#robotics Logs

Dec 10 2015

09:21 adam789654123 hey robotustra
09:25 robotustra how is the progress?
09:25 adam789654123 great
09:25 adam789654123 im working on interpreting my interpreter
09:25 adam789654123 i want to run the multithreaded program as a thread in the multithreaded program
09:26 adam789654123 then i could create a tree of threads
09:26 adam789654123 which seems cool to me
09:26 adam789654123 how about you?
09:26 robotustra did a small site
09:26 robotustra working on robot base
09:26 adam789654123 a small website?
09:27 adam789654123 i gotta get my windowing system up
09:27 adam789654123 ive just been insanely focused on this project
09:27 robotustra so, do it now
09:27 adam789654123 im in the sh*t
09:28 adam789654123 im tracking down a bug concerning two tiers of interpretation
09:29 robotustra on what platform?
09:29 adam789654123 raw common lisp
09:29 robotustra yesterday i recovered my edison
09:29 adam789654123 whats that?
09:29 robotustra for some reason it stopped to load kernel
09:29 adam789654123 oh
09:29 adam789654123 computer problems
09:29 robotustra edison is a small board from intel
09:30 adam789654123 ok
09:30 robotustra I think it's aflash problem
09:30 robotustra but it was passively laying on my table for 1 month
09:30 adam789654123 not sure off hand a good way to troublshoot it
09:30 adam789654123 but you fixed it i guess?
09:31 robotustra i just flash it again and it started to fork
09:31 adam789654123 oh
09:31 adam789654123 great
09:32 robotustra overwrite distro on top
09:32 robotustra but it looks suspisious for me
09:32 adam789654123 could be *so many* things to cause such a malfunction
09:33 adam789654123 from hardware to software
09:33 adam789654123 or both
09:33 adam789654123 you got it working, but maybe dont rely on it too much, eh?
09:35 robotustra it was just laying on my table
09:35 robotustra without power
09:35 robotustra it shouldn't be like this
09:35 robotustra it means that flash is not relyable
09:35 adam789654123 i just had an idea for a troubleshooting tool for my project
09:36 adam789654123 can you change the flash?
09:36 adam789654123 if so, that seems a good idea
09:36 robotustra I don't think I can do it easily
09:37 adam789654123 my idea is this, have a tool that allows for tagging forms such that you can see the evolution of the transformation as it is dissassembled and reassembled by the interpreter
09:37 robotustra what interpreter do you do?
09:38 adam789654123 so if a specific element is wrong, i can trace its evolution
09:38 adam789654123 just something i made
09:38 robotustra show me
09:38 adam789654123 i was thinking about tagging my next snapshot with "chaos" if that gives you any idea of what you would expect to find
09:39 adam789654123 but.... i happen to know how to use this specific chaotic code
09:39 adam789654123 or i should say, chaotic disorganized organization
09:40 adam789654123 im shuffling things around in order to get to a specific functionality
09:40 deshipu adam789654123: do you know what "chaotic" actually means in math?
09:40 adam789654123 i have unworking and working code all mixed together in some files, for example
09:40 adam789654123 no
09:40 adam789654123 i dont know what it means
09:41 deshipu adam789654123: a "chaotic system" is a system which state you cannot compute without actually simulating it
09:41 adam789654123 oh, such a beautiful definition
09:41 adam789654123 :D
09:41 adam789654123 im gonna copy that and put it in my math file
09:41 robotustra I also write my interpreter
09:41 adam789654123 cool
09:41 adam789654123 any details?
09:41 deshipu so, for instance, a cellular automaton may be a chaotic system
09:42 robotustra forth like language
09:42 adam789654123 this is computational automaton, right?
09:42 adam789654123 cellular automaton ?= computational automaton?
09:43 adam789654123 robotustra: well, now i know, and i would like to play with it some time
09:43 adam789654123 but not now, while my hands are tied
09:43 adam789654123 deshipu: i am keenly interested in this subject, btw
09:44 adam789654123 robotustra: can i try it out some time?
09:44 adam789654123 (but not now)
09:44 adam789654123 ?
