#linuxcnc Logs

Oct 05 2022

01:36 AM CaptHindsight[m]: 1:00 am and I found some PCIe extenders in a bin here, looks like some ebay specials, but they are only x1's and they work \0/
01:45 AM CaptHindsight[m]: only a few inches long
01:45 AM SkoFyx: Besides sleeping, I can't think of a better thing to be doing at 1am...
01:50 AM CaptHindsight[m]: late night is usually quiet around here
04:17 AM SkoFyx: So only calculating D when the encoder (and command position) changes, does in fact reduce servo whine/jitter. Unfortunately, on my system, this introduces an unacceptable random/sporadic twitching in the servos. One of the drawbacks of running PID at 10kHz: everything gets jumpy and easily excitable.
04:24 AM Tom_L: morning
04:52 AM JT-Cave: morning
05:07 AM JT-Cave: 41°F
05:24 AM Deejay: hi
06:55 AM unterhausen: If I see a video about pid where it's just a guy with a pencil case bs-ing, I'm not sure it's really a good video about pid
07:00 AM unterhausen: PID doesn't work with all systems. It usually works pretty well with simple systems. Might need a lead lag instead
07:00 AM unterhausen: instead of D
07:29 AM -!- #linuxcnc mode set to +v by ChanServ
07:43 AM JT-Cave: wow 39°F to 79°F today
07:59 AM pcw---home: SkoFyx: (should you come back later) encoder "whine" is expected unless you have a high resolution encoder or good velocity estimation
07:59 AM jpa-: the accurate time stamping on mesa cards & connecting the velocity estimate pin really helped for my system
08:00 AM jpa-: i guess even without timestamping, estimating the velocity based on frequency of encoder pulses could help smooth the D response
08:02 AM pcw---home: its especially severe at higher sample rates where you can get multiple samples of 0 velocity (no count change) and then bang
08:04 AM pcw---home: (assuming no velocity estimation at all)
08:04 AM jpa-: considering that even the software based HAL encoder component does velocity estimation, maybe SkyFox just has the velocity pin not connected between encoder & PID
08:10 AM pcw---home: maybe (or he's not using LinuxCNC at all)
08:59 AM unterhausen: skyfox said he's using his own software.
08:59 AM perry_j1987: moning
08:59 AM perry_j1987: morning
09:00 AM unterhausen: moaning
09:00 AM perry_j1987: that was a fun trip!
09:00 AM cpresser: Has anyone in here used https://github.com/Peter-van-Tol/LiteX-CNC ? Do the stepgens work?
09:03 AM unterhausen: isn't there a known-working lcnc driver for colorlite boards?
09:11 AM pcw---home: Yes, but you would probably need to look it up on the LinuxCNC forum
09:55 AM perry_j1987: pcw---home you play much with SBC testing with the mesa boards?
09:56 AM pcw---home: Only a RPI a while ago
10:06 AM perry_j1987: what did you think of the experience
10:27 AM pcw---home: I started with the RPI3, it was really too slow. The RPI4 is OK, but more trouble in general than X86
10:29 AM pcw---home: Getting fancy GUI stuff to work was a challenge because the libraries lag
10:45 AM CaptHindsight[m]: cpresser: https://forum.linuxcnc.org/27-driver-boards/44422-colorcnc-colorlight-5a-75e-5a-75b-as-fpga-controller-boardis the main thread, and yes they have stepgens working
10:46 AM perry_j1987: think im just going to mount this mini itx board inside the lathe control cabinet and grab a pico psu for it
10:49 AM CaptHindsight[m]: "The RPI4 is OK, but more trouble in general than X86" and now we get similar head aches with RISC-V as they get lowcost and fast enough for LCNC
10:51 AM mrec: did anyone try to adjust the stock in a chuck via camera?
10:52 AM CaptHindsight[m]: it looks like we will have RISC-V boards with 1-2GB of RAM for under $30
10:54 AM CaptHindsight[m]: mrec: you can do it now with openCV
10:57 AM CaptHindsight[m]: mrec: https://forum.linuxcnc.org/38-general-linuxcnc-questions/45583-opencv-with-linuxcnc-2-8-0
10:59 AM CaptHindsight[m]: mrec: you can have you camera at a set Z and edge detect using crosshairs to adjust the XY or..
