Sep 11 2022
01:01 AM XXCoder: 176 now wow
01:10 AM XXCoder: not the worse, but not great
01:30 AM roycroft: it's staying in the 120s here
01:30 AM roycroft: hopefully it won't get worse
01:31 AM roycroft: although it was supposed to
01:31 AM XXCoder: 176 here
01:31 AM XXCoder: doh said it already
01:31 AM XXCoder: yeah hope it goes away soon. im feeling bit ill
01:31 AM XXCoder: im pretty damn sensive to those :(
01:32 AM XXCoder: it was hell when I was working in shop. I had to constantly wear mask. ironically training for covid
02:08 AM Deejay: moin
05:00 AM JT-Cave: morning
06:09 AM Tom_L: morning
06:18 AM Tom_L: 54F
06:23 AM JT-Cave: we get that tomorrow night
06:25 AM mrec: first time hobbing ... fail. not sure what's wrong I think I should better go back to single tooth cutting
06:27 AM mrec: the encoder reported 80 turns of the hobber for a full circle (80 teeth) for real it just removed all the material and did not cut a single tooth
06:27 AM JT-Cave: opps
06:29 AM mrec: single tooth cutting is fool proof
06:38 AM miss0r: mrec: But not idiot proof :D
06:41 AM mrec: I'm fine with it :-) no failures with single gear cutters so far
06:57 AM mrec: well I need to check the encoder settings something's wrong there it seems
06:58 AM mrec: I was just looking at rawcounts it seems strange...
06:59 AM Guest0: Hello, im from germany and new to cnc. I would like to chat about pcb drilling with kicad pcb drill file.
06:59 AM Guest0: Is it ok to diskusss this here?
07:00 AM mrec: let me guess you recently asked in the microcontroller.net forum?
07:00 AM Guest0: at mrec are you asking me?
07:00 AM mrec: yes
07:01 AM Guest0: no. I know the microcontroller forum, but did not ask yet.
07:01 AM Guest0: havent read a thread of this there
07:01 AM mrec: export some excellon drillfile and convert it with flatcam to g-code
07:02 AM Guest0: I converted it with a python script
07:02 AM Guest0: drl2decode or so. that worrked.
07:02 AM Guest0: My Question is more cnc related.
07:02 AM mrec: I did that several times.
07:03 AM Guest0: How do I take care of getting my workpiec (pcb) rectangular.
07:03 AM Guest0: sorry for my english
07:04 AM mrec: mit ner messuhr?
07:05 AM Guest0: is it okay to speak german here?
07:06 AM Guest0: Ich dachte mit der messuhr macht man höchstens die z positionierung
07:06 AM mrec: mit ner Messuhr kannst Du alles machen
07:07 AM Guest0: ich hab ein selbstbau cnc3018.
07:08 AM Guest0: sprechen wir von einer mechanischen messuhr zur tiefen messung? komme aus der etechnik nicht maschinenbau.
07:08 AM mrec: na was besser passt
07:08 AM mrec: https://www.manutan.ch/de/fab/mechanische-messuhr-0-14-mm-verstellbarer-taster
07:10 AM Guest0: Ich hatte gehofft das man das werkstück schief einspannt die fiducials anfährt
07:10 AM Guest0: und dann die Bohrkoordinaten transformiert
07:11 AM Guest0: von der steuerungssoftware
07:12 AM mrec: warum machst du das nicht gleich mit flatcam?
07:12 AM mrec: ich weiß was du meint, nein man misst das so sicher nicht aus; du hast doch das koordinatensystem gespeichert
07:12 AM Guest0: keine Erfahrung. benutze momentan bCNC als erste versuche
07:12 AM mrec: da brauchst du doch nichts ausmessen du weißt doch wo du gebohrt hast und kennst den Ursprung?
07:13 AM Guest0: Ich möchte das pcb ätzen und dann bohren.
07:13 AM Guest0: tonertrnsfer da stören bohrungen extrem.
07:13 AM mrec: warum nicht gleich isolationsfräsen?
07:14 AM mrec: flatcam + autoleveller reicht eigentlich
07:14 AM mrec: ich hab das ätzen schon lange aufgegeben und fräse nur noch wenn ich breakout boards mache.
07:15 AM Guest0: flatcam schau ich mir mal an. mein setting derzeit cnc3018 mit arduino firmware (gcode und linuxpc mit linux + python + bCNC (python gcode sender)
07:16 AM Guest0: Macht soviel dreck :-)
07:17 AM mrec: dann schmierst Du öl auf die Leiterplatte, aber so viel staub kommt beim Isolationsfräsen ja auch nicht runter
07:17 AM Guest0: Wollte gerade das milling vermeiden. Sehe aber das das fast alle so machen.
07:18 AM Guest0: Meine befürchtung ist das ich nicht wirklich so kleine strukturen hinbekomme. und die abstände werden immer kleiner.
07:18 AM Guest0: 0.5mm pinabstand
07:19 AM mrec: deshalb auch den autoleveller verwenden da du die Leiterplatte wohl niemals flach auf das Bett bekommst
07:20 AM mrec: den flatcam output einfach durchschleifen und das erzeugt eine weitere g-code file
07:20 AM Guest0: beim bohren wäre das dann nicht so kritisch
07:21 AM mrec: meine Fräs-Resultate waren eigentlich immer besser als die geätzten
07:21 AM Guest0: was braucht den flatcam als umgebung
07:21 AM mrec: vor allem konsistenter
07:21 AM mrec: Ich denke Windows, habe meine letzten LPs aber vor einem guten Jahr gemacht
07:22 AM Guest0: Ich merke schon du möchtest mich bekehren (scherz)
07:22 AM Guest0: windows ist bei mir killkriterium. bin nur noch mit linux unterwegs.
07:24 AM Guest0: flatcam scheint auch python zu sein, das wäre ok
07:25 AM mrec: ich verwende win/linux/mac was gerade besser passt
07:26 AM Guest0: Wie gesagt windows äuserst ungern, aber ich sehe das linux supported wird.
07:27 AM Guest0: aber nochmal zum vorgang.
07:27 AM mrec: ich denke ich werde auch bald wieder ne leiterplatte machen müssen zur Zeit versuche ich nen single ended encoder mit linuxcnc zu verwenden bin gerade über mein 3m kabel problem gestolpert
07:27 AM mrec: mit 1m klappt's wohl noch
07:28 AM Guest0: was ist ein single ended encoder?
07:29 AM mrec: single ended wo das signal als referenz ground verwendet; differential wo beide Signale sich mehr oder weniger gegenüberstehen A+ A- / B+ B- / I+ I-
07:29 AM Guest0: bidirektionale kommunikation?
07:29 AM mrec: unidirektional
07:30 AM mrec: bei single ended gibt's auch noise auf der Leitung, bei differential gleichen sich die störungen aus
07:31 AM Guest0: kannst du ein bussystemverwenden i2c oder aus der autobranche can bus
07:36 AM Guest0: wenn es um einen drehencoder geht dann würde ich mit einem pcf8574 und i2c arbeiten
07:42 AM Guest0: Ok. werde mich mal mit flatcam auseinandersetzen. Sieht vielversprechend aus. Erstmal Dnke für die Tips.
07:42 AM Guest0: quit
07:44 AM mrec: ok second try with gear hobbing fail again this shit just doesn't work
07:45 AM mrec: isn't it just like the gear hobber needs to run a full rotation * teeth I want in the disc?
07:45 AM mrec: maybe the hob itself is just crap
08:03 AM Tom_L: maybe the operator :)
08:07 AM JT-Shop: :)
08:11 AM mrec: the operator is just putting blanks in and pushing buttons :D
08:11 AM mrec: so it must be linuxcnc
08:11 AM mrec: (kidding, of course it's me)
08:16 AM JT-Shop: https://www.youtube.com/watch?v=ZhICrb0Tbn4
08:22 AM mrec: ok fixed ... that belongs into the category stupid mistakes... the direction was wrong so I set the encoder scale to -1000 instead of 1000
08:22 AM mrec: next is to decrease the encoder counts of the magnetic encoder so I can cut a little bit quicker
09:02 AM Tom_L: JT-Shop2, did you find an img file that worked?
09:20 AM coredump6391[m]: Does anyone has an example of using a touch probe to set tools after tool changes? I never got that to work
09:21 AM JT-Cave: Tom_L, the rpi stock img worked I've not tried any others yet and it was 11GB
09:21 AM Tom_L: k
09:22 AM Tom_L: i found with that buster image i have, it works fine then when i do the upgrades to bring it up to date and lcnc 2.8.3 the USB quits working
09:22 AM Tom_L: the one i found that worked was the 2.8.0 img
09:23 AM Tom_L: i think andy updated that once to 2.8.1 and i can't seem to get that to work here
09:24 AM Tom_L: 11G!