09:44 adam789654123 :)
09:44 robotustra you can, but I'm not finished with that
09:44 robotustra there is no version 1.0 yet
09:45 robotustra more like 0.5
09:45 adam789654123 deshipu: apparently computational automata is more abstract
09:45 robotustra I have to invest 2-3 months to make version 1.0
09:46 adam789654123 https://en.wikipedia.org/wiki/Cellular_automaton
09:46 adam789654123 ive looked into this before, and im interested in it
09:46 adam789654123 im thinking that discrete geometry is the ticket that i want to buy into the world of cellular automaton
09:47 adam789654123 i see robotustra
09:47 adam789654123 as i learn more, maybe its something i can help you with
09:47 adam789654123 but im very limited now, but this is a direction that seems like it could be something of a niche for me
10:52 femtogram ahh, sorry, i totally got distracted by soldering right after i asked a question yesterday... :/ thanks, SpeedEvil, Hyratel, Snert__. that clarified a lot.
14:03 akem hey robot teasers.
14:04 Hyratel hoy!
14:05 Hyratel anyone here do a lot with Neopixels?
14:25 SpeedEvil neopixels are a very expensive way to purchase addressible LEDs in general
14:26 femtogram ^^
14:26 femtogram though the pictures i've seen do look pretty
14:27 Hyratel weeeell
14:27 Hyratel https://dl.dropboxusercontent.com/u/33593534/Editing/Fritzing/NeoString_schem.png
14:27 Hyratel this is what I'm working on
14:27 Hyratel addressable-by-string using WS2811
14:27 Hyratel and a set of mosfets
14:30 robotustra has anybody build gazebo?
14:30 robotustra gazebo simulator?
14:36 Hyratel SpeedEvil, yeah it's expensive. do you have a better option?
14:37 SpeedEvil I've built gazebo
14:37 SpeedEvil Hyratel: buy the individual LEDs?
14:37 SpeedEvil Chop up a rope?
14:38 SpeedEvil Hyratel: Interesting
14:40 SpeedEvil Hyratel: I've wished occasionally for something like that. - Ws* with 1W LEDs
14:41 Hyratel SpeedEvil, my current design is a voltage driver, nota current driver
14:41 Hyratel you've seen the Pixie?
14:41 Hyratel https://www.adafruit.com/products/2741
14:43 Hyratel it's not WS28* protocol though
14:43 Hyratel instead unidirectional 8N1 serial
14:47 SpeedEvil Hyratel: ah
14:47 SpeedEvil Hyratel: doing the otehr is problematic
14:49 Hyratel SpeedEvil, the other?
14:49 Hyratel not sure i follow
14:49 SpeedEvil current mode
14:50 Hyratel also, I'm using through hole optoisolators because I couldn't find any SOIC-4 or SOIC-8 optos in the Fritzing libraries
14:51 Hyratel there's SO-08 2 gang optos
14:51 Hyratel but that's the same package as the WS2811 chip
14:51 Hyratel ... fuck
14:52 Hyratel I could probably have a SMALLER BOARD if I went through-hole except the WS2811 and voltage reg
14:52 Hyratel stand-up on end resistors
14:52 Hyratel TO220 FETs
14:52 Hyratel https://dl.dropboxusercontent.com/u/33593534/Editing/Fritzing/NeoString_pcb_incomplete.png
14:56 SpeedEvil Or you could not do that.
14:57 SpeedEvil And just isolate the WS
14:57 SpeedEvil the data in that is
14:57 SpeedEvil and use it directly as a gate driver
14:57 Hyratel I have a hard time designing in double layer for SMD, still not used to it
14:57 Hyratel it won't
14:57 Hyratel it's a current-mode outputs, so it can't drive FET gates directly
14:57 SpeedEvil It has current limited outputs.
14:58 SpeedEvil If you connect a 100 ohm resistor to V+, it's going to switch from V+ to close enough to 0 to turn the FET off.
14:58 SpeedEvil maybe 470
14:59 Hyratel actually it won't
14:59 Hyratel I did the research, that won't actually work
14:59 SpeedEvil why?
14:59 Hyratel because the resistance of the current-limited WS* output isn't enough to overpower the pullup resistor
15:00 Hyratel also
15:00 Hyratel that makes the FETs default to an ON state
15:00 Hyratel the optos are inverting and buffering the WS outputs
15:01 Hyratel because they /are/ LEDs
15:01 SpeedEvil The current limited WS* doesn't have a resistance, it has a current limit.
