#linuxcnc Logs

Mar 08 2025

01:19 AM Deejay: moin
03:46 AM -!- #linuxcnc mode set to +v by ChanServ
03:58 AM -!- #linuxcnc mode set to +v by ChanServ
04:15 AM Tom_L: morning
04:28 AM -!- #linuxcnc mode set to +v by ChanServ
04:29 AM -!- #linuxcnc mode set to +v by ChanServ
05:37 AM -!- #linuxcnc mode set to +v by ChanServ
05:38 AM -!- #linuxcnc mode set to +v by ChanServ
06:23 AM -!- #linuxcnc mode set to +v by ChanServ
06:23 AM -!- #linuxcnc mode set to +v by ChanServ
07:14 AM JT-Chicken: https://youtu.be/g6RyhbE1K_c
07:41 AM -!- #linuxcnc mode set to +v by ChanServ
07:41 AM -!- #linuxcnc mode set to +v by ChanServ
07:45 AM -!- #linuxcnc mode set to +v by ChanServ
08:26 AM -!- #linuxcnc mode set to +v by ChanServ
08:26 AM -!- #linuxcnc mode set to +v by ChanServ
08:30 AM Tom_L: pcw-home, what do --device RECOVER, --device AUTO, --device LITEHM2 do?
08:44 AM JT-Shop_ is now known as JT-Shop
08:45 AM -!- #linuxcnc mode set to +v by ChanServ
08:48 AM pcw-home: --device recover currently works only on 6I24 and 6I25 and allows you to program a card with a corrupted EEPROM
08:50 AM pcw-home: --device litehm2 is for litemh2 cards (Hostmot2 running on a generic probably originally LED signage FPGA card)
08:51 AM pcw-home: not sure about --auto (auto card detection?)
09:17 AM -!- #linuxcnc mode set to +v by ChanServ
09:17 AM -!- #linuxcnc mode set to +v by ChanServ
09:57 AM rdtsc: skunkworks8841 if you decided to try two SN74LVC2G17 to clean up your encoder signal, here are some tips: you'll need a stable 5.0V. One 0.1uF and 10nF (0.01uF) power bypass caps across each chip, close as possible. Any unused inputs must be tied to 0V or 5V. Be mindful of any differences in GND.
10:08 AM -!- #linuxcnc mode set to +v by ChanServ
10:08 AM -!- #linuxcnc mode set to +v by ChanServ
10:16 AM rdtsc: pcw-home et.al. yes, updated firmware fixed the rpi5, thank you! :)
10:20 AM pcw-home: Glad it's working, and again sorry for the issue.
10:26 AM rdtsc: Ditto - glad I didn't have to send a good Rpi back to vendor only to find out it was actually fine :)
10:26 AM rdtsc: regarding the [ 0.943220] rp1-pio 1f00178000.pio: error -ENOENT: failed to contact RP1 firmware message, others are saying this is because GPIO tries to load before the RP1 is ready - unrelated to SPI
10:39 AM Tom_L: pcw-home, thanks
10:42 AM Tom_L: rdtsc, glad you got it working
10:43 AM Tom_L: pcw-home, would the 7i90 on spi have the same issue?
10:43 AM Tom_L: i've tested it on a rpi4 before
10:45 AM pcw-home: No, it wasn't a RPI5 issue at all, just that the pre-installed firmware on some shipped 7C80s had /CS on the wrong pin
10:46 AM lcnc-relay: <ccatlett1984@> Pcw-home, is there a way to validate what version of firmware is on the 7c80? If so I'm going to add that to the guide I'm building.
10:46 AM Tom_L: ok, i just wasn't sure if the 7i90 needed a firmware update for that new rpi5 spi driver
10:46 AM pcw-home: (this was masked by the fact that on RPI4s the alternate CS pin happened to be at GND potential)
10:47 AM Tom_L: ok
10:48 AM pcw-home: only affects some 7C80s and not a version issue, 7I90 firmware from 5 yeas ago will work on a RPI5)
10:49 AM lcnc-relay: <ccatlett1984@> So is there any way to validate if the firmware on a 7c80 is the affected version?
10:49 AM lcnc-relay: <ccatlett1984@> Other than, it doesn't work.
