#linuxcnc Logs

Jun 03 2019

02:53 AM veek: could someone suggest a book on understanding steel - just enough to buy it.. all the grades.. a bit on manufacturing.. cross sections
02:58 AM XXCoder: i wonder if machinist handbook has that section
05:00 AM Tom_L: morning
05:20 AM jthornton: morning
05:21 AM Tom_L: looking at devel, seems they're getting the kernel stuff better
05:21 AM Tom_L: i took notes on kernel options he pasted there fwiw...
05:23 AM jthornton: on the irc or the mailing list
05:23 AM Tom_L: here
05:23 AM jthornton: ah, just waking up
05:25 AM Tom_L: i wish i had a macro to optimize parts on a plate for maximum material useage
05:25 AM Tom_L: taken cutter width etc into account
05:26 AM jthornton: that gets real complicated sometimes
05:27 AM Tom_L: laen did that for his board making business
05:27 AM Tom_L: not sure how efficient it was
05:44 AM jthornton: if the parts are all the same shape you could use the way I do for plasma parts and use rotation and even row offset to nest them
08:27 AM perry_j1987: Tom_L cabinet shops use software like that for cutting max parts out of plywood etc.
08:27 AM perry_j1987: Tom_L might be able to adapt something from that software prob just need dxf of the outlines to cut etc
08:36 AM Deejay: hi
08:37 AM JT-Shop: hi
08:57 AM perry_j1987: squaring up the drillpress table is on the docket for today it seems
09:16 AM perry_j1987: youch 2.5ish mm out front to back in 100mm or so
09:16 AM perry_j1987: dead on side to side though heh
09:19 AM perry_j1987: i see the issue
09:20 AM perry_j1987: they put the table clamp at the top of the bracket instead of centered in the middle of it
09:20 AM perry_j1987: so the split on back of column is wider at bottom of the bracket than the top
09:20 AM perry_j1987: thereby tilting the bed down in the front
10:49 AM CaptHindsight: Tom_L: he's still trying different kernel configs with preempt_rt....
10:51 AM CaptHindsight: the big change is when you isolate real time from non-real time on cores
11:19 AM mozmck: Tom_L: linuxcnc already puts the rt tasks on the last cpu - that's what find_rt_cpu_number() does is find the last cpu number. mem just deleted that and hardcoded it to 3 - which will only work on a quad-core cpu
11:28 AM CaptHindsight: mozmck: there more that is going on, I was just simplifying for Tom_L
11:29 AM CaptHindsight: it will also work on 2-32cores
11:29 AM CaptHindsight: scheduling on one core and only real time tasks on another
11:30 AM CaptHindsight: he still working it out
11:30 AM CaptHindsight: he is
11:32 AM CaptHindsight: "kernel timekeeping and linuxcnc should be on different cores ideally"
11:33 AM mozmck: CaptHindsight: I think he misunderstood what find_rt_cpu_number() does - I'm pretty sure it puts rt on the last cpu.
11:34 AM mozmck: isolcpus I use all the time, but you only need to isolate the core with the rt stuff along with any cores that share cache with it.
11:35 AM CaptHindsight: I'm sorry I brought it up
11:35 AM mozmck: I have a script which jepler and I made to find all that out and set up isolcpus on the kernel command line. I'll try and dig it up later - I have to run for now.
11:35 AM CaptHindsight: I know, we worked on RTAI for 10 years
11:36 AM CaptHindsight: preempt_rt behaves differently
11:37 AM mozmck: CaptHindsight: I was basing what I said on memfrob saying this: "So for best latency the real-time thread should run on the last core instead of the first. How would I change this?" Unless I am totally mixed up, real-time already runs on the last core.