09:24 AM Tom_L: the img must contain a bunch of blank space
09:26 AM Tom_L: that's more than a xilinx download :)
09:32 AM -!- #linuxcnc mode set to +v by ChanServ
09:32 AM funkenjaeger1683: I have a toolsetter routine in my config:
09:32 AM funkenjaeger1683: https://github.com/Funkenjaeger/fj-lcnc-cfg
09:32 AM funkenjaeger1683: Mine's a little more complicated than usual because I also use an optical beam-break sensor just above the tool setter, but you can ignore that part
09:33 AM funkenjaeger1683: * coredump (@_discord_106527593053847552:jauriarts.org) I have
09:38 AM funkenjaeger1683: measure.ngc is the actual tool length measurement routine which I have mapped to M201 in the INI. I call M201 from within my M6 (tool change) remap, which is implemented in toolchange.ngc
09:42 AM funkenjaeger1683: I have an ATC, but I've also added a toggle button on my GUI to select either automatic or manual tool changes. If you don't have an ATC, then you can also ignore a lot of the code in toolchange.ngc and just focus on the few lines that pertain to manual tool changes
09:42 AM GTKplusplus6432[: https://wiki.printnc.info/en/controllers/linuxcnc/tool-setter-no-probe
09:42 AM GTKplusplus6432[: I have this setup which I quite like
09:42 AM GTKplusplus6432[: you need to call m600 before you load your first tool, then it's seamless
09:45 AM -!- #linuxcnc mode set to +v by ChanServ
09:48 AM mrec: Tom_L: ... xilinx?... when I tried to install it it said it wants 140 gb
09:50 AM roycroft: the cedar creek fire has more than doubled in size overnight, and has jumped all the fire breaks, so it's now 0% contained again
09:50 AM roycroft: and it is just outside of oakridge
10:02 AM Tom_L: mrec, the download
10:04 AM coredump6391[m]: Is there a cheap atc or every solution is expensive
10:06 AM Tom_L: DIY might be cheaper
10:12 AM Tom_L: coredump6391[m], carousel, chain driven or other?
10:17 AM funkenjaeger1683: Generally no, as far as I've ever seen... even the chinese ones on aliexpress are well over $1000
10:18 AM coredump6391[m]: Ouchy
10:18 AM coredump6391[m]: The holders are also expensive i the end
10:18 AM coredump6391[m]: In the end
10:20 AM funkenjaeger1683: they are... ISO30 tool holders tend to go for a bit above $100 each, or you can roll the dice on chinese ones for as low as $30ish.
10:22 AM coredump6391[m]: Ok that’s not a solution now. I’ll just try to set the tool touch thing
10:26 AM mrec: Tom_L: the download is at least 40-50gb xilinx...
10:26 AM mrec: maybe an older version is smaller but I just checked 2 days ago
10:30 AM Tom_L: i was referring to the older webpack
10:36 AM unterhaus_: you can't use rtai with an ethernet mesa board, right?
10:36 AM Tom_L: right
10:38 AM coredump6391[m]: I use the qtdragon ui, does it make a difference on that thing? It has a touch tool option and a probe screen that I also haven’t done much with
10:38 AM GTKplusplus6432[: it shouldn't, but I've never tried it
10:39 AM GTKplusplus6432[: probe-screen v2 remap stops it from working though
10:39 AM Tom_L: unterhaus_, you saw the link about the 5.3.2 RTAI release?
10:40 AM unterhaus_: I'm not sure, I saw there is a new rtai kernel
10:40 AM Tom_L: right
10:40 AM coredump6391[m]: If that’s the standard on git I probably use it since I’m using the git version
10:40 AM unterhaus_: it's just that I have computers that work better with rtai but I want to use ethernet
10:41 AM unterhaus_: I had a glimmer of hope when andy mentioned the uspace rtai
10:41 AM Tom_L: that would be a peter question
10:46 AM coredump6391[m]: https://linuxcnc.org/docs/devel/html/gui/qtdragon.html#_auto_tool_measurement
10:46 AM coredump6391[m]: I think this is what I wanr
10:46 AM coredump6391[m]: Want
10:51 AM funkenjaeger1683: Give some thought to how you really want the process to work, since there are several ways to skin the cat when it comes to tool length measurement. The approach detailed in that link is one way - to me personally it sounds like a hassle to have to manually measure the height of your workpiece from the machine table every time...
10:51 AM GTKplusplus6432[: ^
10:51 AM GTKplusplus6432[: I really don't like that setup
10:51 AM GTKplusplus6432[: which is why I use the M600 one.
10:51 AM GTKplusplus6432[: You set your work offsets, tell the machine that the next tool measurement will act as a reference, and that's it
10:52 AM GTKplusplus6432[: no need to know either sensor or workpiece height since it's calculated on the fly
10:53 AM coredump6391[m]: I’ll do that one then
10:53 AM coredump6391[m]: Will report back with results and broken end mills
10:58 AM funkenjaeger1683: That sounds generally consistent with how I do it in my remap, except I don't have any manual MDI steps (e.g. M600) up front so I'm guessing there are some implementation differences under the hood.
10:59 AM GTKplusplus6432[: Yeah, I was thinking about moving to an absolute offset from zmax or something
10:59 AM GTKplusplus6432[: but this works well enough and my first toolchange for each run is a probe anyway
10:59 AM GTKplusplus6432[: which I call manually anyway
11:02 AM Tom_L: test on a wood block.. wood is cheaper than carbide
11:06 AM funkenjaeger1683: I do have to be sure I've probed the initial tool (to set the tool length offset) before I go touching off my workpiece to set Z0 in the WCS, which is probably similar to the need to run M600 in the approach you're using. That happens automatically if I do a tool change first thing, and I have at times set up my homing routine so that it automatically measures the tool immediately after homing, which makes it automatic enough that
11:06 AM funkenjaeger1683: it's reasonably idiot-proof.
11:10 AM GTKplusplus6432[: yeah, it's exactly the same. I don't have a way to manually probe a tool setup and m600 takes care of setting the reference for me
11:17 AM mrec: finally I figured out mesa has 7I96S in stock... okay ordered one
11:17 AM Tom_L: nice board
11:18 AM mrec: 60$ shipping .. crazy
11:20 AM Tom_L: you in UK?
11:20 AM Tom_L: no overseas stock there?
11:23 AM mrec: nothing in asia it seems
11:23 AM mrec: I want to study it and put that functionality onto my zynq board
11:24 AM Tom_L: good options for an all in one board
11:25 AM mrec: someone in #fpga said he put vivado onto a network share and starts it from there rather than a local system
11:25 AM mrec: ##fpga
11:27 AM mrec: last week I studied the RGMII ethernet interface
11:27 AM Tom_L: you know the 7i96s isn't xilinx right?
11:27 AM Tom_L: due to lack of stock
11:27 AM mrec: the picture says xilinx spartan 6
11:27 AM mrec: http://store.mesanet.com/index.php?route=product/product&path=83_85&product_id=374
11:27 AM Tom_L: not on the 7i96s, the 7i96 is
11:28 AM mrec: do you know what the 7i96s uses?
11:28 AM Tom_L: to get the tools for the new fpga you gotta buy a $30 dev board that now sells for $70ish
11:30 AM Tom_L: as far as the pic, the layout probably didn't change much so he used that pic
11:31 AM mrec: Tom_L: do you know which chip it uses?
11:32 AM Tom_L: hang on
11:32 AM Tom_L: Efinix
11:35 AM Tom_L: Jan 16 16:00:19 <pcw-home> Yeah, we are using a Chinese FPGA in a new design
11:37 AM Tom_L: Dec 11 17:55:39 <pcw--home> Efinix makes you buy a devel card to get the tools
11:37 AM pcw-home: The 7I96S uses a T20F256
11:38 AM Tom_L: i was trying to find that cheap dev board for the software...
11:38 AM pcw-home: Yeah the Xyloni _was_ $35
11:38 AM Tom_L: not anymore :)
11:42 AM Tom_L: https://www.digikey.com/en/products/detail/efinix-inc/XYLONI/13535080?s=N4IgTCBcDaIBoE8A2B7AdgSwAQBECmAbnqgA4C2eaALlgNIZUgC6AvkA
11:47 AM mrec: pcw-home: you're from Mesa?
11:47 AM mrec: my main question is why is shipping so expensive?
