#linuxcnc Logs

Oct 11 2019

12:44 AM taloot: whats the good pulse/mm for router
12:49 AM taloot: for smooth operatiion
12:58 AM sensille: depends on the resolution you want to achieve. i think my cheapo router has 640 steps/mm, not very smooth
12:59 AM sensille: if you can choose, choose the highest value that still gives you the required feed
12:59 AM taloot: im having 400 but also not very smooth
01:00 AM taloot: i will show you video and tell me..
01:00 AM sensille: do you already have that new servo in place?
01:02 AM taloot: no
01:02 AM taloot: this is stepper
01:02 AM taloot: close loop
01:02 AM taloot: but its much smoother than regular one
01:03 AM taloot: but still not like servo
01:03 AM taloot: you have 640 on servo?
01:03 AM sensille: stepper
01:03 AM sensille: 1/16th microstepping
01:04 AM taloot: im having 2000/rev
01:06 AM sensille: 3200 here
01:06 AM taloot: rackk and piinon 10:1 gear
01:07 AM taloot: u have direct drive mybe
01:10 AM miss0r: morning
01:11 AM sensille: taloot: ballscrew, 5mm lead
02:35 AM Deejay: moin
02:39 AM taloot: normal stepper is no go anymore
02:39 AM taloot: price of closed loop simiilar
02:40 AM taloot: 450rmb for 12.5nm nema 34
02:40 AM sensille: should close the loop with linear encoders
02:41 AM taloot: im looking for cheap and good sloutions
02:41 AM taloot: ii use servo delta to benchmark
02:42 AM taloot: my new laser machine wiill be belt driivern
02:42 AM taloot: 45000 max feed
02:43 AM sensille: do you design it yourself?
02:44 AM taloot: yse
02:44 AM taloot: yes
02:45 AM taloot: https://grabcad.com/taloot-1
02:45 AM sensille: hard to be accurate with belts i guess, unless you use linear encoders
02:46 AM taloot: nice advice i will look into it
02:46 AM taloot: but i design it be stright along the axis
02:48 AM taloot: https://www.photobox.co.uk/my/photo/full?photo_id=502224787714
02:49 AM taloot: check the cutting speed
02:49 AM taloot: for 1mm steel
02:49 AM taloot: we do pipe flages for oil pipe
02:49 AM taloot: 8mm
02:49 AM sensille: link doesn't work
02:50 AM taloot: https://ibb.co/51Vvn45
02:58 AM sensille: how accurate do you want to cut?
03:05 AM taloot: 50u
03:05 AM taloot: 50um
03:06 AM taloot: the laser spot
03:06 AM sensille: yeah, and your targeted positional accuracy?
03:06 AM taloot: 0.001
03:07 AM sensille: mm
03:07 AM taloot: yes
03:07 AM taloot: 6000rpm servo
03:07 AM taloot: 3000 but max 6000
03:07 AM taloot: my machine will be 6061 alumnum frame
03:08 AM sensille: 0.001mm with belts? which belts?
03:09 AM taloot: 5M
03:09 AM taloot: HTD timinig belt
03:09 AM taloot: the gear i have it
03:09 AM taloot: 3 acrmin
03:10 AM taloot: i want drag race style axiis
03:10 AM taloot: lol
03:10 AM taloot: 45m/m
03:10 AM sensille: the pulley might be the crucial part to achieve your accuracy
03:11 AM taloot: it will big pulley
03:11 AM taloot: better for motion and responsiveness
03:12 AM taloot: but do you thing iii need tensioner
03:12 AM taloot: thniiik
03:13 AM taloot: now whats better to control the servo in analoge or pulse/dir setup
03:13 AM taloot: pulse /dir = 4mhz max for my driver
03:13 AM taloot: analage ii dont now
03:15 AM sensille: analogue was 15 bit resolution, right? pulse seems to be much better to me
03:15 AM taloot: 24 bit
03:16 AM sensille: Analog Input - 15 bit (from the datasheet)
03:16 AM taloot: aha
03:17 AM taloot: but 15 bit = how many step/dir
03:21 AM taloot: and why pulse is better for you
03:22 AM sensille: just because it's digital, even getting 15 bit out of an analog signal is hard i imagine. but i haven't used it, so i don't know
03:23 AM sensille: an interface where you could just feed the new position would be even better
03:24 AM sensille: given you can properly synchronize the motors
03:25 AM pink_vampire: hi
03:25 AM taloot: so i will stuck with step/dir
03:26 AM pink_vampire: how bad is this thing? https://www.ebay.com/itm/123786693650
03:26 AM sensille: i don't know about laser cutting, is it important to have a constant feed?
