#linuxcnc Logs

Aug 14 2017

#linuxcnc Calendar

01:05 AM rodw: anybody here been playing with Dewey's external offsets?
01:08 AM LeelooMinai_ is now known as LeelooMinai
02:10 AM IchGucksLive: morning from germany
02:10 AM XXCoder: hey
02:11 AM IchGucksLive: long starshower night
02:11 AM IchGucksLive: short sleep
02:11 AM IchGucksLive: heat warning for today
02:11 AM IchGucksLive: rediciulous after a cold night
02:11 AM XXCoder: here finally broke the cap on smokey air
02:12 AM XXCoder: so air is now clear and cooler :)
02:21 AM IchGucksLive: clean air is getting more and more a problem on citys
02:21 AM XXCoder: it wasnt that here
02:22 AM IchGucksLive: im off till later
02:22 AM XXCoder: it was cananda fire smoke traveling south then getting trapped here.
02:22 AM IchGucksLive: wild fire is a bigger problem as people like to burn
02:23 AM Vq: Morning from Sweden.
02:23 AM IchGucksLive: there shoudt be jail on knowing fire starters
02:23 AM IchGucksLive: hi VQ
02:23 AM XXCoder: there is. I think canada fire was natural fire though
02:23 AM XXCoder: caused by nature
02:23 AM IchGucksLive: vq did you got a eye on starshower up there
02:24 AM IchGucksLive: need to mve
02:24 AM IchGucksLive: move
02:25 AM XXCoder: move because canada forest had a fire, and are over 500 miles away? sure why not :P
02:26 AM Deejay: moin
02:26 AM XXCoder: hey
02:35 AM -!- #linuxcnc mode set to +v by ChanServ
02:48 AM Rod_W: HI guys, rod from Australia here
02:48 AM XXCoder: hey
02:49 AM Rod_W: I've been building a plasma cuter and can't get External Offsets to work. I think I might be the first to put this on real hardware any ideas?
02:50 AM XXCoder: maybe someone knows though this time is very slow for this channel.
02:51 AM Rod_W: Yeh I know, our time zone is never good..... Dewey Garrett is probably the man I need ..., he wrote it
02:54 AM XXCoder: yeah
02:54 AM XXCoder: 61% humidity 66f lol wonder if they will meet :P
03:01 AM Rod_W: 33% humidity and 22c here (and my wunderground PWS is down.. grr) 5:30 pm and the beer is cold :)
03:03 AM XXCoder: 19c here
03:03 AM XXCoder: my room was orginially 79f or approx 22c?
03:06 AM Rod_W: So where's here?
03:06 AM XXCoder: wa state, usa
03:08 AM Rod_W: Would be past my bedtime if I was there...
03:09 AM XXCoder: 12:38 am heh
03:09 AM XXCoder: I usually sleep at 4 am
03:10 AM Rod_W: I'm waking up then! I find the US has the most unfriendliest timezone for an Aussie
03:10 AM Rod_W: We get about 2 hours to sort out stuff with suppliers and thats with me stumbling bleary eyed to my home office at 6:00am
03:12 AM Rod_W: Anyway, back to Plasma THC and Dewey's eoffset_pid component. I'll post now in case someone scrolls back while i'm having dinner
03:13 AM XXCoder: yeah
03:14 AM Rod_W: This 15 second video of Hal signals https://youtu.be/Ed4en8a6qAc shows that the zo.active never becomes active.
03:15 AM Rod_W: Various inputs enable (a,b,c and then the zo.enable-out signal goes true
03:18 AM Rod_W: and then zo.enable-out is wired to axis.z.eoffset-enable
03:19 AM Rod_W: and then motion.eoffset-active is wired to zo.active but it never comes on.....
03:20 AM Rod_W: The hal is set up the same as Dewey does in his plasma sim.
03:20 AM archivist: does his sim work
03:21 AM Rod_W: If you watch the Z DRO at the top left and listen, you'll see the torch go up to pierce height of 4mm, then drop to 1.8mm and start cutting.
03:22 AM archivist: I cannot see yt on this box
03:23 AM Rod_W: Yeh, his sim is perfect, I'm at a bit of a loss to see what is wrong. I've poked and prodded in the settings and can't see anything stopping it from enabling.
03:24 AM Rod_W: Video is just of halshow and the component signals so not very exciting. Everything happens so fast, I needed to film it for trouble shooting.
03:24 AM archivist: have you forgotten the top of the hal file where stuff is placed in one of the loops
03:25 AM archivist: if you do forget it just wont do anything :)
03:26 AM gloops: stepconfig wizard - leadscrew pitch, is this simply the distance travelled in one revolution ? im using chain
03:27 AM Rod_W: I think I got it all "loadrt eoffset_pid names=zo" and "addf zo servo-thread 1"
03:28 AM archivist: may be another addf you forgot
03:29 AM Rod_W: maybe, I have been caught before...
03:29 AM archivist: gloops, basically yes
03:32 AM Rod_W: gloops, if you can do the maths manually in your machine units, you can look in the ini file after stepconf finishes, to check it is correct.
03:32 AM archivist: measure the machine movement to verify
03:33 AM archivist: which can be entertaining on some machine types
03:34 AM Rod_W: gloops, and debug from there. I had a non standard rack and pinion adn I measured against a metre ruler until it was right. Then tested in relative mode for 20 moves back and forth with a dial indicator on the strating position.
03:34 AM archivist: my hobbing machine was not funny as measuring angle over many turns
03:35 AM Rod_W: archivist, I have 4 components that feed data in to the thcad which are in the forum plasma section archivist, I have 4 components that feed data in to the thcad which are in the forum plasma section all are loaded ock and addf are there for all; loadrt showzdir names=zdir
03:35 AM Rod_W: loadrt scalethcad names=thcad10
03:35 AM Rod_W: loadrt cornerlock names=c-lock
03:35 AM Rod_W: loadrt kerfcross names=k-cross
03:35 AM Rod_W: loadrt ghold names=ghold
03:36 AM Rod_W: I can imagine. I wrote a Rotary table controller for an Arduino and 4 years after I wrote it, somebody found a bug.
