#linuxcnc-devel Logs

Jan 09 2020

#linuxcnc-devel Calendar

07:27 AM jepler: "I am, honest". the comma strikes again.
07:32 AM skunkworks: every 4 ms
07:33 AM skunkworks: heading to work
08:14 AM skunkworks: not effecting motion at all
08:21 AM jepler: Ddt is too simple to have bugs, surely
08:31 AM skunkworks: I would think so...
08:31 AM skunkworks: but you know me...
08:31 AM skunkworks: http://linuxcnc.org/docs/html/hal/comp.html#cha:hal-component-generator
08:31 AM skunkworks: with out the /period.
08:32 AM skunkworks: Now - if I compiled it on a different pi - and coppied it over to /usr/mumble/mumble... could have that caused a problem?
08:51 AM skunkworks: would anything roll over running overnight?
09:33 AM jepler: shouldn't, shouldn't....
09:34 AM jepler: oh!
09:34 AM jepler: float tmp = in;
09:34 AM jepler: you are losing most of the precision of your value
09:34 AM jepler: should be double tmp = in
09:34 AM jepler: and yeah it would get worse as running time (coordinate) increaes
09:34 AM jepler: increases
09:35 AM skunkworks: Ok!
09:38 AM skunkworks: jepler: could the iptables error make it so something else could interfere with the ethernet port?
09:38 AM jepler: why, what are you seeing (I don't think so)
09:41 AM skunkworks: I just got a random packet error and was wondering.
09:41 AM skunkworks: (just threw it down on the table and started running it)
09:41 AM skunkworks: still testing
09:54 AM jepler: hi roguish!
10:14 AM skunkworks: prevsub looks way better now.. :)
10:15 AM skunkworks: no it is just a bit of noise
10:15 AM jepler: now send us a pull request to fix the docs, and check whether the real ddt also has the same bug
10:19 AM pcw_home: real ddt is correct
10:19 AM jepler: I wonder how much carnage would result from renaming the type to hal_double
10:19 AM jepler: because this trips up everyone, to pair "double" with "hal_float" to do arithmetic or even just storage
10:34 AM skunkworks: http://electronicsam.com/images/greenmachine/2020-01-09-102252_1920x1080_scrot.png
10:34 AM skunkworks: notice the minmax.2
10:51 AM skunkworks: and now it makes sence why prevsub.0.out got worse the longer it ran...
11:21 AM skunkworks: jepler: your image is looking good!
12:31 PM skunkworks: as far as I have read - you can force_turbo=1 without voiding warrantee.. If you start playing with voltages is where the void bit gets tripped
12:32 PM skunkworks: https://www.raspberrypi.org/documentation/configuration/config-txt/overclocking.md
02:11 PM jepler: pcw_mesa: 7i97 requires 5v, not the wide voltage range that 7C80 accepts, right?
02:11 PM jepler: and 7c81 is also 5v only?
02:11 PM jepler: (it's right there on the silk but I'm asking for some reason anyway)
02:12 PM pcw_mesa: Yes 7I97(96,97) and 7C81 are 5V
02:12 PM pcw_mesa: Trying to use orange connectors for wide range
02:13 PM jepler: that's what tipped me off as I considered making a big mistake
02:15 PM pcw_mesa: yeah on way just wont work, the other way is smoky...
08:07 PM skunkworks: pcw_mesa: worse so far... http://electronicsam.com/images/greenmachine/2020-01-09-195713_3840x2160_scrot.png
08:11 PM skunkworks: 53c
08:40 PM jepler: skunkworks: what variable did you .. vary .. that made it worse?