11:39 AM CaptHindsight: he's running custom kernels with all sorts of different options set
11:41 AM CaptHindsight: he's in the middle of it all
12:01 PM Tom_L: no, i get what's going on.. i was just saving the config changes as a reference for the odd time i might build a kernel
12:03 PM Tom_L: keeping the cpu0 separate i understand but that's not part of a kernel anyway is it? that's in the linuxcnc codebase. at least that's my understanding of it
12:03 PM Tom_L: perry_j1987, if i had to batch parts alot i'd look into it more. i was just complaining :)
12:04 PM perry_j1987: heh
12:05 PM perry_j1987: two taps both labeled the same... .35mm off from eachother
12:06 PM Tom_L: is the H value the same?
12:06 PM Tom_L: H1 = Basic to Basic plus .0005 H2 = Basic plus .0005 to Basic plus .0010 H3 = Basic plus .0010 to Basic plus .0015 H4 = Basic plus .0015 to Basic plus .0020 H5 = Basic plus .0020 to Basic plus .0025 H6 = Basic plus .0025 to Basic
12:07 PM Tom_L: taps are not all created equal
12:08 PM perry_j1987: these are from two different tap sets from harbor freight heh
12:08 PM perry_j1987: one of them wont fit in the correct size tap drill hole the other works fine
12:09 PM perry_j1987: heh
12:09 PM Tom_L: the only thing harbor freight knows about H value is it's the first letter of their name
12:09 PM perry_j1987: good times
12:09 PM perry_j1987: lol
12:18 PM perry_j1987: why does the toolpost wrench have to fall behind unaccessable places
12:18 PM Tom_L: because it's monday
12:20 PM MarkusBec: pcw_home: Tom_L is it possible to use a 7I85S as a rs485 interface like the 7i96
12:21 PM Tom_L: it's got a rs422 output... isn't the difference just distance?
12:22 PM Tom_L: i'm not sure
12:23 PM Tom_L: The RS485 Standard is similar to the RS422 standard upon which it is based. The main difference is that up to 32 transmitter receiver pairs may be present on the line at one time. A 120 Ohm resistor should be used to terminate either end of the main line.
12:23 PM MarkusBec: rs422 is rx and tx on 2 lines
12:24 PM MarkusBec: rs485 is a and b (rx/tx) on 1 line
12:25 PM MarkusBec: with the 7i96 you can just connect rx and tx of the r422 interface to get rs485
12:25 PM MarkusBec: but manual dont mention if this is possible with the 7i85
12:25 PM Tom_L: yeah it may be a different chip
12:25 PM Tom_L: better ask da boss... pcw_home!
12:26 PM MarkusBec: add line pairs :)
12:28 PM pcw_home: 7I96 supports RS-485 (driver disable) 7I85 does not
12:28 PM Tom_L: and there you have it...
12:30 PM pcw_home: for low speeds you could use series resistors in the TX lines (say 100 ohm each) and get way without disabling TX
12:31 PM MarkusBec: speed is 9600 boud and 2m of cable
12:31 PM pcw_home: the series resistors would probably be OK then
12:34 PM pcw_home: the issue on the 7I85/7I85S is that all 17 I/O pins are used, so none free for TXEN
12:35 PM MarkusBec: yep
12:35 PM MarkusBec: sometimes you have somethink like disable tx if tx is idle
12:35 PM MarkusBec: g
12:36 PM pcw_home: yeah some adapters do that
12:36 PM pcw_home: basically a re-triggerable oneshot
01:25 PM perry_j1987: do you guys have a preference on safety glasses
01:26 PM perry_j1987: my only pair left are ilfitting and end up hurting after little bit
01:35 PM Tom_L: only ones i have are the free ones i got when i went on a tour at the vo'tech for my kids
01:36 PM Tom_L: bearkat brand
01:44 PM perry_j1987: great now im logged asking about safety glasses
01:44 PM perry_j1987: :P
01:46 PM FinboySlick: perry_j1987: Safety glasses are used in when handling dangerous chemicals too... You're on every terrorist watch list now.
01:46 PM perry_j1987: haha
01:47 PM CaptHindsight: https://images-na.ssl-images-amazon.com/images/I/415SOcCjcpL._SX425_.jpg keeps the splash out of your eyes
01:47 PM CaptHindsight: still gets all over your face though
01:48 PM CaptHindsight: perry_j1987: are you near a Home Depot?
