#robotics Logs

Jan 14 2021

01:13 AM Jak_o_Shadows: I normally use matplotlib
01:30 AM Jak_o_Shadows: https://gist.github.com/Jak-o-Shadows/d0ab7ae9c8af76f21b08a5f9fe58d247 should do it
01:30 AM Jak_o_Shadows: use "ax.grid()" to turn the grid on
01:32 AM rue_mohr: ooo
01:32 AM rue_mohr: thanks, I can give it a go tommorow
01:32 AM Jak_o_Shadows: No worries
01:32 AM Jak_o_Shadows: I use matplotlib a LOT
02:08 AM flav0r: HEY HEY !
02:08 AM flav0r: whats new dudes ???
09:01 AM rue_mohr: PID filter analysis
09:02 AM rue_mohr: the bode plot is interesting
10:05 AM BitEvil is now known as SpeedEvil
04:56 PM Tom_itx is now known as Tom_L
07:50 PM flav0r: DUDES !!!!
08:08 PM FLHerne: what
08:30 PM rue_mohr: it would be nice to be able to adjust this PID realtime
08:30 PM rue_mohr: I cant work out why the I gain wont flatten out
08:30 PM rue_mohr: how much IS 50db anyhow?
08:30 PM rue_mohr: 316
08:31 PM rue_mohr: I pulled it down by 1/1024 tho and it was still there
08:33 PM rue_mohr: idea
08:44 PM mrdata-: +50 db is 100000x the power
08:45 PM mrdata-: you took it down about 30 dB
08:46 PM rue_mohr: !?
08:47 PM rue_mohr: 50(db) / 20 -> 10^x = 316?
08:49 PM rue_mohr: I got my program to generate all 4 profiles, P, I, D and the sum
08:52 PM rue_mohr: http://ruemohr.org/~ircjunk/tempimage/foo102.png
08:52 PM rue_mohr: heh, Kp is 2
08:53 PM rue_mohr: #define Kp 2
08:53 PM rue_mohr: #define Ki 1.0/64.0
08:53 PM rue_mohr: #define Kd 6
08:53 PM rue_mohr: 20000.000123 SPS
08:54 PM rue_mohr: oh wait, I forgot the multiplier
08:55 PM rue_mohr: try that again...
08:57 PM rue_mohr: http://ruemohr.org/~ircjunk/tempimage/foo102.png <-- hit refresh
08:57 PM rue_mohr: thats better
10:15 PM Tom_L: self tuning pid
10:15 PM mrdata-: rue_mohr, ok, sure, power is V^2/R so
10:15 PM mrdata-: the factor 316 is about voltage
10:16 PM rue_mohr: its the ratio 1:316
10:16 PM rue_mohr: ?
10:16 PM mrdata-: if you took voltage down 1/1024 then thats about -60 dB i guess
10:17 PM rue_mohr: I think I didn't notice the auto-scale
10:17 PM rue_mohr: the remaining signal might have been 0.4db and looked like a huge up-swing