#avr Logs

Nov 28 2017

12:27 AM anonnumberanon: when i want to write assembly (without embedding it in the .c source file) do I write it as a .o file, and possibly first compile a .c to a .o to already give the main assembly structure then edit the .o to make changes?
12:30 AM anonnumberanon: apparently there is also the .asm file format
12:30 AM anonnumberanon: oh i found the avr assembler guide pdf that will help
01:00 AM rue_mohr: you write a .S file, that is assembled to an .o file
01:00 AM rue_mohr: and then linked into the project
01:01 AM polprog: anonnumberanon: http://polprog.net/blog/asm
01:04 AM anonnumberanon: polprog, rue_mohr thanks
01:04 AM rue_mohr: polprog, nice.
01:04 AM polprog: :)
01:04 AM anonnumberanon: oh its yours
01:04 AM polprog: wanted to try my hand at writing
01:04 AM anonnumberanon: very nice layout
01:05 AM rue_mohr: now I' know I'm not the only one making "awefull" webpages that are more content than they are social media and blipverts
01:06 AM polprog: thanks ;)
01:08 AM rue_mohr: hmm
01:08 AM rue_mohr: if I could make a content search system for my website, I could makea search
01:08 AM rue_mohr: (no, I dont want to use google to search my own website)
01:08 AM rue_mohr: mostly cause google dosnt' search for what people ask it to search for
01:22 AM Casper: rue_mohr: a bad search is better than no search, go with google
01:24 AM rue_mohr: no, I dotn want a search my site that comes up with things from other peoples websites that dosn't even have the keywords they asked for
01:26 AM Casper: ow well, people just need to learn the site: keyword then
01:44 AM anonnumberanon: rue_mohr, next time i go to your server I'll wget -rf
01:44 AM anonnumberanon: and grep
01:44 AM anonnumberanon: the wget flag is probably wrong though
01:45 AM anonnumberanon: grep makes me happy most of the time
05:05 AM nuxil: rue_mohr, what are you up too ,, http://ruemohr.org/~ircjunk/doc/bomb.txt :p
05:08 AM thardin: excellent posadist praxis
05:11 AM thardin: The best way to avoid inhaling plutonium is to hold your breath while handling it
05:12 AM cehteh: you have to die anyway, why not just now :D
05:18 AM Ameisen: What's the minimum latency when having boards talk to one another over serial pins?
05:20 AM thardin: somewhere between 1-2 bytes worth of time?
05:20 AM polprog: depends on many factors
05:27 AM Haohmaru: minimum?
05:27 AM Haohmaru: zero
05:27 AM Haohmaru: [/theworldisperfect]
05:29 AM thardin: cehteh: what are you trying to do?
05:31 AM cehteh: me?
05:34 AM thardin: herp, I meant Ameisen of course
05:35 AM Ameisen: have two boards talk to one another - want one board to be a 'master' board, and delegate certain sensor reading tasks to the daughter board
05:35 AM Ameisen: and possible _another_ daughter board to drive steppers at a high frequency
05:36 AM Haohmaru: so you mean synchronous communication then?
05:36 AM thardin: spi or uart? or i2c?
05:37 AM Ameisen: whichever is best
05:37 AM thardin: spi then
05:37 AM thardin: max speed is F_CPU/2 I think
05:37 AM thardin: make your packets and replies as small as possible
05:38 AM cehteh: f_cpu/2 per bit
05:38 AM thardin: ye
05:38 AM cehteh: and dont forget about isr, parsing and all shit overhead
05:38 AM thardin: or 2/f_cpu for time (unit, units)
05:39 AM cehteh: that easily sums up to some hundred cycles
05:39 AM thardin: if both request and reply is a single byte and you poll SPIR you can get the latency pretty low
05:39 AM thardin: should detect the SPI byte coming in within 3 cycles or so
05:41 AM cehteh: sensor for positioning?
05:42 AM Ameisen: thermal
05:42 AM cehteh: thermal is slow anyway
05:42 AM thardin: thermal is slow
05:42 AM thardin: lol
05:43 AM cehteh: unless you want to focus a laser beam :D
05:43 AM thardin: even in laser stuff thermal is ages
05:43 AM cehteh: think about DVD writer
05:44 AM thardin: yeah but it's not measuring a temp really
05:44 AM Ameisen: also want to drive steppers... which is not slow
05:44 AM cehteh: decouple that
05:44 AM Haohmaru: it's a concentric book with a small eye reading it very fast
05:44 AM cehteh: pid controler
05:44 AM * Haohmaru hides
08:53 AM rue_mohr: mhm, whats t he scale for the d factor
08:53 AM rue_mohr: ?
09:29 AM Guest30926 is now known as A001
09:30 AM A001 is now known as Guest49989
12:04 PM polprog: https://www.youtube.com/watch?v=XPU72VmtHuY
03:14 PM Ameisen: if I get a chance, I will try to build avr-gcc 8, possibly with an attempted fix for 8-bit shifts in g++
03:14 PM Ameisen: I don't have a gcc build environment atm :(
03:15 PM polprog: cant wait for my pcbs
03:15 PM polprog: maybe i would put together a quick 8 digit multiplex prototype on a breadboard
03:16 PM polprog: to check how fast i have to strobe
03:18 PM Ameisen: I reworked the X and Y axes of my printer last night
03:18 PM Ameisen: neea
03:18 PM Ameisen: err
03:18 PM Ameisen: X and Z
03:18 PM polprog: Y is up-down?
