
04:21 PM gloops: youre have some fun with this machine lcvette
04:21 PM lcvette_: no
04:21 PM lcvette_: not this part
04:22 PM lcvette_: this sucks
04:22 PM lcvette_: halscope locked up on me
04:22 PM lcvette_: had to be shut down
04:22 PM lcvette_: not sure what thats about
04:23 PM lcvette_: not sure if thats why im not seeing what JT's screens look like or if its a setting
04:24 PM lcvette_: but sliders and what not aren't covered very well in the tutorial so again im staring at the screen not certain what to do
04:24 PM lcvette_: thinking to myself
04:24 PM gloops: ive no idea with mesa and servos
04:24 PM lcvette_: all the times everyone said man Linuxcnc is the best its so powerful and tunable and configurable
04:24 PM lcvette_: and its free
04:25 PM lcvette_: what they don't tell you is how difficult it is to work with and that all that powerful tunability comes with a very steep learning curve
04:26 PM lcvette_: what amazes me is there isn't a single freakin video on how to do this
04:27 PM tjb1: and you have to learn Linux on top of it haha
04:27 PM andypugh: Have you found the halscope tutorial? http://linuxcnc.org/docs/html/hal/tutorial.html#sec:tutorial-halscope
04:27 PM gloops: lcvette you can make on when you work it out lol
04:27 PM andypugh: Personally I hate trying to learn how to do stuff from videos. I can understand that others have different opinions.
04:28 PM andypugh: Halscope does have an odd interface.
04:28 PM lcvette_: im reading it now
04:29 PM gloops: ive only done simple stepper configs and vfd, i blamed linuxcnc - but i had no experience of cnc at all - that was the real problem
04:29 PM lcvette_: but it was fasirly self explanateroy while i was trying to bring the graph lines into view just wasn't working
04:32 PM tjb1: Anyone experienced with Arduino want to take a look at this and tell me if there are any glaring issues? https://pastebin.com/i6n1EmNm
04:33 PM andypugh: I found a PID and halscope tutorial on Youtube. But it is in German: https://www.youtube.com/watch?v=0NqzwDKqwV8
04:33 PM lcvette_: lol
04:33 PM lcvette_: yeah i was just watching it to try and see if i could make out anything
04:33 PM lcvette_: oddly enough it made more sense than the language I was interpereting fromt he link you sent
04:34 PM pcw_mesa: Halscope assumes familiarity with an oscilloscope plus is a bit buggy/quirky
04:34 PM lcvette_: im not an oscope guy
04:35 PM lcvette_: i don't want to be an oscope guy
04:35 PM lcvette_: i don't want to be a programmer
04:35 PM lcvette_: lol
04:35 PM lcvette_: i want to be a designer and part manufacturer
04:36 PM lcvette_: and maybe one day will be able to get back to that
04:36 PM pcw_mesa: well if you are doing retrofits you will get dragged into machine design...
04:36 PM pcw_mesa: unless someone make a pre-setup config for a specific machine
04:37 PM lcvette_: yeah well, when the decision to go this route was made there was someone guiding me
04:37 PM lcvette_: actually pushing me
04:38 PM lcvette_: quite convincingly
04:38 PM lcvette_: who vanished
04:38 PM gloops: lol, who was that lcvette?
04:38 PM andypugh: I think it’s the right choice for any real machine.
04:38 PM lcvette_: unfortunately at that point the machine had already been shipped
04:39 PM pcw_mesa: if limit switch homing is good enough (and your not paranoid about losing steps) and tuning is this painful for you, just use open loop
04:39 PM lcvette_: had i known the headaches that were going to follow this decision i would have gone with a siemens controller from the manufacturer installed and ready to make parts on arrival
04:39 PM andypugh: tjb1: Your quadrature decoder seems unconventional.
04:39 PM tjb1: andypugh: Its all I found that didnt require an interrupt and was able to get working
04:40 PM lcvette_: lose steps?
04:41 PM lcvette_: will linuxcnc lose steps?
04:41 PM pcw_mesa: If you have a step/dir only machine its possible to lose/gain steps from electrical noise and you have no feedback to check
04:41 PM andypugh: No. Steppers can. (mechanically).