01:48 PM perry_j1987: ya
01:49 PM CaptHindsight: they have all sorts of styles
01:49 PM CaptHindsight: wrap around, side shields, anti-fog, x-ray etc
01:49 PM CaptHindsight: https://www.homedepot.com/b/Safety-Equipment-Protective-Eyewear-Safety-Glasses-Sunglasses/N-5yc1vZc1xt
01:50 PM CaptHindsight: ANSI, OSHA, CSA, BVD etc certified
01:51 PM CaptHindsight: if you're accident prone I'd suggest Fire Resistant as well
01:51 PM perry_j1987: so funny i never even thought to look at home depot heh
01:54 PM CaptHindsight: yeah, everyone got so safety conscious in the past 20+ years
01:55 PM CaptHindsight: in the old days we'd build an entire house without ever touching safety glasses
01:55 PM CaptHindsight: now you can't start up a power tool without them on
01:57 PM perry_j1987: i'd rather loose a finger than an eye :P
02:00 PM perry_j1987: always fun loosing screw into the chip bin
02:25 PM elmo40: if it is aluminum chips and a steel screw, easy enough with a magnet.
02:52 PM perry_j1987: that'd be bad if it were alum screw in steel chips :P
03:04 PM Deejay: gn8
03:19 PM tumdedum_ is now known as tumdedum
03:21 PM CaptHindsight: so huzgot the best deal now on tapezoidal lead screws?
03:22 PM elmo40: why trapezoidal? go straight to ball screw. :-)
03:22 PM elmo40: more accurate and less friction.
03:22 PM CaptHindsight: slow moving application
03:23 PM CaptHindsight: mm/minute
03:24 PM CaptHindsight: but with the discount balls crew sources the prices might be very similar
03:25 PM CaptHindsight: http://www.automation-overstock.com/categories.asp?cat=11
03:26 PM CaptHindsight: http://www.automation-overstock.com/proddetail.asp?prod=BSFU1605%2D0350%2DFS
03:26 PM CaptHindsight: 16mm Ø x 5mm lead x 350mm length rolled ball screw assembly $88.65
03:28 PM CaptHindsight: https://www.roton.com/family/trapezoidal-right-screws-and-nuts-7061118/
03:39 PM CaptHindsight: http://www.automation4less.com/store/proddetail.asp?prod=BSFU1605%2D0350%2DFS same price as above, might be the same place but different website
03:42 PM * perry_j1987 wants an extremely expensive cnc machine :P
03:46 PM CaptHindsight: perry_j1987: I'll sell you mine for $10 million :)
03:47 PM XXCoder: just expensive? not fancy? ill sell my cnc royter for one million usd
03:47 PM perry_j1987: haha
03:48 PM perry_j1987: for 10million it better keep me from making bone headed mistakes too :P
03:48 PM CaptHindsight: I'll come and operate it for free for 1 year
03:49 PM CaptHindsight: I'll even throw in a Cambodian family for 10 years
03:51 PM perry_j1987: heh
03:51 PM elmo40: O_o
03:54 PM elmo40: https://www.roton.com/family/recirculating-right-screws-and-nuts-7059321/
03:54 PM elmo40: difference of Price: $16.91 /ft. VS. $14.79 /ft.
03:56 PM elmo40: CaptHindsight, even the metric brass nut costs more than the recirculating ball!
03:56 PM elmo40: that makes no sense...
03:57 PM CaptHindsight: I didn't see any accuracy specs on the roton.com site
03:58 PM CaptHindsight: C? grade, just trust us :)
03:58 PM elmo40: sure... why not? what's the worst that can happen...
04:00 PM elmo40: besides, on ~14" how inaccurate can it get?
04:00 PM elmo40: C5 is the worst. At 350mm you'll get ~25microns
04:00 PM elmo40: C1 is best, with only 5micron inaccuracy within 350mm. For rolled ball screws.
04:01 PM elmo40: you don't want to look into C7 or C10... but who knows, they could also be one of them!
