#robotics Logs

Apr 15 2020

06:41 AM Jak_o_Shadows: Well, I have pretty much built http://stubby.digitalcave.ca/stubby/
06:47 AM Jak_o_Shadows: need to figure out how to drive the last 2 servos
06:47 AM Jak_o_Shadows: the i2c PWM chip I lazily use only has 16
06:47 AM Jak_o_Shadows: also, 2-3 of the servos might jitter too much
02:42 PM rue_mohr: I have that code nobody will use
02:42 PM rue_mohr: for completely solid timing of 8 servos
02:42 PM rue_mohr: at really high resolution
02:43 PM veverak: lol
03:01 PM mrdata: nice
03:01 PM mrdata: why will nobody use that
03:02 PM mrdata: 8 servos is kind of what you need for a multi-legged robot
03:02 PM rue_mohr: cause were makers
03:02 PM rue_mohr: we never use a solution that already exists
03:02 PM mrdata: perhaps even 12
03:02 PM rue_mohr: we start with designing a wheel, every time
03:02 PM rue_mohr: it does _perfect_ timing
03:03 PM rue_mohr: no jitter
03:03 PM mrdata: i would use a solution that already exists if one were available that fit my needs
03:03 PM rue_mohr: (pas source clock jitter)
03:03 PM rue_mohr: I used it for all my robot arms
03:04 PM rue_mohr: runs on an arduino pro mini
03:04 PM mrdata: sweet
03:04 PM rue_mohr: which is a really handy board to fit in things as a servo controller
03:05 PM mrdata: it's tough to get multiple servos running on the same arduino
03:05 PM mrdata: timing issues abound
03:05 PM rue_mohr: yea, I got it with nothing to delay the timers
03:06 PM rue_mohr: https://github.com/ruenahcmohr/m328hobbydriver
03:06 PM rue_mohr: if you add any code to it, it wont have proper timing anymore }:|
03:07 PM rue_mohr: I wrote it cause I wanted high resolution on the servo positions (beyond the deadband) and I wanted _NO_ jitter, cause some of the servos I have are REALLY sensitive to timing jitter
03:10 PM mrdata: nice
03:21 PM veverak: rue_mohr: how do you communicate with it?
03:21 PM veverak: i2c? uarT? spi?
03:22 PM rue_mohr: serial
03:22 PM rue_mohr: plain ole 9600 N81
03:23 PM rue_mohr: ... unless I changed it to 19200, but I think 9600 was fast enough to update multiple times between servo updates
03:23 PM rue_mohr: if you fire data at it too fast and have lag problems cause of your fifo, dont blame me
03:28 PM veverak: I managed to fillup the buffer of mine robot tday
03:28 PM veverak: I am really glad I actually wrote proper error errporting for full input buffer
03:28 PM veverak: saved a lot of debugging time :)
03:55 PM weyland|yutani: well veverak is a good robodad
03:57 PM weyland|yutani: but in my days we had to climb infinite stairs while managing full buffers and segfaults while hanging upside down in space ;D
03:59 PM veverak: :D
04:00 PM veverak: I found out that proper error handling is something you want
04:10 PM weyland|yutani: every dev wants that if he doesnt want to go down the find the error hurrdurr rabbit hole
10:15 PM rue_mohr: It was neat how slowing down the datarate on a serial link resulted in less latency
10:18 PM rue_mohr: https://www.youtube.com/watch?v=3N7DNfg4p9I
10:18 PM rue_mohr: parallel streams, you can see the effect of the latency
10:44 PM rue_mohr: fix was to slow down the sample rate