#garfield Logs
May 24 2021
#garfield Calendar
02:42 AM rue_mohr: polprog, the results are interesting
02:49 AM rue_mohr: I need to make a chart
02:49 AM rue_mohr: but
02:49 AM polprog: morning rue
02:49 AM rue_mohr: twitter goes out to about 6 people initially
02:49 AM polprog: what's that tape youre playing?
02:49 AM rue_mohr: and it goes out to about 1 person every 2.5 minutes
02:49 AM rue_mohr: its steppinwolf
02:49 AM polprog: hmm
02:49 AM polprog: wanted to make a musical stepper?
02:49 AM rue_mohr: it was suggested
02:49 AM rue_mohr: I wanted to use a 4017 to break up the zero crossings and send them out to a bunch of steppers
02:49 AM rue_mohr: but the audio cant just be fed into a comparator
02:49 AM rue_mohr: you lose all the low amplitude signals
02:49 AM rue_mohr: they get "washed out" by the larger ones
02:49 AM polprog: omg this is *that* track??
02:49 AM polprog: https://www.youtube.com/watch?v=egMWlD3fLJ8
02:49 AM polprog: i was looking for it for ages
02:49 AM rue_mohr: An error occurred. Please try again later. (Playback ID: 26QxgqCJXFzy9moy)
02:49 AM polprog: (it's Born To Be Wild with video from Easy Rider)
02:51 AM polprog: oh crap, its like the 90s again
02:51 AM polprog: steppenwolf is nice
02:53 AM rue_mohr: :)
02:53 AM rue_mohr: !BigTruck Steppenwolf - Born To Be Wild.mp3
02:53 AM rue_mohr: bit loud in there
02:53 AM rue_mohr: ther are two chanenls
02:53 AM rue_mohr: #mp3passion and #mp3_collective
02:54 AM rue_mohr: if you cant find a song, try the other one
02:55 AM polprog: thanks
02:55 AM polprog: :)
02:56 AM rue_mohr: did you get anything?
02:56 AM rue_mohr: use stars and things usually come up
02:56 AM rue_mohr: @find word*word*word*word
02:58 AM polprog: yea im receiving one thing
02:59 AM rue_mohr: ok, I hope its a good copy!
03:00 AM rue_mohr: I'll often download a few and check them before adding them to my collection
03:02 AM rue_mohr: I think I have finished wire-wrapping the timer into the z80
03:04 AM polprog: ive got classes in a moment
03:05 AM polprog: exam from electronics (again) this friday
03:05 AM rue_mohr: later!
03:05 AM polprog: the guy is a ham and the only excercises and examples he shows are transmission lines, and waveguides, lol
03:06 AM polprog: like, this is a general class on electronics, not an arrl exam dude
03:06 AM rue_mohr: hmm
03:10 AM polprog: hmm, ill have to set up a local client for that
03:10 AM polprog: scp'ing from the server is a bit cumbersome
03:12 AM polprog: ill keep that in mind
03:15 AM polprog: mp3_collective -do i @find there too?
03:15 AM polprog: it's silent
03:16 AM rue_mohr: sielnt? you sure!?
03:16 AM polprog: ah
03:16 AM rue_mohr: yes, same commands
03:16 AM rue_mohr: you sure?
03:16 AM polprog: it was, for a couple sec :p
03:16 AM rue_mohr: strange
03:29 AM rue_mohr: ok, that should do it
03:30 AM polprog: good
03:32 AM rue_mohr: I think I have it wire-wrapped in
03:32 AM rue_mohr: lets see if it still boots
03:33 AM rue_mohr: after I add a comm port, I'll need to move 2x 28 dips of wire-wrap wiring
03:34 AM rue_mohr: and I want a beeper
03:34 AM rue_mohr: I want it to give a click when I type
03:34 AM rue_mohr: and a startup beep
03:34 AM polprog: probably easier to put a mezzanine board
03:35 AM rue_mohr: eek no
03:35 AM rue_mohr: no headroom
03:36 AM rue_mohr: oh no!!!
03:37 AM rue_mohr: the new timer chip has caused all the text on the screen to be upside down!!!!
03:37 AM polprog: haha
03:37 AM rue_mohr: :)
03:38 AM rue_mohr: retire
03:38 AM rue_mohr: hmm
03:38 AM rue_mohr: the people I know now cant retire
03:38 AM rue_mohr: I know people working again at 80
03:38 AM rue_mohr: dear god, one day their not gonna let the good chap doing maintenance at the old folks home out
03:45 AM rue_mohr: cool, the crash protection worke
03:45 AM rue_mohr: d
03:46 AM rue_mohr: I called a fn that wasn't here yet and it did a restart
03:50 AM polprog: neat
03:51 AM rue_mohr: when I boot up, I push the address of the cold reset address on the stack
03:51 AM rue_mohr: an over-return will land it there
03:54 AM rue_mohr: hmm, isn't working
03:57 AM rue_mohr: oops
03:57 AM rue_mohr: almost 2am, I should prolyl go to bed
03:58 AM rue_mohr: the code should have written to 2003, then 2002, ...