04:42 PM hazzy-lab: For the 3d folks on here, this super high res benchy was just spotted minutes ago by my brother in Ireland: https://i.imgur.com/FXuv2gj.png
04:42 PM andypugh: it has some fairly obvious layering.
04:42 PM hazzy-lab: lol
04:43 PM tjb1: multi material print, wooow
04:43 PM andypugh: I was amused by” https://www.thingiverse.com/thing:2866134
04:44 PM hazzy-lab: Oh, I'll have to print one of those!
04:44 PM lcvette_: well, i wasn't worried about losing steps
04:44 PM lcvette_: before
04:44 PM lcvette_: geeeez
04:44 PM lcvette_: never ending
04:45 PM pcw_mesa: Its very unlikely especially with decent drives and interfaces
04:45 PM lcvette_: not sure if my drives are decent
04:45 PM tjb1: andypugh: do you see any problems with the way im using float?
04:45 PM lcvette_: but i did install filters
04:46 PM lcvette_: on all the drives
04:46 PM Roguish: lcvette: if you're really worried about loosing steps, watch your wiring. shield like hell.
04:46 PM lcvette_: including the spindle drive
04:46 PM lcvette_: everything is shielded
04:46 PM lcvette_: control cabinet is very cleanly done
04:46 PM pcw_mesa: Or use differential signals if the driver accepts them
04:47 PM lcvette_: they are
04:47 PM Roguish: especially the encoder lines. differential signal TOO
04:47 PM lcvette_: 4 wire
04:47 PM Roguish: like PCW said....
04:47 PM lcvette_: TOO?
04:47 PM andypugh: tjb1: I think there is an encoder library. It probably does use interrupts, but that seems like the best way to do it anyway.
04:47 PM Roguish: ALSO
04:48 PM Roguish: in a loud voice.
04:48 PM lcvette_: the drives are wired diffrential
04:48 PM tjb1: andypugh: interrupts interfere with the frequency counter
04:48 PM lcvette_: 4 wires
04:48 PM lcvette_: right
04:48 PM andypugh: Hmm, OK.
04:48 PM lcvette_: step+ step- dire+ dir-
04:49 PM pcw_mesa: I would just go back to open loop for now at least to get up and running and then worry about encoder feedback if you need better homing or want to know what the drives are really doing
04:49 PM lcvette_: thats the thing its up and running now
04:50 PM lcvette_: just has a little bit of an abrupt shudder at higher rapids when stopping
04:50 PM lcvette_: when im at 708ipm
04:50 PM lcvette_: its fine under that
04:50 PM lcvette_: 600 and under its smooth as silk
04:51 PM gloops: you wanting more than 600 then lcvette?
04:51 PM lcvette_: but thats with the settings pcw said were not good
04:51 PM lcvette_: P-3000
04:52 PM pcw_mesa: its running in a very broken mode
04:52 PM lcvette_: right
04:52 PM Roguish: just because the speedo goes up to 180 doesn't mean ya have go that fast.
04:52 PM lcvette_: agreed, but like pcw stated it lets me know its not right
04:52 PM lcvette_: it shouldn't have a problem at that speed
04:52 PM Roguish: crawl, walk, then try running.....
04:52 PM lcvette_: its within the rated speed of the servos
04:53 PM pcw_mesa: p=3000 and pid_maxerr =.0005 means you have bang-bang PID
04:53 PM lcvette_: right
04:53 PM andypugh: Yikes.
04:53 PM lcvette_: so theres a better way
04:54 PM lcvette_: but im not capable of getting it there
04:54 PM andypugh: You got it here…
04:54 PM lcvette_: im standing on the shoulders of many
04:55 PM pcw_mesa: You mean you cant try different P values via "Calibrate"
04:55 PM lcvette_: whats "calibrate"
04:55 PM lcvette_: see
04:55 PM andypugh: It’s a menu item in the Axis GUI
04:55 PM lcvette_: im using gmoccapy
04:55 PM pcw_mesa: and a way to change tuning values with restarting LinuxCNC
04:56 PM pcw_mesa: gmoccapy also has calibrate AFAIK
04:56 PM pcw_mesa: or you can just set the P value from the commnd line
04:56 PM lcvette_: ah found it
04:56 PM lcvette_: yes i can change the p value
04:57 PM Roguish: lcvette: whether you like it or not you're going to have to learn a bit of Linuxcnc, and that includes controls. not all that difficult. RTFM and then ask questions here an in the forum.
