#robotics Logs

Oct 12 2017

12:00 AM orlock: a falling knife has no handle
12:03 AM rue_mohr: yea, a manual I had for an alarm install used twist+tape, I shuddered
12:03 AM orlock: i had to patch the camera into the reverse lights
12:03 AM rue_mohr: lots of times its also better to see WHERE a small screw goes, than to try to catch it
12:03 AM orlock: i tried uding those blue crimp-an-extra-wire things
12:03 AM orlock: didnt bite through the insualation, no worky
12:03 AM rue_mohr: IDC, yea ~
12:03 AM orlock: so i scraped off 1cm or so of insulation, bit of a pain.. solder + etc etc
12:03 AM orlock: all good, now our bloated pig has a reverse camera
12:09 AM theBear: mmm, good question... i still dabble in foreign language (mostly via film-watching and mumbling to myself :] ) mmmm.... do the odd circuit-mathing here and there as usual, and of course occasional repairs more interesting than soldering a power-socket
12:09 AM rue_mohr: hmm
12:10 AM rue_mohr: how would you like some freelance php work?
12:10 AM rue_mohr: I know an artist who said they would like help, I dont know if they already found any
12:12 AM theBear: this year i've not been much good for any work... between unusually high % of flat-on-back time, and some family-matters like alzeimerey dad and far away deceased estates to help with, i'm WAY slow/behind even on little jobs at the moment
12:13 AM rue_mohr: ok
12:13 AM theBear: also from what i gather, php has changed kinda focus/paradigm semi-radically in the last one or two major versions, which i know nothing agbout
12:13 AM * rue_mohr looks for someone else to throw a bone at
12:13 AM * rue_mohr gives up
12:14 AM theBear: sure, i might "own" the undernet php channel, but that doesn't automatically make me less than a decade or so up2date :)
12:15 AM rue_mohr: ?
12:15 AM rue_mohr: do you!?
12:15 AM theBear: if he pops up (appears around irc here irregularly, irl VERY seldom) i'll ask my buddy.... he's a true php NINJA
12:15 AM theBear: yeah, took it over maybe 5 years ago now
12:15 AM rue_mohr: hah, good show
12:15 AM theBear: when i first met him he had 99% implemented a bash-like shell in pure php
12:16 AM theBear: just for something to do in his spare time kinda thing
12:17 AM theBear: he lived with me fer a while, young and without knowledge of how to care for kitchen-ware, and occasionally makes a irc-dick of himself, but more like pepsi than flyback, if ya know what i mean ... kinda as a curiosity rather than an obsession
12:18 AM theBear: oh, that sentence was sposed to include that he's a good guy
12:18 AM theBear: heh, without that, it looked a lot worse than it was sposed to
12:19 AM rue_mohr: hahaha
12:19 AM rue_mohr: pepsi was a smart nit till something went wrong and then he was just an asshole
12:20 AM rue_mohr: iirc there was speculation about drugs
12:20 AM rue_mohr: your not on drugs right!
12:21 AM * rue_mohr checks theBears cabinet... tylenol, morphine, advil, some green stuff... nope, nothing here
12:21 AM theBear: maybe i blinker myself more than i know when i like someone, and i know pepsi was a super-dick now and again, but i always thought it was occasional and interspersed with useful/smartness, and it wasn't just speculation, he openly admits he had a bit of a drug problem for a while there
12:22 AM rue_mohr: yea, someone can be an ass, but if they exhibit smart often enough, its ok
12:22 AM theBear: heh, i'm a special case... antidepressants don't agree with my head, and my head always had a bit of small-smallmedium chemistry issue, so carefull application of liquor and maybe the odd funny looking cigarette keep society safe ;-)
12:23 AM rue_mohr: k, well, I'm gonna go lay in bed
12:24 AM theBear: rue_mohr, heh, at the same time, you just described flyback, and we all know he's slightly on the other side of the ok-line ... poor feller, if my head is chaos, must be hell inside his
12:25 AM rue_mohr: yes, flyback is over the line
12:25 AM orlock: irc also has an effect
12:25 AM orlock: https://www.penny-arcade.com/comic/2004/03/19
12:25 AM theBear: hmm, if i gonna keep using all these empty medication boxes for other purposes, i gotta start marking them appropriately... it's getting way too easy to misplace something on the desk and have to open a bunch of boxes before i hit the right one
12:26 AM theBear: orlock, indeed, and <grin> at chosen example...