03:36 AM archivist: loadrt is only enabling that hal item, it does not add to a thread
03:36 AM Rod_W: yeh, but i also have addf zdir servo-thread
03:36 AM Rod_W: addf thcad10 servo-thread
03:36 AM Rod_W: addf c-lock servo-thread
03:36 AM Rod_W: addf k-cross servo-thread
03:37 AM archivist: and order can matter
03:37 AM Rod_W: All of that code (except kerf-cross) has been tested extensively without the torch on.
03:39 AM Rod_W: dewey adds a parameter to his component to load it first. addf zo servo-thread 1
03:39 AM gloops: ok thanks, i can calculate it, probably end up with a mixture of caculation and measuring yes
03:40 AM Rod_W: gloops: I never used stepconf (or pcnconf in my case as I use Mesa) I just did everything in the files. You might as well start now as you will need to edit by habd eventually :)
03:40 AM gloops: looking at it after a bit of testing my max speed is going to be about 70 inch/minute heh
03:40 AM Rod_W: *hbnd = hand
03:41 AM gloops: yes Rod i will be looking at that, first ever build, first ever motion i just wanted to play quickly
03:41 AM gloops: handy feature for the test axis
03:41 AM archivist: the wizards or ak for basic get you going, then edit
03:42 AM Rod_W: gloops: playing quickly! I staerted 13 months ago and only just turned on my torch!. Should never have done plasma as a first project
03:43 AM gloops: i have built far too big and far too heavy i think, 4x2 router with box section, it is working though so far so good, the next one will be better
03:43 AM Rod_W: gloops: I still remember how excited I was to see the very first glimmer of movement on a miscalibrated stepper motor. Ten I used a cheap rev counter to work out what the maximum unloaded speed could be. Was very useful
03:44 AM gloops: 6 week, i made everything bearing housings bosses for sprockets the lot, used bike chain and sprockets lol, not seeing any backlash yet - not bad anyway, but not got router on
03:44 AM archivist: then you get stalls, never use max speed on a stepper :)
03:45 AM gloops: its a lot to take in quickly, i guess ill probably know enough to make a machine and run it next year
03:45 AM archivist: I lurked in here a year or more before starting my build
03:45 AM gloops: ive been watching with interest for a while yeah
03:46 AM Rod_W: yeh, I backed off a bit and top speed is 21 metres a minute with 18 m/minute for rapids
03:46 AM gloops: a lot of it doesnt mean anythihng until you actually have a machine working
03:46 AM gloops: 21 metres a minute haha, servo motors?
03:47 AM Rod_W: Nah. Mesa with stepper and rack and pinion. Mesa pushes them to about 700 rpm with 20x stepping and i could only ever get about 400 rpm out of my arduino code
03:47 AM gloops: im dissapointed with speed, unless theres something im not configging right, 425 oz motors not pulling so fast at all
03:48 AM gloops: whats mesa?
03:48 AM Rod_W: This stuff http://mesaus.com/ I am using a 7i76e ethernet board
03:49 AM archivist: gloops, larger motors mean lower speed, also have you got sufficient power supply voltage
03:49 AM gloops: bit over my head at the momemnt
03:50 AM Rod_W: The board has its own stepgens and generates the step frequency LCNC tells it to and it does that until LCNC changes the speed. Can do up to 10 Mhz
03:50 AM gloops: im putting 36v to them, just ordered a 60v pack
03:50 AM gloops: about 80mm per revolution, should be flying
03:51 AM Rod_W: Too low. I run 48v. I started with small Nema 23 and a 19v laptop supply on my rotary table and it missed steps. BOught a new stepper ant a 48 v power supply but never needed to change the stepper.
03:52 AM gloops: well the 36v supply is actually reading about 45v with multimeter, so dont know whats happening there, its ok..ish anyway if i can cut 60 inch/min thats about typical for diy machines i suppose
03:52 AM Rod_W: what type of machine?
03:53 AM gloops: its a router
03:53 AM gloops: i wish id built smaller tbh for first go
03:53 AM archivist: too small was my error :)
03:54 AM gloops: haha thats what i thought, with anything you usually wish youd built bigger
03:54 AM Rod_W: I think you will need to gear it back a bit. Should be 18-30mm per rev. Search "Do you really need a spring loaded pinion drive" on the forum and it will tell you about my mistakes....
03:55 AM gloops: ok will do, already thinking about the gearing yeah, everyone says gear it down with chain , probably would have got away with 1-1
03:55 AM Rod_W: https://forum.linuxcnc.org/30-cnc-machines/31509-do-you-really-need-a-spring-loaded-pinion-drive?limitstart=0
03:56 AM gloops: anyway got stuff to do bbl, will have a look
03:57 AM Rod_W: Thats my point. I had 1:1 with 93mm per rev and the guys made me redo it...
04:35 AM -!- #linuxcnc mode set to +v by ChanServ
05:53 AM jthornton: morning
05:54 AM XXCoder: hey
05:56 AM Deejay: yo
05:56 AM jthornton: flew the tiny whoop a few times last night with fpv
05:56 AM Vq: afternoon
05:57 AM XXCoder: how was it
06:00 AM jthornton: the trim was way off for some reason once I got that sorta close I was able to fly it around the shop bumping into things from time to time lol
06:05 AM XXCoder: fun
06:26 AM Rod_W: hey JohnT, did you ever get Dewey's external offset code going on your plasma table?
06:27 AM jthornton: no, the pc is dead and I need to get it back alive and try to sort it out
06:27 AM jthornton: I think I have a working configuration but time and other things like coyotes have my attention at the moment
06:31 AM Rod_W: John, thanks, I've got my table working but the external offsets are not enabling
06:31 AM Rod_W: I know I'm got at stuffing things up but usually I work it out myself. but this time I'm stumped.
06:32 AM Rod_W: *I'm good at stuffing things up
06:33 AM jthornton: it's been a few weeks since I looked at it but IIRC you have to enable 2 things to make it work
06:34 AM Rod_W: you men the enable-in-a, b &c?
06:34 AM jthornton: yea
06:35 AM Rod_W: Yeh, I got that far and the enable-out signal goes true but offsets don't enable.
06:35 AM Rod_W: if youhave got a spare 15 seconds, could you watch this? https://youtu.be/Ed4en8a6qAc
06:35 AM archivist: trace the signal itself
06:36 AM -!- #linuxcnc mode set to +v by ChanServ
06:37 AM Rod_W: archivist, I've done that in hal show. The enable out signal goes straight to axis.z.eoffset-enable
06:39 AM Rod_W: So then I think what should happen is the external offsets are turned on and motion.eoffset-active should go true
06:40 AM Rod_W: Which then should come back into the component as zo.active in the video. But its not happening.