04:57 PM lcvette_: I have read alot of the manual
04:57 PM lcvette_: but alot of it doesn't click for me
04:57 PM lcvette_: some does
04:57 PM lcvette_: alot doesn't
04:57 PM * JT-Shop wonders why a sim for gommcappy locks you out of the settings?
04:58 PM Roguish: repetition helps.
04:58 PM pcw_mesa: you have to hit "test" in calibrate for the value to take effect
04:59 PM andypugh: I wonder how big naxerr needs to be? I think I have always ised zero (no limit)
04:59 PM lcvette_: i just tested it at 1000
04:59 PM Roguish: it'll cure your insomnia
04:59 PM lcvette_: seemed ok
05:00 PM lcvette_: my hal file is back at.0005 though
05:00 PM pcw_mesa: yeah that has to go
05:00 PM andypugh: I would try a P of maybe 1 and zero maxerror and move up from there, I think.
05:01 PM pcw_mesa: start with P low or you will get violent oscillations!
05:01 PM andypugh: And make sure that f-error is larger than normal, but not enormous.
05:01 PM lcvette_: ok changed hal back to 0
05:01 PM andypugh: What are I and D?
05:01 PM pcw_mesa: 0 to start
05:02 PM lcvette_: ok i tried 100 for P
05:02 PM lcvette_: with hal seemed to do ok
05:02 PM lcvette_: jogging meaning no faults
05:02 PM lcvette_: not using halscope
05:03 PM lcvette_: 20 faults
05:03 PM andypugh: You don’t have to use Halscope.
05:03 PM pcw_mesa: keep bumping up P until it oscillates
05:04 PM lcvette_: increment?
05:04 PM lcvette_: 1, 5, 10?
05:04 PM andypugh: I don’t recall if I have used it to tune my machines. But then I don’t think my machines are all that well tuned. But they do what I want, so I stopped tinkering.
05:04 PM pcw_mesa: if 100 doesnt oscillate, start there
05:05 PM lcvette_: 100 seems to be ok
05:05 PM lcvette_: not getting a fault at max jog velocity
05:06 PM lcvette_: not sure what oscillate means
05:06 PM lcvette_: mean at idle state?
05:06 PM pcw_mesa: but you may get better performance if its higher and you wont know how high you can go until get there
05:06 PM andypugh: If you want to use Science, perhaps try E6 like resistors: 100, 150, 220, 330, 470, 680, 1000, 1500 and so on
05:07 PM pcw_mesa: Yeah that's a good sequence
05:07 PM andypugh: Then when it oscillates, back off downwards
05:07 PM lcvette_: oscillates?
05:08 PM lcvette_: help me identify what im looking for
05:08 PM pcw_mesa: You will know when it oscillates...
05:08 PM andypugh: Yeah, shaking violently back and forth.
05:08 PM pcw_mesa: running for the estop
05:08 PM andypugh: The F-error limits should stop it before it gets out of hand.
05:09 PM lcvette_: i put those back to where they were is that ok?
05:09 PM lcvette_: .1 and .01
05:09 PM pcw_mesa: not for tuning, you are still tuinig
05:10 PM lcvette_: ok better change them then
05:10 PM pcw_mesa: You may be ok now if its that close
05:11 PM lcvette_: ok how about 1 and .1 then
05:12 PM pcw_mesa: That is, if the Ferror settings don't give you a following error you can bump up P untill it get s unstable and you will know by the following error...
05:12 PM andypugh: Are you metric or imperial?
05:12 PM lcvette_: metric
05:12 PM pcw_mesa: did P of 100 fix the high speed issues?
05:13 PM andypugh: OK, 1 and .1 seems quite small. 1 inch would be too big.
05:13 PM lcvette_: seems to clean it up
05:13 PM andypugh: You can probably go much faster then.
05:14 PM andypugh: P I D of 100, 0, 0 is rather unlikely to be the best you can possbily have.
05:14 PM lcvette_: so 10 and 1?
