#linuxcnc-devel Logs

Jan 12 2018

09:37 AM speedy: related to yesterday's talk about halui.feed-override.value: changing this value definitely affects jogging speed via axis.N.jog-* (in vel-mode being both: true or false). it is machinekit -2.6.0~pre1. is this a known issue to linuxcnc?
10:37 AM dgarr: speedy: longstanding behavior in control.c:handle_jogwheels(), (2.7 branch):
10:37 AM dgarr: https://github.com/LinuxCNC/linuxcnc/blob/1140fc156648b8e522e4b1f0e23c950b5bcc2981/src/emc/motion/control.c#L932
10:37 AM dgarr: and (for velmode honoring net_feed_scale):
10:37 AM dgarr: https://github.com/LinuxCNC/linuxcnc/blob/1140fc156648b8e522e4b1f0e23c950b5bcc2981/src/emc/motion/control.c#L963
10:37 AM dgarr: same behavior exists in master branch for *joint* jogging but feed settings not honored for *axis* letter (coordinate) jogging
10:40 AM dgarr: not sure what is best behavior should be but the implementation has not changed in years:
10:40 AM dgarr: https://github.com/LinuxCNC/linuxcnc/commit/57e8a63e31
10:40 AM dgarr: https://github.com/LinuxCNC/linuxcnc/commit/bc268d0361
10:40 AM dgarr: not sure what is best behavior should be but the implementation has not changed in years:
10:40 AM dgarr: https://github.com/LinuxCNC/linuxcnc/commit/57e8a63e31
10:40 AM dgarr: https://github.com/LinuxCNC/linuxcnc/commit/bc268d0361
01:26 PM seb_kuzminsky: skunkworks: i think that should fix the hm2 stepgen runaway reported on the forum (and by me here on irc)
01:26 PM skunkworks: Yay! thanks
01:26 PM seb_kuzminsky: jepler: can you take a look at that commit and see if you agree?
01:26 PM seb_kuzminsky: bbl
04:21 PM jepler: someone should figure out if this can make linuxcnc tests faster https://idea.popcount.org/2013-07-19-how-to-sleep-a-million-years/ (MIT license, home at https://github.com/majek/fluxcapacitor#fluxcapacitor)
04:22 PM jepler: maybe I'll try to do it this weekend .. if I can find the time (heh)
04:28 PM jepler: seb_kuzminsky: argh did I break it? typical
04:28 PM jepler: looks like I did. sigh, thanks for the fix
04:37 PM jepler: errrr not sure about the removal of the size suffix
04:37 PM jepler: from the "64" constants in particular
04:38 PM seb_kuzminsky: ii didn't touch the 64-bit constsnts, did i? i didn't mean to!
04:38 PM seb_kuzminsky: i made the 32-bit ones look like they do in stdint.h
04:40 PM seb_kuzminsky: oops i did
04:40 PM seb_kuzminsky: derp
04:40 PM seb_kuzminsky: (i should mention before i forget, rtapi_print() silently does the wrong thing with %lld and %llu, tied my head in knots for like an hour)
04:48 PM seb_kuzminsky: it *looks* right, but i'm open to the argument that it's still wrong somehow
04:48 PM seb_kuzminsky: http://paste.debian.net/1004912/
04:48 PM seb_kuzminsky: it also works right with -std=gnu11
05:03 PM seb_kuzminsky: my reading of the freely-available drafts of the ANSI C standards (90 and 99) are that a decimal integer constant consist of a non-zero digit [1-9] followed by zero or more digits [0-9], and are of the smallest type that can represent it (from the set {int, long int, and long long int} for C99 or from the set { int, long int, unsigned long int } for C90)
05:13 PM pcw_mesa_ is now known as pcw_mesa
06:08 PM mozmck: Did someone in here convert a bridgeport R2E3 to linuxcnc once?
06:22 PM seb_kuzminsky: cradek and i did
06:22 PM seb_kuzminsky: well, cradek did and i was in the room
06:24 PM Tom_L: heh
06:25 PM seb_kuzminsky: rebuilt the power supply to run on single phase ac, replaced the air motor that controlled the CVT with a stepper motor
06:25 PM seb_kuzminsky: kept the servo amps
06:25 PM seb_kuzminsky: replaced the control electronics with an old PC and a couple of mesa boards
06:25 PM seb_kuzminsky: replaced the control panel with a shop-made one
06:26 PM mozmck: Well, I have one to convert, and might need some pointers!
06:26 PM seb_kuzminsky: i'd be happy to help
06:26 PM mozmck: It was running, but seems that the controller is acting up and giving weird errors (they keep changing).
06:27 PM seb_kuzminsky: heh yeah
06:27 PM mozmck: We were wanting to convert it anyhow, so it's probably time.
06:27 PM seb_kuzminsky: cradek had been running the machine with the original Boss8(?) controller for years, and i remember he was so happy when it homed with linuxcnc the first time
06:27 PM seb_kuzminsky: (though we called it emc2 back then)
06:28 PM seb_kuzminsky: simultaneous X & Y homing :-)
06:28 PM seb_kuzminsky: he said "it's already moving better than it ever has!"