11:47 AM mrec: 60$ ... we ship boards oversea for less than 10$
11:48 AM mrec: and a few years ago I shipped from Germany for like 6-7$ worldwide
11:53 AM Tom_L: mrec, he _is_ mesa
11:53 AM Tom_L: and that was a few years ago :)
11:54 AM Tom_L: i was shipping stuff worldwide for $3ish too
11:55 AM mrec: ya but 60$ shipping is too much, if it's a hidden product price just add it to the product price story done
11:56 AM pcw-home: Shipping cost are direct from shippers we don't have any real control (USPS is usually cheapest)
11:57 AM Tom_L: now that you can't find anything in the stores and get it online, the shippers feel the need to rape you
11:59 AM pcw-home: shipping costs have almost tripled in the last 10 years especially USA --> international
12:00 PM roycroft: grizzly just sent out an email informing their customers that prices are being reduced across the board
12:00 PM -!- #linuxcnc mode set to +v by ChanServ
12:00 PM mrec: well but never 60$
12:00 PM roycroft: they said that a couple years ago it cost $3500 to ship a container to the us from china
12:00 PM mrec: for a simple pcb
12:00 PM roycroft: that went up to $25000, and they had to increase prices accordingly
12:00 PM roycroft: now the cost has gone back down, so they're reducing prices
12:03 PM jpa-: 60 USD is fairly typical if you don't have the volume for any special agreement with shipping companies
12:05 PM pcw-home: Fedex intl economy from US to Taiwan for example (1 lb 8"x8"x2") is 135.93
12:06 PM JT-Cave: USPS is using Global Post now
12:07 PM jpa-: it's silly how digikey can offer free fast shipping for 50 USD orders, while for non-contract customers even just the shipping costs would be way over 50 USD
12:08 PM pcw-home: Digikey has UPS and FEDEX branches on site
12:15 PM mrec: we're talking about postal office no? USPS?
12:15 PM mrec: couriers are expensive, but there are brokers out there which collect up parcels
12:26 PM mrec: there are a few of them in Europe, also Asia .. transglobalexpress, jumingo, etc.
12:34 PM mrec: Trion T20 BGA256 Board
12:34 PM mrec: this seems to be a nice board
01:07 PM skunkworks: CaptHindsight[m]: you around?
01:08 PM skunkworks: andypugh: https://github.com/LinuxCNC/linuxcnc/issues/2024
01:11 PM coredump6391[m]: Ok I got it setup and dry runs seem to work fine. Just need to make sure fusion is configured correctly. Does everyone just use whatever Autodesk has on their cloud as the post or is there a better one
01:12 PM CaptHindsight: mrec: shipping from the USA has been far more expensive than too the USA for decades, the shipping and trade deals are designed to favor shipping from Asia
01:12 PM skunkworks: CaptHindsight[m]: found it. https://github.com/LinuxCNC/linuxcnc/issues/2024
01:12 PM CaptHindsight: check shipping rates from the EU to Taiwan, they aren't low like Taiwan to the EU
01:14 PM CaptHindsight: skunkworks: thanks, he was working on RTAI, the encoder issue was next in line
01:15 PM skunkworks: andypugh: I was hoping to use that on the lathe to save a printer port pin :)
01:15 PM CaptHindsight: https://github.com/NTULINUX/RTAI/releases/tag/v5.3.2 tom posted this earlier I think
01:17 PM CaptHindsight: was going to test the new RTAI on Bullseye and see if the encoder issue was still there, looks like it still would be
01:20 PM GTKplusplus6432[: I have a custom one
01:21 PM GTKplusplus6432[: includes a property for m600 (if you use your first tool to set WCS instead of a probe), enables the inspect surface cycle and it has configurable g64
01:22 PM mrec: CaptHindsight: please remember I'm from Europe/Germany and shipping is not that expensive from Germany.
01:23 PM CaptHindsight: mrec: apparently the EU makes better trade deals
01:24 PM coredump6391[m]: well, I will see what the default one does
01:25 PM CaptHindsight: mrec: I used to ask customers to send me their shipping labels if they could find better deals
01:25 PM mrec: 2kg shipping from Germany to US current rate (no insurance, no tracking): 10.99$ 5kg 47.99$ incl. 500$ insurance
01:25 PM mrec: that's DHL Paket
01:26 PM CaptHindsight: DHL went bankrupt in the USA about 13 years ago
01:26 PM mrec: DHL in Germany is more or less deutsche post
01:26 PM CaptHindsight: very poor coverage here, I use to have to drive 20 miles to the airport for DHL service
01:27 PM mrec: USA to Germany 43$ express
01:28 PM mrec: https://snipboard.io/TX3YBS.jpg
01:28 PM CaptHindsight: DHL has one drop off point somewhat near me and it's a 20 minute drive
01:28 PM mrec: express is about right.
01:29 PM Tom_L: CaptHindsight, did you post the info about the new RTAI?
01:29 PM mrec: 60$ scares off potential customers and they will keep looking further; it's better to have the shipping cost in the product price and just say like 10-40$ various options
01:29 PM mrec: especially since this is just a relatively cheap PCB
01:30 PM GTKplusplus6432[: I posted mine on the printNC discord if you want to give it a try
01:30 PM CaptHindsight: I wish it was $60 for an international <20lb box, I think it's closer to $75 now
01:30 PM GTKplusplus6432[: aside from the edits I've done to it it's the default one
01:30 PM coredump6391[m]: where?
01:30 PM coredump6391[m]: I mean on what channel
01:30 PM CaptHindsight: mrec: what choice do we have?
01:31 PM GTKplusplus6432[: https://github.com/GTKplusplus/linuxcncPost
01:31 PM coredump6391[m]: k thanks
01:31 PM GTKplusplus6432[: ah, it has some 4 axis stuff
01:31 PM mrec: https://about.usps.com/postal-bulletin/2021/pb22578/html/info_005.htm#:~:text=n%20Registered%20Mail%20%C2%AE%20service,service%20will%20increase%20to%20%2417.15.
01:31 PM mrec: USPS registered mail
01:31 PM GTKplusplus6432[: which I'm not sure linuxcnc will handle if you don't have the axis setup
01:32 PM CaptHindsight: Fedex and UPS are even higher for international, and if USPS damages or loses the packages it takes a couple months to get reimbursed by the carrier
01:33 PM mrec: registered mail seems to be pretty much okay
01:33 PM GTKplusplus6432[: delete lines 303 to 309 if it becomes an issue
01:33 PM mrec: if you send packages to africa ya they might be lost sometimes (did that one time never arrived)
01:33 PM mrec: but europe and taiwan is usually fine
01:33 PM GTKplusplus6432[: s/303/304/, s/to/and/, s/309/305/
01:34 PM CaptHindsight: mrec: that is for letters and flats, I get charged extra for shippers when they send me non flats, my local post office will charge me the extra for trying to get away with flat service for a non flat item, even though it fits in the envelope
01:34 PM coredump6391[m]: eh I will give it a try
01:34 PM GTKplusplus6432[: yep, if it doesn't work tell me, it's a quick edit
01:34 PM mrec: CaptHindsight: we use registered mail from Taiwan and put stuff into it
01:34 PM mrec: did the same in Germany (but they changed that last year)
01:35 PM GTKplusplus6432[: I wanted to tell you the specific lines to delete (303) but I'm not sure if just deleting the lines will cause issues off the top of my head
01:35 PM mrec: they now need warensendungen I think they stripped the insurance
01:35 PM CaptHindsight: so don't blame the manufacturer or seller for the poor trade deals the US makes with international shipping
01:36 PM mrec: CaptHindsight: that rate looks okay?
01:36 PM CaptHindsight: well you can but it's at the voting level vs them adding extra profit to a sale
01:36 PM mrec: 15.994 oz 13$ looks pretty much similar like in Germany
01:37 PM CaptHindsight: for a flat item <2mm
01:37 PM CaptHindsight: a blank PCB might make it
01:38 PM mrec: it will be put into a box anyway
01:38 PM mrec: 0.45kg letters?
01:38 PM CaptHindsight: magazine, documents etc
01:39 PM CaptHindsight: a pcb with parts needs bubble packing, that will get flagged as non flat
01:39 PM coredump6391[m]: It seems to have posted correctly
01:39 PM coredump6391[m]: what's the point of preloading the next tool?
01:40 PM mrec: The U.S. Postal Service offers Registered Mail for customers sending valuable or irreplaceable items through the mail. Registered Mail provides added protection for valuable and important customer and internal mail with evidence of mailing and delivery.