15:01 SpeedEvil This is 18.5mA. It will pull the inputs all the way down to at least 1.5V at 18mA - it has to in order to work with blue LEDs.
15:02 SpeedEvil If the pull-down resistor is 3.5V/18mA or higher, it will pull it to 1.5V, which is plenty to turn the FET off.
15:02 Hyratel down to 1.5v, don't you mena red LEDs?
15:02 SpeedEvil yes, there is still the issue of the FETs being backwards.
15:02 SpeedEvil no.
15:02 SpeedEvil 5-3.5 (blue) = 1.5
15:02 Hyratel so what would a better way to invert the outputs be?
15:02 Hyratel oh
15:02 SpeedEvil I'm not saying the opto solution won't work
15:03 Hyratel but it's not the most elegant?
15:03 SpeedEvil Well, you could use one opto, not three.
15:03 SpeedEvil Just from a cheapness perspective
15:03 SpeedEvil It's not more or less elegant
15:03 Hyratel ??? what
15:04 Hyratel how do you use one opto to run 3 FETs
15:04 SpeedEvil because the WS* has a serial data bus.
15:04 SpeedEvil you put the opto on that bus
15:04 Hyratel inverty the bus?
15:04 SpeedEvil no, that won't work
15:04 Hyratel taht still doens't deal with the power-on inversion
15:05 SpeedEvil Sure.
15:05 SpeedEvil Oh.
15:05 Hyratel because you'd have the FETs pulled up
15:05 SpeedEvil Also, you can't daisy-chain them like this
15:05 Hyratel how so
15:06 SpeedEvil (I mean with the proposed one isolator solution)
15:06 Hyratel oh
15:06 SpeedEvil because you'd need to also optoisolate the otuptu
15:06 Hyratel I was trying to design around using them as daisychained
15:06 SpeedEvil There is nothing at all wrong with the above design
15:07 Hyratel so you'd have NP*** NP***
15:07 Hyratel (because non-Addr strings are built in three series with current resistors
15:07 SpeedEvil Except I would add a 10uF cap across it.
15:08 Hyratel S*** P S*** P S***
15:08 SpeedEvil the power bus, at least.
15:08 Hyratel across power?
15:08 Hyratel yeah
15:08 SpeedEvil probably more than 10U actually
15:08 SpeedEvil Or 10u across the WS
15:09 Hyratel 10u on the WS power to ground
15:09 SpeedEvil Also - 12V is the absolute max
15:09 Hyratel absolute max for which part
15:09 SpeedEvil At that voltage, the WS will be dissipating ~11V*54mA or so = 600mW
15:10 SpeedEvil that's quite toasty
15:10 Hyratel oh, through each Opto LED?
15:10 SpeedEvil yes, when all on
15:10 Hyratel so I should add resistors to those to trim that back
15:11 Hyratel what do you think,... 680R?
15:11 SpeedEvil What is the current transfer of that opto
15:11 Hyratel checking
15:13 SpeedEvil Also. If you're feeling silly.
15:14 SpeedEvil these will drive 12V motors just fine.
15:14 SpeedEvil (add a diode for back EMF)
15:14 Hyratel that would be neat
15:14 Hyratel that'll up the board size though
15:14 Hyratel as is, it's a bit under 2" square
15:14 SpeedEvil Or use avalanche rated FETs, or use FETs with a voltage large enough to not catch the uncaught spike
15:15 Hyratel that would be neat though
15:15 Hyratel you could use it for
15:15 Hyratel ooooh
15:15 Hyratel projector appertures
15:16 Hyratel huh. 1206 SMD is a larger boardspace than 100mil THT
15:16 Hyratel hmmm
15:16 Hyratel trying to think in 2-sided construction
15:17 Hyratel it's actually a new thing to me
15:17 Hyratel SpeedEvil, Vf = 1.15
15:17 Hyratel IR LED
15:18 SpeedEvil 1206 is _huge_
15:18 SpeedEvil Hyratel: transfer current is the maximum output current you get for a transistor versus input current.
15:18 SpeedEvil ^opto
15:19 SpeedEvil It's like 'hFe' for optoisolators
15:19 Hyratel oh
15:19 Hyratel the 'gain'?