06:28 PM mozmck: Ah, yes! A lot of things will be nice! The "screen" on the control is something like 40 characters by 2 lines high.
06:29 PM mozmck: Do the servos take 0-10V analog in?
06:29 PM seb_kuzminsky: i bought the machine from him and i've been using it for the past 10ish years, 0 problems
06:29 PM seb_kuzminsky: yep
06:30 PM mozmck: I'll have to figure out what mesa cards I need. I have 7i92 and 7i80
06:30 PM seb_kuzminsky: i'm driving it with a 5i22 and a 7i48 iirc
06:30 PM seb_kuzminsky: i wanted a fourth encoder channel for spindle feedback, though i haven't gotten around to making that happen yet
06:30 PM mozmck: I assume I need some DACs for servo control and something for the encoders.
06:31 PM seb_kuzminsky: the 7i48 is all the interface hardware you need for the servos
06:31 PM seb_kuzminsky: or a 7i33, either way
06:32 PM mozmck: I bet the 7i33 would be enough - 3 axes + spindle. Unless we wanted to put a 4th axis on the table...
06:33 PM mozmck: But we probably won't.
06:33 PM seb_kuzminsky: yeah, i dont remember now why i opted for the 7i48
06:33 PM seb_kuzminsky: to try out muxed encoders maybe?
06:33 PM seb_kuzminsky: i use an Automation Direct GS2-23P0 VFD to drive the spindle motor
06:33 PM mozmck: Oh, not the original stuff?
06:34 PM seb_kuzminsky: the original spindle drive expected 3-phase input power, which i dont have
06:34 PM mozmck: Ah, we generate 3-phase with a rotary converter.
06:35 PM seb_kuzminsky: years ago i somehow ended up with a cardboard box full of small stepper motors with integrated drive electronics, just the thing for the VCVT control
06:35 PM mozmck: We run a Haas VF1 on it, and had to add an extra idler to run the bridgeport at the same time.
06:36 PM mozmck: VCVT control?
06:36 PM mozmck: The second character shows as a box with numbers in it here.
06:37 PM seb_kuzminsky: i meant CVT, continuously variable transmission
06:37 PM seb_kuzminsky: this is the stepper i use: https://www.linengineering.com/products/integrated_motors/silverpak-17d/
06:37 PM seb_kuzminsky: though if you have 3-phase i bet you have compressed air too
06:37 PM mozmck: Oh, the variable speed pulley drive?
06:37 PM seb_kuzminsky: yeah execatly
06:37 PM seb_kuzminsky: exactly
06:37 PM seb_kuzminsky: i don't know if CVT is the right name for it
06:38 PM mozmck: Yes compressed air, and even got a dryer for it.
06:38 PM mozmck: I don't know either - I just call it variable speed pulleys!
06:38 PM seb_kuzminsky: two conical pulleys with the points pointing at each other, and you can change the offset between them, and a belt runs at whatever diameter is implied by their distance from each other
06:38 PM mozmck: yep
06:41 PM seb_kuzminsky: https://github.com/SebKuzminsky/linuxcnc-config-bridgeport-r2e3
06:42 PM seb_kuzminsky: that's for 2.7, i haven't jumped on the JA bandwagon yet :0/
06:42 PM mozmck: Thanks!
06:42 PM seb_kuzminsky: works well wiwith both Wheezy/rtai and Stretch/rtpreempt
06:42 PM mozmck: I haven't either on JA mostly from lack of time.
06:43 PM seb_kuzminsky: i run ja on some of the machines at the hackspace
07:09 PM mozmck: The 7i77 looks interesting - it has all the I/O as well as 6 servo interfaces. I wonder how he gets all that through a 25 pin connection? smart-serial?
07:44 PM pcw_mesa: latest stretch ISO on a H97/g3258:
07:44 PM pcw_mesa: http://freeby.mesanet.com/h97-stretch.png
07:46 PM pcw_mesa: only 5 minutes or so (youtube video running +glxgears) , will leave running over weekend
07:47 PM mozmck: Hi pcw_mesa! Is the 7i77 the only analog servo board you have that would interface with a 7i92?
07:48 PM pcw_mesa: Yes
07:48 PM pcw_mesa: well 7I97 but a bit unsupported yet
07:48 PM mozmck: Ok. How does it have 48 I/O and all through the 25 pin connector?
07:48 PM pcw_mesa: sserial for analog and digital I/O
07:49 PM mozmck: I figured it was something like that.
07:49 PM pcw_mesa: so 7 wires for 3 sserial channels (analog/digital/expansion) and 10 for 6 encoder channels
07:51 PM mozmck: That gives a bit more I/O than a 7i37, but I don't know how much we will need right now.
07:57 PM pcw_mesa: you need a 50 pin FPGA card for the 7I37 (or some ugly cabling)
07:58 PM mozmck: Yes, looking at either a 7i93 + 7i33 + 7i37, or possibly 7i92 + 7i77
08:02 PM pcw_mesa: bbl dinnertime
08:12 PM skunkworks: we are running the matsuura 7i80 + 7i48 + some smart serial cards.
08:14 PM skunkworks: way more I/O than I need :)