01:40 PM * coredump6391[m] sent a gcode code block: https://libera.ems.host/_matrix/media/r0/download/libera.chat/3e2cda60f5e3f7ab778a8349df4657e4a3a0093e
01:40 PM CaptHindsight: mrec: the old scam here was shipping and handling fees for 2-10X the actual shipping rate for items sold on TV
01:40 PM mrec: yes that's known
01:41 PM mrec: also free shipping where the shipping price is added to every single item
01:41 PM Tom_L: anything to steal a buck
01:41 PM coredump6391[m]: In this case I should start with the tool that I will use on the first operation, run the job, it will start with the M600 and then the M6 will use it to touch and set the offset for the next ones then
01:42 PM CaptHindsight: $60 for a pcb from USA to Taiwan is the just the reality of shipping from the USA
01:42 PM mrec: not with USPS sorry... registered mail is like 15$
01:42 PM Tom_L: CaptHindsight, lemme know how that RTAI test goes
01:42 PM coredump6391[m]: Oh, it does have `G0` to the A axis
01:42 PM coredump6391[m]: I don t have that setup on the machine
01:43 PM mrec: I would never choose the postal service for express delivery
01:44 PM CaptHindsight: i used to ship 6Kg cans of liquid internationally, I ended up finding a rectangular metal can that just fit into the flat rate USPS box or it would cost >$100 for the same weight if it didn't fit into the flat rate box to 20Kg
01:45 PM mrec: the boxes which I used for PCBs are usually between 3-5cm (roughly 1-2 inch thick)
01:45 PM CaptHindsight: USPS and FEDEX don't offer a flat rate it would be ~$120 if we used them for the same, yet I could ship ~75Kg by air for only ~$350 to just about anywhere on Earth
01:46 PM mrec: if you ship with fedex or ups look for a broker
01:46 PM CaptHindsight: mrec: the problem with brokers is when things go wrong
01:46 PM coredump6391[m]: welp just changed from `if true` to `if false` around line 303 and the A axis seems not to be around anymore on the code
01:47 PM CaptHindsight: the y don't accept responsibility for loses or damage
01:47 PM CaptHindsight: you pray that nothing goes wrong
01:48 PM mrec: CaptHindsight: ups/dhl/etc. is behind the brokers in Europe/Asia as mentioned
01:48 PM CaptHindsight: oh sorry that YOU packed it wrong, WE don't see you package in our system, did you wait 30 days to see if it arrives, etc etc
01:48 PM mrec: www.transglobalexpress.com
01:49 PM roycroft: the ups strike is going to be interesting
01:49 PM mrec: this is not a shipping company they just forward everything to ups/dhl/tnt/etc.
01:49 PM CaptHindsight: the best deals with fewest problems was always asking the customer to send us a shipping label with the "deal" that they found
01:49 PM roycroft: especially if it lasts for several weeks, as it well may
01:50 PM mrec: yes
01:50 PM CaptHindsight: so when it leaves our dock we are no longer responsible
01:51 PM mrec: anyway I'll wait for my mesa card and I also ordered a development kit
01:52 PM mrec: I have some other fpga design here but lattice doesn't seem to be a good deal at the moment
01:52 PM CaptHindsight: https://i.imgur.com/nFAsIhU.jpg good luck getting a low cost broker to fix this problem
01:53 PM mrec: it well depends what you're shipping
01:53 PM mrec: a single pcb inside foam .. you can kick it and nothing will happen
01:54 PM mrec: my lathe also arrived slightly damaged back then
01:55 PM CaptHindsight: standard ocean shipping insurance from the USA also has a $10k deductable
01:57 PM coredump6391[m]: well it did go nowhere near my WCS but I think I know why
01:58 PM CaptHindsight: shipping UPS for a ~35 KG box international is like $800 by parcel service (max weight of the service)
01:58 PM CaptHindsight: 75Kg by air freight UPS international ~$350
02:01 PM mrec: 75kg ..350$ ... vs <1kg pcb 60$ :D
02:01 PM * roycroft is a fan of making stuff near where it's going to be used, and not messing around with all this international shipping nonsense
02:02 PM CaptHindsight: yes, I'd put a single box of 20Kg on a small pallet and ship by air freight vs parcel service to save $
02:05 PM CaptHindsight: for some reason shipping from the USA is always higher than any other direction, must be that all the headwinds are towards the USA vs every other country
02:05 PM CaptHindsight: it just takes way more effort to get things out of the USA :)
02:05 PM mrec: well Mesa has similar prices within the US :D
02:05 PM mrec: https://snipboard.io/4HiULb.jpg
02:06 PM mrec: ok fedex ground is an exception
02:06 PM CaptHindsight: it's not conspiracy by Mesa, US shipping just costs more
02:06 PM mrec: you can get bad rates if you pick the wrong service for your product
02:07 PM mrec: anyway I'm done with that topic ... I paid my price already and just wanted to point out but not to discuss so much about it.
02:07 PM CaptHindsight: we also have regional issues with mail and parcels, some carriers just have poor service in some areas
02:08 PM CaptHindsight: Mesa will even warn you if you want to choose a carrier that they have issues with
02:08 PM CaptHindsight: they will ship the way YOU ask , but you have been warned
02:08 PM Tom_L: i suppose crying over postal rates is better than politics but really... haven't we covered it?
02:09 PM mrec: haha
02:09 PM mrec: post office politics
02:09 PM Tom_L: just suck it up and buy the board! :)
02:10 PM mrec: did already as mentioned
02:10 PM CaptHindsight: I'm sorry that your life has been so dark
02:10 PM mrec: time to talk about what to do until it arrives :D
02:10 PM Tom_L: heh
02:10 PM Tom_L: mrec, go get one of those dev boards too
02:10 PM mrec: did already ... waiting for it to arrive
02:11 PM Tom_L: they were $30 when pcw mentioned them now they're $75 here
02:11 PM mrec: https://www.digikey.com/en/products/detail/efinix-inc/T20F256C-DK/10654491
02:11 PM mrec: I have no need for small fpgas anymore
02:11 PM mrec: got enough of them
02:12 PM mrec: ICE40UP5K - nice chip and we have some stock
02:13 PM Rab: CaptHindsight, I've tested a fresh Debian 11 installation with a scratch 2.9 build + machine config and everything worked fine with both of the CH382 cards I have. No funny kernel options or other tweaks required.
02:13 PM roycroft: usps rates are all about politics
02:13 PM roycroft: private courier rates are all about profit
02:14 PM Tom_L: Rab, is that with the stock kernel?
02:14 PM Rab: It wasn't exactly a high-performance test, I ran the LinuxCNC test job on my machine a couple of times. I also didn't push the input very hard--no encoders or anything--but the limit switches worked.
02:14 PM Rab: Tom_L, stock RT from the Debian repo, 5.10.0-17-rt-amd64.
02:14 PM Tom_L: huh
02:14 PM mrec: no wonder that lattice wanted to be taken over... efinix is killing them
02:15 PM Tom_L: i just built a kernel for i think debian 11
02:15 PM Tom_L: 5.15.64 iirc
02:15 PM CaptHindsight: Rab: I still have to see what is up with the ch382, thanks for testing
02:16 PM Rab: This is on an HP 8300 SFF, but I don't see how the platform would make a difference...unless CH382 is somehow broken with some PCIe implementations.
02:17 PM CaptHindsight: Rab, do you recall the LSPCI wit the ch382?
02:17 PM CaptHindsight: I'd be interested at what you get and which base address works
02:18 PM Rab: CaptHindsight, it's not very verbose:
02:18 PM Rab: 01:00.0 Serial controller: Device 1c00:3250 (rev 10)
02:18 PM Rab: 02:00.0 Serial controller: Device 1c00:3250 (rev 10)
02:18 PM Rab: (Two cards installed)
02:18 PM CaptHindsight: lspci -vv and just post the section for the ch382
02:18 PM CaptHindsight: yeah those are the pci id's
02:18 PM CaptHindsight: same as mine
02:19 PM Rab: http://reboots.g-cipher.net/linuxcnc/lspci.txt
02:19 PM CaptHindsight: ok similar
02:20 PM CaptHindsight: which base address works, the lower or the higher of the two?
02:20 PM CaptHindsight: for example: Region 0: I/O ports at e000 [size=256]
02:20 PM CaptHindsight: Region 1: Memory at f0100000 (32-bit, prefetchable) [size=32K]
02:20 PM CaptHindsight: Region 2: I/O ports at e100 [size=4]
02:20 PM CaptHindsight: Expansion ROM at f7d00000 [disabled] [size=32K]
02:20 PM CaptHindsight: e000 or e100?
02:21 PM Rab: I'm getting these kernel messages: parport0: PC-style at 0xe100, irq 16 [PCSPP,TRISTATE]
02:21 PM CaptHindsight: ok the higher of the two
02:21 PM Rab: parport1: PC-style at 0xd100, irq 16 [PCSPP,TRISTATE]
02:21 PM CaptHindsight: great, thanks
02:22 PM Rab: Sure, let me know if there's anything I can provide in more detail.
02:23 PM CaptHindsight: OK give me a few minutes, I have to swap PC's over at the bench
02:24 PM Rab: All the serial ports work for both cards as well, but I have to pass a kernel option to support UARTS >4 (IIRC).
02:25 PM coredump6391[m]: ok, worked almost perfectly
02:25 PM coredump6391[m]: for some reason the last tool was cutting waaaay deeper it was supposed to
02:34 PM CaptHindsight: ah ok mine are single port LPT cards, 1c00:3050
02:47 PM Rab: MDSEL being pin 63 of the LQFP-64.