10:50 AM Tom_L: pretty good validation if you ask me
10:50 AM Tom_L: :)
10:50 AM lcnc-relay: <ccatlett1984@> /me sighs
10:50 AM Tom_L: heh
10:51 AM lcnc-relay: <ccatlett1984@> I'm trying to make as much of a foolproof doc from beginning to end as possible, without having to rely on any "institutional" knowledge.
10:51 AM lcnc-relay: <ccatlett1984@> Hence, I want to cover as many edge cases as possible.
10:51 AM Tom_L: just knowing it's out there is worth running an update on one that doesn't work
10:51 AM lcnc-relay: <ccatlett1984@> The problem being though, that means you have to have a additional/older pie to be able to run the firmware update from.
10:52 AM lcnc-relay: <ccatlett1984@> Pcw, can you confirm would you be able to use Mesa flash from a pi5 to apply the fixed firmware?
10:52 AM lcnc-relay: <ccatlett1984@> I'm assuming no, because of the communication failure.
10:52 AM Tom_L: if you can't connect to it you wouldn't be able to
10:53 AM Tom_L: maybe with jtag but that gets messy
10:54 AM lcnc-relay: <ccatlett1984@> The best way that I think of handling this, in this case, would be to request that the firmware be updated before being shipped from mesa.
10:55 AM rdtsc: to be fair, the 7c80 was released far before the rpi5 - just request in your doc that anyone considering updating to rpi5, do a firmware update first, even if things seem to be working fine. speaking of doc, made a notes document and will share it with you soon ccatlett1984
10:56 AM rdtsc: for 6.12.11
10:56 AM lcnc-relay: <ccatlett1984@> I'm not disputing that, the big thing is the requirement for having additional hardware to be able to flash it, if you have an affected device.
10:56 AM pcw-home: Once the firmware issue was discovered (after RPI 5 support was released) there we no more 7C80s shipped with the old firmware
10:57 AM lcnc-relay: <ccatlett1984@> Oh awesome, so I can just make a note that if you've already had your 780 for a while, it's best to update the firmware.
10:58 AM pcw-home: also people that have older 7C80s used them with RPI4s so may have changed the firmware at some point
10:59 AM pcw-home: I suspect it can also be fixed by temporarily grounding the alternate CS pin and updating via mesaflash on the RPI5
11:00 AM pcw-home: (none of the distributed firmware files have this issue)
11:01 AM pcw-home: (wrong CS pin is RPIO25 right is RPIO8)
11:01 AM Tom_L: one of you guys could test that theory since you have rpi4 in case it didn't work
11:02 AM Tom_L: i suppose if it didn't work you wouldn't change the firmware anyway
11:03 AM Tom_L: you'd have to load the old firmware for the test
11:05 AM -!- #linuxcnc mode set to +v by ChanServ
11:05 AM Tom_L: i'd be willing to bet i have an old version of the firmware
11:05 AM Tom_L: (dated 2022)
11:07 AM Tom_L: actually the bit files are dated 2020
11:15 AM pcw-home: If you have a 7C80 from 2022, then yes it has the old firmware
11:16 AM pcw-home: bit files are all OK
11:16 AM pcw-home: no distributed bit files had the issue
11:17 AM pcw-home: only bad bit file was on our PC used for initialization
11:18 AM Tom_L: oh
11:20 AM pcw-home: also this does not affect 7C81 or 7I90
11:21 AM pcw-home: only 7C80s with the original pre-installed firmware
11:25 AM -!- #linuxcnc mode set to +v by ChanServ
12:11 PM -!- #linuxcnc mode set to +v by ChanServ
12:12 PM lcnc-relay: <skunkworks8841@> pcw-home: did you see the trace from the cui encdoer?
12:12 PM -!- #linuxcnc mode set to +v by ChanServ
12:18 PM rdtsc: ccatlett1984: https://paste.debian.net/1361993/
12:33 PM xxcoder: ouch. this might hurt https://www.bleepingcomputer.com/news/security/undocumented-backdoor-found-in-bluetooth-chip-used-by-a-billion-devices/
12:41 PM pcw-home: Skunkworks: Yes, pretty bad velocity estimation, what resolution?
