#robotics Logs

May 04 2019

03:19 AM veverak: hmm
03:24 AM veverak: quite pricey
03:24 AM veverak: deshipu: we got LX15D from lewansoul for 14$
03:24 AM veverak: frined of mine designed custom PCB with a magnet encoder for additional 10$ (now)
03:24 AM veverak: so
03:24 AM deshipu: why do you need additional encoder?
03:26 AM deshipu: I should really figure out encoders and a driver for some cheap motors, and make a micro-rhex
03:27 AM Jak_o_Shadows: Once the printers working, i've been thinking I should jump to the brushless motors. ODrive type stuff
03:28 AM deshipu: is it just me, or does odrive seem hugely overpriced and overhyped?
03:35 AM veverak: deshipu: not additional
03:35 AM veverak: replacing the inner electronics compeltely
03:35 AM veverak: ;)
03:35 AM veverak: *completely
03:35 AM veverak: magnetic encoder is fined because it's 360 deg.
03:35 AM veverak: so you can use it to drive wheels
03:35 AM veverak: original in LX15D is only 240D
03:35 AM veverak: *240 deg
03:35 AM veverak: deshipu: odrive is awesome if you got the usage for it
03:58 AM deshipu: nice, the 240° limitation is the main reason I don't like them
04:12 AM mrdata: my shoulder has about 240° limitation
04:13 AM deshipu: I don't like your shoulder either
04:14 AM mrdata: fine
05:04 AM rue_mohr: veverak, those magnetic encoders aren't cheap
05:06 AM rue_mohr: not that I have a good solution
05:06 AM rue_mohr: I'm playing with the idea of two channel resolvers
05:09 AM veverak: yeah, they make most of the price of custom PCB
05:28 AM rue_mohr: wtf
05:28 AM rue_mohr: -3db = "0.707" amplitude, BUT, its ACTAULLY 0.70794... so why the hell not use 0.708!!!!
06:58 AM Jak_o_Shadows: I thought 3dB was half?
06:58 AM Jak_o_Shadows: ah, ok, googling it helps
08:35 AM mumptai: -3dB is half power
09:43 AM durrf is now known as drrfy
09:44 AM drrfy is now known as durrfy
11:57 AM rue_mohr: 0.707 is 3db
11:57 AM rue_mohr: (-3db)
12:26 PM mrdata: rue_mohr, mumptai is right
12:27 PM mrdata: oh you were referring to amplitude
12:28 PM rue_mohr: yea
12:28 PM rue_mohr: wtf is power, point to it on my 'scope
12:28 PM rue_mohr: :)
12:30 PM mrdata: but .70794 isnt right.
12:30 PM mrdata: sqrt(0.5) = 0.70710678118654752440084436210485
12:31 PM rue_mohr: I was told 20Log()
12:31 PM rue_mohr: wtf is it?
12:31 PM mrdata: hang on, checking extraneous digits
12:32 PM rue_mohr: if you do it in reverse with log, its -3.01db
12:32 PM rue_mohr: the .01db is where 0.7071 becomes 0.7079
12:33 PM rue_mohr: https://en.wikipedia.org/wiki/Decibel
12:33 PM rue_mohr: -3 .... 0.708
12:34 PM rue_mohr: (approx equal to sqrt(0.5)
12:34 PM rue_mohr: )
12:34 PM mrdata: do you want half the power or do you want -3.0000000 dB
12:34 PM rue_mohr: I want to know amplitude
12:34 PM rue_mohr: cause I'm doing filters
12:34 PM mrdata: amplitude of half the power is sqrt(0.5)
12:35 PM rue_mohr: no its approx sqrt(0.5)
12:35 PM mrdata: no it's exactly
12:35 PM rue_mohr: https://en.wikipedia.org/wiki/Decibel
12:35 PM rue_mohr: its based on Log
12:35 PM mrdata: decibel is log based but so what
12:36 PM mrdata: we are talking about power and amplitude
12:36 PM rue_mohr: SO, there are all sorts of stupid approximations out there
12:36 PM mrdata: exactly half the power is sqrt(0.5) amplitude exactly
12:36 PM rue_mohr: 4 decimal places should not be too much to ask for today
12:36 PM mrdata: \no approximations
12:36 PM rue_mohr: no, its 0.501
12:36 PM mrdata: wrong
12:37 PM rue_mohr: you saw that chart, right?
12:37 PM mrdata: the chart doesnt matter; especially if it's wikipedia
12:37 PM mrdata: the math matters
12:37 PM rue_mohr: ok, so
12:38 PM rue_mohr: ugh, all my notes are in the shop
12:38 PM rue_mohr: dB = 20*log(a) yes?
12:38 PM mrdata: yes
12:39 PM rue_mohr: so a = 10^(dB/20) yes?
12:39 PM mrdata: yes
12:39 PM rue_mohr: so whats 10^(-3/20)
12:39 PM mrdata: the -3 is rounded
12:40 PM rue_mohr: _I_m saying whats the amplitude at exactly -3db
12:40 PM rue_mohr: not -3.01
12:40 PM mrdata: thats why i asked above, do you want half the power or do you want -3.0000000 dB
12:40 PM rue_mohr: because, on a chart, I'm not looking at -3.01, I'm looking at -3
12:40 PM rue_mohr: hmm
12:41 PM rue_mohr: I dont even care about power
12:41 PM rue_mohr: I care about amplitude and phase angle
12:41 PM rue_mohr: so, who is at -45 degrees?
12:41 PM mrdata: you care abou the values for things when building filters
12:41 PM rue_mohr: (first order)
12:41 PM mrdata: the filter design is based on the math not on dB approximations
12:42 PM mrdata: you dont want exactly -3.000000 dB
12:42 PM rue_mohr: and I cant tell who is approximating what
12:42 PM rue_mohr: ah
12:42 PM rue_mohr: cute. :-/
12:43 PM rue_mohr: I wonder if that explains the phase angle being out about 5 degrees
12:43 PM rue_mohr: prolly just 20% caps
12:43 PM mrdata: yeah probably the 20% caps
12:44 PM rue_mohr: I want to make up a filter profiler
12:44 PM rue_mohr: from like 1Hz to 5Khz
12:44 PM rue_mohr: sound card is an idea, but I think I want to go manual on it, maybe an stm32
12:45 PM rue_mohr: oscillator is a problem
12:45 PM rue_mohr: kinda
12:46 PM rue_mohr: I suppose I could use an external analog sinewave osc and have the uC just watch whats going in
12:46 PM rue_mohr: but an stm32 should be able to push 5Khz sine audio to a DAC
12:47 PM mrdata: yes
12:50 PM rue_mohr: looking at the dma stuff
12:50 PM rue_mohr: looks like its best if I have an i2c DAC
12:50 PM rue_mohr: dunno if I do
12:51 PM rue_mohr: something that I can fire off a console program and get a filter profile via a serial port
12:52 PM rue_mohr: tell it the range and how many steps, get back amplitude and phase
04:02 PM rue_shop1: haha, gas lawn mower started on the first pull!
04:03 PM rue_shop1: (granted to move the wheels up a notch I had to re-machine two of the axle bolts cause the impact driver snapped them off, but)