02:48 PM CaptHindsight: there a kernel PCI ID issue with the ch382
02:48 PM CaptHindsight: was and maybe it's back
02:49 PM CaptHindsight: I have to fix the issue that skunkworks found as well with 2.9
02:49 PM CaptHindsight: right now I can see the encoders effect on the spindle GUI in Axis but I can't see the encoder signals in halscope
02:50 PM CaptHindsight: spinning the encoder actually generates the signal 8 error
02:50 PM CaptHindsight: but I get NO steps out of the LPT port
02:51 PM CaptHindsight: the ch384 may be pin strapped to 4 different PCI id's
02:51 PM CaptHindsight: the kernel was missing a couple of the options
02:51 PM Rab: Indeed, pin 63 on my card goes to pads for an unpopulated resistor to GND (R4 on these cards). I guess I could perform the experiment in reverse, add a jumper to GND and see if the now-single-parport card magically stops working.
02:52 PM CaptHindsight: so i get inputs, but no outputs here
02:52 PM Rab: No EEPROM on these cards, which I guess is typical.
02:52 PM CaptHindsight: yeah, same here
02:53 PM CaptHindsight: ok so this looks fixable, have to make one fix at a time
02:53 PM CaptHindsight: i can also strap this card to get your version PCI id
02:54 PM CaptHindsight: but we found a kernel fix for this that was backported years ago
02:54 PM CaptHindsight: maybe they dropped the fix or something else changed
02:55 PM CaptHindsight: but now I am wondering why some signals get into the GUI but not into halscope or vice versa
02:55 PM Rab: https://i.imgur.com/FH9pgu0.jpg https://i.imgur.com/zYm4MkO.jpg
02:56 PM Rab: (Single serial card, I populated the footprints for the second port)
02:56 PM CaptHindsight: yeah mine are the same, just missing the serial port parts
02:59 PM Rab: Unfortunately this year's cards are cosmetically different (and cost somewhat more), so I'm not sure this is working toward anything that could go in the wiki beyond "buy a card, and maybe it'll work".
03:01 PM perry_j1987: man lots of rain
03:01 PM perry_j1987: lucky my shop's not flooded heh
03:01 PM CaptHindsight: http://www.wch-ic.com/downloads/file/140.html page #4
03:02 PM XXCoder: perry_j1987: hope it rains here soon lol smokey now
03:02 PM XXCoder: not too bad, 66 now
03:03 PM perry_j1987: we're getting straight rain from last night till sometime tomorrow nonstop
03:03 PM CaptHindsight: http://www.wch-ic.com/downloads/file/320.html page #16
03:03 PM CaptHindsight: Rab: i have bunch from over several years, they used to work, always the 1 LPT 0 serial version
03:04 PM CaptHindsight: something must have changed with device ID list in the kernel
03:04 PM CaptHindsight: I'll figure it out again
03:05 PM Rab: I had thought that the issue with the multi-port cards a few years ago was that the kernel tried to bind a special driver for multi-port, and that had to be disallowed. That's part of why I did an install from scratch, to make sure there were no tweaks I had done and forgotten about (e.g. blacklisted modules).
03:05 PM CaptHindsight: https://linux.kernel.narkive.com/ifx1Ejfx/patch-parport-serial-add-support-for-wch-ch382l-pci-e-single-parallel-port-card-adapter
03:07 PM Rab: So your cards have no serial ports but are showing as 1c00:3250 rather than :3050?
03:07 PM Tom_L: CH3510 chip any good?
03:07 PM CaptHindsight: Rab: yeah I never had your problem, though other had mentioned it being the issue they also had
03:07 PM Tom_L: just found one of those
03:09 PM CaptHindsight: Rab : 1c00:3050 1p 0s
03:09 PM Rab: Ahh OK
03:11 PM CaptHindsight: I have never seen a CH3510
03:11 PM Tom_L: PCI
03:11 PM CaptHindsight: I have some CH351
03:11 PM CaptHindsight: yes, double check the part number
03:11 PM Tom_L: hard to read but i think it's got an '0' on the end
03:11 PM CaptHindsight: and yes CH351 PCI LPT works great
03:12 PM CaptHindsight: http://www.wch-ic.com/products/CH351.html
03:13 PM Rab: I'm gonna shut down and add this jumper.
03:13 PM Tom_L: CH351Q
03:15 PM Tom_L: also UL7512EQ
03:19 PM CaptHindsight: http://www.wch-ic.com/downloads/file/173.html
03:20 PM CaptHindsight: ch351 data sheet ^^
03:22 PM coredump6391[m]: Ok so it work 99% perfectly. All the cuts measured and tool changes and etc, but not the last operation. My first try, the last operation was off by a lot, and then I fixed what I thought it was wrong (the stock size) and tried again, but the first contour pass was fine, 0.5 mm, and then when it goes for the second pass at 1mm it plunged all the way to the MDF table, and looking at the coordinates it look like it was 10mm off
03:22 PM coredump6391[m]: I don't understand why it changes only on the last of the ops tho
03:22 PM GTKplusplus6432[: no idea
03:23 PM GTKplusplus6432[: I've had that happen a couple times and I never figured out what causes it
03:23 PM GTKplusplus6432[: but it's not only on the last operation
03:23 PM GTKplusplus6432[: also in my case it stay higher
03:23 PM GTKplusplus6432[: and it's very sporadic
03:23 PM coredump6391[m]: well, I guess I was unlucky
03:23 PM GTKplusplus6432[: s/stay/stays/
03:23 PM coredump6391[m]: maybe I put a dummy last operation 😄
03:23 PM Tom_L: did you touch off a surface you just cut? :)
03:24 PM GTKplusplus6432[: I'd test it with a few different files and dummy safe wcs
03:24 PM GTKplusplus6432[: just to figure out what causes this
03:38 PM CaptHindsight: 60F and rain, is this Canada?
03:41 PM Tom_L: 77 & sunny here today
03:41 PM Tom_L: yesterday was in the 60's and light rain
03:47 PM CaptHindsight: our Maples are already changing colors
03:47 PM CaptHindsight: it's been so dry
03:49 PM XXCoder: still moneky hellscope here
03:49 PM XXCoder: not too bad though 60 or so now
03:53 PM Rab: CaptHindsight, well...I jumpered R4 and the card showed up as a 3050 single-port card. But it worked just as well as before, steps out on all three axes and limit switch/estop inputs. http://reboots.g-cipher.net/linuxcnc/lspci-3050.txt
03:53 PM CaptHindsight: ok weird
03:57 PM Rab: There are some interesting differences in lspci output.
03:57 PM Rab: BusMaster+ (3259) and BusMaster- (3050)
03:57 PM Rab: er 3250
03:58 PM Rab: 3250: Kernel driver in use: parport_serial, Kernel modules: parport_serial
03:58 PM Rab: 3050: Kernel driver in use: parport_pc Kernel modules: parport_pc, parport_serial
03:59 PM Rab: Anyway...
04:15 PM CaptHindsight: have only been testing one ch382 card here today, will try the other later
04:16 PM CaptHindsight: so far no outputs or inputs even with 2.8 and 5.10 kernel
04:17 PM CaptHindsight: I bet it's a typo in the driver now
04:19 PM CaptHindsight: anyway long day yesterday getting new RTAI for 2.9
04:20 PM CaptHindsight: Rab: I think we have enough to go on now, thank you
04:21 PM Rab: CaptHindsight, happy to help.
04:29 PM CaptHindsight: ok I jump back to 2.9 and 5.18 kernel and all I get is inputs on the ch382, no outs work :(
04:31 PM CaptHindsight: I have encoder inputs to GUI and in halscope
04:32 PM CaptHindsight: I see steps in halscope generated but no motor movement, will have to verify with a scope on LPT pins
04:35 PM CaptHindsight: halscope looks correct, I can even see the accel and decel of the steps
04:39 PM skunkworks: printer ports are vogue again?
04:39 PM Tom_L: ppl must really be bored
04:39 PM skunkworks: lol
04:53 PM CaptHindsight: I don't think that anything has changed. People just make commits without testing.
04:55 PM CaptHindsight: and they don't stomp on Peters hm2
04:56 PM CaptHindsight: and Jon supports his own drivers
04:56 PM CaptHindsight: looks we have to do similar here
04:57 PM CaptHindsight: yet LPT is not dropped from LCNC
04:57 PM CaptHindsight: and people waste time on Rpi4 images
04:58 PM skunkworks: mach 4 will run the printer port. but I think you need 32 bit windows 10.. And it doesn't thread anymore..
04:58 PM CaptHindsight: are they bored or just have personality disorders?
04:59 PM XXCoder: mach yuck
04:59 PM XXCoder: but why drop lpt from lcnc?