06:40 AM archivist: did he make it so an actual offset as well not just enables means active
06:40 AM Rod_W: I wondered if real hardware needed another enabling signal but I could not see any reference.
06:42 AM Rod_W: he's built in a pid element with a command and feedback signal to create the offset under PID control. They both exist and the feedback is coming in correctly
06:44 AM Rod_W: And n seconds into the video, you see the commanded volts is 80 volts and the actual feedback is 82 volts
06:44 AM archivist: I am not able to see vids
06:46 AM Rod_W: Never mind, Maybe John can. The only thing I can see that is different is that external offsets for the Z axis are not enabling and I don't know why.
06:47 AM archivist: pastebin entire hal so people can see the connections
06:56 AM Rod_W: Sorry, not sure how to do that. can you explain
06:56 AM jthornton: http://paste.ubuntu.com/
06:59 AM Rod_W: done now from Rod Webster
06:59 AM Rod_W: pasted the gmoccappy plasma.hal
06:59 AM Rod_W: maybe you need the URL http://paste.ubuntu.com/25311826/
07:01 AM Rod_W: And the main hal file. http://paste.ubuntu.com/25311837/
07:03 AM Rod_W: YOu should not need it but this is Gmoccappy Signals.hal http://paste.ubuntu.com/25311844/
07:15 AM gloops: cnc ocd lol
07:23 AM Rod_W: Been down in my shedd getting a bit more stuff. The sim hal file I used as the template http://paste.ubuntu.com/25311907/
07:24 AM archivist: diff that with yours
07:25 AM Rod_W: And component man file http://paste.ubuntu.com/25311913/
07:25 AM Rod_W: sorry to besiege you guys.
07:27 AM archivist: you may have the order inverted you have stuff in servo-thread 1 and the rest in servo-thread
07:28 AM archivist: I dont know if servo-thread is 0 or 2 :)
07:29 AM archivist: dunno why he set up two threads when one thread in the right order should be similar
07:32 AM jthornton: I don't remember two threads in my configuration only that things had to be in order
07:32 AM Rod_W: I don't think that is a separate thread. From what I understand that puts the eoffset_pid first in the load order. Check the comments line 15 of the sim hal file I posted.
07:33 AM archivist: you have the 1 but not the 2
07:33 AM Rod_W: I remember a post by Andy about an undocumented feature. BUt heck, you guys know more than me
07:34 AM Rod_W: Yes but summer is for the sim, so its not used in my live config
07:34 AM archivist: there is a feature not well known that execution order is similar to the text order, so if not done correctly things can be forced to wait a cycle
07:36 AM jthornton: I just realized who you are Rod :) saw "spaceship" in the hal file
07:36 AM Rod_W: ;) and a client of yours too :)
07:37 AM jthornton: I wonder if the power supply from the server I took apart will work in my plasma pc
07:40 AM Rod_W: Yes, the number means position. Now documented in Master (which I'm using) http://linuxcnc.org/docs/devel/html/hal/basic-hal.html#_addf
07:44 AM jthornton: not seeing anything that sticks out in the hal files
07:45 AM archivist: needs to be an automated addf error checker
07:47 AM Rod_W: Just wondering how addf motion-command-handler servo-thread
07:47 AM Rod_W: addf motion-controller servo-thread get loaded?
07:47 AM Rod_W: they are in the addf's but where does the loadrt take place?
07:47 AM jthornton: everything is loaded in order except those with a specific number
07:48 AM jthornton: the addf is the only thing that is ordered
07:48 AM archivist: not sure about that last comment
07:50 AM Rod_W: so do the motion-command-handler and motion-controller get loaded with Linuxcnc system. I could not really see where they get loaded but maybe thats not required.
07:50 AM archivist: I thought nets are read from outputs in order
07:56 AM Rod_W: THere could be a bug in the addf commands for motion-* perhaps as this is a Mesa 7i76e?
07:57 AM Rod_W: When I look at the docs for motion, it says its its loaded with loadrt motmod which we don't use
08:01 AM archivist: look in your ini for that
08:07 AM Rod_W: Just went and commented out those addfs and now I know they are certainly needed so no bug there
08:10 AM Rod_W: The only other thing I could think of is that the axis e-offsets needed another signal to wake them up that was outside of the eoffset_pid component. But I could not see anything
08:13 AM jthornton: whew finally got that power supply out of the dang dell server so off to the shower see you guys in the shop
08:13 AM Rod_W: Another thought was to see if I could graft Dewey's simulated torch onto my working config so I could work on this without firing up my plasma cutter.
08:13 AM Rod_W: cool...
08:14 AM jthornton: Rod_W: when I trouble shoot I try to have the most minimal configuration
08:20 AM IchGucksLive: hi from germany
08:21 AM IchGucksLive: Rod_W: what thc do you got most thc can simulate
08:29 AM Rod_W: John, thanks, I suspect that will be my next step.... Tear down 9 months work will hurt when this is the only par tleft to do. I've been so careful to add one part at a time and debug all the way.
08:30 AM IchGucksLive: rod i run 4plasmas in store and build 6 more that is a real manual config you use but genus as i will say
08:30 AM Rod_W: IchGucksLiv, I'm using LinuxCNC as the THC using an experimental branch... https://github.com/LinuxCNC/linuxcnc/tree/dgarr/external_offsets
08:30 AM IchGucksLive: i go for just eagle thc as i run most more then 10 sheets same part
08:32 AM Rod_W: This one will be cool. jthornton was meant to debug this for us months ago but his PC died. John, you realise you must get this working now to claim line honours in the race to use external offsets for plasma :)
08:33 AM IchGucksLive: i realy like the THCUD as it is simple and does a good job for most need
08:33 AM IchGucksLive: so no external ofsets needed
08:34 AM IchGucksLive: just 2 hallines and 2 settings
08:35 AM Rod_W: Yes, THCAD (for mesa) that John wrote is an option but all of us are excited to have pid control over our torch and someone has to try it out :)
08:35 AM IchGucksLive: my personel expiriance is as the thc and the internal plasma elecronics torch contol electronic synch is the goal to get best clean cuts
08:36 AM IchGucksLive: Rod_W: the hal you presented is way beond starting peoples mind
08:37 AM IchGucksLive: Rod_W: http://www.schweden-feuer.de/feuersauelen.htm scrol down
08:38 AM IchGucksLive: rod and as hal is the key to belive the system it getting for starters hard to belive
08:38 AM JT-Shop: well crumb that power supply won't fit the plasma pc case
08:38 AM IchGucksLive: on 10+ components loaded
08:39 AM IchGucksLive: JT-Shop: so the space to use on plasma http://heimwerkermarkt-tretter.de/schrank.jpg
08:40 AM IchGucksLive: ok im off till later
08:40 AM JT-Shop: nice
08:41 AM -!- #linuxcnc mode set to +v by ChanServ
08:43 AM Rod_W: Well thats annoying. Thanks for trying. Maybe if you can flag to Dewey that somebody is ready to test his new components. That would be appreciated. I did send him an email via the forum.