01:00 PM lcnc-relay: <skunkworks8841@> pcw-home: it is set to 4096
01:00 PM pcw-home: Skunkworks: Is that ripple one per turn?
01:00 PM lcnc-relay: <skunkworks8841@> looks like it
01:01 PM lcnc-relay: <skunkworks8841@> Jon on the mailing list said they have to be perfectly centered. But I put my finger on the back of the encoder and me kinda flexing things around doesn't make it worse or better.
01:01 PM NetPipe: are we living in a cuckoo clock ?
01:02 PM pcw-home: That might make sense since I think they use a 2 pole magnet and the rest is interpolation (so you would think centering would be critical)
01:03 PM NetPipe: https://www.youtube.com/watch?v=UrXtuwQfGw4
01:03 PM pcw-home: Does that mean the angular accuracy is only 1% or so?
01:04 PM lcnc-relay: <skunkworks8841@> at 1700rpm it is +/- 125 rpm
01:05 PM pcw-home: Yow!
01:05 PM lcnc-relay: <skunkworks8841@> https://photos.app.goo.gl/D1J5xbqTk92WDNTZA
01:06 PM lcnc-relay: <skunkworks8841@> very top trace is velocity-rpm from mesa encoder counter
01:07 PM lcnc-relay: <skunkworks8841@> https://photos.app.goo.gl/bDpgs6vkLMMjDEDR6
01:07 PM lcnc-relay: <skunkworks8841@> that is zoomed in - 50rpm per div
01:08 PM pcw-home: Even without encoder velocity estimation, at 1700 RPM and 4096 CPT that 116 counts per ms
01:08 PM pcw-home: so no excuse for more the ~1% ripple
01:09 PM lcnc-relay: <skunkworks8841@> at 100ms - that is 3 cycles - which I think comes to every rotation.
01:09 PM lcnc-relay: <skunkworks8841@> at 1700rpm
01:09 PM lcnc-relay: <skunkworks8841@> ish
01:11 PM pcw-home: Yeah if the velocity is that bad, how bad is the angular error?
01:11 PM lcnc-relay: <skunkworks8841@> right - that is why I demoted these to spindle encoders.. Now I don't even know about that.
01:12 PM pcw-home: Possibly just a bad one?
01:12 PM rdtsc: how long and what type of wire?
01:15 PM lcnc-relay: <skunkworks8841@> I had bought about a half a dozen of these.. They all seem to have issues.
01:15 PM rdtsc: maybe they are incapable of driving any significant load
01:15 PM lcnc-relay: <skunkworks8841@> If you try to use them as closed loop servo encoders - they grumble and stuff..
01:16 PM pcw-home: Yeah I think their interpolation is slow
01:17 PM lcnc-relay: <skunkworks8841@> these are pretty old.. Maybe their new ones are better.
01:17 PM lcnc-relay: <skunkworks8841@> rdtsc: less than 4ft.
01:17 PM lcnc-relay: <skunkworks8841@> shielded
01:18 PM pcw-home: It interesting that the velocity ripple get worse with increasing speed, that's opposite of normal encoders
01:18 PM lcnc-relay: <skunkworks8841@> right!?
01:19 PM lcnc-relay: <skunkworks8841@> https://www.youtube.com/watch?v=8RWqHW2AtrA
01:19 PM lcnc-relay: <skunkworks8841@> I replaced that with an optical encoder and the noise went way.
01:20 PM pcw-home: sounds like noise from the interpolator
01:33 PM rdtsc: incidentally, the 7c80 is happy running gantry-mill air-cuts on a Rpi5 with keyboard, mouse, touchscreen, and USB pendant all connected - no power issues at all and cpu at 40C. When installed, will only have one USB device connected; a powered industrial hub, hoping for no hiccups there.
01:41 PM -!- #linuxcnc mode set to +v by ChanServ
01:41 PM -!- #linuxcnc mode set to +v by ChanServ
02:26 PM -!- #linuxcnc mode set to +v by ChanServ
02:26 PM -!- #linuxcnc mode set to +v by ChanServ
02:28 PM lcnc-relay: <skunkworks8841@> the rpi5 has been pretty darn good latency wise..
02:29 PM memleak: that's good!
02:32 PM memleak: are there any relatively new arm64 boards with pci express 16x so you can hook up a PCIE GPU up to it?