04:02 PM CaptHindsight: https://postimg.cc/K3LNV5tp
04:02 PM elmo40: ok, i'm all messed up. was looking at the wrong chart...
04:02 PM CaptHindsight: I only need ~300 of travel, guess i could cut this down :)
04:03 PM elmo40: sure. why not? chop saws are great for that.
04:03 PM elmo40: According to THK, rolled ball screws are all shit... C7, C8 and C10.
04:04 PM CaptHindsight: 40mm dia 5mm pitch
04:04 PM elmo40: +\- 50micron, 100 and 210micron, relative.
04:04 PM CaptHindsight: yeah was just checking stock at THK
04:04 PM elmo40: 40mm diameter? That is one strong ball screw!!
04:04 PM elmo40: Ground ball screws are from C0 to C5, with rejects in the C7 range.
04:05 PM CaptHindsight: considered using a linear servo for this application just for the accuracy
04:06 PM CaptHindsight: but it's a Z and then I have to have a brake
05:18 PM perry_j1987: any reason i should talk myself out of a vertical mill slide for my mini lathe?
05:18 PM Tom_shop: if you need one
05:18 PM Tom_shop: by all means...
05:19 PM perry_j1987: im thinking it might have a better chance than this 6040 :P
05:22 PM jthornton: 6040 pretty shaky?
05:23 PM perry_j1987: its pretty nice as far as i can tell
05:23 PM perry_j1987: seems real solid
05:23 PM perry_j1987: the spindle has no torque at low rpm though
05:28 PM jthornton: that can be fixed with a belt driven spindle
05:28 PM jthornton: prob easier than fitting a mill slide to a lathe
05:29 PM perry_j1987: treadmill motor driven spindle or something?
05:32 PM JT-Shop: they are pretty noisy lol, motor selection depends on the size of the spindle and machine
05:33 PM JT-Shop: for real small an outrigger motor with a esc might work
05:57 PM perry_j1987: ah ya thats right
05:57 PM perry_j1987: i seen someone did that on their mini lathe too
06:38 PM jthornton: XXCoder: you awake?
06:40 PM _unreal_: tonights project get the step controller setup and working
06:40 PM _unreal_: different cnc machine not using turbocnc :(
06:40 PM _unreal_: err linuxcnc
06:40 PM _unreal_: using... ahh
06:41 PM jthornton: after you no you first no you first mother would never permit that ok after you
07:48 PM skunkworks: http://electronicsam.com/images/greenmachine/IMG_20190603_183551.jpg
07:58 PM skunkworks: That chuck has a 5 inch through hole...
08:31 PM skunkworks: 15 inch
09:07 PM elmo40: skunkworks, appears someone drilled the hole off center... ;-)
09:08 PM elmo40: perry_j1987, i've used a treadmill motor to power a lathe before. made it easier using the original wiring harness.
09:09 PM elmo40: it was a manual lathe with variable speed chuck :-)
09:09 PM perry_j1987: ya i was thinking that might be a decent upgrade for this mini lathe
09:09 PM elmo40: probably find one for free on the side of the road, or kijiji
09:15 PM perry_j1987: aye
09:15 PM elmo40: most are ~1.5Hp. If you are lucky you can find 2.0 to 2.5Hp
09:16 PM elmo40: i collected a set within the past 5 years. broke the control board (dropped a brick on it...) now i need to see how it operates to figure out how to LinuxCNC it
09:17 PM elmo40: i figure they have a magnetic sensor for the speed. can't be too complicated. just a pulse counter.
09:18 PM elmo40: could be an opto-coupler with a little tab cutting the beam at each rev.
09:18 PM elmo40: how do i use that kind of signal to give feedback to LinuxCNC?
09:21 PM pcw_home: you can use a single pulse for threading (as long as the speed remains fairly constant)
09:38 PM skunkworks: elmo40: that is a welded up chamber - turned on centers and a way to pressurize it to make sure it is air tight
09:42 PM perry_j1987: i threw a rotary encoder on my mini lathe and a inductive probe for single index pulse