03:27 AM taloot: yes
03:29 AM sensille: hey, you're spending 14k on a laser and want to save $1 on a power supply?
03:29 AM taloot: lol not me
03:29 AM sensille: oops, sorry :)
03:30 AM sensille: pink_vampire: probably good for 50mA :)
03:30 AM taloot: emi
03:30 AM taloot: problem wiith those
03:31 AM pink_vampire: i need it for arduino and a relay
03:31 AM taloot: i guess it enough
03:33 AM pink_vampire: I hope it will not burn the house down..
03:33 AM taloot: lol
03:33 AM sensille: probably not
03:33 AM taloot: dont worry
03:33 AM taloot: looks good product
03:34 AM sensille: mostly they just can't supply the rated current
03:34 AM taloot: i have small grow room
03:35 AM taloot: and i have led
03:35 AM pink_vampire: let me see how much current it take
03:35 AM taloot: driiver cost me more than chips it sels
04:04 AM taloot: 7i96
04:04 AM taloot: i order it,,, im very excited to try it
04:04 AM taloot: do i need to flash fiirmware or somthng
04:19 AM XXCoder: man went though quite small fog and external temperate went from 36f to 19f near instantly lolk
04:19 AM XXCoder: returned to 38f when entered freeway tho
04:57 AM taloot: why there is no mach3 plugin for mesa
04:57 AM taloot: or mach4
04:57 AM taloot: i think its doable
04:58 AM XXCoder: probably because most people prefer linuxcnc
04:58 AM XXCoder: in least emsa buying peoplke
04:58 AM taloot: gr8 answer
04:58 AM XXCoder: *mesa
04:58 AM taloot: o agree with u 100%
04:58 AM taloot: but keep in mind somthing
04:58 AM taloot: all newbie go fro mach3
04:59 AM taloot: linuxnc required more skillful competer user
04:59 AM taloot: i wasted alot of time learning mach3
05:00 AM taloot: after i learned... i disocver problems in mach3... so i shifted to linuxcnc
05:01 AM taloot: linuxcnc more robust ..
05:02 AM taloot: much more... i feel cconfortable wiith it from day 1
05:03 AM taloot: now i will make small production line
05:03 AM taloot: based on linuxccncc
05:03 AM taloot: it will be a gcode,, that run
05:03 AM XXCoder: cool :) glad its working out well!
05:05 AM taloot: i guess so
05:05 AM taloot: it shot
05:05 AM taloot: should
05:28 AM Tom_L: morning
05:28 AM XXCoder: hey
05:30 AM jthornton: morning
05:32 AM Tom_L: 36�F
05:33 AM jthornton: 73f
05:33 AM jthornton: high of 66f today
05:33 AM Tom_L: must be nice :)
05:33 AM jthornton: haven't been outside yet
05:33 AM jthornton: low tonight is 35
05:33 AM Tom_L: yeah that's what we got yesterday
05:34 AM Tom_L: started out nice and went downhill from there
05:34 AM jthornton: yea that's what it's going to do here today
05:35 AM Tom_L: windy as heck last evening
05:35 AM Tom_L: somewhat calmed so far this morning
05:36 AM XXCoder: heh work was at 36f, driving out into road I go though small fog wall, temperate goes all way down to 19f
05:36 AM XXCoder: then recovers to 40f on freeway, gpoes down to 35f at home
05:42 AM _unreal_: jthornton, like 80f here :)
06:44 AM jthornton: and the rain has started
06:46 AM XXCoder: yay water
08:41 AM jthornton: temperature is going down now...