05:14 PM andypugh: Depends how big the machine is so an extent. 10 and 1 should be fine
05:15 PM pcw_mesa: yeah bang-bang PID is ok for thermostats not so much for servos (unless the bang-bangs are at a MHZ or so)
05:15 PM andypugh: Enough room to wobble in, but not enough to get out of hand and hit the end stops.
05:16 PM pcw_mesa: with maxerror 0f .0005 and P=3000 the only likely velocity corrections are +- 1.5mm/sec
05:17 PM pcw_mesa: (unless you are within .5 u of the commanded position which is unlikely)
05:17 PM rmu: sorry to interrupt, but i have a stupid question. given a well-tuned servo drive that gets step/dir signals from linuxcnc, IMHO there should never be any appreciable difference between feedback and command (from linuxcnc), right?
05:18 PM pcw_mesa: with a perfect drive sure but none are
05:18 PM lcvette_: ok, question
05:19 PM lcvette_: in the Y axis it seems like the jog goes longer then it should
05:19 PM lcvette_: as in it goes further then im hitting the button
05:19 PM lcvette_: is that normal?
05:19 PM pcw_mesa: it can depending on jog mode
05:20 PM lcvette_: and yes i now know what an ascillation is
05:20 PM andypugh: It means that the commanded position is getting ahead of the actual position. but thai might be between the motion module and the step-generator
05:20 PM lcvette_: i went to jog the x axis while still at 3000 p and it oscillated
05:21 PM Tom_itx: you sure?
05:21 PM * Tom_itx smiles
05:23 PM rmu: what good is a servo drive in position mode if it can't track position
05:24 PM rmu: nevermind
05:26 PM pcw_mesa: It may follow local feedback correctly by you may want to say add slow correction feedback from a linear scale
05:26 PM lcvette_: im at 1000 now
05:26 PM lcvette_: still no oscillations
05:26 PM lcvette_: soooo
05:26 PM lcvette_: thought big numbers were bad?
05:27 PM andypugh: rmu: If it can’t do it at all, not a lot. But some drives have quite an unsophisticated internal controller (with the advantage of being very close to the hardware) and LinuxCNC might be able to squeeze more out of it by being closer to the motion planner (velocity feed-forward, for example).
05:27 PM andypugh: No, numbers that are _too_ big are bad.
05:28 PM andypugh: I have controllers at work that are unstable at P = 0.1. And some like P = 10000. Amongst other things it depends on the units of the feedback and output.
05:29 PM lcvette_: ok
05:29 PM Tom_itx: every loop is different
05:29 PM lcvette_: mdoes it matter what my drives are using for the steps per revolution?
05:30 PM lcvette_: 6000 steps per revolution
05:31 PM pcw_mesa: steps/encoder counts/rev should not make any difference as long as the step scale and encoder scale are right (so the PID works on mm only)
05:32 PM Roguish: lcvette_ : what is the distance per step?
05:32 PM lcvette_: .001mm
05:32 PM Roguish: nice.
05:33 PM lcvette_: ok
05:33 PM lcvette_: that was what the drive manual example had
05:33 PM lcvette_: for my 6mm screws
05:33 PM lcvette_: crap
05:33 PM lcvette_: i am stuck on the limit switch
05:33 PM lcvette_: cant get linux to start
05:34 PM Tom_itx: axis has an override checkbox, not sure what gmoccapy has
05:34 PM lcvette_: it does, but its grayed out
05:34 PM lcvette_: it has worked before
05:34 PM lcvette_: not sure whats going on
05:37 PM andypugh: Simple solution might be to turn the screw by hand to back off the limit.
05:38 PM Roguish: jumper on the switch
05:38 PM lcvette_: fixed it
05:38 PM andypugh: (I can do this on my lathe by pulling back the spiral screw cover on Z but it’s not an option on X)
05:38 PM lcvette_: dorgot i had an old test config that didn't have limits setup
05:41 PM lcvette_: 2000 no oscillations
05:53 PM andypugh: Did he get oscillated off the IRC?
05:55 PM lcvette_: no still here
05:55 PM lcvette_: that was my phone
05:55 PM lcvette_: think it died
05:55 PM lcvette_: i was checking a drive setting
05:55 PM lcvette_: my Y axis was seeming to be a bit of a different animal than x and z
05:58 PM Roguish: yes it will be. different mass (weight)
05:59 PM Roguish: different damping and friction also.