08:45 AM Rod_W: I'm off then getting close to midnight here.. If anybody has some ideas, maybe you can post them on the forum here https://forum.linuxcnc.org/10-advanced-configuration/33140-help-with-external-offsets-on-a-real-plasma-machine
08:45 AM JT-Shop: well it's not the power supply
08:45 AM JT-Shop: hang on a second
08:45 AM Rod_W: waiting....
08:46 AM JT-Shop: pm
08:47 AM Rod_W: where do I look for it?
08:47 AM JT-Shop: in hexchat it just shows up as a new tab
08:48 AM Rod_W: thats what I'm using.
08:48 AM JT-Shop: right click on my name and open a dialog window
08:53 AM JT-Shop: lol now the PC is working
09:34 AM gloops: looking at homing and step rate ...haha this isnt a 2 minute job is it, whats a good sequence for setting up the machine - im not even getting near with units - like 10mm command doesnt travel 10mm
09:35 AM gloops: in fact ill probably watch some youtube vids before much else to get more familair with the terms
09:39 AM Polymorphism: This damn crunchy ballscrew
09:39 AM Polymorphism: need to get this thing up and running
09:39 AM Polymorphism: this painful 4000usd loss
09:43 AM gloops: 4000 dollars man?
09:43 AM gloops: what are you making
09:56 AM -!- #linuxcnc mode set to +v by ChanServ
10:43 AM -!- #linuxcnc mode set to +v by ChanServ
11:08 AM IchGucksLive: hi
11:08 AM IchGucksLive: gloops: still around
11:09 AM IchGucksLive: Polymorphism: i agree with you cnc can cost you alot of money
11:10 AM gloops: in and out for a bit
11:11 AM IchGucksLive: ok i will help you seting things up correct if you want to just double click hard my nick
11:26 AM IchGucksLive: till later
12:21 PM IchGucksLive: back
12:40 PM IchGucksLive: gloops: around
12:59 PM phipli: evening
01:01 PM IchGucksLive: ;9-
01:01 PM IchGucksLive: ;-)
01:02 PM IchGucksLive: its clearing up here so Starshowernight ahed
01:03 PM IchGucksLive: Loetmichel: one more convertibal day to go the weather service saying winter will come
01:16 PM IchGucksLive: im off Gn8
01:16 PM phipli: night
03:40 PM gloops: ok, setting pitch and pulley teeth is not giving me 1 inch = 1 inch, what other variables affect this
03:42 PM andypugh: gloops: Just work out how many steps-per-inch you really need and edit that directly into the INI file.
03:42 PM andypugh: I think the stepconf calculator tries too hard.
03:43 PM andypugh: How far out is it?
04:00 PM gloops: sorry was distracted, tried various combos - stepconfig doesnt seem to be giving any rational reponse, its all over the place
04:01 PM gloops: so just trial and error?
04:01 PM andypugh: No, absolutely not
04:02 PM gloops: it is chain drive with reducing sprockets, i also dont have the data for my driver so nothing really concrete to put in
04:02 PM andypugh: You should be able to work out, from what you know, exactly how many steps it should take to move the machine one unit.
04:03 PM gloops: i can calculate that but cant input that in stepconfig, it only asks for pitch - revs/distance
04:05 PM phipli: do you have the angle of a step on the motors?
04:05 PM andypugh: Yes, but you can edit your HAL file to put the right number in.
04:05 PM phipli: is it 1.8 or something?
04:06 PM phipli: can you measure the diameter of the two pulleys?
04:06 PM andypugh: (or, in fact, lie to stepconf by setting everything except motor steps to 1)
04:06 PM gloops: yeah standard 1.8 - 200 steps
04:06 PM phipli: (or the two numbers of teeth)
04:06 PM andypugh: Measuring isn’t the way, you should count the teeth.
04:07 PM gloops: well working on chain pitch of 12.5 and number of sprocket teeth i can calculate the distance 1 rev produces
04:07 PM phipli: do you know the pitch and number of start points on the leadscrew?
04:07 PM gloops: there arent any screws
04:07 PM phipli: ah right
04:07 PM phipli: (/me reconfigures the mental machine)
04:08 PM phipli: 12.5 whats?
04:09 PM gloops: but even measuring the distance travelled by physically rotating the sprocket and measuring the distance - input that in stepconfig doesnt reproduce anything near, probably missing something very simple
04:09 PM phipli: microstepping?
04:09 PM gloops: 12.5mm - thats the distance between sprocket teeth
04:09 PM gloops: have had microsteps at value of 2
04:10 PM phipli: so 400 steps per rev
04:10 PM gloops: so basically if you have 14 tooth sprocker - 1 rotation pulls 87.5 - 12.5 x 14
04:11 PM gloops: yes 400 steps
04:11 PM phipli: and it is direct drive?
04:11 PM gloops: no has a reducer at 3-1
04:11 PM phipli: just a stepper with a gear directly driving the chain?
04:12 PM gloops: or 1-3
04:12 PM phipli: so it does 1/3 rotations per stepper rotation?
04:12 PM gloops: yes
04:13 PM phipli: so it is 1200 steps per 175mm?