04:59 PM skunkworks: oh - windows 7
04:59 PM skunkworks: 32
04:59 PM skunkworks: bit
05:00 PM skunkworks: or xp
05:01 PM CaptHindsight: no good Mesa solution in stock for closed loop steppers
05:01 PM Tom_L: CaptHindsight, bored :)
05:01 PM CaptHindsight: can't fool me
05:02 PM Tom_L: for my purpose was more a learning thing
05:02 PM Tom_L: i have no plans to use the rpi4
05:03 PM Tom_L: so far i learned how to freeze the bios
05:03 PM XXCoder: maybe you can do one of those alternates capt linked to
05:03 PM CaptHindsight: NVEM are being replaced by EC300 that uses an IMX micro, will takes weeks to get support
05:03 PM XXCoder: ones thats good and far stronger than rpi4
05:03 PM XXCoder: and for now, cheaper
05:04 PM Tom_L: i have none to test on
05:04 PM Tom_L: i had this already
05:04 PM CaptHindsight: https://forum.linuxcnc.org/18-computer/44828-remora-ethernet-nvem-cnc-board?start=380#251551
05:05 PM Tom_L: i wonder how bullseye/rtai would do on old hardware
05:05 PM XXCoder: tom lol yea
05:05 PM JT-Shop: just because the rpi4 is unobtainum don't mean people don't already have some, I have 3 or 4
05:06 PM CaptHindsight: oh great, I got mine, go get yours again
05:06 PM Tom_L: mkay
05:06 PM pcw-home: You could use a 7I96S+encoder daughter card for step/dir + secondary feedback
05:06 PM XXCoder: my rpi4 predates scalping too
05:06 PM XXCoder: got it to control my laser machine
05:07 PM pcw-home: we should have 7I95s with Efinix FPGA in a bit
05:07 PM JT-Shop: cool
05:07 PM pcw-home: (so step/dir + encoder on on card)
05:07 PM pcw-home: on one
05:07 PM CaptHindsight: 7I96S Availability: 9
05:07 PM JT-Shop: I have 20
05:08 PM JT-Shop: that is if I can find the one I misplaced...
05:09 PM CaptHindsight: pcw-home: 7I95s with Efinix FPGA in quantity and regularly?
05:09 PM pcw-home: Yep no issue (DK has 36K currently)
05:09 PM CaptHindsight: great
05:10 PM CaptHindsight: I was actually considering sending you used and reballed Xilinx to make me some boards
05:10 PM pcw-home: we have ~650 7I96S in process, 7I92 next then 7I95,7I97 and 7IU6E replacements
05:11 PM pcw-home: 7I76E replacement will use new output driver chip that supports source,sink,push-pull
05:11 PM JT-Shop: neat
05:12 PM pcw-home: so replaces 7I76E and 7I76ED
05:12 PM JT-Shop: will it be a 7i76ES?
05:13 PM pcw-home: glad we didn't wait for Xilinx, Lattice, or Microchip
05:13 PM pcw-home: maybe 7I76EU
05:13 PM JT-Shop: U for universal?
05:14 PM pcw-home: Yes (also 7I84U)
05:14 PM XXCoder: pcw those boards less sensive to chip issues?
05:15 PM pcw-home: Well, they use available FPGAs (OPTOs and Ethernet chips are still an issue)
05:16 PM XXCoder: hmm ok
05:16 PM JT-Shop: are you making one for the 7i71/7i72 with the new output driver chip?
05:16 PM pcw-home: we are getting chips (With 2022 date codes) but demand so far exceeds supply so long lead times
05:16 PM pcw-home: Yes 7I71U Eventually
05:17 PM CaptHindsight: https://www.mouser.com/ProductDetail/Microchip-Technology-Atmel/LAN8720A-CP?qs=pA5MXup5wxF9mfEVLnGX8g%3D%3D
05:17 PM CaptHindsight: 39,200
05:17 PM CaptHindsight: Expected 01-Dec-23
05:18 PM pcw-home: yep Almost all Ethernet chips are in short supply
05:19 PM XXCoder: sucks
05:19 PM pcw-home: OPTOs, MOSFETs also
05:20 PM CaptHindsight: too risky to fly to Shenzhen and shop since the lockdowns, land run to the shops and get stuck there for a month
05:20 PM pcw-home: They are being made but most dont trickle down to the peons buying through distribution
05:20 PM CaptHindsight: sorry we nailed your hotel room door shut for your safety
05:21 PM CaptHindsight: grey market
05:24 PM CaptHindsight: pcw-home: which is the matching daughter card for 7I96S for encoder inputs?
05:25 PM pcw-home: Either 7I85, 7I85S or 7I89 (4,4,8 encoders, 5,1,1 RS-422, 0,4,0 Step/Dir)
05:28 PM CaptHindsight: ok db25
05:29 PM CaptHindsight: pcw-home: is there a chart like JT's that has more of the models available? http://mesaus.com/ethernet-cnc-solutions/
05:29 PM Tom_L: we had a guy living there that would do orders but i don't think he's there now
05:29 PM JT-Shop: and I'll need to update that soon
05:30 PM Tom_L: in fact i think he just got outta prison :)
05:30 PM JT-Shop: that is a pretty cool chart if I do say so
05:31 PM Tom_L: didn't he drop the 7i96?
05:31 PM Tom_L: with no parts available
05:32 PM Tom_L: the s being the first with the new chips
05:34 PM JT-Shop: I assume the 7i96 is a goner
05:34 PM JT-Shop: tis why I need to update the chart if I can remember how I made it lol
05:35 PM JT-Shop: bad link too
05:35 PM pcw-home: Well if SP6 parts become available (say in 2 years...)
05:36 PM JT-Shop: I better leave it just in case :)
05:36 PM JT-Shop: did they stop making the spartan 6?
05:37 PM pcw-home: I had planned to change to Artix/SP7 before the chipageddon
05:37 PM pcw-home: Hard to tell, its not been announced and the still have SP3 and some SP2 chips available
05:38 PM pcw-home: they still
05:38 PM JT-Shop: is the sp7 available
05:38 PM pcw-home: nope
05:38 PM pcw-home: well maybe in a year
05:38 PM Tom_L: sp2 & 3 smaller chips or just older?
05:39 PM pcw-home: I think they have keep the high volume ones around (but still 52 week or so leadtime)
05:41 PM JT-Shop: wow
05:46 PM CaptHindsight: everyone wants the low cost parts with enough gates to get the job done, I see no problem getting the $200+ FPGA's
05:46 PM -!- #linuxcnc mode set to +v by ChanServ
05:46 PM Tom_L: JT-Shop, what rpi4 image was that you downloaded?
05:49 PM JT-Shop: pretty sure it was https://downloads.raspberrypi.org/raspios_full_armhf/images/raspios_full_armhf-2022-09-07/2022-09-06-raspios-bullseye-armhf-full.img.xz
05:50 PM Tom_dev: 32bit
05:50 PM Tom_dev: https://www.raspberrypi.com/software/operating-systems/#raspberry-pi-os-64-bit
05:50 PM Tom_dev: i got the 64bit bullseye on a card
05:52 PM Tom_dev: built what i though was a rt kernel for it but now i'm not so sure
05:56 PM JT-Shop: BB Good Vibrations
05:57 PM Tom_dev: i figured the rt kernel patches from kernel org would be realtime
05:57 PM JT-Shop: making a cat play house... was 4' x 8' now it's 4' x 12' 4' x 8' screen panels with 2x4 PT is a bit awkward to handle
05:57 PM JT-Shop: run the latency test, you can do that in mesact now
05:58 PM Tom_dev: none of the lcnc is installed on it
05:58 PM JT-Shop: ah
05:58 PM JT-Shop: what does it do when you try and run the latency test from mesact?
05:58 PM Tom_dev: at least i don't think it is... i've been messin with several SD images
05:58 PM Tom_dev: i can check
05:59 PM Tom_dev: bullseye or bookworm?
06:00 PM JT-Shop: should not make a diff
06:00 PM Tom_dev: i'll start with bullseye
06:00 PM JT-Shop: I expect a crash now that I think about it
06:00 PM Tom_dev: it probably doesn't have the latest installed either
06:01 PM -!- #linuxcnc mode set to +v by ChanServ
06:01 PM Tom_dev: where's the tool located?