11:00 AM mrec: I think about writing my own program for it to match my application
11:01 AM mrec: I see I have a little bit runout on my spur gears, still good enough to be used but I want it as good as possible
11:01 AM CaptHindsight[m]: have a set Z (since Z will effect accuracy) and you could ballpark the stock in the chuck, then move XY to find the corner or edge using the crosshairs and then have a known offset
11:01 AM mrec: nono, my applicatin is a chuck (A axis)
11:01 AM mrec: so I'd capture it in front of it
11:01 AM CaptHindsight[m]: similar for a lathe
11:03 AM mrec: yes I also need it on my lathe
11:03 AM mrec: I have one annoying gear with an 1.5mm center pin
11:05 AM CaptHindsight[m]: https://i.imgur.com/IFdlrj3.jpg sobel edge detect filter, ~50um gap
11:06 AM CaptHindsight[m]: https://i.imgur.com/QU0ka4r.jpg
11:07 AM CaptHindsight[m]: the calipers were dirty (dust)
11:07 AM mrec: which camera do you use with it?
11:07 AM mrec: aliexpress.com/item/1005004310905185.html I have just ordered this one
11:07 AM CaptHindsight[m]: this was with a $30 USB microscope
11:07 AM mrec: I still have another black white usb camera here which I'm going to test first
11:11 AM mrec: thanks for that linuxcnc forum post!
11:11 AM CaptHindsight[m]: the better the camera and lens the faster you can analyze the image
11:12 AM CaptHindsight[m]: I've used some really cheap USB cams and they all work but they do so much frame averaging that you have the make really slow moves to get accurate results
11:13 AM mrec: I looked at some mjpeg cameras that one was fast enough for SMD component detection in the past
11:14 AM CaptHindsight[m]: something like that should allow you to detect edges in one frame vs several
11:15 AM CaptHindsight[m]: "real" camera and lens on a fraction of a second, cheap webcam has similar results but in maybe a couple of seconds
11:16 AM mrec: I think it's very much up to how the chipset is programmed, 30 fps should be fine it's realtime mjpeg so each frame can be decoded immediately
11:19 AM CaptHindsight[m]: speed might only matter if you are automating the process
11:19 AM CaptHindsight[m]: using the cheap webcams you mostly get a blur when positioning,you have to stop between moves to get any accuracy
11:20 AM CaptHindsight[m]: real cam can get accurate results on the fly
11:22 AM CaptHindsight[m]: https://i.imgur.com/ssEuxB7.jpg
11:23 AM CaptHindsight[m]: you can scan a known line width or gap at a set Z position, use that to calibrate the distance per pixel
11:24 AM mrec: your microscope is lacking a diffuser
11:24 AM CaptHindsight[m]: then using the same Z distance from the part to lens
11:24 AM CaptHindsight[m]: $30 USB "microscope"
11:25 AM mrec: wrapping some chip tray foil around the light also helps to improve the lightning
11:28 AM CaptHindsight[m]: I saved that as an example pic
11:28 AM CaptHindsight[m]: yeah lighting matters
11:28 AM CaptHindsight[m]: https://www.amazon.com/Vividia-Handheld-Endoscope-Microscope-Diameter/dp/B005P40OXY these have led's built into the end
11:29 AM CaptHindsight[m]: you can see all four in the pic :)
11:40 AM mrec: next on my list will be a small band saw
11:55 AM roguish[m]: good morning.... has unreal been around? he lives in Florida. wonder if he's still alive.....
12:27 PM Tom_L: hasn't said much since he wished his x was in the storm
12:28 PM roycroft: probably because he said too much then
12:28 PM roycroft: but he lives on the other side of the peninsula, and was not directly affected by the hurricane
12:29 PM Tom_L: i'm sure he'll be back to tell us about all the salvaged equipment he got from the storm
12:29 PM roycroft: yeah, at some point
12:32 PM Tom_L: i looked at the portable bandsaws and they just didn't have enough reach for me
12:36 PM Tom_L: kid is lovin his new job. they put him on 2 brand new Starrag 6 axis mills
12:38 PM Tom_L: had him on that cincinatti 3 head for a couple days then move him once they figured out what he could do
12:46 PM roycroft: i don't have a portable band saw, but if i ever start doing field work again i'll probably get one
12:46 PM roycroft: i used to lug my 4x6 around with me when i was doing installs at central offices and fiber huts
12:46 PM roycroft: and that was always a lot of work
12:47 PM roycroft: but now i have a 7x12, and it's not leaving my shop
01:31 PM JT-Shop: Tom_L, does your kid program the 6 axis mill or run it?
01:32 PM pere: perhaps they clean it?
01:56 PM Rowow[m] is now known as Rowow5457[m]
02:02 PM xxcoder[m]: 6 axis programming probably would need in least journeyman
02:17 PM JT-Shop2: 18.5° bevel on 12.5' PT 2x6 done and real easy too
02:19 PM Tom_L: runs it
02:19 PM Tom_L: they use mastercam
02:19 PM Tom_L: he's trained with catia and mastercam but they have several programmers already
02:21 PM roguish[m]: Mastercam has some pretty good simulation. Hopefully it's used. crashing 6 axes is way bad.