02:33 PM memleak: AMD came out with one around 2016 apparently
02:33 PM memleak: https://www.phoronix.com/news/AMD-LeMaker-Cello
02:38 PM memleak: or at least an arm64 board that has a possible expansion add-on board that has PCIE 16X?
02:43 PM memleak: there's nothing x86 specific to radeon cards such as atombios according to the radeon devs so should theoretically work
03:23 PM -!- #linuxcnc mode set to +v by ChanServ
04:31 PM -!- #linuxcnc mode set to +v by ChanServ
04:41 PM -!- #linuxcnc mode set to +v by ChanServ
04:41 PM memleak: JT-Cave, how do you have so many different external IPs?
04:42 PM JT-Machine-Shop: I don't know why it keeps rebooting, but I have 4 pcs with hexchat on them
04:43 PM memleak: each with their own external IP?
04:44 PM JT-Machine-Shop: they are all on the same lan so no clue about the ips
04:44 PM memleak: oh, they're all hooked up directly to the modem?
04:44 PM memleak: oh.. weird.
04:44 PM memleak: lol
04:45 PM JT-Machine-Shop: about to go reset starlink to see if that fixes it
04:45 PM memleak: hows the speed with starlink?
04:45 PM memleak: we get 350~ megabits down with comcast business
04:46 PM JT-Machine-Shop: 75 Mbps down 4.26 up
04:46 PM memleak: ~35 up
04:46 PM memleak: thats not bad, how much is it?
04:46 PM JT-Machine-Shop: if you live in a rural area like I do it's $120 a month
04:47 PM memleak: is it cheaper or more expensive in the cities?
04:47 PM JT-Machine-Shop: cheaper in the city due to competition from other isp's
04:47 PM memleak: ah gotcha
04:48 PM JT-Machine-Shop: oboma cable is 1.5 miles from my house but not hooked to anything so a huge waste of our money
04:49 PM JT-Machine-Shop: actually fiber cable
04:52 PM xxcoder: isps took money to build em, but terms didnt inclue reqirement to use em
04:52 PM xxcoder: so theres tons of dark fiber
04:53 PM JT-Machine-Shop: so a not well thought out goverment plan as usual
04:54 PM xxcoder: most works fine. theres few flops yeah but thats life
04:54 PM xxcoder: well in bunch of cases, till recently working fine
05:00 PM JT-Machine-Shop: while it's good to have that infrastructure in place the money would have been better spent developing a low cost electric commuter vehicle
05:01 PM xxcoder: I just hope telo takes off
05:01 PM JT-Machine-Shop: telo?
05:01 PM xxcoder: it would be something I want, mostly.
05:01 PM xxcoder: yeah new ev truck company
05:01 PM xxcoder: they will also make vans also later
05:02 PM xxcoder: https://www.telotrucks.com/
05:03 PM JT-Machine-Shop: they are too expensive for a commuter vehicle
05:03 PM xxcoder: yeah its more for working trucks than just commute
05:04 PM xxcoder: still less than half price of that other ev truck
05:04 PM xxcoder: and quality probably double
05:05 PM -!- #linuxcnc mode set to +v by ChanServ
05:13 PM lcnc-relay: <ccatlett1984@> still looks like a 4ft box on the back
05:14 PM xxcoder: 6 if I recall right, it has mid hatch to open for full 8
05:16 PM lcnc-relay: <ccatlett1984@> 60" without the mid-hatch, so 4ft
05:16 PM lcnc-relay: <ccatlett1984@> 5ft
05:16 PM xxcoder: ah yes 5/3 yeah
05:16 PM lcnc-relay: <ccatlett1984@> looks like it's trying to be a van and a truck
05:17 PM xxcoder: yeah
05:17 PM xxcoder: I dont plan to replace my focus for a while but I do want ev vehicle
05:18 PM xxcoder: telo has solar tops options which is awesome, I dont drive all that much
05:19 PM xxcoder: 27 miles a day for main body only if I recall right.