06:02 PM lcvette_: oscillations start at 2700
06:02 PM lcvette_: on x axis
06:05 PM andypugh: You probably want to back off a bit from there, but you can try seeing if adding some D-gain lets you go higher. Start very small, 0.01 for example.
06:06 PM lcvette_: nope
06:06 PM lcvette_: didnt like that
06:07 PM lcvette_: that was scary
06:08 PM lcvette_: oscillated for a good 3-4 seconds after stopping
06:09 PM lcvette_: was waiting for it to decide to go somewhere
06:09 PM lcvette_: lol
06:12 PM andypugh: Try 0.1 then, or 0.001...
06:12 PM andypugh: (Or just settle on 2500 and see how that works with G-code)
06:12 PM andypugh: What is your steady-state error?
06:13 PM andypugh: (difference between commanded and actual when the axis stops moving)
06:14 PM lcvette_: im in halscope
06:14 PM lcvette_: ok
06:14 PM lcvette_: what do you want?
06:15 PM andypugh: F-error is a good place to start
06:19 PM lcvette_: ok
06:19 PM lcvette_: its all over the plce
06:20 PM andypugh: It will be.
06:21 PM andypugh: How big does it get, and what does it settle at when you stop moving?
06:21 PM lcvette_: f (0.482000) = -0.02289 (ddt -1.51166)
06:22 PM lcvette_: thats the lowest negative spike
06:22 PM lcvette_: it never really settles
06:23 PM lcvette_: im unfamiliar with linux, how do i print screen
06:23 PM lcvette_: or screen capture an image?
06:24 PM andypugh: I think you can use the print-screen button.
06:24 PM andypugh: But you can also save the halscope data file, from the halscope menu
06:24 PM andypugh: Which channels do you have in the scope?
06:25 PM JT-Shop-2: menu accessories take screen shot in most distros
06:25 PM andypugh: Yes, the LinuxCNC ISO I have in my VM has Applications->Accessories->Screenshot
06:26 PM lcvette_: http://i.imgur.com/8JUjhh2.png
06:26 PM lcvette_: that easier?
06:27 PM pcw_mesa: looks like P is too high
06:27 PM lcvette_: ok
06:27 PM lcvette_: so lower P until it looks like JT's?
06:28 PM andypugh: 5m per division, that’s 5 steps peak error...
06:28 PM pcw_mesa: till it looks calmer
06:28 PM pcw_mesa: 5u per division
06:28 PM andypugh: Sorry, 25 steps (25 microns)
06:30 PM lcvette_: im not sure what you guys are now discussing
06:30 PM andypugh: That’s 0.001” for the mediaeval folks.
06:30 PM lcvette_: do i need to make a change?
06:30 PM pcw_mesa: the scale of the error
06:30 PM andypugh: You have some ringing, but the absolute position error is pretty small.
06:31 PM pcw_mesa: I would lower the P gain so the ringing is reduced
06:31 PM lcvette_: is ringing a number?
06:31 PM pcw_mesa: its on the edge of oscillation
06:31 PM lcvette_: in the halscope?
06:31 PM andypugh: The steady-state error is astonishngly small. A few nm.
06:31 PM lcvette_: a target i can shoot to hit?
06:32 PM pcw_mesa: just set the P to 1/2 what is is now and replot
06:32 PM lcvette_: yes thast was just under the oscillation P value
06:32 PM lcvette_: it oscillated at 2700 that was set at 2600
06:32 PM pcw_mesa: 1/2 should be a lot better
06:32 PM pcw_mesa: or 1000
06:35 PM lcvette_: http://i.imgur.com/FXPj4lo.png
06:35 PM lcvette_: thats at p-1300
06:35 PM lcvette_: i set the curser on the highest peak before screenshot capture
06:36 PM andypugh: Can you zoom out a little, so we can see the steady-state error?
06:36 PM pcw_mesa: ok ~10 times better. since the baseline error is still very small you could probably go even lower (say 500)
06:36 PM lcvette_: ok i'll try 500
06:37 PM pcw_mesa: that will increase the baseline error but reduce the oscillations
06:37 PM andypugh: Might be instructive to to stay at 1000 and add a little bit of D-gain to see what difference it makes.
