a few hundred $$ isn't going to be a big deal. an extra few thousand would be
I said simple hw - $1000 + $1000 isn't
is this a one-off project?
no matter what you are gonna have to deal with a couple hundred wires
right now, all we need is the ability to capture some images with some timing
yes, there are 75 cameras
what form are those wires in? is there a standard camera cable/connector?
so minimum 225 connections, and an additional 150 for feedback
bare ends, maybe a miniphone stereo plug
oh, I guess I should answer the phine
well, bare ends means you might as well just use screw terminals
one sec - finding out the details now
looks like the physical wiring can be handled by 7i42T - 24 screw terminals for $45
still adds up tho, you are talking about 375 wires if I understand properly, at ~$2 each thats $750
not sure I followed 75 cameras = 225 connections + 150 for feedback
I thought it was two out and one back per camera
ground + prerelease + release for the outputs, ground + contact for feedback
= 225+150 terminations, but only 225 total signals
still on the phone
the 7i42T has screw terms for ground and signal
so 225 signals at $2 per signal, $450
I may try to do it with what I have - I've got a couple of 7i37T, a few 7i37, and a 5i22 here
I have enough 50-pin terminal blocks for the 7i37's
proof of concept with fewer cameras?
that also extends my reach - I need ~40 feet end to end
small shoot - only 75 cameras
later, it'll be at least 120
I don't see how you can get close 75 cameras with one board
the first shot doesn't need all the fancy prerelease and feedback stuff
oh, just GPIO?
which is why I might be willing to do it with my own hardware instead of making the client buy his own
eventually, it'll be FPGA based timing (and maybe other functions), with custom hardware on the outside
so what do you actually need for phase 1 - jyst 75 outputs?
and maybe a trigger input ;)
do they need to be isolated?
I'd be happier with feedback for confirmation that the cameras fired
if not, spring for the 7i42s, they let you use all 24 pins in the cable as outs (or ins, your choice)
not really, except that people sometimes do funny things with generators and long metal trusses
they don't have buffers
I don't know that the FPGA can directly trigger a camera
especially at the end of a long cable
Pete recommended the same thing L(
so you are basically gonna be limited to 8 cameras per 7i37
oh shit. I forgot about that
this is probably too complex for phase 1, but lemme ramble a bit
ok - gotta pee ;)
20 cameras per cable...
20 input pins, direct to FPGA (put whatever isolation on the board you want - yes, I'm talking about a board)
the other 4 pins are: serial data out, serial clock, release vector strobe, pre-release-vector-strobe
the only logic on the boards is some 8 bit serial in, parallel out shift regs, 16 pin soics IIRC
the rest is isolation and termination, which you need regardless
copy peter's clever isolated output circuit from the 7i37, if you can tolerate the inherent delays
although if you know the camera characteristics, you can probably select a simple opto that will be faster and cheaper
the design is simple, layout simply boring (same circuits x20)
soldering them (by Friday evening) is the big issue
nothing like waiting till the last minute
heh, don't I know it
I didn't realise that "fast" was _that_ heavily weighted
I just found out that the job could happen yesterday (at 4:00 PM)
pretty much need to be 100% off the shelf then
yeah. unfortunately, even 2x 5i22 isn't enough for 75 cameras with off-the-shelf
and you are talking 10x 7i37s too, that is $800 right there
well, they're $65 at that quantity, but still
argh. gotta eat. after buying plane ticket ...
I thought of a way to use fewer 5i22s, at the expense of a very very simple board
three 50 pin connectors
one goes to the 5i22
I'm not so concerned with reducing the number of 5i22 boards
the other goes to two 7i37s
eventually there will be more complex functions in there
that would give you 16 out and 8 in (plus another 24 unused ins)
just multiplex somehow?
actually, you could do the same thing by splitting and offsetting the ribbon cable (I think)
now I gets it ;)
just take 12 signals from each 7i37, 8 out and 4 in, and merge them
right - I could split a ribbon cable into thirds and use 3x 7i37, for 24 outputs
the ribbon cable trick might even be viable
(in the available time)
I may even be able to do it with simple connectors - use RA headers pointing in opposite directions for two of them, and straight for the third
then just use 4 cables
one from 5i22 to 3x pin headers, one each from the 3x pin neaders to the 7i37
but you have to move signals around
not really, hm2 lets me use all pins as outputs
and I can move the header over so I'm always using the "left hand side"
pin 1 of 7i37 A goes to pin 1 of ribbon, pin 1 of 7i37 B goes to pin 13 of ribbon, etc
that's why I use RA and straight headers
oh, offsetting the headers
are you talking about just soldering the headers to each other in space, no board?
