#robotics Logs

Dec 29 2020

11:14 AM veverak: hmm
11:20 AM veverak: seems my PID hack works
11:20 AM veverak: I have position control of the servo via velocity commands
11:20 AM rue_mohr: ah, were you the one working on the PID?
11:20 AM veverak: yes
11:22 AM rue_mohr: ok lets say this
11:23 AM rue_mohr: you have a big speaker, and you stick a sine wave in it
11:23 AM rue_mohr: at low freqs it responds great (cause it has a big heavy cone)
11:23 AM veverak: again
11:23 AM veverak: stop
11:23 AM rue_mohr: but as you go up, it cant move fast enough to hit the same amplitude it can at a low freq
11:24 AM rue_mohr: oooh, you dont want to TRY to understand, sorry.
11:24 AM rue_mohr: so for the rest of us
11:24 AM Tom_L: that's where the crossover kicks in and the tweeter takes over
11:24 AM veverak: I have low audio knowledge, so your explanation is like different language to me
11:24 AM rue_mohr: if you plotted the response, you might get soemthing like
11:24 AM rue_mohr: ---------\__________
11:25 AM rue_mohr: so your PID (EQ) filter, needs to boost the highs
11:25 AM rue_mohr: HOW high and where the boost is needed, is a function of the system your trying to control
11:37 AM rue_mohr: This is why I got a digital scope, I can do a freq sweep on a hardware system to find its natural curve, and then know what freqs I need to boost and by how much
11:42 AM rue_mohr: With what we have available now for processor power, PID should be obsolete, Its not hard to make some pretty darn complex filters
11:42 AM rue_mohr: apparently I'd had the digital scope for a year now, and I have not had a chance to get back to my servo projects
11:42 AM veverak: I mean, "available computing power" is not an argument
11:43 AM veverak: you may want to use it for different stuff than filter
11:43 AM rue_mohr: looking forward to attacking a servo with a FIR filter
11:43 AM rue_mohr: PID is a filter
11:43 AM rue_mohr: its a filter with a high and low pass
11:43 AM rue_mohr: and a bypass
11:44 AM veverak: as "you would rther invest the extra computing power into something else, than better filter than PID"
11:44 AM veverak: FW for quadcopters uses something better than PID I think? (or some advanced version of it)
11:44 AM veverak: but for a lot of stuff it's "good enough"
11:46 AM rue_mohr: P is the bypass, I is low pass, D is high pass
12:01 PM rue_mohr: https://d2vlcm61l7u1fs.cloudfront.net/media%2Fb38%2Fb3869e0b-520f-4787-a855-29af3e0e33df%2FphprXjbEG.png
12:02 PM rue_mohr: veverak, have you ever seen a graph like that ^^ ?
12:04 PM veverak: nope
12:04 PM rue_mohr: have you ever adjusted the bass/treble on a sound system?
12:11 PM rue_mohr: maybe I should ask you to select all the images that have bicycles..
12:11 PM rue_mohr: [][][]
12:11 PM rue_mohr: [][][]
12:11 PM rue_mohr: [][][]
03:02 PM veverak: hmm
03:02 PM veverak: deshipu no longer here
05:10 PM mumptai: veverak, dunno, at least not at the moment
11:25 PM mrdata-: https://youtu.be/fn3KWM1kuAw
11:25 PM mrdata-: boston dynamics
11:25 PM mrdata-: new video