06:39 PM lcvette_: http://i.imgur.com/VcZXRQs.png
06:39 PM lcvette_: 500 p
06:40 PM andypugh: That has increased steady-state error but not really improved stability.
06:40 PM andypugh: So perhaps try going back to 1000 and adding a little D.
06:43 PM pcw_mesa: Ha! You wont suffer from lack of advice... I would use 500 and add some I :-)
06:44 PM andypugh: I was going to suggest adding I next, I was only suggesting D to see what difference it makes to the wiggles.
06:45 PM lcvette_: http://i.imgur.com/wA9D8f0.png
06:45 PM lcvette_: this is p-1000 no d
06:45 PM lcvette_: the other one was p-1300 i never did a p-1000 before this one
06:46 PM lcvette_: ok i or d?
06:46 PM pcw_mesa: since your peak error is only~ 2 u now you could probably call it good
06:48 PM pcw_mesa: I would get rid of the static error a tiny bit of D might calm the wiggles
06:48 PM Roguish: lcvette: hey, you're doing good. hang in there. none of us learned this stuff overnight !!!!
06:49 PM MrHindsight: too make a long story short, did you put the servo in velocity mode and then tune it pretty quickly?
06:49 PM lcvette_: lol thanks i have great help
06:49 PM lcvette_: tiny bit meaning.01?
06:49 PM pcw_mesa: actually there really is no static error, the error other than little peaks is all during accel
06:50 PM pcw_mesa: so could be fixed with a tiny bit of FF2
06:50 PM andypugh: Indeed. http://i.imgur.com/VcZXRQs.png seems to show steady-state error but that’s just the time-zoom not showing any static position.
06:51 PM lcvette_: MrHindsight they are position only drives
06:51 PM lcvette_: tiny bit
06:51 PM pcw_mesa: thats a technical term...
06:51 PM lcvette_: pcw_mesa: im not quite current on your scale system just yet
06:51 PM lcvette_: is tiny bit .01?
06:52 PM andypugh: A tiny bit is about twice a smidgeon.
06:52 PM pcw_mesa: for D I would try .001
06:52 PM andypugh: 0.001, then if you see no change, 0.01, and so on.
06:53 PM lcvette_: http://i.imgur.com/eZXmk3t.png
06:53 PM lcvette_: :)
06:53 PM lcvette_: thought we were gonna do ff2
06:54 PM MrHindsight: reads the 6 hours of backlog
06:54 PM pcw_mesa: for FF2 maybe .0001 I should be able to calculate it
06:56 PM lcvette_: http://i.imgur.com/g8exTJM.png
06:56 PM lcvette_: ff2 .0001
06:56 PM MrHindsight: <pcw_mesa> We are treating your servos as if they are running in velocity mode
06:57 PM pcw_mesa: .00014
06:58 PM lcvette_: http://i.imgur.com/FRwrZkT.png
06:58 PM lcvette_: .00014
07:00 PM lcvette_: MrHindsight: don't let the manual fool you, if you look on page 9 i think at the model number it will tell you that the drive ending in AA is position only
07:00 PM lcvette_: those are mine
07:00 PM lcvette_: even though it speaks of being able to set it in analog mode
07:00 PM MrHindsight: yeah, thats the drive itself
07:01 PM lcvette_: believe me we tried to run in analog mode
07:02 PM lcvette_: then called peter to get 7i76's
07:02 PM lcvette_: lol
07:03 PM lcvette_: I? D?
07:04 PM andypugh: You can probably stop. Your error is less than 1 step.
07:04 PM andypugh: (Peak error is 2 steps, but steady state is less)
07:05 PM andypugh: Try some D to see what it does, if you want to see
07:12 PM lcvette_: http://i.imgur.com/EWiJvJO.png
07:12 PM lcvette_: D.001
07:16 PM lcvette_: think it was better with no D
07:23 PM andypugh: I am not sure it has made much difference, I would keep increasing it until it is definitely better or worse.
07:25 PM pcw_mesa: This is so good I have a sneaking suspicion that the drive is returning the step/dir position counter not the real encoder...