no solder, just use male pin headers, but plug both ends into connectors instead of one end into a PCB
then tape the hell out of it ;)
ugly but possibly functional
and use the separate power pins to power the 7i37's
I'd have to do that anyway, given the cable lengths
(I had planned to anyway)
pins 33 thru 47 of the 7i37 are outputs
how about this for perverse (but I bet more robust than the connector tricks
get some 16 pin IDC connectors
split the ribbon into three 16 conductor chunks, crimp one connector each
snip off pins 49 and 50 of the ribbon
and plug the 16 pin connector into the right place on the 7i37
the male IDC connectors are a pain
what male IDC connectors?
and the females (and males) can't plug next to pins, the sides are too big
so there would be bent pins on the other side
ok, use 50 pin connectors, but only stick 16 pins of cable into them
yep, that can work too
hydra cables - 50 pin connector and cable on one end, splitting into three 16 conductor branches terminating in three 50 pin connectors
that's a good idea
less likely to frotz something by plugging in off-by-one too
ok, time to eat. I'll be back later - maybe even get a DigiKey order in before the cutoff ;)
food time here too
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/hal/ (basic_hal.lyx drivers.lyx): move servo to go to own file and add a bit on net
I saw what you did skunkworks :)
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/drivers/servo_to_go.lyx: move servo to go to own file
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ (Master_Integrator.lyx Submakefile docs.xml index.tmpl): add servo to go to html
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/hal/images/HAL_Configuration.png: add graphic
Hi - how is it going?
good here didn't get much ice
urk - ice?
bit of an ice storm yesterday near by
maybe I should have checked the weather forecast before choosing flight routing
we have a bit of that going on right now
SWPadnos: you get the rust in check?
we had ice saturday night.
haven't had a chance to look at the machine today
and probably won't until next Thursday :(
crap can't see the keyboard'
one key off'
whee, nominations right and left
I have not checked in a while...
cool, looks like you're one of them
I saw that :)
wow - a petabyte of downloads per month at SF
(for at least a few months)
that would be well over my 10 TB limit ;)
hey - my weller solder station auto-shuts off. That is a plus for me.. :)
must be a new one, mine just stays on I think...
huh - how does this view on Linux systems? http://www.hammondmfg.com/pdf/1598B.pdf
on this Windows machine, I have a spinning model of the enclosure, which I can drag to change perspective
(though if I stop it, I get only a black rectangle where it should be)
can you move it?
that's, uh, I forget the name of it, but acroread for linux can do it
acrobat 3D or some such
I wonder if I'll get lucky and find an enclosure that a 7i37T will scre directly into
you mean 10 of them?
sure, that would be fine :)
you need to find someone with a CNC mill
hmmm. could be tough
you can use mine SWPadnos
can you fed-ex it to me
I like the UPS guys better
the fedex guys are from Sikeston
they get lost over here :)
I may be an evil genius
the 7i31 has buffer chips, which may be able to drive my outputs
err - inputs
how are you gonna reverse the drive direction?
oh, 7i31 - that is the LED board?
with 4 hex buffers
and I have 3 of them, which means 3 of these + 1 7i37 = 80 outputs and 16 inputs
how will you connect to the 7i31?
um. details details ;)
lots of little wires and soldering? ick
(probably soldering lots of wires to it)
slightly better than designing and building several boards with hundreds of components
hey seb_kuzminsky, got a question or two for you
EMC: 03cradek 07concave_comp2 * 10emc2/src/emc/rs274ngc/interp_convert.cc: unionize the queue because it's going to have a lot of types. but man is this all a pain.