12:28 AM theBear: kinda strange that equation.... some people it's 100% correct, others (like us i guess) just don't seem to lose control ever (having gained anonymity etc)
12:32 AM rue_bed: https://sketch.io/render/sk-60ac789314efa7c77750c998cb8c925a.jpeg
12:32 AM rue_bed: thats all I came up with
12:32 AM rue_bed: its parallel line math
12:36 AM theBear: heh, paralell is a bit rich, they more like wiggly scribbly lines <grin<>
12:37 AM theBear: hmm, are those low-number (like C1 and C2) there just variable names (as opposed to exponents/indices or similar) ?
12:39 AM theBear: erg, not used to sitting and typing.... i must excuse myself so i can flexible a bit ;-)
01:21 AM rue_bed: the two lines are parallel , so the two slope numbers are the samefor the two formulas, but their offset is different
01:21 AM rue_bed: hence c1 and c2
01:21 AM rue_bed: you cant use y = mx+b because it will flip out on verticle lines
01:22 AM rue_bed: you have to use ax+by+c = 0
02:26 AM rue_bed: I need an audio filter that removes all the tones and leaves the static
02:34 AM jason__: you usually need some model or assumptions about the signal or noise
02:35 AM jason__: like "the signal is human speech so everything outside this frequency range is probably noise"
03:43 AM theBear: indeed... define "tones" and "static" <grin>
03:44 AM weyland|yutani: why leaving the static?
03:46 AM deshipu: you have to leave something
03:48 AM weyland|yutani: sure but ami misunderstanding? does he want static or the tones?
03:52 AM theBear: the static hallegedly, but like we say, you gotta define tones and static for that to mean anything, or at least give a context
03:54 AM Jak_o_Shadows: noise cancelling headphones are great
04:29 AM deshipu: noisy people-cancelling curare darts are even better
04:30 AM theBear: wtf is a curare ?
04:51 AM deshipu: oh, if we only had some kind of encyclopedia website where we could check the words we don't understand
04:52 AM theBear: heh, rough, course only a fool could argue :)
09:28 AM rue_mohr: the tones in speech are useless
09:28 AM rue_mohr: if you drop them, you have whispering, if you leave them on, you have singing
09:29 AM rue_mohr: technically, since you mention it, there are prolly only two freqs to sort out, the harmonic of each vocal chord
09:30 AM z64555: try messing with an FM tuner
09:31 AM z64555: since tone can be thought as a modulation of frequency/pitch
09:31 AM z64555: (If that's what I think you're talking about)
09:31 AM rue_mohr: pll the two freqs and then subtract them
09:31 AM rue_mohr: hmmm
09:58 AM deshipu: rue_mohr: that's just for vowels
09:58 AM deshipu: rue_mohr: for consonants you need all the noise in there
09:59 AM rue_mohr: yea, all you need is the noise tho
10:01 AM rue_mohr: Tom_L, I think the lower bearing on my new cnc is gradually giving out
10:01 AM rue_mohr: dont have any other explination for the chatter
12:13 PM Tom_L: misaligned or too light a bearing?
01:24 PM veverak: rue_mohr: was thinking
01:24 PM veverak: assert(I have proper ARM cpu)
01:24 PM veverak: with the technique you described for measuring LOAD on PWM servos
01:24 PM veverak: I should also be able to measure servo position?
02:53 PM Snert___: I don't think you can query those MG90 servos for position. Not without an external sensor of some sort.
