Back
[09:10:28] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/moveoff_touchy 1dc5885 06linuxcnc 10configs/sim/axis/moveoff/hookup_moveoff.tcl 03configs/sim/hookup_moveoff.tcl 10scripts/moveoff_gui moveoff_gui: provisions for non-axis ui * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=1dc5885
[09:10:28] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/moveoff_touchy 6067d8b 06linuxcnc 03configs/sim/touchy/ngcgui/hookup_moveoff.tcl 03configs/sim/touchy/ngcgui/pyngcgui_touchy_moveoff.ini 03configs/sim/touchy/ngcgui/pyngcgui_touchy_moveoff.txt pyngcgui_touchy_moveoff.ini: touchy moveoff demo * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=6067d8b
[09:20:12] <dgarr> i will test deb for this new branch and if ok add to 2.7 axis demo
https://www.youtube.com/watch?v=KY6hx7WBkO8
[09:20:18] <dgarr> touchy demo
https://www.youtube.com/watch?v=XGDq6620fPQ
[09:23:14] <cradek> dgarr: awesome
[09:24:05] <skunkworks> dgarr, is the waypoint limit a hard limit or settable by a hal pin?
[09:26:12] <dgarr> hard coded at 1000 points, the distance /time between points is controlled by hal pins waypoint-sample-secs and waypoint-threshold
[09:26:58] <dgarr> it may take some experimentation to set the hal pins but for the use-cases that seem reasonable, i imageine 1000 is enough
[09:27:13] <skunkworks> ah - cool
[09:27:55] <dgarr> Dans ses écrits, un sage Italien Dit que le mieux est l'ennemi du bien.
[09:29:07] <skunkworks> sound about right :)
[09:29:08] <cradek> very true
[10:33:31] <kwallace> Smithy doesn't use LinuxCNC anymore?
[10:35:28] <skunkworks> looks like they went siemens
[10:37:05] <kwallace> Their small CNC machines look a bit on the larger size too.
[10:38:55] <skunkworks> maybe that is why we have not seen matt around :(
[10:54:33] <kwallace> Just thinking out load. Let's say I have a 100 ppr spindle encoder for a 10k RPM spindle. I use three fairly quick I/O pins to get A B and Z into a counter at base thread rates. On the other hand, what if I had the counter at the sensor. I could send count data at servo rates and be just as well off. Maybe? Then I could use Modbus RTU or Ethernet?
[10:57:10] <skunkworks> yes. is modbus realtime?
[10:57:40] * skunkworks hasn't played with modbus
[10:58:15] <kwallace> Oh crud, no, oops.
[10:58:53] <skunkworks> mesa already did all this stuff for you... ;)
[10:59:34] <kwallace> Yeah, I think so. I need to start looking in that direction.
[11:01:05] <skunkworks> if you are converting from the printer port - the 5i25 would be a painless step as you could initally use your existing breakout board(s) (it can emulate 2 printer ports in effect)
[11:01:42] <skunkworks> But still give you high speed encoder counting, pwm and step generation.
[11:02:01] <skunkworks> (up to the limit of single ended encoders..)
[11:03:51] <skunkworks> and a butt ton of expantion options
[11:04:04] * skunkworks likes mesa...
[11:04:22] <KGB-linuxcnc> 03Dewey Garrett 052.7 62b2ab9 06linuxcnc 10configs/sim/axis/moveoff/hookup_moveoff.tcl 03configs/sim/hookup_moveoff.tcl 10scripts/moveoff_gui moveoff_gui: provisions for non-axis ui * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=62b2ab9
[11:04:22] <KGB-linuxcnc> 03Dewey Garrett 052.7 8298b47 06linuxcnc 03configs/sim/touchy/ngcgui/hookup_moveoff.tcl 03configs/sim/touchy/ngcgui/pyngcgui_touchy_moveoff.ini 03configs/sim/touchy/ngcgui/pyngcgui_touchy_moveoff.txt pyngcgui_touchy_moveoff.ini: touchy moveoff demo * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=8298b47
[11:07:51] <kwallace> We have a 5i25 onboard. I'm exploring having the counter at the sensor so that the connection can be slower and maybe have more than one device on the wire.