08:46 AM taloot: the good part of the year
08:46 AM taloot: for us in arabia.
08:56 AM * JT-Shop2 makes a thermoforming thingy
09:48 AM as is now known as Guest51938
10:30 AM jym: Morning folks. Blackout day 3 - OH DA HORROR!!!
10:33 AM jym: sensille: 1/16th?! Why?
10:46 AM jthornton: jym, build a fire and take a nap
10:47 AM jthornton: wow the temperature has dropped 16f already this morning
10:51 AM jym: jthornton: I wish! Got a smog the other day, the lighter socket shorted out when they plugged into it, blew the fuse, and I spent half the day ripping out a FSCKING lighter socket, NFC why, but I literally had to destory it to remove it, sucha a PITA, But now that I have my OB2 port back (same fuse), I can now work on the MAF sensor today
10:51 AM jym: ODB2*
10:52 AM jym: jthornton: Ouch =(
10:52 AM jym: jthornton: snow yet?
11:52 AM jthornton: no snow in sight
11:54 AM fdarling: does anyone whether or not LinuxCNC, when working with servos, "slows down" faster axes to match slower axes when using position feedback? In other words, does it try to avoid a following error (positionally) by slowing down the feedrate to match the lowest common denominator?
11:55 AM fdarling: or does the trajectory planner simply issue waypoints and velocities, and if the velocities aren't met, it just gives up with a following error?
12:05 PM jym: jthornton: How are the chickens doing in 16F weather?
12:05 PM jthornton: it's 57f
12:06 PM jthornton: fdarling, no the move is coordinated with the slowest axis AFAIK
12:06 PM fdarling: jthornton: the axis configured to be the slowest, the the axis that ends up being the slowest?
12:07 PM fdarling: jthornton: I am asking these questions because I was explaining to people why servo based systems can be faster, but I want to make sure I'm not telling lies about the ability to course correct on the fly
12:11 PM jthornton: usually x and y can move at the same speed
12:12 PM fdarling: yes, but I am asking if the servo cannot actually keep up, does it throw a following error, or does it determine that the path can still be followed, just at a lower feedrate
12:12 PM fdarling: I am not sure where in the trajectory planner's sourcecode I can look to find my answer
12:12 PM jthornton: if a servo gets far enough out of position you get a following error
12:13 PM fdarling: yes, but does it do it per-axis, or for the combination of axes?
12:13 PM jthornton: servo drives can add more power to keep up, stepper drives can't
12:13 PM fdarling: hypothetically, let's say that you were moving at a 45 degree angle at 100 IPM
12:14 PM jthornton: the drive does what it has to in order to go where it's told to go and that is between the drive and the servo
12:14 PM fdarling: if one axis was only really capable of 80 IPM instead of 100 IPM, would it get behind during the move and then fault out (because it deviated by say more than 0.010")
12:14 PM jthornton: if for some physical reason the motor can't keep up like it it a wall then you get a following error
12:14 PM fdarling: or would everytime it started to deviate by 0.001", would it slow the other axis down to at least keep the 45 degree angle consistent, in other words to "stay on the beeline" of the path even though it's having to slow down to do so
12:15 PM jthornton: if it falls behind what the drive tells it to do then there is a following error
12:15 PM fdarling: instead of sort of drifting like a racecar and allowing that to go on for too long
12:15 PM jthornton: the drive don't know about each other
12:15 PM jthornton: servo drives
12:15 PM fdarling: I am talking about closed loop servo control using LinuxCNC
12:15 PM fdarling: like velocity mode servos
12:15 PM jthornton: so am I
12:15 PM fdarling: so the servo drives don't know about each other, but LinuxCNC has all the information presented to it
12:16 PM fdarling: from the encoder feedback signals
12:16 PM jthornton: linuxcnc tells the drive to move and reads the encoder if it gets too far out then the machine stops with an error
12:16 PM fdarling: yes, I get that
12:16 PM fdarling: but I am asking whether or not LinuxCNC is intelligent enough to attempt to prevent the following error by slowing down and course correcting
12:16 PM jthornton: will linuxcnc slow down because an axis can't keep up? no
12:17 PM jthornton: no
