#linuxcnc-devel Logs

Jul 05 2018

02:17 PM pcw_mesa: So if you dont move more than 1/2 turn when off you can pretend its multiturn absolute?
03:59 PM rene-dev_: pcw_mesa yes, I think thats the idea.
04:07 PM pcw_mesa: I think some older machines did this (Fadal?)
04:09 PM andypugh: jepler: Your patch seems to fix the crash on loading.
04:10 PM andypugh: But: I am not sure if it doesn’t bring the watchdog bite bug back.
04:11 PM andypugh: pcw_home: I am not seeing any analogue output, and I am seeing mesa-7i65.watchdog.has-bit as true. There might, or might not, be a connection?
04:14 PM pcw_mesa: Yes if the WD has bit, analog outs are set to 0V
04:14 PM pcw_mesa: (and you get a red light on the 7I65)
04:39 PM andypugh: I don’t think I have ever seen the red LED go off.
04:41 PM pcw_mesa: set the WD for the max time
04:42 PM pcw_mesa: I,ll try here in a bit, is the change merged into 2.8?
04:44 PM andypugh: Not yet.
04:45 PM andypugh: But it’s a couple of very wimple edits to tram.c
04:45 PM andypugh: How do you reset the watchdog on the 7i65?
04:45 PM pcw_mesa: werry werry wimple?
04:46 PM andypugh: (I get the red light even with both jumpers up, for no watchdog)
04:46 PM andypugh: You need to get a nun to do it for you,
04:46 PM pcw_mesa: :-)
04:48 PM pcw_mesa: The WD is reset via a CPLD SPI command bit, not sure if the driver knows how to reset it if tripped while running
04:49 PM rene-dev_: andypugh speaking of things I have but dont need: a 5 axis wire edm machine. https://imgur.com/a/TAwnFKn
04:50 PM rene-dev_: even comes with gremlin https://imgur.com/a/TK6SK6v
04:51 PM andypugh: You just can’t resist anything painted RAL 6011 can you?
04:52 PM andypugh: pcw_home: I think it is tripped permanently. Surely if the 7i65 is powered up before the driver is loaded, it will bite?
04:55 PM rene-dev_: I have no idea what to do with it. it was cheap, and around the corner.
04:55 PM andypugh: I might have broken my 7i65 though. One input shows 6.5 volts all the time
04:56 PM andypugh: The others show -4V, but they are disconnected and do flicker
04:56 PM andypugh: rene-dev_: LinuxCNC. Set it up as a foam cutter config.
04:56 PM rene-dev_: thats only 4 axis
04:57 PM pcw_mesa: that does sound bad Although your is so old the CPLD firmware may be obsolete
04:58 PM andypugh: pcw_home: I won’t worry if your setup is working with the jepler patch then
04:58 PM andypugh: People are using them, i believe?
04:58 PM rene-dev_: I do have the original controller, and see how far I get with it. even with a retrofit, I still need the generator.
04:58 PM andypugh: You can help test the reverse-run branch
04:59 PM rene-dev_: :D
05:03 PM pcw_mesa: andypugh yes around 100 (low qty because of price I think)
05:04 PM rene-dev_: andypugh I want to do the same thing with stmbl. the pseudo-abs
05:04 PM andypugh: That means that I have to up my estimate of linuxCNC users from 40 (that I have met) to 100 :-)
05:04 PM rene-dev_: is there any reason the reverse run branch is not in master? seems pretty outdated
05:05 PM andypugh: rene-dev_: I was about to start nagging you to put it in STMBL
05:05 PM andypugh: Needs to be optonal, some axes drop
05:05 PM andypugh: Would STMBL store full-turns on-board?
05:09 PM pcw_mesa: One of our older encoder interface cards could store the current count in Ferro-electric memory (FRAM)
05:09 PM pcw_mesa: not sure those are even a thing anymore
05:09 PM andypugh: Cool!
05:10 PM andypugh: Reminds me that our envronmental test cells use Tramsputers
05:10 PM andypugh: Sorry, Transputers
05:10 PM pcw_mesa: Wow thats a blast from the past
05:11 PM andypugh: Yes, every time I see the message “initialising Transputer” scroll past I am back in the 1980s
05:12 PM pcw_mesa: I think the Transputer people make some cheap relatively current high count multiprocessor chips now
05:12 PM rene-dev_: andypugh stmbl cant store anything.
05:13 PM rene-dev_: andypugh it will report the position with turns reset to 0 after a powercycle
05:13 PM rene-dev_: if you tell me what pins you need, I can put that in
05:15 PM andypugh: It’s not how I would want to do it, but the resolver code watches motor-pos-fb to see when the values read from position.txt are stuffed in there.
05:15 PM andypugh: Sorry, joint-pos-fb
05:16 PM andypugh: So I guess another float input. Seems a waste, but there is room.
05:16 PM pcw_mesa: andypugh do you have jeplers changes to tram.c ?
05:16 PM pcw_mesa: I can test that it works here
05:17 PM andypugh: https://github.com/LinuxCNC/linuxcnc/pull/456/files#diff-0
05:17 PM andypugh: Two additional “(char*)” casts
05:18 PM pcw_mesa: that is werry wimple
05:21 PM rene-dev_: I dont think there is room
05:43 PM andypugh: You can use a little float, it only needs to be accurate enough to guess full turns
05:44 PM andypugh: I put in support for 8-bit miniflat, that is too small, but float16 is probably enough
06:05 PM pcw_mesa: OK 7I65 seems to be fixed (launched/exited LinuxCNC about 30 times without a splat)
06:10 PM andypugh: Did the analogue outs work in your tests?
06:11 PM pcw_mesa: yes
06:11 PM andypugh: Great
06:11 PM andypugh: That’s a fix then
06:12 PM andypugh: And I guess my 7i65 would need an update if I needed it to work, but I don’t. (I can prove outputs as the enables work)
06:14 PM pcw_mesa: Yeah I think there was a CPLD change at some (very early) point
07:19 PM pcw_mesa: RPI didn't explode when plugged in to the 7C80 so that's good
07:19 PM jepler: forum update & reboot will be coming shortly
07:20 PM jepler: andypugh: I don't think 2.6 is affected. did you see different?
07:21 PM andypugh: I don’t know when the driver went in, to be honest.
07:21 PM andypugh: Let me check.
07:22 PM andypugh: 2.6 has the driver but not the bug-fix that caused the new bug.
07:22 PM andypugh: (It does no memset).
07:22 PM jepler: OK
07:23 PM jepler: thanks for clarifying that
07:23 PM jepler: andypugh, pcw_mesa: thanks for testing the fix
07:23 PM andypugh: In my testing just commenting out the memset was generally OK, dependng on what was in the memory. So 2.6 can probably be left.
07:24 PM andypugh: This was certainly an interesting one.
07:24 PM jepler: yes let's agree to call it interesting
07:24 PM andypugh: Valgrind showed it’s worth, so it is probably worth working out why it works for you and not me.
07:25 PM pcw_mesa: jepler: thanks for spotting that bug!