06:01 PM JT-Shop: options I think
06:02 PM Tom_dev: also no option to dl the arm64 file
06:02 PM Tom_dev: armhf is it
06:02 PM Tom_dev: i can pull it off my server for now unless you changed it since then
06:03 PM JT-Shop: I need to add that
06:04 PM Tom_itx: don't see it in this ver
06:04 PM JT-Shop: I added it 13 days ago
06:05 PM Tom_itx: k, i'll pull it off my server for now
06:05 PM JT-Shop: maybe pc tab
06:05 PM Tom_itx: yes
06:05 PM Tom_itx: latency histogram test
06:06 PM JT-Shop: yep
06:06 PM Tom_itx: servo thread test
06:06 PM JT-Shop: just accept defaults
06:06 PM CaptHindsight: Rab one of my ch382 is bad, no outputs, the other ch382 is fine now
06:06 PM Tom_itx: woopsie
06:06 PM CaptHindsight: INs and OUTs work
06:06 PM Tom_itx: PE in subprocess.py ln 1823
06:06 PM JT-Shop: crash?
06:06 PM Tom_itx: no such file or dir
06:06 PM JT-Shop: I thought so, I need to disable if lcnc is not found
06:07 PM Tom_itx: k, gonna load bookworm now, i think it may have lcnc on it
06:07 PM JT-Shop: ok
06:08 PM Tom_dev: and i just grabbed their version of buster 32bit to see if it has the USB issue
06:12 PM JT-Shop: Van Morrison Moondance
06:12 PM * JT-Shop wanders inside
06:13 PM -!- #linuxcnc mode set to +v by ChanServ
06:13 PM Tom_dev: nope, not on bookworm either
06:13 PM JT-Shop: ok
06:13 PM JT-Shop: I'll fix that in the morning
06:14 PM Tom_dev: installing on bookworm
06:14 PM Tom_dev: linuxcnc-uspace
06:14 PM Tom_dev: on bullseye will probably have to be built
06:14 PM JT-Shop: wow there are more Tom's than JT
06:14 PM JT-Shop: 's
06:15 PM Tom_dev: hah
06:15 PM Tom_dev: gonna run outta nicks soon
06:15 PM JT-Shop: I need a blue dot for JT-Shop2
06:16 PM Tom_L: bad thing is they're all right in front of me instead of scattered across the property
06:16 PM Tom_L: associate it with your main nick and it will
06:18 PM Tom_shop: yeah, this one has the kernel i built too
06:18 PM Tom_shop: be kinda interesting to test that
06:19 PM Tom_shop: Linux raspberrypi 5.15.56-v8+ #1575 SMP PREEMPT Fri Jul 22 20:31:26 BST 2022 aarch64 GNU/Linux
06:28 PM Tom_dev: ok installing mesact on it now
06:28 PM Tom_dev: lcnc installed ok
06:31 PM -!- #linuxcnc mode set to +v by ChanServ
06:31 PM Tom_shop: JT-Shop, it failed to find linuxcnc on this copy
06:31 PM Tom_shop: it did find mesaflash
06:38 PM Tom_shop: and found the latency test
06:38 PM Tom_shop: no Linuxcnc version in the Machine tab though
06:59 PM skunkworks: lol - andy already figured out what line in the encoder code it was..
07:03 PM CaptHindsight: yeah looking at it now
07:11 PM CaptHindsight: skunkworks: what version debian did you use to bisect 2.9 LCNC?
07:15 PM skunkworks: uh..
07:16 PM skunkworks: CaptHindsight: buster
07:16 PM CaptHindsight: discovered the LCNC 2.9 repo for Buster the other day
07:16 PM CaptHindsight: yeah we only had 2.9 on Bullseye and newer
07:17 PM CaptHindsight: thank you
07:19 PM CaptHindsight: ok so ch382 issue solved, bad hardware and you have to manually edit hal with the LPT base address
07:20 PM CaptHindsight: and we need to back out not being able to manually input LPT addresses into Stepconf
07:20 PM XXCoder: couldnt you find removal in history of git
07:20 PM XXCoder: see what rationale was
07:21 PM Tom_L: yeah that was an odd one
07:21 PM CaptHindsight: C Morel though maybe that the kernel should be able to find all the LPT ports and label them as ), 1,2,3, etc
07:21 PM XXCoder: that reminds me, I really need to push new mascot artowkr to git
07:21 PM CaptHindsight: but that is not what happens with all LPT cards
07:21 PM Tom_L: not willing to undo the changes?
07:21 PM XXCoder: maybe ill take chance to fix small amount of fixes to mascot
07:21 PM CaptHindsight: C Morely thought
07:21 PM skunkworks: right - I have a few that still require manual address input
07:22 PM CaptHindsight: C Morely said to make it a bug report so that they don't forget to revert it
07:23 PM CaptHindsight: and we can wait for Andy to fix his line that broke the encoders with LPT
07:23 PM skunkworks: kinda glad it was the hal encoder and not some weird realtime issue.
07:23 PM CaptHindsight: yeah
07:23 PM CaptHindsight: that was actually one reason why Alec worked on RTAI for 2.9
07:23 PM CaptHindsight: RTAI uses hardware for floating point
07:24 PM CaptHindsight: so it's very different than preempt_rt way with soft floats
07:24 PM skunkworks: If he can get the bug fixed - I hope to use the missing slot encoder on the emco - then I can have a real jog wheel.
07:25 PM skunkworks: (with 1 printer port) I could always add another one.. but.
07:25 PM CaptHindsight: well it looks like it compares two variables and gets their fifference
07:25 PM CaptHindsight: if the difference is Zero then there is a divide by Zero error
07:26 PM CaptHindsight: so the logic might be get the difference and if =0, do nothing
07:27 PM CaptHindsight: if different do blah, if not then do forget about it
07:27 PM CaptHindsight: or nothing
07:30 PM CaptHindsight: another option might be using an available CPLD to do the counting and then just EPP a value every 1mS back to LCNC
07:31 PM CaptHindsight: put the hal encoder into a CPLD and still use the parallel port
07:31 PM CaptHindsight: but if Peter is getting FPGA's in finally then the new cards will also do it
07:33 PM CaptHindsight: I was considering a CPLD or microcontroller on LPT port to do the encoder hal work
07:48 PM mrec: CaptHindsight: are you still awake?
07:48 PM CaptHindsight: sort of
07:49 PM Tom_L: who you kiddin.. you'll be up half the night
07:49 PM CaptHindsight: 19:49 here
07:49 PM mrec: I wonder does anyone know which ethernet card can be used reliably with linuxcnc?
07:49 PM Tom_L: most of them
07:49 PM skunkworks: agreed
07:49 PM mrec: 00:07.0 Bridge: NVIDIA Corporation MCP61 Ethernet (rev a2)
07:49 PM Tom_L: some you have to turn something off on
07:49 PM CaptHindsight: I have not heard of any that don't work
07:50 PM Tom_L: umm intel chips
07:50 PM CaptHindsight: but it's been a while
07:50 PM Tom_L: ask pcw-home about the intel chip ethernet cards
07:50 PM mrec: I just read that intel might not be so good :-)
07:50 PM CaptHindsight: are they still in production?
07:50 PM Tom_L: they're ok but you have to do something with them
07:50 PM mrec: "I suspect that we'll discover that certain NICs are better than others
07:50 PM mrec: at giving realtime performance, just like with the PCs themselves. For
07:50 PM mrec: instance, Sam has seen on one system that the onboard RTL-chipset NIC
07:50 PM mrec: worked reliably, and a PCI NIC with Intel chipset did not. I've seen
07:50 PM mrec: good performance with the only NIC I tried, a PCI-E NIC with an Intel
07:50 PM CaptHindsight: oh I haven't used anything intel since the 90's
07:50 PM mrec: chipset.
07:50 PM mrec: - jeff epler
07:51 PM Tom_L: realtek is more prevalent now
07:51 PM mrec: 8 years ago..
07:51 PM CaptHindsight: intel makes crappy stuff
07:51 PM CaptHindsight: and they lie
07:51 PM mrec: they all do...
07:51 PM skunkworks: hardware-irq-coalesce-rx-usecs 0
07:51 PM skunkworks: http://linuxcnc.org/docs/stable/html/man/man9/hm2_eth.9.html
07:51 PM CaptHindsight: and they don't share docs
07:52 PM mrec: I worked at AMD a long time ago... there was a chipset bug and they got around it with some microcode update which made it slower
07:52 PM Tom_L: yeah what skunkworks said...
07:52 PM mrec: before they figured out about that bug all the items were shipped to google already things were burning back then
07:52 PM CaptHindsight: yeah they ofetn hide fixes in BIOS
07:52 PM skunkworks: lol - I am the sam...
07:53 PM skunkworks: that was a long time ago
07:53 PM mrec: so the mesa ethernet cards should be directly connected or is it okay to go via some router?