12:17 PM fdarling: instead of just letting the error on one axis get bigger and bigger while the other axis is succeeding in keeping up
12:17 PM fdarling: hmm, disappointing
12:17 PM jthornton: it just waves it's hand and says you have a following error and stops
12:18 PM jthornton: well the integrator should not allow an axis to go faster than it can that's just poor planning
12:18 PM fdarling: for me, the position should take priority over the planned feedrate, so it should course correct and slow down, and you could have two different error thresholds -- positional error (even when slowing down), and what is an acceptable "slowdown ratio" before it considers it a fault
12:18 PM fdarling: for example if you had it set to 0.9 (90%) then it would slow axes down by 10% to stay with the slower axes, but beyond that it would error down
12:18 PM * jthornton heads to the shop
12:18 PM fdarling: *error out
12:19 PM jthornton: there is two following errors, feed and rapid and they can be set differently
12:19 PM fdarling: jthornton: yes
12:19 PM fdarling: jthornton: again, that is forward planning and not really analyzing feedback to attempt to fix a problem before it causes a part to be mis-cut
12:19 PM SpeedEvil: If you have an actual machine, setting comedically off limits and cutting circles might be faster than trying to read the code
12:20 PM fdarling: if you are making small motions in rapid succession to cut out a part, and an axis starts to get behind, the safer thing to do would be to slow down the other axes to match it rather than allow the cart to be cut out with an incorrect contour
12:20 PM fdarling: an example would be if a leadscrew had a rough spot on it where the axis slowed down unexpectedly, the other good axes could potentially react fast enough to compensate and maintain the contour
12:21 PM fdarling: I know I might be speaking of hypothetical situations here, but I am trying to see how the algorithm works
12:21 PM fdarling: if it doesn't work the way I am saying, I might be interested in trying to implement that in the future and make test cases where an axis is artificially slowed down from the commanded velocity to test it
12:22 PM JT-Shop: the smart thing to do is to not configure an axis max speed faster than it can go
12:22 PM fdarling: yes I know
12:22 PM fdarling: but it's like telling people forget about backlash compensation, just fix your mechanics
12:22 PM fdarling: there are situations where it's relevant, like laser cutting -- no forces on the cutter
12:23 PM fdarling: what I am talking about is differentiating per-axis following errors, and the overall following error
12:23 PM JT-Shop: yea backlash compensation is evil
12:23 PM fdarling: I don't think it is evil :-P
12:24 PM fdarling: I think it's not ideal, but it can be extremely useful
12:24 PM fdarling: imagine if you have a dual Y motor gantry and you didn't support backing off the limit switches by different amounts to straighten out the gantry (due to them homing to different distances, not because the gantry was originally twisted)
12:25 PM fdarling: one might be able to simply say, "physically adjust your home switches to match each other!" but really a software solution would work too and be a lot easier to calibrate
12:26 PM fdarling: backlash compensation works with repeatability if you don't have too much "inertia" with an axis to where it can get thrown forward (instead of being pushed), and if there aren't cutting forces strong enough to allow the axis to pull itself forward either
12:26 PM fdarling: the other issue is that when switching directions the axis (or axes) have to make small fast motions to take up the slop, and if other axes aren't at zero speed, that could be an issue...
12:27 PM fdarling: the time it takes to make the quick motion isn't 0, but if other axes are in motion at the time, then it means there will be a slight undesired artifact in the real world position
12:27 PM fdarling: that isn't easily solved, if it can be solved at all, and for that I agree that backlash compensation isn't ideal
12:28 PM fdarling: however if you are working with a drilling machine for instance, it's always doing full stops in X/Y before drilling in Z, and backlash compensation is 100% useful
12:28 PM fdarling: what are your thoughts?