02:22 PM roguish[m]: and expensive
02:38 PM JT-Shop: hmm the contrinex prox switches I got from AD are made in the usa
02:41 PM JT-Shop: https://www.starrag.com/en-us/machine/stc-1600170/183
02:42 PM xxcoder[m]: tom not surpised. I have never setp 5 axis, and 4 axis only simple kind (like rod on rotatary)
03:06 PM Tom_L: i'm not sure they have the 6 axis shown yet
03:06 PM Tom_L: but you can see the head in the pics jt linked
03:26 PM Tom_L: JT-Shop, 0.7.3 installs over 0.7.2?
03:26 PM Tom_L: didn't ask to reinstall
03:30 PM Tom_L: are you keeping the old debs on the site for a reason?
03:36 PM Tom_L: i guess it doesn't hurt anything having them
03:42 PM JT-Shop: Tom_L, yes if you install 0.7.3 and have 0.7.2 installed then 0.7.2 gets replaced
03:43 PM JT-Shop: I keep at least one previous deb in case I foul something up, then the user can back down one to get going
03:43 PM Tom_L: how can i push the deb if it doesn't show up in git gui?
03:43 PM Tom_L: something errored and now it doesn't show
03:43 PM Tom_L: but the file is built and ready to push to my repo
03:43 PM JT-Shop: if you want to try ini3 branch it doesn't write the original ini but creates a test.ini in the same directory
03:44 PM Tom_L: i can later, need to go for a while but trying to push these files
03:44 PM JT-Shop: I assume you did a rescan in that case the file has not changed
03:44 PM Tom_L: it showed up at first and i tried to push and got some sort of error and now it doesn't show up
03:44 PM * JT-Shop it trying to get the 12.5' 2x6 in position for screws
03:45 PM Tom_L: the armhf is on my repo, the arm64 is the one i'm trying to fix
03:47 PM roycroft: there is a track saw on festoolrecon.com right now, btw
03:47 PM roycroft: it's the old model, but i prefer the old model to the new one
03:47 PM roycroft: the old one has a 2.3mm thick blade, while the new one has a 1.8mm thick blade
03:48 PM roycroft: that makes the new one seem more powerful, but it doesn't cut as cleanly as the old one
03:48 PM Tom_L: i'll have to deal with it later i suppose..
04:02 PM JT-Shop: does that track saw use the same track as the new one?
04:02 PM roycroft: yes
04:02 PM roycroft: and it can use the makita or mafell tracks as well
04:03 PM roycroft: but they cost more than the festool ones
04:05 PM JT-Shop: thanks
04:12 PM justache is now known as justHaunted
04:26 PM solarwind: How do they make glass syringes?
04:27 PM JT[m]1: Some reviews complain that the guide rail is different widths but don't say how much
04:27 PM solarwind: Do they grind the ID and OD?
04:29 PM roycroft: there is an adjustment on the saw to make it run smoothly on the track
04:30 PM roycroft: you can snug it up to ensure there is no play
04:31 PM roycroft: and there is a strip you can replace that goes on the cutting side of the rail
04:32 PM roycroft: it is overwidth, so when you do your first cut on a rail it gets trimmed perfectly
04:32 PM JT[m]1: Thanks
04:33 PM roycroft: there are some cheaper track saws, like the grizzly, but they're not very accurate
04:34 PM JT[m]1: Just need to get a couple of rails
04:34 PM roycroft: yeah
04:34 PM JT[m]1: 1400mm?