03:58 AM rue_mohr: but it started at 2000 and went down
01:38 PM rue_mohr: no sunshine today
02:01 PM rue_mohr: its just hiding while I'm off from work
02:01 PM rue_mohr: it'll be out tommorow in full force
02:06 PM Tom_L: of course
02:06 PM Tom_L: especially if you have attic work
02:09 PM rue_mohr: hah
03:33 PM Tom_L: why we off today?
03:33 PM rue_mohr: holiday
03:34 PM * Tom_L checks his Canadian calendar...
03:34 PM Tom_L: hmmmmmmmmmmmm
03:35 PM Tom_L: so what you gettin done today?
03:35 PM Tom_L: fix the mower?
03:43 PM rue_mohr: no
03:43 PM rue_mohr: its about to rain
03:43 PM rue_mohr: I wanted to do some welding too
03:43 PM rue_mohr: that sait
03:44 PM rue_mohr: if I just start welding so the rain can start maybe there will be a bit of usefull day left at the end
03:45 PM rue_shop1: oh there, it started raining
03:45 PM rue_shop1: I didn't even have to haul out the welder
03:47 PM Tom_L: hah just mentioning it !
03:47 PM Tom_L: you know, i bet if i tried that cut that stalled the spindel it would handle it fine now
03:48 PM Tom_L: not that i'm going to try it
03:51 PM rue_mohr: did you find out about the linuxcnc pid tuning?
03:53 PM Tom_L: no, i've not done any more on it since yesterday
03:54 PM Tom_L: after i reverted settings, i did tweak the KP KI just a bit to see
03:54 PM Tom_L: it's still alot better than it was
03:55 PM Tom_L: i've got better response and can go down to probably 100 rpm
03:55 PM Tom_L: it does overshoot if i command like 3000 rpm from a dead stop
03:55 PM Tom_L: it snaps right up to 1500 or 2k
03:56 PM Tom_L: i think no matter how close i get it, there will either be compromise on the top end or on the bottom
04:18 PM rue_shop1: those settings in the middle were pretty badly out
04:21 PM Tom_L: which?
04:22 PM rue_mohr: the linuxcnc settings, and the vfd settings
04:22 PM rue_mohr: that do the analog conveyance
04:22 PM Tom_L: yeah
04:23 PM Tom_L: i think he sells the vfd combo with very mild settings so nobody will have trouble out of the box
04:23 PM Tom_L: but they are very mushy settings
04:43 PM rue_shop1: we had those settings right, just needed linuxcnc adjustments
04:47 PM rue_shop1: I suppose a good start to taking apart the boiler is getting the crane up here
04:48 PM Tom_L: i know and plan to use them once i find time enough to tweak the linuxcnc settings
04:49 PM Tom_L: i think those factory default KP and KI settings may have been too agressive along with the lcnc pid
05:47 PM rue_mohr: as long as it wasn't overshooting it should be ok
05:47 PM rue_mohr: the PID should do nothing if it doesn't need to
05:47 PM rue_mohr: thats why I dont understand why the gain of 1 didn't work
05:47 PM rue_mohr: it should have been the same as open loop
05:48 PM Tom_L: i thought i understood pid to a degree..
05:49 PM rue_mohr: what I'v learned is that its as long as a string
05:49 PM rue_mohr: there are so many ways to implement it, that it can be totally different
06:02 PM Tom_L: gets confusing for sure
06:08 PM Tom_L: i think as a test i'll leave the KP KI backed off from the 15xx and 320 numbers and try the rest of the parameters with the pid i have in place in linuxcnc and see what that does
06:08 PM Tom_L: i think with KP KI set that high it just fights the linuxcnc ones
06:08 PM Tom_L: and the linuxcnc ones is what gets me the bottom end rpm
06:41 PM rue_mohr: I dont understand why Kp of 1 is not straight feedthru
06:45 PM Tom_shop: damn this is getting good
06:45 PM Tom_shop: setp pid.0.Pgain 2.20
06:45 PM Tom_shop: setp pid.0.Igain 4.5
06:45 PM Tom_shop: setp pid.0.FF0 1.0
06:45 PM Tom_shop: setp pid.0.FF1 .75
06:45 PM Tom_shop: setp pid.0.maxerror 200
06:46 PM Tom_shop: nearly perfect with your stuff from yesterday
06:46 PM Tom_shop: except the KP KI i set to 870 & 100
06:46 PM Tom_shop: NO overshoot at any rpm
06:46 PM Tom_shop: and snaps to speed real quick
06:47 PM Tom_shop: the low 300ish takes a sec but i'm ok with that
06:48 PM rue_mohr: nice
06:49 PM rue_mohr: so, hows it hold under load then
06:49 PM Tom_shop: i think the reason KP o the linux side didn't work was because KP KI were too high on the vfd
06:49 PM rue_mohr: ok
06:49 PM Tom_shop: haven't tried a load yet
06:49 PM Tom_shop: i just wandered out here a few min ago
06:49 PM Tom_shop: pretty happy though
06:50 PM Tom_shop: you remember how it would overshoot coming up to speed?