EMC: 03cradek 07concave_comp2 * 10emc2/src/emc/rs274ngc/interp_convert.cc: allow dwells in concave corners
whee, this actually works
sigh. this doesn't
I can't get the hostmot2 driver to recognize my 5i22
or give me any error messages
the module loads, but exports nothing
loadrt hm2_5i20 config="firmware=hm2-trunk/5i20/SVST8_4IM2.BIT num_encoders=6 num_pwmgens=4 num_stepgens=0 enable_raw"
yep, that's pretty close to what I've got
so you have both lines
and I have num_*=0
and in that order
I don't have enable_raw
how recent is your source? num_*=0 have been bugs for some values of *, I think
your firmware path is surely different too - this is a local hack because of RIP
fresh checkout from an hour ago or less
yep. I tried 2.2.8 and also symlinked hm2-2 to mu trunk checkout
are there any jumpers of any sort that need to be set?
loadrt hm2_pci, not hm2_5i20
yes, got that
lspci shows the board, but hm2_pci doesnt detect it
well, hard to tell - there are no errors of any sort from hm2_pci
pastebin lspci -nnvv pls
one sec - I shut that machine down to look at jumpers :)
it emits nothing if it finds nothing
is this a 5i22-1.0 or -1.5?
I'm using the B variants
tried a couple
donno what B is
"Big", isn't it?
they're the larger bitfiles
oh, the *B.BIT firwmares, right, that's for the bigs
ok so that sounds good
someone else tried a -1.5 and reported it working
i dont have a -1.5, just a -1.0, so i havent seen it work, myself
ok, I guess this will be the first :)
the bug cradek mentions is in 2.2.8 (and the v2_2 branch), but not in trunk
it bites if num_encoders=0
ok, I was wondering about that
any other module can be set to 0 without problem
and in trunk the encoders can be totally turned off too
[04:32:44] <SWPLinux> http://pastebin.ca/1302288
I only copied out the PLX info. let me know if you need more
this has the 3131 ID
no big deal, we'll add it to the pci table, hold on
I think only the first few 5i22 had that
I guess you can still tell since the 5i22 uses a 9054 and the 5i20 uses a 9030
yea this is an easy fix
yep. I can add another entry to the ID table
hmm. that doesn't seem to have worked. maybe I did it wrong
ah, the big switch in hm2_pci_probe also needs fixing
which is a pain, since the only thing used to determine what board is present is the subsystem device
which was unique, until now
is the plx info also copied over?
oh, it's a dev struct - good
so I should have acces to dev->device as well
so, should I add a goto or do a block copy? :)
i'm working on a nested switch right now
ok, check bridge device then SS ID?
[ 1497.422791] hm2_pci: loading Mesa AnyIO HostMot2 driver version 0.6
[ 1497.424359] hm2_pci: discovered 5i22-1.5M (old) at 0000:00:09.0
[ 1497.424607] hm2/hm2_5i22.0: invalid token in config string: "hm2-2/5i22/SV16B.BIT"
[ 1497.424612] hm2_5i22.0: board fails HM2 registration
[ 1497.424825] hm2_pci: probe of 0000:00:09.0 failed with error -22
[ 1504.972578] hm2_pci: driver unloaded
that's good! it finds the board now :-)
should be firmware=hm2-2/blah/blah.bit
steve@goal3-1:/lib/firmware/hm2-2/5i22$ ls -l
drwxr-xr-x 2 steve steve 4096 2009-01-06 22:27 CVS
-rw-r--r-- 1 steve steve 651926 2008-10-29 00:18 SV16B.BIT
-rw-r--r-- 1 steve steve 8515 2008-10-25 00:54 SV16.PIN
-rw-r--r-- 1 steve steve 403014 2008-10-29 00:18 SV16S.BIT
-rw-r--r-- 1 steve steve 651926 2008-12-22 00:04 SVST824B.BIT
doesn't look it
SWP: how about just updating your EEPROM?
now that It's working? ;)
about how many boards went out with the 5i20 ID?
Dont want to spread embarrassing wrong subsystem IDs throughout the world...
only a few, maybe 5?
oh. the 5 I bought? ;)
would you prefer i not commit this update?
* seb_kuzminsky pauses over the Enter key
I think its better to update the EEPROM...
what's involved in doing that?
ok. what's the procedure to do that on Linux?
assuming no access to a WIndows machine that I can put the hardware into
DOS unforch, unless some nice Linux programmer wants to write a EEPROM utility
do you have the source?