07:27 PM andypugh: Maybe worth jogging up to a dial indicatior?
07:28 PM pcw_mesa: and that the spikey noise is sampling jitter on the encoder (i suspect the encoder is not using the DPLL but it could)
07:29 PM lcvette_: ?
07:29 PM lcvette_: oh man
07:29 PM lcvette_: now what
07:29 PM lcvette_: more oddball issues for me?
07:30 PM pcw_mesa: can you post your hal file again?
07:30 PM lcvette_: i was already on to my Y axis
07:30 PM lcvette_: sure
07:31 PM lcvette_: Y axis
07:31 PM lcvette_: http://i.imgur.com/naiCb0Z.png
07:31 PM lcvette_: before i lose the paste
07:32 PM lcvette_: https://pastebin.com/0wb43KFZ
07:32 PM lcvette_: Hal
07:36 PM pcw_mesa: These are not really oddball issues (though invisible unless you have feedback)
07:36 PM lcvette_: ok
07:36 PM lcvette_: is it a drive setting i need to change?
07:37 PM lcvette_: i just tried changing the D to .2
07:37 PM lcvette_: it didn't make alot of difference
07:38 PM andypugh: You have quite large P so the D is likely to be larger than we are used to
07:39 PM lcvette_: D1?
07:39 PM lcvette_: P is at 1100 on Y
07:39 PM lcvette_: seemed happiest there on the halscope
07:39 PM lcvette_: below started drifting away at start and stop from the line
07:40 PM lcvette_: and above started getting too spiky
07:40 PM lcvette_: thats the balance im looking for right?
07:40 PM pcw_mesa: I would add
07:40 PM pcw_mesa: setp hm2_7i92.0.dpll.01.timer-us -100
07:40 PM pcw_mesa: setp hm2_7i92.0.stepgen.timer-number 1
07:40 PM pcw_mesa: setp hm2_7i92.0.encoder.timer-number 1
07:40 PM pcw_mesa: to your hal file
07:40 PM lcvette_: ok
07:40 PM lcvette_: anywhere?
07:40 PM pcw_mesa: anywhere
07:45 PM lcvette_: now things look totally different
07:46 PM lcvette_: http://i.imgur.com/301N9l6.png
07:47 PM pcw_mesa: yeah you will need to retune FF2
07:48 PM pcw_mesa: I think add .0001 (100 usecs)
07:49 PM lcvette_: http://i.imgur.com/kwJlSPv.png
07:49 PM lcvette_: .00035 ff2
07:50 PM pcw_mesa: ahh I go it backwards
07:50 PM pcw_mesa: no thats correct
07:51 PM lcvette_: coulkd maybe stand .00036
07:52 PM lcvette_: http://i.imgur.com/Vc06SzU.png
07:52 PM lcvette_: .00037
07:52 PM pcw_mesa: I think we've arrived to the land of endless futzing...
07:53 PM lcvette_: its tuned?
07:54 PM pcw_mesa: Yeah I'd say though I still suspect the drive is returning its position counter, not its rotor position
07:55 PM pcw_mesa: since it looks so much better :-)
07:55 PM lcvette_: i think its outputting the gear ratio
07:55 PM lcvette_: no?
07:56 PM lcvette_: not a raw signal
07:56 PM pcw_mesa: right but is that coming from its pulse counter or encoder?
07:57 PM pcw_mesa: I guess you could disable the drive and see if the DRO still reads correctly if you manually rotate the motor shaft
07:58 PM pcw_mesa: (it should)
07:59 PM pcw_mesa: bbl Dinner time!
07:59 PM pcw_mesa: at least it should be running well now and home to index should workl
08:01 PM lcvette_: sweet
08:01 PM lcvette_: i looked through the manual and i see it shows the encoder output going out
08:02 PM pcw_mesa: you cant hide your lying drives...
08:02 PM lcvette_: that didn't make sense
08:02 PM lcvette_: lol
08:02 PM lcvette_: the picture shows the encoder line coming from the motor intot he drive but also continuing through to the output to the control
08:06 PM lcvette_: you've got mail, sent over the drive manual *if* you got curious about it
08:21 PM ziper: the isel went for 210 plus fees
08:24 PM Tom_itx: lcvette_ gettin a workout tonight?