[11:09:35] <skunkworks> andy has used ,umm, smart serial is it? to communicate realtime betweeen and arduino and mesa I think.
[11:10:24] <skunkworks> you would have to ask him
[11:11:04] <cradek> lots of the mesa and ppmc products do exactly this (count in specialized hardware, and then the servo cycle just reads the count at its leisure)
[11:11:57] <cradek> so you're not inventing anything new, unless you decide to try to reimplement the idea badly in an arduino or pic or something, as people do
[11:15:37] <bjmorel_work> Signal isolator for KBIC-125 from KB is about $75.00 , Spinx1 from Mesa, $24.00
[11:15:46] <bjmorel_work> Not really a hard choice
[11:16:55] <skunkworks> And you know which one is probably build better :)
[11:17:08] <cradek> I think the kbic stuff is nice
[11:17:21] <cradek> sherline uses a kbic drive on their spindles and it's great
[11:17:39] <kwallace> I'm thinking I need to come up with an accessory I/O port that is enough of an improvement in order to jump over the current I/O port.
[11:17:53] <cradek> iirc, I bought the isolator board, but by the time I got it I had already installed an opto myself
[11:18:09] <bjmorel_work> I think they make nice stuff, but it's not 3x as nice for sure.
[11:18:36] <cradek> http://timeguy.com/cradek-files/cnc/lathe/DSCN6292.JPG
[11:18:54] <cradek> heh here's the "isolator board" on my lathe
[11:19:03] <cradek> I'll sell you the plans for $75
[11:19:21] <cradek> from
http://timeguy.com/cradek/cnc/lathe
[11:20:44] <cradek> sucky part now is that to use the drawbar you have to take off the spindle encoder:
http://timeguy.com/cradek-files/cnc/lathe/DSCN6300.JPG
[11:21:43] <CaptHindsight> http://linuxgizmos.com/35-dollar-quad-core-hacker-sbc-offers-rpi-like-expansion/
[11:24:18] <kwallace> The problem, for me at least, with the Pi and Arduino format is not enough I/O pins. I rather like the Due though:
http://arduino.cc/en/Main/ArduinoBoardDue
[11:26:40] <bjmorel_work> Ya, I want to do an encoder directly on the spindle of my mill, but with the MT3 spindle I need the drawbar all the time. I think I'm going to try to fab up an encoder disk like people have done on the 7x lathes.
[11:28:14] <archivist> I mounted one on the back of a pulley (taper lock type) on one of my spindles
[11:35:08] <kwallace> I have about a half inch space between the current belt pulley and the bearing housing. The plan is to add a steel disk to the back of the pulley with gear like teeth and Allegro gear sensors. The problems are to come up with a pcb to mount the sensors, then to get the pulses to the PC through a single bit I/O, or come up with a better I/O.
[11:35:39] <bjmorel_work> I think I'm going to go the treadmill motor route w/ the KBIC controller. I should be able to eliminate the two speed belt drive and replace it with a small timing belt drive.
[11:35:54] <bjmorel_work> Might have the room to add the encoder disk right on the spindle.
[11:57:13] <kwallace> In case anyone might be interested, I made some machinable wax yesterday using ice cream buckets (#2, HDPE) and candles:
http://wallacecompany.com/tmp/IMG_8432-1a.jpg . It was easy to make and seems to work:
http://wallacecompany.com/tmp/IMG_8427-1a.jpg
[12:17:23] <skunkworks> kwallace, How is the mill interface coming along?
[12:24:34] <kwallace> skunkworks, I couldn't say, but I'm starting to look for the next thing to do.