04:13 PM phipli: (12.5*14=175)
04:13 PM gloops: yes thats right
04:14 PM phipli: you wrote 87.5
04:15 PM phipli: which have you entered in your settings
04:15 PM gloops: but stepconfig should calculate that if i enter reducer values, pitch, microsteps and so on, i will have to take a close look at ini and hal files, havent even opened hal file yet
04:16 PM gloops: yes sorry bad maths on the spot lol, there is a 14 tooth sprocket on one axis but was just an example
04:17 PM phipli: so that axis is 174.1714286 steps per inch
04:17 PM gloops: stepconfig doesnt ask for that calculation - it asks for step angle, microsteps, pitch and reduction pulley
04:18 PM phipli: andypugh recommended that stepconfig gets it wrong and suggested directly modifying the INI file
04:18 PM gloops: right
04:19 PM andypugh: Stepconf doesn’t actually get it _wrong_ but some of the questions can be hard to answer for some drives.
04:19 PM phipli: is it rounding?
04:20 PM phipli: both metric and imperial values come out... irrational
04:20 PM phipli: (in the mathematical sense)
04:20 PM andypugh: That’s not unlikely
04:20 PM phipli: *is it excessively rounding?
04:20 PM andypugh: Is the machine direct chain drive?
04:21 PM phipli: 3:1 reduction
04:21 PM andypugh: Or does it torn screws?
04:21 PM andypugh: (turn screws)
04:21 PM gloops: not sure - i am not getting close enough to be looking at fractions really, the results i have had have been bizarre, like just random changing pitch doesnt seem to give a consistent result, so i can say yes keep reducing or increasing getting closer
04:21 PM gloops: no no screws at all
04:23 PM andypugh: That’s really not the way to go
04:23 PM gloops: anyway i will have another look tommorrow, as i say probably something simple im overlooking
04:23 PM andypugh: What pitch are you using, and what wort of chain is it?
04:24 PM gloops: 12.5 - cycle chain
04:24 PM andypugh: In your case “pitch” is the distance the carriage travels for one turn of the last sprocket.
04:24 PM andypugh: How many teeth on the last sprocket?
04:25 PM gloops: ahh yes, i will get the exact ficgure tommorrow - it is about 80mm - repeatability is fine with dial gauge it isnt a mechanical problem
04:25 PM andypugh: (This is a tensioned-chain drive? No screws, no belts, no racks?
04:26 PM gloops: yes just chain
04:26 PM andypugh: How many teeth?
04:26 PM gloops: 14 on the final sprocket, 10-30 on reducer
04:26 PM andypugh: Then just multiply by 0.5” and that’s the “pitch” for this cacluation.
04:27 PM phipli: 177.8
04:27 PM phipli: (using 12.7)
04:27 PM andypugh: So, 7” pitch, reduction is whatever you actually have, steps is 200 and microsteps is wjhatever you have
04:28 PM gloops: right, but remember stepconfig asks for the reducer values, if i do the full calculation ill be entering one value twice
04:28 PM andypugh: I don’t think so.
04:28 PM gloops: yes it is 12.7 - not 12.5 - i have used the proper numbers, im just giving figures here for examples sake
04:28 PM andypugh: Unless you have two chains round one sprocket...
04:29 PM phipli: Pitch : 177.8
04:29 PM phipli: Reduction : 3:1
04:29 PM phipli: Microsteps : 2
04:29 PM phipli: Steps : 200
04:30 PM gloops: ok i will try that tommorrow
04:30 PM phipli: never seen a machine using bike chain - is it common?
04:32 PM gloops: lol, probably not, i think chain was quite popular in the earlier days of diy cnc, before ballscrews were readily available, it probably isnt worth the effort, but i fancied a go at a machine and didnt want to spend too much on the first attempt
04:33 PM phipli: There is a bike shop down the road from me...
04:33 PM andypugh: My lathe is driven by motorcycle camchains.
04:33 PM phipli: I wonder how much old chain they have :)
04:34 PM phipli: I have some motorbike drive chain in the shed somewhere
04:34 PM andypugh: Old chain has innaccurate pitch.
04:34 PM andypugh: At some point you might have to calibrate round that.
04:34 PM gloops: yeah probably better with smaller pitch chain, about 8mm
04:34 PM phipli: you can make awesome anamatronic hands with motorbike chain
04:35 PM gloops: thats probably what ill end up doing with the parts from this lol
04:35 PM phipli: (andypuch : yeah, I know bike chain stretches, and you'd only get it free if it had...)
04:35 PM phipli: it would be nice if good stuff was free
04:35 PM andypugh: phipli: Spot-weld bowden cable to the rollrs?
04:36 PM phipli: I think so
04:36 PM phipli: something my brother was talking about
04:36 PM phipli: they'd put cables on it and fix it to a plate
04:36 PM phipli: then stick a glove on it and remotely operate the cables
04:36 PM gloops: the backlash inherent in chain is realised - its under tension
04:36 PM gloops: sorry - ISNT realised
04:36 PM phipli: gives quite a natrual look
04:37 PM phipli: natural*
04:37 PM phipli: guess it isn't that different to how hands work
04:37 PM andypugh: No, practically identical in fact
04:38 PM andypugh: You can even get welds in the wrong place
04:38 PM phipli: https://www.youtube.com/watch?v=ONPkB-TmphE
04:39 PM andypugh: A friend of mine had trouble with the strings in his fingers getting stuck. Apparently known as “Vikings disease” https://en.wikipedia.org/wiki/Dupuytren%27s_contracture
04:41 PM sync: andypugh: don't you get a polygonal effect with camchains as well?
04:42 PM phipli: we have a friend who's tendons are too short
04:42 PM phipli: makes their legs very stiff and unwieldy
04:43 PM phipli: they're bloody minded though and do better at getting around than they should do!
04:43 PM andypugh: sync: The camchains are of the inverted tooth type (also known as “silent chains” and are extremely smooth
04:43 PM phipli: fall over several times a day
04:46 PM sync: andypugh: yeah I know, but I remember that even those have effects due to the teeth and the drive angle
04:47 PM andypugh: No more so than a gear, as I understand it.
04:47 PM gloops: chain drive xcarve - https://www.youtube.com/watch?v=oMWowEu5MiY
04:47 PM andypugh: (I think that they are almost exactly equivalent to a gear drive with the advantage that you can adjust backlash easiliy)
04:49 PM andypugh: gloops: Is your drive system like that? If it is then you _don’t_ have a 3:1 reduction.