04:34 PM roycroft: i started out with a couple shorter ones, and connected them together for long rips
04:34 PM roycroft: 1400 would be a good one to start with
04:34 PM roycroft: i just got a 2700mm one this summer
04:35 PM JT[m]1: Nice
04:35 PM roycroft: the old festool connectors did not align the two rails, so i had to get a device to do that
04:35 PM roycroft: but they came out with some new connectors last spring that do a much better job of alinging the rails
04:36 PM roycroft: a lot of folks use a track saw in lieu of a table saw
04:36 PM roycroft: and in a tiny shop, or for job sites, that might be ok
04:36 PM roycroft: i would never want to get rid of my table saw, though
04:36 PM roycroft: but the track saw is so useful for breaking down sheet goods and doing things like rips on long pieces of lumber
04:38 PM roycroft: one really nice thing about that track saw is that it pivots exactly on the guide strip like
04:38 PM roycroft: line
04:38 PM JT[m]1: Been wanting a track saw for a while
04:38 PM roycroft: so if you're doing a bevel cut you can use the same track with the same setup as a 90 degree cut - there is no offset at all
04:38 PM JT[m]1: Nice
04:38 PM roycroft: yeah, that's a huge time saver
04:39 PM JT[m]1: Thats sweet
04:39 PM roycroft: the festool also has a riving knife
04:39 PM roycroft: which many track saws do not
04:41 PM JT[m]1: Wonder if it will arrive by my birthday lol
04:43 PM xxcoder[m]: happy birthday soonish lol
04:43 PM * roycroft is waiting for a router to appear on the recon site
04:43 PM JT[m]1: Thanks
04:44 PM JT[m]1: Starlink arrives tomorrow
04:44 PM roycroft: they must have multiple track saws to sell, because it's still on the site
04:44 PM JT[m]1: I saw that
04:44 PM roycroft: if you're going to rip with it you should get a rip blade
04:45 PM JT[m]1: Ok
04:45 PM roycroft: the 48 tooth blade that comes with it is great for sheet goods, but it's pretty wimpy in hardwood, especially for rippign
04:45 PM roycroft: you would have to take several passes
04:46 PM JT[m]1: Ok
04:46 PM JT[m]1: Good to know
04:46 PM roycroft: the new tracks are really nice in the shorter lengths
04:47 PM * JT[m]1 uploaded an image: (340KiB) < https://libera.ems.host/_matrix/media/r0/download/matrix.org/vOWyfuKKNXgMyFgiVUvQmzUf/20221005_164521_5569708640941153238.jpg >
04:47 PM roycroft: they have some cutouts in the middle of the track where you can insert some holders to stick it up on a wall
04:47 PM roycroft: that turned out nice
04:47 PM roycroft: and it would have been easier with the track saw
04:48 PM roycroft: you'll probably want to get a pair of dovetail clamps for holding the track in place - that can be really helpful at times
04:49 PM roycroft: you don't have to get the festool ones - several other brands will fit - but the cost difference is not that much
04:49 PM JT[m]1: It turned out to be pretty easy with the jig I made
05:07 PM Tom_L: angled for the siding?
05:09 PM JT[m]1: Angled for the roof tin
05:10 PM Tom_L: working on arm64 deb
05:11 PM JT[m]1: Making a slightly unconventional frame for the metal roof
05:11 PM JT[m]1: Working on a glass of cabinet
05:12 PM Tom_L: i think what happened was the version change screwed up it recognizing it as part of the repo
05:12 PM Tom_L: it showed up the first time in git gui but not after that
05:15 PM Tom_L: ok #29 is up
05:17 PM JT[m]1: Ok
06:35 PM roycroft: the track saws are gone
06:36 PM roycroft: now they're back to hawking extractors
06:37 PM roycroft: there's a router on cl right now, which is very rare, but it's the one i already have
08:02 PM -!- #linuxcnc mode set to +v by ChanServ
08:15 PM -!- #linuxcnc mode set to +v by ChanServ
08:15 PM -!- #linuxcnc mode set to +v by ChanServ
10:48 PM CaptHindsight[m]: solarwind: https://www.youtube.com/watch?v=xEfQ_Pj6Pjo
10:49 PM solarwind: CaptainHindsight thanks. THey could have done so much better with the video if they weren't trying to be hip and cool
10:49 PM solarwind: CaptHindsight[m] are you on two accounts
10:49 PM CaptHindsight[m]: only on Matrix right now
10:50 PM CaptHindsight[m]: not sure how the bots handle it all
10:50 PM CaptHindsight[m]: solarwind: I'm building machine like that right now
10:51 PM CaptHindsight[m]: it also add a glass dot to keep the glass tube from rolling
10:54 PM * CaptHindsight[m] uploaded an image: (20KiB) < https://libera.ems.host/_matrix/media/r0/download/matrix.org/bwJOaQTEDJnffhpojTprjTjv/bead001.jpg >
10:56 PM xxcoder[m]: is it low fps crap or is my rpi4 simply not keeping up?
10:59 PM solarwind: xxcoder[m] it's part of the video
10:59 PM solarwind: it's horrible
11:01 PM CaptHindsight[m]: look for glass forming videos, they are all similar
11:01 PM CaptHindsight[m]: https://www.youtube.com/watch?v=GsEg9DRAO98
11:05 PM xxcoder[m]: ok hard to tell for me because video lags so much
11:17 PM CaptHindsight[m]: https://www.youtube.com/watch?v=CB7cz0Bz_Jw sound off or down
11:29 PM xxcoder[m]: pretty cool
11:30 PM xxcoder[m]: barely could play it lol