I can probably translate
ok i'll revert this here
git --throw-away-scratchpad or something ;)
The utility just writes one word to EEPROM, the batchfile has all the data and comments
that should be relatively easy to port
gotta install dos2unix ;)
If you download the 5i22 support zipfile you want 5I22EE15.BAT
it uses 9054EEP.EXE
source is all in UTILS/DOS/PAS
Should run under FreeDOS
yep. that's the easy way, most likely
(or I can just use the modified hm2_pci for a few days)
oh, I had an evil thought earlier - I have a few 7i31. how do you think those would fare as non-isolated 24 output buffer boards?
(assuming I want to tack solder many many wires to the buffer side of the LEDs)
Might want to pull the LEDS
Or I could send you some bare boards
bare would be not so good
it's easier to remove the LEDs off a few boards I think
Not that many parts...
well, if you can send a parts kit too, that would work ;)
I'm pretty sure the cameras only need a few mA sink to trigger
Yah clip the lead, blow out the solder...
I'd have to check if we have bare boards I know we have assembled ones
ok. that's good to know. I can order 5 tomorrow, with whatever else I need
Parts are just hex inverters or 14s
(in case I break some)
and 50 pin header
huh. the LEDs may be over the load rating of the HCT14
it looks like 330 ohm resistors, and the LED voltage is ~2.1, so the current is 2.9V/330 ohm = ~8.8 mA. the chips are rated for 4 or 6 mA (10 or 15 LSTTL loads @ 0.4mA each)
no worries... leds dont worry about valid ttl thresholds
sub the 74act series.... big honkin' bus drivers
well off to bed for this boy - cheers
EMC: 03cmorley 07TRUNK * 10emc2/src/emc/usr_intf/stepconf/stepconf.py:
EMC: Fix so stepconf copies whatever type ladder program the radio buttons specify
EMC: Regardless if you edited a program or not. Add to warning dialog that custom.clp
EMC: will be backed up. Add new dialog if a ladder program was edited but not choosen
EMC: to copy to the config file. Some minor changes to clear things up
EMC: 03cmorley 07TRUNK * 10emc2/src/emc/usr_intf/stepconf/stepconf.glade: minor name change from 'new ladder program' to 'blank ladder program'
[Global Notice] Hi all! I'm about to do a tiny bit of rehubbing to try sort out some desync issues with one of our client servers, affected users ~2500. It will be somewhat bumpy but should be quick! Apologies for the inconvenience and have a nice day.
hmm.. do you also get the impression I talk a lot?
I just counted > 700 emails on the mailing lists
over how long a period?
all of it
I assume using your mailbox search you can find out more exactly/faster
oh sort of
I tried to count for myself but there were lots of duplicates (because I have both sent and received mail in the archive)
ah, I see
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/drivers/motenc.lyx: add file
for you I have 1599 messages since oct 13 2005
I have roughly 15*58,38 => 900
that might be counting commit messages too
there don't seem to be many duplicates but it does count commit messages
taking a guess at how many I have without duplicates, I think about 1000 in the same period
I counted roughly 900
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ (Master_Integrator.lyx Submakefile docs.xml index.tmpl): move motenc to own file and add a bit of info to faq
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/common/ (Getting_EMC.lyx Linux_FAQ.lyx): move motenc to own file and add a bit of info to faq
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/hal/drivers.lyx: move motenc to own file and add a bit of info to faq
Hey jepler would it be okay if I forward your post re the LA linux expo to Sherline?
rayh: I think the invitation is valid, but not under the same terms (the original invite was for non-commercial participants)
Okay. I'll say nothing to Sherline.
are they in the area around CA ?
hmm.. I see they are
maybe they are interested in going as a commercial participant
rayh: best would be to ask Gareth first
Okay. I'll do that.
did you talk recently to sherline? how do they feel about changing to emc2?
I worked up a test machine for them but they decided not to change just yet.
There is some new hardware in their development path that my ease that transition.
there were a couple of emc1 users asking for help on the lists lately
they all converted to emc2
The Debian they are using is getting very dated.
is it sarge?
It was but I don't know what PC has done with it of late.
the last version I know of is 2006
That sounds about right.
I've not tried to run the BDI since I switched to Ubuntu.
ok, too bad that sherline doesn't want to go an easier way, but it's their choice ..
rayh: I never heard any more about the lathe project. are they still selling a cnc lathe with emc1?