04:49 PM sync: hmm, I would have to look up my engine design course stuff again
04:50 PM gloops: no hes got twin motors, - ive got a driveshaft running accross, 1 motor with reducer
04:51 PM hazzy2 is now known as hazzy
04:53 PM gloops: the pick up method is the same, you need to get the chain round as many teeth as possible to stop it shifting on the sprocket
04:58 PM andypugh: Have you seen the Servobelt? The same setup but with belts, and a clever twist, a mating belt bonded to the machine frame.
05:00 PM phipli: can get long lengths of timing belt for half nothing in china
05:00 PM gloops: not seen it, cant really picture it, what does the stuck belt do?
05:00 PM hazzy2 is now known as hazzy
05:01 PM gloops: chain and belt is used to drive almost everything in industry, cheap standard components
05:01 PM andypugh: https://www.youtube.com/watch?v=xh-AUPAWbPc
05:01 PM andypugh: The bonded belt makes the drive much stiffer, no belt spring.
05:03 PM gloops: yeah good stuff, they have a very long cnc circular saw at the local pallet factory, runs on belt, its like lightning
05:03 PM gloops: kid just throws wood on, the wood stays still and the saw rips along it
05:04 PM sync: oh, that's neat andypugh
05:04 PM phipli: https://www.banggood.com/10M-Rubber-GT2-Open-Timing-Belt-Width-6mm-For-3D-Printer-RepRap-p-1148503.html?rmmds=search
05:05 PM phipli: gloops : only 6mm wide, but still - cheap for 10m!
05:05 PM gloops: youre into problems with screws at a certain length
05:06 PM phipli: dual gantry!
05:06 PM phipli: :)
05:06 PM sync: you can often just drive the nut instead of the screw
05:06 PM gloops: thats cheap enough to tie plants up in the garden lol
05:06 PM phipli: gloops: :)
05:07 PM gloops: sync yes but then the strengths of the ballscrew start to get undermined
05:07 PM gloops: youre gonna be turning that nut with a belt
05:07 PM LeelooMinai_ is now known as LeelooMinai
05:08 PM sync: yes, but that doesn't have to be a disadvantage
05:08 PM sync: you can also use a hollow motor
05:08 PM gloops: your build starts to get complicated
05:08 PM gloops: so why not drive the whole thing with belt
05:08 PM sync: less stiffness
05:09 PM gloops: stiffness how?
05:09 PM sync: a belt will stretch more than a ballscrew
05:09 PM phipli: especially a long one
05:10 PM phipli: also wear and strength I guess?
05:10 PM gloops: yeah but you arent using a single belt, youve got a belt on each side of the gantry, each side is indexed physically
05:10 PM gloops: well the screws will wear as well
05:10 PM sync: I'd use two ballscrews on a fast gantry
05:12 PM gloops: well nobody can say that wouldnt be a good system
05:12 PM gloops: thats your first 2k gone though lol
05:14 PM gloops: and everything you said about belt - applies to the belt driving those nuts
05:14 PM sync: yes and no
05:14 PM sync: significantly less so than if you'd put the gantry on it
05:15 PM sync: as there is a lot less force transmitted through it
05:16 PM phipli: there is a wormgear between the belt and the machining forces
05:16 PM gloops: well, my car uses a timing belt, needs changing every 70k miles, pretty horrendous stresses, keeps 2 cams times to perfection, belt is pretty durable and accurate stuff
05:17 PM sync: your timing belt stretches a suprising amount depending on the position of the cam
05:17 PM sync: it is not a lot, but it can be an issue
05:17 PM sync: also an engine needs a lot less precise valve control than the positiioning of a cnc machine
05:18 PM gloops: possible but i mean ive built a heavy gantry wityh box section steel, a driveshaft linked with chain on both sides, i can turn the sprockets with my fingers, the forces involved are negligible, screws are actually working much harder
05:18 PM enleth: gloops: you can use a huge ass belt to drive the nut, the kind you wouldn't be able to use for pulling the whole gantry along
05:19 PM sync: the forces are not small, you are just not acclerating it fast enough
05:19 PM gloops: haha, i guess im not going to convert anyone to chain here
05:19 PM gloops: belt i mean
05:19 PM phipli: some of the guys use belt
05:20 PM enleth: gloops: my Bridgeport does that - drive a ball nut with a toothed belt - but the belts are 50mm wide, I can't remember the thickness but it was substantial as well
05:20 PM phipli: main advantage is price though
05:20 PM enleth: *30mm, not 50mm, derp
05:20 PM gloops: enleth well your mill isnt using screws 3 metres long though
05:20 PM phipli: 50mm sounds like a supercharger belt :)
05:21 PM phipli: off something /big/
05:21 PM phipli: gloops: your machine is 3m long?
05:22 PM phipli: blimey
05:22 PM phipli: what are you planning to cut?
05:22 PM phipli: boats in one go?
05:22 PM gloops: no, but i thought we were talking about long traverse
05:22 PM phipli: (which would be awesome... if wasteful)
05:23 PM phipli: big machines for wood tend to use belts
05:23 PM gloops: for shorter distance you turn the screw, not the nut
05:23 PM phipli: some at least...
05:23 PM phipli: sigh. every machine is different
05:24 PM enleth: gloops: yes, it's not, but I think I'd still consider this setup for a plotter. My point is that you *can* use a hugely oversized belt as a part of the intermediate drivetrain, the sort that would be too expensive or too difficult to tension properly as an actual axis drive
05:24 PM gloops: these industrial routers are cutting at 500 inch a minute, its hardly worth loading a 1 metre machine
05:26 PM enleth: gloops: besides, isn't the biggest problem with using belts directly that you need a much finer angle control at the motor?
05:26 PM gloops: enleth that depends really what youre doing, what resolution you want
05:26 PM gloops: belt can be much more aggressively geared than screws, very fast
05:27 PM sync: a screw can also be very fast
05:27 PM sync: the problem is always how stiff the setup is
05:27 PM sync: as you will be loosing position during accel
05:28 PM sync: or decel...
05:29 PM gloops: how is position lost?