02:53 PM veverak: I think, that maybe I can
02:53 PM veverak: :)
02:55 PM veverak: Snert___: do you know how simple servos _can_ work?
02:57 PM veverak: internal chip, produces PWM value based on _internal potentiometer_
02:57 PM veverak: that PWM value is compared with PWM value sent to the servo as control signal
02:57 PM veverak: if both are HIGH, nothing happens
02:58 PM veverak: if control is HIGH and potentiometer is LOW, it moves motor on full in one direction
02:58 PM veverak: if control is LOW and potentiometer is HIGH, it moves motor on full in other direction
02:58 PM Snert___: yes, but you can't ask it a question.
02:58 PM veverak: when both are low, nothing happens
02:59 PM veverak: things is, you should be able to detect this with single digital pin attached to the voltage pin of servo (with proper circuity)
02:59 PM veverak: to detect when motor is powered and when not
03:00 PM Snert___: Yes, but that's not sending the servo a query asking for position. That's tracking/calculation.
03:00 PM veverak: and?
03:00 PM veverak: :)
03:01 PM veverak: with precise timing, I should be able to detect actual position based on the load
03:01 PM Snert___: be that as it may...I spose one can track it sucessfully. Good luck.
03:01 PM veverak: assert, this servo works as described
03:01 PM veverak: Snert___: in theory, I should be able to
03:02 PM deshipu: in practice it's much easier to just add two wires, one to the pot and one to the motor
03:02 PM Snert___: you can query your data object that's related to the servo, but not directly ask the servo where it's at.
03:02 PM veverak: I've got this 50Mhz FPGA to play with ;)
03:02 PM deshipu: ugh
03:02 PM veverak: Snert___: I never said I would query the servo
03:02 PM deshipu: veverak: you will need 12 of them
03:03 PM veverak: deshipu: why?
03:03 PM veverak: :)
03:03 PM deshipu: veverak: all those legs
03:03 PM veverak: deshipu: it should be able to measure all those legs
03:04 PM veverak: thing is, we are still measuring HIGH/LOW logic
03:08 PM veverak: to explain
03:09 PM veverak: to generate PWM pulse, I setup internal FPGA counter
03:09 PM veverak: I set OUTPWM to HIGH and counter to 0
03:09 PM veverak: when counter == 'pulse_wanted_to_sent', the OUTPWM goes to LOW
03:09 PM veverak: -> pulse is sent
03:10 PM veverak: so I simply have to expand this logic, by measuring the LOAD pin, and there are two cases:
03:10 PM veverak: if load goes UP before OUTPWM is LOW, store that as actual position of potentiometer
03:11 PM veverak: if load goes UP when OUTPWM is LOW, you simply wait for the case when load goes DOWN, and mark that counter value as actual position of potentiometer
04:14 PM robotustra: evening
07:18 PM rue_shop3: veverak, are you using my code for measuring servo load via the servos internal feedback loop?
07:25 PM Snert___ is now known as Snert_
07:42 PM veverak: nope, do not even have a link
07:43 PM veverak: it's just idea now
07:43 PM veverak: :)
08:03 PM rue_shop3: you can measure the servo load by looking at the time the onboard driver drives the motor
08:04 PM rue_shop3: you can pick up on it by doing a simple current detector on ther servo power line
08:04 PM rue_shop3: but if you want to know direction, you need to tap off from the motor leads
08:10 PM rue_mohr: Tom_L, turns out I cant find any play in the new cnc spindle
08:10 PM rue_mohr: not sure why it chatters
09:16 PM veverak: rue_mohr: do no think so
09:17 PM veverak: if I know that load exists before the control signal goes low -> it shou go in one direction
09:18 PM veverak: if it exists only after the signal goes low -> opposite direction
09:18 PM Tom_L: rue_mohr, is the table loose?
09:38 PM robotustra: not rigid table
09:39 PM robotustra: or spindle mount
09:39 PM robotustra: or frame
09:39 PM robotustra: or whatever