06:50 PM Tom_shop: i'll shoot a video, listen to the sound
06:52 PM Tom_shop: bak inside in a sec
07:02 PM Tom_L: https://www.youtube.com/watch?v=iruAYR-V3Ro
07:03 PM Tom_L: i think it's pretty close
07:04 PM Tom_L: i might bump the decel up a bit to match the accel
07:08 PM rue_mohr: be mindfull that brakeing backpowers the DC bus on the VFD
07:08 PM rue_mohr: some of them blow up if you brake too hard
07:08 PM Tom_L: that's why i haven't changed it yet
07:08 PM Tom_L: i may just leave it alone
07:09 PM Tom_L: i _am_ using the external resistor though
07:09 PM Tom_L: and it hasn't even gotten warm at all
07:09 PM Tom_L: i'd rather play it safe
07:09 PM Tom_L: what's your thoughts on the response?
07:10 PM Tom_L: i don't hear any overshoot now
07:10 PM Tom_L: nothing like it was
07:11 PM Tom_L: i'm gonna stuff a chunk in the vise and see what it does with a manual cut in a bit
07:46 PM Tom_shop: ok, .375 cutter probably .1 deep, half the cutter width @ 2500rpm F15 spindle doesn't bog at all
07:47 PM Tom_shop: i'd say we did a fine job
07:47 PM Tom_shop: thanks for the help
08:14 PM rue_mohr: :)
08:14 PM Tom_L: i don't think the machine would take much more of a cut
08:15 PM rue_mohr: I think I know why I keep getting t e h instead of the, I think my keyboard, between slow scan times and wrong priorities casues it
08:15 PM Tom_L: i just don't type as good as i used to
08:49 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/configs/NEW_MILL_SHOP_2.8.1/New_Mill_IO_Pinout.pdf
08:49 PM Tom_L: last page shows what we did
09:06 PM rue_mohr: how long were we at it?
09:08 PM Tom_L: about an evening and a half maybe?
09:08 PM Tom_L: i never would have done those calcs
09:09 PM Tom_L: but at least i keep good records of the mill
09:20 PM rue_mohr: it was a mess
09:21 PM rue_mohr: doesn't help that it wouldn't go to 10
09:38 PM Tom_L: still not sure about that other than the vfd KP KI were too high
09:39 PM rue_mohr: ok
09:39 PM Tom_L: other than the balance bot that's all the PID i've messed with hands on
09:40 PM rue_mohr: ;)
09:40 PM rue_mohr: I learned not that long ago that its a mess
09:41 PM Tom_L: not so bad if you remember a few simple things about it
09:41 PM rue_mohr: everything about how a PID loop is done seems to be totally different from one implementation to the next
09:41 PM Tom_L: true
09:42 PM Tom_L: one more under the belt
09:43 PM Tom_L: toward the end, the FF0 & FF1 helped quite a bit
09:44 PM Tom_L: on the bottom end especially
09:48 PM rue_mohr: what are they?
09:50 PM Tom_L: Zero order feed-forward term. Produces a contribution to the output that is FF0 multiplied by the commanded value. For position loops, it should usually be left at zero. For velocity loops, FF0 can compensate for friction or motor counter-EMF and may permit better tuning if used properly.
09:50 PM Tom_L: they take it from FF0 to FF3
09:50 PM Tom_L: but i only used FF0-1
09:51 PM Tom_L: First order feed-forward term. Produces a contribution to the output that FF1 multiplied by the derivative of the commanded value. For position loops, the contribution is proportional to speed, and can be used to compensate for friction or motor CEMF. For velocity loops, it is proportional to acceleration and can compensate for inertia. In both cases, it can result in better tuning if used properly.
09:52 PM Tom_L: http://linuxcnc.org/docs/2.8/html/man/man9/pid.9.html
09:54 PM Tom_L: 2-3 not used for velocity
10:28 PM rue_mohr: hmm