05:29 PM enleth: oh, I just remembered reading about pressure oiled ball nuts somewhere, not sure what that was for exactly but it needed extreme traverse speeds
05:29 PM enleth: as in, actively pumped with oil at high pressure
05:29 PM sync: due to the belt stretch
05:29 PM sync: or strech of your screw
05:29 PM sync: or your bearings
05:30 PM sync: high accel and high position accuracy is difficult
05:30 PM phipli: night folks
05:31 PM gloops: goodnight
05:32 PM enleth: sync: that's probably the point where you want to start modeling the machine as a second order system to counter that
05:32 PM HighInBC: lol @ pressure oiled ball nuts
05:33 PM sync: enleth: usually you compensate by chosing adequate sized screws
05:33 PM sync: you can also run pressurized oil through the screw to dampen the harmonics
05:34 PM sync: which also has the benefit of allowing you to control its temperature
05:36 PM gloops: pretty fast - https://www.youtube.com/watch?v=YZLK-UUqJvs
05:36 PM enleth: I don't have a "feel" for machines this big, but I imagine there are practical limits to upping the size of the drivetrain, due to its own increased inertia, where you just have to live with the fact that the damn thing is springy and needs to be driven as such
05:37 PM sync: the inertia can be countered by a properly sized servo
05:37 PM sync: 50mm ballscrews are not uncommon
05:38 PM sync: I notice the whip very much when I drive one of my test axis to 68m/min
05:38 PM sync: I'd need a larger screw
05:39 PM enleth: biggest I've had my hands on was 30mm, but I can imagine those gantry mills that you could drive a truck onto need something a lot more substantial
05:40 PM sync: those are driven quite slow, so the forces are actually now
05:40 PM sync: highly dynamic machines need to be driven very stiff
05:40 PM sync: or a lot of matlab simulink needs to be applied
05:42 PM enleth: yes, that's where a second order model comes into play
05:44 PM enleth: I know many (not sure if most) 6R robot arms are being modeled and driven as second order systems to counter their lack of rigidity,
05:44 PM gloops: well, i dont know, if you are trying to build a large machine with industrial specs its going to cost, so you will have an expensive toy, which is fair enough, surely if you have work for such a machine it would be as well to just buy one and start making money straight away
05:46 PM enleth: as a result, they can pull off moves that would fail completely with horrible vibration and overshoots if they were driven in a "naive" manner
05:47 PM gloops: like state of the art fighter aircraft couldnt fly without the compensation forces applied by the computer enleth
05:47 PM sync: no
05:47 PM sync: a modern fighter could not be flown manually
05:47 PM sync: as it is an inherently unstable system
05:47 PM sync: and control theory makes it happen
05:48 PM Loetmichel: sync: as far as i know it COULD be flown by hand
05:49 PM sync: no, if the FC fails in the eurofighter the pilot has to eject
05:49 PM enleth: Loetmichel: that probably depends, supposedly some fighter jets are beyond the abilities of a human pilot
05:49 PM Loetmichel: but it woud require quite a "hellsichtiger" Pilot
05:49 PM gloops: they will eventually do away with the 'machine' as a fixed platform, the robot will roam the workpeice doing the cutting
05:49 PM Loetmichel: because of human reaction time way to slow to compensate the unstable behaviour
05:50 PM Loetmichel: the pilot had to know in which way the plane will deviate BEFORE it happens
05:50 PM sync: yes, the pilot has not enough phase margin to stabilize the system
05:50 PM Loetmichel: s/had/would have
05:51 PM enleth: gloops: you could run a 6R arm with just the IK matrices and no jacobians etc., but only at very slow speeds compared to what the limits of the arm are as designed for proper dynamic control
05:52 PM gloops: we have driverless delivery vans here now, i mean there are people building cnc machines - when cnc as it is now will be obsolete before they finish building it lol
05:52 PM Loetmichel: gloops: already happened... https://www.youtube.com/watch?v=quN37YskoaM
05:52 PM pink_vampire: olllllllllllllllllllld
05:52 PM Loetmichel: pink_vampire: thats the point ;)
05:54 PM gloops: that almost looks organic
05:55 PM Loetmichel: <- is a sf fan. it reminds me awfully of an SG-atlantis replicator ;)
05:56 PM gloops: when human operators and designers are obsolete, only the machines will be able to build machines
05:56 PM enleth: the robotic overlords have mastered milling slightly heated butter with no chatter, we are surely doomed
05:56 PM gloops: haha
05:58 PM Loetmichel: enleth: wait a few interations of soft and hardware... lets see who laughs then... ;)
05:58 PM Loetmichel: -n
05:59 PM gloops: ‎was something they said would never happen - https://www.youtube.com/watch?v=SQ2PdpLEAG0
06:02 PM gloops: and the former bricklayers - they just bring the bricks for the robot
06:05 PM enleth: they already put something like this on a crane boom with machine vision and a brick conveyor
06:06 PM enleth: but I haven't heard of any real "production" deployments yet
06:06 PM gloops: yeah i saw that, all the problems of getting round the workspace have been overcome
06:06 PM enleth: still in testing I think
06:06 PM gloops: in the next 20 years a lot of jobs are going down the pan
06:08 PM enleth: someone needs to stay as the domain expert to work with robot programmers
06:10 PM enleth: besides, concrete high-rise construction doesn't seem to lend well to on-site automation
06:11 PM gloops: well, those places are built for workers - but there wont be any workers
06:14 PM enleth: best I've seen was a bulding that had an identical core and prerimeter cross-section across its whole height, so they constructed two sets of formwork that were just hydraulically lifted up one floor after the previous one finished setting
06:14 PM gloops: yeah have seen that - they built a winding tower for a coal mine near me like that years ago
06:15 PM gloops: continuous pouring the concrete, the hydraulic shutter just moved up gradually as the concrete went in
06:16 PM enleth: oh, and lifting/placing stuff with cranes at height doesn't seem to be easy to automate
06:17 PM enleth: the crane sways all the time, the load swings randomly, it has to be manually caught from below and pulled into place at some point
06:18 PM enleth: half of the work is grabbing the stuff with a hooked boom and pulling it against the wind through the last couple meters
06:18 PM gloops: or is that because human crane drivers arent good enough
06:20 PM enleth: I'm not saying it can't be done, but it won't be *easy* when you have wind randomly messing things up by a significant margin
06:20 PM Loetmichel: thats classing "self teaching AI" territory there
06:20 PM Loetmichel: the crane thing
06:21 PM gloops: well yeah there will always be those kind of jobs, probably communication based jobs as well for a good while
06:43 PM Frank___: hellooo
06:45 PM enleth: BTW, I'll be putting the Bridgeport back together after a move and I'm wondering if it's worth capturing it on a video, even if it's probably not going to be narrated or anything
06:45 PM Frank___: guys im having a problem with my toolpath, on some concave paths, the machine stutters, like when you have g64 to a very low tolerance, i have it to 0.2mm, but still encounter this issue, any advice?