03:18 PM Ameisen: still need to rework Y, but I'm doing a test print to test X and Z
03:18 PM Ameisen: Y is front-back
03:18 PM polprog: mm
03:18 PM polprog: im used to blender convention
03:19 PM Ameisen: I have a bedflinger, so X is mounted on Z
03:19 PM Ameisen: it sounds quieter, at least
03:19 PM Ameisen: question is if it prints better.
03:19 PM polprog: what kinda printer you have?
03:19 PM Ameisen: wanhao i3 plus clone
03:19 PM Ameisen: fairly modified at this point
03:19 PM Ameisen: as said before, will be building a new control board for it
03:19 PM Ameisen: I have the parts
03:19 PM Ameisen: need to assemble them and write the firmware
03:20 PM Ameisen: I have stepper drivers, need to figure out a good way to control them
03:20 PM Ameisen: you 'pulse' stepper drivers at their frequency
03:20 PM Ameisen: and they control the stepper motors
03:20 PM Ameisen: the problem is that I don't want to occupy a ton of the CPUs time doing the pulsing
03:20 PM Ameisen: I'd rather send linear/cubic equations to a daughter board to do that
03:20 PM Ameisen: and have the daughter board do the pulsing along the curve
03:20 PM polprog: math coprocessor
03:20 PM Ameisen: sorta.
03:20 PM Ameisen: more like 'pulsing coprocessor'
03:21 PM Ameisen: I'm not a huge fan of using ISRs like that on single-core systems
03:21 PM polprog: TI89 style
03:21 PM Ameisen: but i haven't found good multicore boards with many GPIO
03:21 PM polprog: im all for coprocessors
03:21 PM Ameisen: the only one I've found isn't on Amazon anymore
03:21 PM Ameisen: :(
03:21 PM polprog: breaking stuff into logical modules
03:21 PM Ameisen: https://www.amazon.com/gp/product/B01FVY8OOS/ref=ox_sc_sfl_title_1?ie=UTF8&psc=1&smid=A2NAP359VON41Y
03:21 PM Ameisen: I would love that
03:21 PM Ameisen: could dedicate an entire core to _just_ pulsing
03:21 PM polprog: http://www.datamath.org/Graphing/Images/TI-89_PCB2.jpg ti89 pcb
03:22 PM Ameisen: yup
03:22 PM Ameisen: there's a few printers that are ARM plus AVR coprocessors
03:22 PM Ameisen: I'm using serial to do it, but that's prolly what those boards do, too
03:22 PM Ameisen: they just do it with a single board and traces
03:22 PM Ameisen: that's why I was asking about latency last night
03:22 PM Ameisen: wanted to make sure I could actually push the data to the daughter boards fast enough to actually make it work
03:22 PM polprog: if you were realy hardcore you could make it with discrete parts
03:23 PM Ameisen: that wouldn't be fun
03:23 PM Ameisen: that would also be incredibly slow
03:23 PM polprog: depends on the fun definition
03:23 PM polprog: ;)
03:23 PM Ameisen: though technically you could implement the algorithms themselves with discrete parts
03:23 PM Ameisen: that's how older furnaces work
03:23 PM Ameisen: furnaces before the 2000s don't have processors
03:23 PM Ameisen: they are entirely discrete
03:23 PM polprog: i was thinking about, say, digital pot with a 555
03:23 PM Ameisen: they were discrete because they had to be _provable_
03:23 PM Ameisen: so they wouldn't start fires
03:23 PM Ameisen: around the 2000s, they managed to get approval to use microprocessors
03:23 PM polprog: anyway, i dont know much about in this field
03:24 PM Ameisen: which made the controller boards for furnaces MUCH cheaper and simpler
03:24 PM polprog: that''s interesting
03:24 PM Ameisen: but... question is
03:24 PM Ameisen: if your furnace might be in a situation where your house could burn down
03:24 PM Ameisen: do you trust C / C++ ona microprocessor more
03:24 PM Ameisen: or a bunch of mathematically proven discrete parts?
03:24 PM Ameisen: other issue I'm dealing with is the bed on this printer is... designed poorly
03:24 PM Ameisen: a plane is defined by three points
03:25 PM Ameisen: it has four bearings
03:25 PM Ameisen: which is overconstrained
03:25 PM polprog: i have to take some sleep now
03:25 PM Ameisen: so, you want three bearings
03:25 PM polprog: night
03:25 PM Ameisen: however... you want the 'force' (where the belt is attached) on the side with two bearings
03:25 PM Ameisen: however... they added the mount holes for a single 'center' bearing... on the side with the belt
03:25 PM Ameisen: not the other sidfe
03:25 PM Ameisen: :|
03:25 PM polprog: :)
03:25 PM Ameisen: so
03:25 PM Ameisen: I need to bring the plate to a metal shop
03:25 PM Ameisen: have them either make me a better one, or just drill holes
03:25 PM Ameisen: it's aluminum
03:25 PM Ameisen: I dunno if there's better, lighter alloys?
03:26 PM Ameisen: carbon fiber is better, but nobody has made carbon fiber ones for this system
03:26 PM polprog: have fun
03:26 PM polprog: i have to go
03:27 PM polprog: to be able to function at 700 tomorrow
03:27 PM polprog: ;)
06:59 PM enh: hi. good night
11:34 PM day__ is now known as daey