12:29 PM JT-Shop: gantry homing works in 2.8
12:34 PM JT-Shop: compensation won't do much good in a drilling machine the forces of drilling will make it drift around
12:38 PM fdarling: the forces of drilling might not be great enough, and also theoretically there could be brakes that would be enabled (not on the screws, but for example on the ways)
02:20 PM Loetmichel: *giggle* just played a state lottery online... "why do i have a login there? And why is there still 36 eur on that account?" *search the gaming history* "ah, apparently i played that lottery in 2015... and the sole ticket i did back then was a "4 out of 6" correct winner..." never bothered to check if i won something... wouldnt that be funny if that would have been a jackpot ticket?
03:02 PM bjorkintosh: hah. hah.
03:03 PM gloops: ??
03:18 PM CaptHindsight: I yustah run XY at many many m/sec with a lser and galvos
03:19 PM JT-Shop: wow I just did an automatic take off and landing
03:20 PM CaptHindsight: JT-Shop: whatcha werkin on?
03:28 PM JT-Shop: figuring out my DJI Phantom 3
03:28 PM JT-Shop: while I wait for the tablet so the screen is bigger
04:03 PM Deejay: gn8
04:04 PM JT-Shop: night
05:48 PM jym: POWER IS BACK
05:49 PM SpeedEvil: yay
05:53 PM jym: No doubt =)
05:54 PM Tom_L: yippie!
05:55 PM Tom_L: is the whole of kalifornia under smog alert due to all the generators?
05:56 PM jym: I'm gonna do laundry before they figure out they fucked up! lol
05:56 PM jym: Tom_L It is?
05:57 PM Tom_L: or is that just an every day thing?
05:57 PM jym: In SoCal it is
06:33 PM pcw_mesa: CaptHindsight: got the xy100 module driver basically working, still need the accel scaling but I think Ive figured that out
07:57 PM SpeedEvil: https://www.youtube.com/watch?v=xrZiY3_HC5k - not strictly on topic
08:10 PM Kevin`_ is now known as Kevin`
09:33 PM _unreal_: Tom_L, shhhhh or they will pass a law to ban there umpteenth right to bare generators
10:08 PM CaptHindsight: pcw_home: interesting, been under the weather here so not much has happened other than RTAI getting fixed
10:09 PM CaptHindsight: have to finish the hazzy GUI and python to HAL for DLP/LCD printing
11:21 PM drdoc: _unreal_: did you get your lcnc install worked out?
11:22 PM _unreal_: HAAAA
11:22 PM _unreal_: zero time to do anything
11:22 PM drdoc: ah
11:22 PM _unreal_: I dont get time to do anything till tomrrow
11:22 PM _unreal_: I only got home at 10pm ish after dropping my daughter off
11:23 PM _unreal_: right now I'm drawing a CAD drawing
11:23 PM _unreal_: for work
11:23 PM _unreal_: so I can cut the part on my cnc for wokr
11:23 PM drdoc: both Stretch and Buster isos will boot UEFI native
11:23 PM drdoc: so worst case, just install Debian and add LCNC
11:25 PM drdoc: I went to Fry's today to pick up a DB25 cable. Storewide, maybe one out of four items were stocked. 7-8 cars in the parking lot.
11:27 PM CaptHindsight: was at microcenter earlier and also Frys
11:27 PM CaptHindsight: same number of cars
11:27 PM CaptHindsight: Microcenter had everything I was looking for
11:28 PM drdoc: Were Fry's shelves stocked?
11:28 PM CaptHindsight: low but
11:28 PM CaptHindsight: they seem to carry less than they used to
11:29 PM drdoc: I want exaggerating; 3/4 if they're shelves were empty
11:29 PM drdoc: of their*
11:29 PM drdoc: can you tell I'm IRCing from my phone?