I don't think they are.
oh, ok. maybe the questions we've had are from people who set it up on their own.
cradek_ is now known as cradek
I'd guess that is what is happening.
I'm not very close to their work these days but let me see if I can develop a little more interest there in EMC2.
one of their users contributed some emc2 configurations for us to put in the distribution. that will make it much easier on sherline customers.
cradek: did you see the link from mach.. Looks like your work on comp is above and beyond.
[15:04:13] <skunkworks_> http://www.machsupport.com/forum/index.php/topic,9915.0.html
I think we will have limitations #3 and #6 but none of the rest
#4 seems daft
but I don't want to read 5 pages of that :-/
not #2? What's G30 in mach?
cradek: the rest is not interesting
jepler: might work without comping
hahah "1. 2. 3. 4. 5. 6." then "If you follow these 5 rules, you should be all set"
jepler: no, as long as the cutter can follow the path, #2 is not a requirement
I think g20/g21 will be fine in my implementation, but I will not allow g30
(no idea if their g30 is the same as ours)
g30 seems a bit silly
[15:12:52] <cradek> http://timeguy.com/cradek-files/emc/trouble-case-fixed.png
I could send him some of my test files :-)
cradek: send him one of your fixes (GPL of course :)
although I can't possibly imagine who would want to read the sourcecode
well I know I feel that way about it
is that Chinese cradek ?
BigJohnT: possibly, but I don't know
it is a test path that has some possible trouble cases to test my algorithm
I was just kidding
it's a subroutine called three times, with no compensation, then left and right compensation
cradek: you should use some chinese + ttt to generate testcases
alex_joni: not so easy - I would need to add entry moves...
I imagine a short text (couple lines) would cover most any line/circle issues
I assume you happily accept ttt patches that add that...
but it's much less picky about entry moves now too - for instance in that picture you can see the "left" entry move (starting at the origin) is concave
ttt needs full pocketing, not just offsetting
shouldn't a couple of offsets be equal to a pocket?
it's too bad cgal isn't gpl-compatible :(
alex_joni: not when the path falls apart
jepler: iirc, you started working on this
yeah I got to about this point then decided my approach sucked and would never work: http://emergent.unpy.net/files/sandbox/offs-one-more-bug-fixed.png
(the roughly trapezoidal area at the top center should be all filled, but offs "forgot" about it
hm, if wishing would make that work, it would be working
[15:28:42] <BigJohnT> http://imagebin.ca/view/1OYo5p8.html
BigJohnT: what's that from?
this is how one popluar software does it in zigzag
you can pick one of 8 methods to cut out a pocket
this is constant spiral http://imagebin.ca/view/Ph4ygPE.html
BigJohnT: can you load in a gcode file and have it pocket?
yes, I think so
it is more for creating the g code
[15:36:12] <cradek> http://timeguy.com/cradek-files/emc/t3-test.ngc
I think I took out all emc-specific stuff
I used a .15 diameter tool to generate my paths
I'm just curious whether it can pocket something complex
nope it must have been bobcrap that could do that looks like mc needs the cad file
thanks anyway, just a curiousity
however I can run it on my Anilam for sure
EMC: 03cradek 07concave_comp2 * 10emc2/src/emc/rs274ngc/interp_convert.cc: prettyfication only, no functional change
my Anilam ran it but complained about adjusting circles beyond maximum limit what ever that means
with cutter comp?
BigJohnT: got an URL for mc?
alex_joni: not off hand
cradek: let me look at the generated conversational file to see what it made
BigJohnT: don't go to any trouble...
mc = mastercam? http://www.mastercam.com/
hm, even it "goes crazy" sometimes: http://www.cnczone.com/forums/showthread.php?t=52802
"I have about 20 islands with many that need 1/16 endmill to machine small
pockets .. I tried to rough with a 3/8 first but now the tool paths are crazy when I go back with a 1/16 ..
yes mc == mastercam
cradek: this is the Anilam conversion http://pastebin.ca/1302797
It did not pick up on any tool comp
sorry, I took out comp, so you could pocket it with the other program
jepler: when I do that I pocket with the roughing then profile with the small end mill
if I had a dxf of that pocket I could run it mc
BigJohnT: ah, ok then
let me look - maybe I have a dxf of it.