15:19 SpeedEvil yes
15:19 SpeedEvil at minimum, you want about output resistance / current transfer as a input resistor
15:20 SpeedEvil (but you want to drive it 10* harder than that in reality)
15:20 Hyratel ... ok, thaat's beyond my brain right now
15:22 Hyratel SpeedEvil, I have tr9ouble keeping track of numbers
15:22 Hyratel when you explained that my brain just went *buah?*
15:23 Hyratel thank you for vetting this though
15:33 Hyratel SpeedEvil, ran some numbers, does 1K sound right to you? for 12mA forward on the optos
15:33 Hyratel they're rated for 50mA each
15:33 SpeedEvil What does the current transfer ratio say
15:33 SpeedEvil it will be '600%' or something
15:34 Hyratel for all variants - 50-600
15:35 Hyratel oh
15:35 Hyratel ohhh
15:35 Hyratel ohkaaay
15:35 Hyratel this will inform the pulldown resistor
15:35 Hyratel because it forms one side of a voltage divider between VCC and GND
15:35 SpeedEvil Call it 5*.
15:36 SpeedEvil If it's 10%, then 1K on the input acts like 10K on the output
15:36 Hyratel wiat
15:36 Hyratel something ins't connecting
15:36 SpeedEvil Sorry I'm not explaining this - mostly asleep - ##electronics
15:37 Hyratel so a 1k diode CLR means the output transistor is at 10kR when 'active'?
15:37 SpeedEvil 'yes' if the current transfer is 10% and more conditions are met
15:37 SpeedEvil sort-of
16:48 akem https://www.youtube.com/watch?v=qX9PN9k_d_Q
16:49 akem one of the last version bigger like asimo transform into a car if i m not mistaken.
17:21 Hyratel SpeedEvil, you still up?
18:00 SpeedEvil no
18:01 LiohAu SpeedEvil: got you !
18:02 LiohAu Any idea of how I can determine the diameter needed for the linear guide/shaft that I'll use for the shoulder of my scara arm ?
18:04 Snert_ are you machining your own parts?
18:05 LiohAu hm if I have no choice yes
18:05 LiohAu no other choice*
18:05 Snert_ lucky you.
18:05 LiohAu but in that case, misumi.com => Linear Shafts
18:06 Snert_ Quite often I have to go with stock sizes.
18:06 Snert_ then throw out that which is too expensive.
18:06 Snert_ what I'm left with are the only 2 choices.
18:07 Snert_ kindalike.
18:07 orlock LiohAu: looked at Thomson?
18:07 LiohAu orlock: no
18:07 orlock Thomson, Kollmorgen
18:08 orlock Portescap
18:08 LiohAu i'm not needing brands or sellers :P
18:08 LiohAu i'm just looking for a way to determine the diameter needed for these shafts
18:09 orlock http://www.roymech.co.uk/Useful_Tables/Drive/Shaft_deflections.html
18:09 orlock Theres the formulas, plug in your numbers
18:09 LiohAu i almost opened the same link at the same time :P
18:10 LiohAu (i'm on the nook website)
18:10 orlock LiohAu: I only mention those brands as they are all owned by the parent company of my employer
18:11 LiohAu :D
18:11 LiohAu I hope you'll get a bonus ^^
18:11 orlock and apparently they are all high quality, etc etc
18:11 orlock Heh
18:11 orlock They are actually splitting into a new company
18:11 orlock just annoinced the name this week
18:11 orlock http://www.fortive.com/
18:12 LiohAu I see a lot of things related to horizontal shafts, but when the shaft is vertical, forces are not so big..
18:13 orlock Hopefully
18:13 orlock Ideally, none!
18:13 orlock The drive mechanism would be more important then i guess
18:15 LiohAu drive mecanism is almost ok (ball screw diameter, lead ...)
19:11 psYbR :) http://tanglesgaming.com/esp8266_test/
19:35 orlock psYbR: Page looks nice - that all native on an esp?
19:36 psYbR no, it uses ajax to query the ESP which just serves up a JSON string
19:36 psYbR thx :)
19:36 orlock hmm reminds me
19:37 orlock i think i want to try get to jaycar on the way home
19:37 orlock capacitors and resistors to put between ESP8266 and Neopixels