05:19 PM lcnc-relay: <ccatlett1984@> solar is going to charge so little, as to be pointless
05:19 PM xxcoder: if its driven a lot sure
05:20 PM lcnc-relay: <ccatlett1984@> i drive ~ 18 miles a day, but it wouldn't get full sun
05:20 PM lcnc-relay: <ccatlett1984@> and i'm at a higher latitude anyways
05:20 PM xxcoder: yeah? where you at? im at tacoma, wa
05:20 PM lcnc-relay: <ccatlett1984@> MI
05:21 PM xxcoder: cool
05:21 PM lcnc-relay: <ccatlett1984@> i'd rather have the option without it, its wasted cost and another thing to worry about with hail
05:23 PM xxcoder: thats true, it much depends on region
05:23 PM xxcoder: here it would be just fine
05:24 PM JT-Shop: if you had a carport with solar panels charging all day while your at work then transfer from storage batteries to the vehicle that would be cool
05:24 PM xxcoder: definitely
05:25 PM JT-Shop: transfer at night
05:30 PM JT-Shop: ev only makes sense if charged from solar, wind or water otherwise hybrid is more efficient
05:38 PM -!- #linuxcnc mode set to +v by ChanServ
05:42 PM lcnc-relay: <ccatlett1984@> @ JT, fixed the merging errors
05:42 PM lcnc-relay: https://github.com/jethornton/mesact/pull/135
05:42 PM JT-Machine-Shop: cool
05:43 PM lcnc-relay: <ccatlett1984@> didn't see them in my local branch, so it must have been a merge thing
05:43 PM lcnc-relay: <ccatlett1984@> but i moved stuff around until it stopped complaining
05:43 PM JT-Machine-Shop: I almost never understand the merge errors but that time I "think" I understood
05:44 PM JT-Machine-Shop: more than once I've renamed my local and cloned again
05:49 PM -!- #linuxcnc mode set to +v by ChanServ
05:53 PM lcnc-relay: <ccatlett1984@> yeah, i was close to that point
05:53 PM JT-Shop: it still has a conflict :(
05:53 PM lcnc-relay: <ccatlett1984@> look, blank lines
05:53 PM JT-Shop: looks like one line out of place
05:55 PM JT-Shop: between 7i76eu and 7i80db-16
05:56 PM lcnc-relay: <ccatlett1984@> but both lines are blank
05:56 PM JT-Shop: maybe an extra blank line?
05:56 PM lcnc-relay: <ccatlett1984@> i only have 1
05:58 PM lcnc-relay: <ccatlett1984@> side by side, i don't see any differences
06:00 PM JT-Shop: https://pasteboard.co/dYKpSnGC2Ivq.png
06:00 PM JT-Shop: that's what I see
06:01 PM lcnc-relay: <ccatlett1984@> yeah, and both repos have a single damned blank line.....
06:01 PM lcnc-relay: <ccatlett1984@> i don't see how to resolve that
06:01 PM lcnc-relay: <ccatlett1984@> I have both repos open, and i dont see anything different in that section
06:01 PM JT-Shop: I'll look in the morning... getting pretty tired and ready to call it a day
06:02 PM lcnc-relay: <ccatlett1984@> kk
06:05 PM JT-Shop: it will be cool to have the 7c80 support in mesact
06:31 PM lcnc-relay: <ccatlett1984@> i think there will still be a bug or two, as I haven't finished my config. but this will at least prevent further merge issues down the road
06:53 PM rdtsc: re: solar, there have been lots of technological advancements in solar panel efficiency... quantum dots, perovskites, even more... but unlike nuclear fusion, we're actually likely to see these soon. Cells with double the efficiency of silicon have already been proven; the production logistics always lag behind.
06:55 PM rdtsc: ccatlett1984, is your config using PID loops? I've been using the one from https://github.com/LinuxCNC/linuxcnc/blob/master/configs/by_interface/mesa/hm2-stepper/hm2-stepper-rpspi.hal with no PID loops - seems nice and snappy, but haven't cut anything with it yet.
07:24 PM lcnc-relay: <ccatlett1984@> I'm still waiting on driver timings for my stepper drivers
07:24 PM lcnc-relay: <ccatlett1984@> it hasn't moved yet
08:53 PM memleak: LMFAO @ git merge conflict convo
10:34 PM NetPipe: https://arxiv.org/abs/2208.00854
10:39 PM NetPipe: for windturbines maybe