11:29 PM drdoc: :/
11:29 PM CaptHindsight: the one here has more space than they need
11:30 PM CaptHindsight: everywhere seems to have far less customers except around the holidays
11:30 PM drdoc: The store employees claim that they're having to renegotiate deals with all their suppliers
11:30 PM drdoc: "the tariffs"
11:31 PM CaptHindsight: heh
11:31 PM drdoc: and that they *can't* restock
11:31 PM CaptHindsight: lets get china to make everything and ship it here for cheap
11:31 PM drdoc: there Internet is pretty sure they're going out of business
11:31 PM drdoc: ;-)
11:32 PM CaptHindsight: 30 years later, hey what happened
11:32 PM drdoc: yup
11:32 PM drdoc: and it's all on us, the consumer
11:32 PM drdoc: WE made it this way
11:32 PM CaptHindsight: well there was abit of incentive from up top but yeah, well trained consumers
11:33 PM CaptHindsight: consumers allowed themselves to be well trained
11:33 PM drdoc: racing healing to the bottom of the quality barrel
11:33 PM drdoc: headlong
11:34 PM drdoc: I have a friend who keeps buying electronics, like Arduino and pin headers and stuff, from banggood,
11:35 PM drdoc: as well as bearings and small parts for his printers
11:35 PM CaptHindsight: https://books.google.com/books/about/Born_to_Shop.html?id=MHfmPgAACAAJ&source=kp_book_description
11:35 PM drdoc: whatever he saves, here spends back chilling the parts that are wildly out of spec
11:35 PM drdoc: culling
11:36 PM drdoc: not only I can't type, I can't proofread either
11:36 PM CaptHindsight: sorry wrong link
11:36 PM CaptHindsight: https://www.amazon.com/Born-Buy-Commercialized-Consumer-Culture/dp/0684870568
11:36 PM drdoc: and he stays pissed off about getting "cheated"
11:37 PM CaptHindsight: https://www.amazon.com/Overspent-American-Want-What-Dont/dp/0060977582/ref=pd_cp_14_1/142-6943608-9145317?_encoding=UTF8&pd_rd_i=0060977582&pd_rd_r=f4a4fc81-ac59-4f23-8974-8a1545c53b4d&pd_rd_w=hPPrY&pd_rd_wg=irzhi&pf_rd_p=0e5324e1-c848-4872-bbd5-5be6baedf80e&pf_rd_r=3RAFRAWDXQB68XDWP73H&psc=1&refRID=3RAFRAWDXQB68XDWP73H
11:37 PM drdoc: jeebus, man!
11:37 PM CaptHindsight: the author: Schor is a professor of sociology at Boston College, a former member of the Harvard economics department, and a Guggenheim Fellowship recipient.
11:38 PM drdoc: cool
11:38 PM CaptHindsight: you might like skimming them
11:38 PM CaptHindsight: she gets into the science behind training consumers from early age
11:38 PM drdoc: I've actually read reviews of Born to Buy
11:38 PM CaptHindsight: it's not by accident
11:39 PM drdoc: I know a linguist, a PhD researcher at Princeton
11:39 PM drdoc: he closing that our very language had tilted into consumerism
11:40 PM drdoc: and status according to bling
11:40 PM drdoc: he claims
11:41 PM drdoc: I've been building one-off product all my life
11:41 PM CaptHindsight: I'm off for the night
11:41 PM drdoc: this cheaper is better offends me
11:41 PM drdoc: later
11:41 PM CaptHindsight: hasta banana
11:52 PM _unreal_: Just finished drawing this https://drive.google.com/open?id=1kHcbgsNlaQKISq4RqY4vy-0jbfZZ5LH0 https://drive.google.com/open?id=1LRopVdVZw-7N887gpE1Ze_dOfnC-rK7u
11:53 PM _unreal_: That if your wondering is an Anchor holder
11:55 PM _unreal_: the 3/4" dowel goes into the holder of those 2 images a linked https://www.westmarine.com/buy/tie-down-engineering--16lb-standard-fluke-anchor--P005_153_001_500
11:57 PM _unreal_: I have not thought about fry's in a LONG ass time
11:58 PM _unreal_: ? I wonder if florida has one