06:47 PM DaViruz: do you have any spindle commands in the move?
06:50 PM BeachBumPete: Check your code to see if the curvature toolpath is broken up into MANY MANY straight moves instead of one fluid movement. Sometimes geometry created by certain CAD programs makes curvatures like that.
06:55 PM Frank___: weird, because most of the movement are smooth
06:55 PM Frank___: its on fusion 360
06:56 PM enleth: Frank___: try disabling smoothing
06:56 PM Frank___: on fusion or on lcnc
06:57 PM enleth: I don't have F360 open right now, so I can't tell you which tab is that, but I distinctly remember *dis*abling path smoothing helped with stuttering
06:57 PM Frank___: i remember that option, but if anything i would have thought it would help the path, lol
06:57 PM Frank___: ill try that thank!
06:57 PM enleth: it was stuttering like crazy on some tight inside curves
06:58 PM enleth: jerking half a ton of cast iron around in the process
06:58 PM Frank___: hehe
06:58 PM Frank___: here only 150kg
06:59 PM Frank___: btw, anyone ever bought something from moore gear?
06:59 PM enleth: AFAIR "smoothing" in F360 means "generate a shitload of short moves for controls that can't do proper curves", and for some inconceivable reason it's selected at the operation level instead of being handled by the postprocessor
07:02 PM Frank___: how do you manage the outer corner mode option?
07:02 PM Frank___: roll around corner/keep sharp corner
07:03 PM Frank___: now that i am checking, smoothing is off, thats odd
07:04 PM enleth: pretty sure I use "keep sharp"
07:05 PM enleth: but I can't remember specifics, all my machines are packed up on pallets since early June for the move so obviously I haven't been doing much in the way of CAM in the meantime
07:07 PM Frank___: okey, thanks you anyway!
07:07 PM Frank___: im cutting mdf at 8000mm/min
07:07 PM Frank___: so happy
07:34 PM Tom_itx is now known as Guest24493
08:51 PM BeachBumPete: Jeez man I need some more Trim sol E206 and Enco no longer exists. Now the price has gone up considerably.... GAAAAHHHHhhhhhhhhh
08:53 PM Tom_itx: use that stuff my bud uses
08:54 PM BeachBumPete: shit's spensive?
08:54 PM Tom_itx: i dunno, he get 5gal at a time nowdays
08:54 PM BeachBumPete: yeah so did I
08:55 PM Tom_itx: i had a pic of it on my phone... dunno if i posted it or not
08:55 PM BeachBumPete: but I bought it from Useenco for like $75 plus like 20 shipping
08:55 PM BeachBumPete: now its like $150 PLUS shipping
08:56 PM Tom_itx: Ultracut 380R
08:56 PM Tom_itx: Rustlick
08:57 PM Tom_itx: right now i'm soldering up some optos for my relays coming from JT
08:59 PM Tom_itx: one of the few times i actually have everything i need
09:05 PM Polymorphism: zlog
09:05 PM Polymorphism: .zlog
09:05 PM Polymorphism: !log
09:05 PM Polymorphism: !zlog
09:07 PM Tom_itx: simple.
09:12 PM BeachBumPete: thats great man
09:12 PM BeachBumPete: sorry I had to go attend to the machine ;)
09:14 PM BeachBumPete: that rustlick appears to be about the same price or higher generally
09:23 PM nitrxgen_: so my 3020t arrived... comes with a software disk... idk what it is, but is there someone to suggest something better for it?
09:23 PM nitrxgen_: or just use the software it comes with?
09:26 PM jdh: linuxcnc
09:26 PM jdh: I assume it comes with a pirated mach
09:28 PM nitrxgen_: idk
09:28 PM nitrxgen_: i dont have a working cdrom anyway
09:41 PM BeachBumPete: OOh I like infinity pool environment in F360 ;)
10:17 PM nitrxgen_: well, spindle tested my 3020T, that spindle controller can fuckin go lol, the pwm is far too audible
11:00 PM nitrxgen_: hmm :? literally just running planetcnc, drivers, not controlling my 3020t
11:00 PM CaptHindsight: brick laying robot HAH! https://www.youtube.com/watch?v=0-0ChljgWCQ Bridge section laying robot!!
11:20 PM hazzy: CaptHindsight: That's amazing
11:21 PM hazzy: and those are some big bricks!
11:26 PM SpeedEvil: I like big bricks and I cannot lie.
11:46 PM RyanS: I made an electronic diving head like thus https://drive.google.com/file/d/0B8kMGqIWhRiHcHlBcGh6VDMzMDg/view
11:46 PM RyanS: currently it doesn't have enough holding torque to mill steel
11:46 PM RyanS: it's a 200 step per rev, 1.8 Nm motor set to 1/8 microstepping with a 4:1 timing pulley ratio. so 6,400 steps per rev. If I disabled microstepping would I get more holding torque? I don't really understand how steps per revolution translates to accuracy, but surely 800 steps would not be very accurate?
11:47 PM SpeedEvil: The 800 steps are at 0 torque
11:47 PM SpeedEvil: They may be (somewhat) accurate at no torque, but as you push or pull on it, it's going to go towards the nearest full step
11:48 PM SpeedEvil: If you have it in the full step position, then the 'stiction' of the detent adds to the holding torque
11:48 PM RyanS: yeah so microstepping is 'weaker'..
11:52 PM RyanS: hmm, 800 steps is going to be pretty inaccurate if I am cutting eg, a 120 tooth gear? (I suck at maths, I'm just following this http://www.liming.org/millindex/)
11:56 PM hazzy: Yeah, about 15% error for 120 tooth gear, not good
11:56 PM hazzy: Your unite looks really nice though
11:57 PM hazzy: maybe add a lock/break so you could use microstepping?
11:58 PM hazzy: just have to insert an M1 between positioning moves to unlock/lock
11:59 PM RyanS: yeah I overengineer stuff :-) the block is 4030 steel.. tough to mill
11:59 PM RyanS: yeah I was thinking of this https://www.omc-stepperonline.com/stepper-motor-brake/dc-electromagnetic-brake-24v-20nm283ozin-for-nema-23-and-24-stepper-motor-swb-03.html
11:59 PM hazzy: What type of bearings? Size?