07:53 PM CaptHindsight: best i recall was direct
07:53 PM skunkworks: yes - direct
07:54 PM Tom_L: no routers
07:54 PM CaptHindsight: but you can test
07:54 PM mrec: that's why I ask for ethernet cards, I'd have to put a second one into it
07:54 PM roycroft: connect it directly, and use a dedicated ethernet port for it
07:54 PM Tom_L: better off using the 10.10.10.xx addy
07:54 PM roycroft: do not put it on the same network as your lan connection
07:54 PM Tom_L: lots less confusion later
07:54 PM CaptHindsight: mrec: use USB-Ethernet if you run out of PCI/e
07:55 PM CaptHindsight: for networking and save the onboard NIC for Mesa
07:55 PM roycroft: what capthindsight said
07:55 PM roycroft: or use wifi for your lan
07:55 PM skunkworks: that is what I do
07:55 PM Tom_L: rpi uses wifi for lan and the wired for mesa
07:56 PM skunkworks: yep
07:56 PM Tom_L: yeah what he said
07:56 PM roycroft: that's what i do on the rpi and on the pc
07:56 PM skunkworks: The matsurra has 2 nics onboard.. :)
08:07 PM mrec: usb ethernet? it couldn't be more unreliable
08:07 PM mrec: usb > 1ms latency
08:08 PM mrec: I'm still curious about:
08:08 PM mrec: - gbit latency? any improvement there?
08:08 PM mrec: - USB 3 / USB 4
08:08 PM mrec: I'm very firm with usb 2.0, I have used USB 3.0 to do large transfers but did not pay attention to the latency itself
08:09 PM Tom_L: i personally would avoid usb
08:09 PM mrec: they have specified usb 4.0 v2 up to 80 gbit
08:09 PM mrec: that will be interesting
08:12 PM Tom_L: mmm last rt for rpi was 4.19
08:14 PM mrec: what step rate do you get with the RPI?
08:14 PM skunkworks: 10khz
08:15 PM mrec: ok for some applications...
08:15 PM skunkworks: out of gpio...
08:15 PM skunkworks: sure
08:15 PM skunkworks: mesa - different story :)
08:15 PM mrec: I just somewhat refuse to program my encoder since it's only OTP
08:15 PM mrec: the default is like 1000 ppr / 4000 steps
08:15 PM mrec: x4
08:16 PM skunkworks: otp?
08:16 PM mrec: andypugh used less than 100 for his hobbing work back then
08:16 PM mrec: one time programmable
08:16 PM skunkworks: huh
08:16 PM mrec: the allegro encoder can be programmed a few times at least
08:16 PM mrec: but that one is now built into my lathe
08:17 PM mrec: I'm loosing steps at 380 RPM when doing hobbing due to the high encoder resolution
08:17 PM skunkworks: The emco has a 100 line encoder in it.
08:17 PM mrec: (parallel port)
08:17 PM mrec: 180 rpm is the minimum of my mill
08:18 PM mrec: the index pulse of the AS5047P is very short sometimes
08:18 PM mrec: allegro is a little bit more adjustable
08:19 PM skunkworks: mrec: what is your base thread?
08:19 PM mrec: something like 20k
08:20 PM skunkworks: should be able to approch 750 rpm
08:20 PM mrec: ah actually 30k on the mill
08:20 PM skunkworks: ok
08:21 PM mrec: the index pulse was like 25-50 microseconds (average what I saw)
08:21 PM skunkworks: 380 seems plausable.. :)
08:22 PM mrec: and single ended with noise definitely not the best setup
08:22 PM mrec: the hobbed gear looks like a gear I just need a better fixture
08:23 PM Tom_L: 3rd time's a charm
08:23 PM mrec: the pressure on a 0.8mm disc is too much if the fixture is too small it will bend a bit
08:25 PM skunkworks: I just did over 2 million counts with the emco - running 90ipm or 47625 pulses per second
08:25 PM mrec: I think the first thing that messes up here is the index pulse when the speed is higher
08:25 PM skunkworks: 1428rpm at the servo - 2000 counts per rev
08:26 PM skunkworks: sure - no index here
08:26 PM mrec: my lathe is also better with the allegro encoder
09:03 PM CaptHindsight: i just did >100k cycles of a glass bar feeder on LPT steppers with 1K rev encoders, without the encoders I was only off by 0.25 in at the end
09:05 PM CaptHindsight: so that was over 1 billion counts
09:05 PM CaptHindsight: 20uS base thread
09:06 PM CaptHindsight: 18m/min feed rates
09:08 PM Rab: Plus or minus? If minus, doesn't that 0.25 in imply lost steps? And how many would fit in that distance?
09:09 PM CaptHindsight: ~25 steps, most likely due to the bars slipping
09:09 PM CaptHindsight: since the encoder count was right on
09:10 PM CaptHindsight: i think the accel and decell was slightly too high
09:10 PM Rab: 40cm/sec is cooking.
09:12 PM CaptHindsight: glass rods/bars
09:12 PM CaptHindsight: I was trying to see how much anything that contacts the glass will wear
09:15 PM CaptHindsight: Rab: closer to 30cm/sec
09:16 PM CaptHindsight: I'm limited by the encoder counts of a 20uS base thread
09:37 PM mrec: I'll get another ethernet card today and do some FPGA tests with the de2-115
10:22 PM mrec: found a usb ethernet ... that might be used for internet so I can use the onboard for the de2-115
10:36 PM mrec: I have a roundtrip time of 280 microseconds (ping) average
10:58 PM CaptainHindsight: the bridges to IRC worked near flawlessly for over a year
10:58 PM CaptainHindsight: now down for the past 3 days
11:00 PM roycroft: it became a bridge too far, mayhap
11:01 PM XXCoder: CaptainHindsight: yeah matrix issue apparently., I went to libera channel and found that out
11:01 PM CaptHindsight: ah thought it was Libera bridge side
11:02 PM CaptHindsight: my messages from a minute ago were from Discord
11:02 PM XXCoder: yeah its working one way only
11:03 PM CaptHindsight: if it's a matrix issue why the Libera --> Discord direction is down?
11:03 PM XXCoder: matrix is where bridge bot is on
11:03 PM CaptHindsight: how does discord connect?
11:03 PM XXCoder: via bot in discord, connecting to matrix
11:04 PM XXCoder: its matrix input from irc thats kinda broken
11:04 PM XXCoder: it can still send to irc
11:04 PM CaptHindsight: Discord-->Matrix-->Libera
11:04 PM XXCoder: basically...
11:05 PM CaptHindsight: and Libera-->Matrix is down so nothing reaches back to Discord either
11:05 PM xxcoder[m]: nah discord to matrix to irc works
11:05 PM CaptHindsight: does matrix to discord work?
11:05 PM XXCoder: its reverse thats broken. matrix can send to discord, but not irc to matrix
11:06 PM XXCoder: rougish was last guy to say something on matrix, yeserday
11:08 PM CaptHindsight: so he broke it?
11:08 PM CaptHindsight: what did he say?
11:09 PM XXCoder: @roguish:matrix.orgBOT
11:09 PM XXCoder: —
11:09 PM XXCoder: 09/09/2022
11:09 PM XXCoder: for you rPi guys............... https://www.aaeon.com/en/p/subcompact-boards-gene-adp6/
11:09 PM CaptHindsight: i saw that
11:09 PM XXCoder: you got matrix account to test it?
11:10 PM CaptHindsight[m]: for you rPi guys............... https://www.aaeon.com/en/p/subcompact-boards-gene-adp6/
11:11 PM xxcoder[m]: yeah i see on both irc and discord
11:11 PM CaptainHindsight: for you rPi guys............... https://www.aaeon.com/en/p/subcompact-boards-gene-adp6/
11:11 PM xxcoder[m]: looks like its very specifically discord to irc, via mtrix thats broken
11:12 PM XXCoder: irc to discord, via matrix also
11:12 PM XXCoder: bit weird since matrix to either works lol
11:13 PM CaptHindsight: glad it didn't break anything that time
11:15 PM CaptHindsight: the Discord but should put a D after the nicks like matrix puts an m
11:16 PM CaptHindsight: [D}
11:16 PM XXCoder: it lasted pretty long without serious problems for a while besides short disconnects, after initial issues
11:16 PM CaptHindsight: a whole year
11:16 PM XXCoder: capt yeah it goes though matrix so all gets m
11:17 PM CaptHindsight: my message from Discord to IRC did not get a m with the nick
11:17 PM XXCoder: probably because your matrix one already exists
11:17 PM CaptHindsight: CaptainHindsight is on Discord olny
11:17 PM XXCoder: while mine its only discord or irc
11:17 PM XXCoder: ah maybe nick cut off
11:18 PM CaptHindsight: I had to use matrix niskserve to change my password on Libera
11:18 PM CaptHindsight: since I was not logged in and logged in at the same time
11:19 PM CaptHindsight: the libera admins were a bit confused
11:19 PM XXCoder: lol fun
11:19 PM CaptHindsight: I don't make the rules
11:19 PM XXCoder: so you can say "yes" when someone says if you was logged in or not
11:20 PM CaptHindsight: uh huh