08:25 PM Tom_itx is now known as Tom_L
08:27 PM lcvette_: just finished tuning the last axis
08:27 PM Tom_L: z will probably be different since it's got weight on it
08:27 PM Tom_L: does that machine have a counterweight?
08:30 PM Tom_L: i had one on my sherline but the new mill doesn't need one
08:33 PM lcvette_: yes its counterweighted
08:34 PM lcvette_: and has a different servo then x and y
08:34 PM lcvette_: and different rated speed
08:34 PM Tom_L: it should
08:34 PM lcvette_: it tuned easier then x and y surprisingly
08:34 PM lcvette_: THANKYOU to Peter and ANDY for all your help tonight guys!
08:35 PM lcvette_: I seriously seriously appreciate it
09:13 PM Roguish: lcvette: ya having fun yet?
10:03 PM Isleape1 is now known as Isleape
10:12 PM Lcvette: Roguish: I am now!!!
10:12 PM Lcvette: Tomorrow I'm going to try turning on index homing
10:13 PM Lcvette: :o
10:22 PM hazzy-lab: Lcvette: Nice! homing to index is satisfying
10:23 PM hazzy-lab: My RF45 has encoders but no indexes, lame :(
10:25 PM hazzy-lab: The BP homes to indexes, and since one of the limit switches does no work I have to trip it by hand, and if I trip it just right it homes to the *right* index :D
10:28 PM Roguish: hazzy-lab: hey, just got a new monitor.. old one was crapping out occasionally. 32", 4K.
10:29 PM * hazzy-lab is jelly
10:30 PM Roguish: I pay a lot of bills with a 'cash back credit card' so I used the 'cash back'
10:30 PM Roguish: and watched sales for a long time.
10:31 PM hazzy-lab: Very nice! That is a good way to use the rewards
10:35 PM roycroft: i have a citi double cash back card, which gives me 2% back on everything, and a discover card that gives me 5% rewards on various categories, by quarter
10:35 PM roycroft: i also use merchant-specific cards such as lowes and target that give me a 5% discount at point of sale
10:36 PM roycroft: i don't believe in paying credit card interest, but i use credit cards all the time and get a lot of rewards
10:36 PM Roguish: best is paying my $1500 / month health insurance with the 2% CC
10:37 PM roycroft: i pay all my utilities with the citi double cash back card
10:37 PM Roguish: there ya go.
10:37 PM roycroft: this quarter, discover are doing 5% back on groceries, so i'm using my discover card at the market 100% of the time
10:37 PM roycroft: next quarter it will be something different, and i'd only get 1% on groceries, so i'll go back to using the citi card for that
10:38 PM Lcvette: hazzy-lab: sweet! I haven't tried it yet. I'm guessing I need to change some things in my ini file?
10:39 PM roycroft: i think it's funny that some folks are so down on credit cards
10:39 PM roycroft: with a little self-discipline they can be very adventageous
10:39 PM roycroft: carrying a large balance and paying double-digit interest definitely sucks
10:40 PM roycroft: but all i have to say about that is: don't do that
10:45 PM Lcvette: hazzy-lab: not sure if I told you but on our earlier discussion of the ATC pockets, the tool numbers are stored in the linuxcnc.var file
10:46 PM Lcvette: Not sure if it's hard to pull information from there?
10:48 PM hazzy-lab: Lcvette: To home to index all you have to do is add HOME_USE_INDEX = 1 for each joint in the INI
10:48 PM hazzy-lab: It should be easy to read info from the var file, one way or another
10:49 PM hazzy-lab: dID YOU GET THE SERVOS TUNED?
10:49 PM Lcvette: Yup
10:49 PM Lcvette: Tuned
10:49 PM hazzy-lab: opps, sorry
10:49 PM hazzy-lab: Hurray!
10:49 PM Lcvette: Yay
10:49 PM Lcvette: Lol
11:05 PM Lcvette: hazzy-lab: HOME_USE_INDEX = no
11:05 PM Lcvette: I currently have that there now
11:05 PM Lcvette: So just change those to 1
11:05 PM Lcvette: HOME_USE_INDEX = 1
11:09 PM aircooled is now known as hazzy-lab