[12:33:30] <skunkworks> this is random...
https://code.google.com/p/dxf2gcode/source/browse/trunk/source/tooltable/tables/lathe-cradek.tbl?r=241
[12:37:11] <kme> is anyone watching the PocketNC development?
http://www.pocketnc.com/products/
[13:09:09] <archivist> putting two steppers in parallel is somewhat silly
[13:22:45] <cradek> skunkworks: weird
[13:29:16] <kme> yeah I couldn't think of a good reason to do that -
[13:32:49] <kme> worm drives or anti-backlash spur gears may have been a better choice
[13:35:05] <kme> most industrial rotary tables are direct drive I believe -
[14:07:04] <kwallace> The steppers could be adjustable such that you can loosen the mount on one stepper and rotate it to take out any backlash then tighten. From then on, there will be a preload. During start up, one stepper would need to be run against the other to restore the preload. Just a guess.
[14:19:53] <PCW> kwallace: if you are running out of I//O pins on FPGA config I would _not_ move the encoder
[14:19:54] <PCW> elsewhere. I would use the FPGA encoder interface and use serial for digital I/O expansion
[14:23:01] <PCW> (sserial/modbus/SPI/muxed pins whatever)
[14:26:56] <kwallace> PCW, I have started looking at Smart Serial, but one application is to have the spindle encoder counter reside with the sensors on the spindle. The 7i73 has an encoder, but I need something about the size of an RJ-45 connector.
[14:53:45] <PCW> There are major tradeoffs doing this
[14:54:20] <PCW> if the encoder is used for anything other than a speed sensor
[14:54:50] <PCW> if its for real time feedback, serial is tough
[14:56:29] <PCW> and assuming you want synchronous readings from linuxcnc you still need 2 wires: request and data
[15:32:26] <kwallace2> PCW, I would need the encoder for rigid tapping. The number of wires isn't a problem.
[15:33:43] <PCW> well serial will be a problem because you either have to request the data and wait for it or always be 1 cycle late
[15:34:06] <PCW> neither of which a great for real time
[15:34:23] <PCW> are great
[15:34:48] <kwallace2> How about Ethernet?
[15:34:55] <PCW> same issue
[15:36:06] <kwallace2> Okay, back to the drawing board, thank you for your help.
[15:36:30] <PCW> now with fast serial or Ethernet you may not have to wait very long so at a 1 KHz sample rate that may be acceptable
[15:40:02] <PCW> at say 10 MBits/s a RS422 link with request and data returned might be OK, only taking a few usec (assuming say a Cortex M3 or something decent at the remote end)
[15:41:59] <kwallace2> The normal servo rate is 1kHz. If I can keep that, I should be in good shape, I think.
[15:59:50] <PCW> Is this just for wiring ease?, seems like quite a project to save a few wires
[16:11:28] <kwallace2> PCW, not really. I want to add a spindle encoder, but try to stay within the requirements of the current hardware. I can't, so I'm looking at alternatives. I think I really need to nail down what I want and what I have.
[16:13:38] <PCW> Like I said If you are making any changes in the breakout I would let real time stuff have priority and use serial for non real time
[16:15:10] <PCW> if you have stepgens, muxing the stepgens is a possibility (drops 4 axis from 8 pins to 5)
[17:15:03] <PCW> kwallace2: I guess if you had fast hardware/uproc the uproc could send absolute position at a high rate (say 2 byte position 1 byte CRC) at 10 mbits/sec
[17:15:05] <PCW> with maybe a 3 usec gap so 166 KHz update rate (the position code could have a reserved number for unknown pos (untill index hit))
[17:16:34] <PCW> thats no worse Jitter than normal for PCs
[17:16:58] <PCW> (and only 1 wire)
[17:19:29] <PCW> you could also encode the absolute encoder position as a frequency and measure this
[17:20:00] <PCW> (better noise immunity than serial)
[17:37:15] <kwallace2> PCW, I'm sorry, I should not have said anything before doing my homework. Please don't fret too much on this.
[21:56:54] <skunkworks> dad has a replacement servo coming from ebay.. (a bit of a gamble - but cheap)
[23:53:57] <memfrob> with kernel 3.18 my radeon r9 290 works with 2d and 3d hw accel, first time i ever had this working