[16:07:02] <cradek> http://timeguy.com/cradek-files/emc/t4-test.dxf
I *think* this is the same path
looks like it
I might have to edit it a bit as it is not one single chain...
it is - it's a polyline
I see that in acad but it chokes in mc :/
I just had to explode it :)
some software doesn't like polylines
heh - that opened right up in the browser here. (a really old version of acad view is installed)
ok now I'm getting somewhere
this is an outside profile http://imagebin.ca/view/N5FodCA.html
inside and outside http://imagebin.ca/view/NR-4ZPjA.html
neat. can it pocket it?
coming right up
cradek_ is now known as cradek
pockets come in 8 different flavors
the one where it follows around the edge - I think it was the last one you posted before
[16:23:23] <BigJohnT> http://imagebin.ca/view/n-4WAxp.html
that's not a really good algorithm
why is that?
duplicate moves, etc
this is a different one http://imagebin.ca/view/qKRaG_J.html
does it do cycle time estimations?
it is set to rough then make a finish pass so near the edge will be two paths close to each other
can you compare the 2 ?
just a second or three
the last one too 18m 46s
the first one took 21m 49s
* alex_joni loves to be right :)
another one just took 17m 58s
BigJohnT: btw.. using png to save helps really much
I was using windoz paint :/
or even jpeg.. If that isn't an option
paint does jpeg.. If it is xp or newer
ha it even does png
cool, it works
anyone have a wallet I can borrow?
you can have my Dad's old one
it is empty...
afraid that won't work
EMC: 03cradek 07concave_comp2 * 10emc2/src/emc/rs274ngc/interp_convert.cc: handle perpendicular moves in a concave corner
EMC: 03cradek 07concave_comp2 * 10emc2/src/emc/rs274ngc/interp_convert.cc: in the home stretch: support feed mode, spindle mode, spindle speed, spindle cw, spindle ccw, spindle stop, mist/flood on/off
* skunkworks_ cheers from the cheap seats
you got a seat? I'm stuck sitting on the berm!
I'm down to the obscure stuff not being handled
user defined mcodes, override override enable/disables, motion output/input bits
I'm pretty sure you have to forbid reading inputs
(uhh, and lathes and inverse time modes)
oh, G18/G19 yet to do?
nothing hard, just fiddly
going to do G19 or is that silly?
in the, uh, future rewrite, it might make sense to handle all (six) planes and all (nine) tool offsets
but adding that to this project would be a mistake I think
what about XW, surely you'll allow that!
xy, yz, zx, uv, vw, wu
I'm pretty close to calling for testers...
Runtest: 27 tests run, 26 successful, 0 failed + 1 expected
well, the good news is that I can now use my 5i22-1.5
the bad news is that the shoot next Teusday was just cancelled
this is, of course, just after I went out and bought $6400 worth of cables
completely cancelled or postponed?
cancelled, as far as I can tell
so the bad news is that I have to sit on some cables for a little while, but the good news is that I no longer have to do a seat-of-the-pants system in a week
too bad :)
I think this was close enought that we'll actually go forward with the system (finally, after 6 years), so that is good news
we means you and some associates?
excuse me while I take a moment to breathe again ;)
ah, sure.. go right ahead
my customer will finally pay for the development, which I'll do
I hear breathing is quite usefull (from time to time)
* SWPadnos exhales
yes, I had heard that
now I get to play Guitar Hero this Sunday :)
do we have a desire to include a Linux version of the Mesa EEPROM programming utility?
with EMC2, that is
EMC: 03cradek 07concave_comp2 * 10emc2/tests/oword/sub.0/expected: reinstate interp's debug comments, restore comment handling in sai, sync regression tests with trunk, enqueue comments so they don't appear out of order, and then notice that the tests pass!
EMC: 03cradek 07concave_comp2 * 10emc2/tests/interp/flowsnake/expected: reinstate interp's debug comments, restore comment handling in sai, sync regression tests with trunk, enqueue comments so they don't appear out of order, and then notice that the tests pass!
EMC: 03cradek 07concave_comp2 * 10emc2/tests/interp/cam-nisley/expected: reinstate interp's debug comments, restore comment handling in sai, sync regression tests with trunk, enqueue comments so they don't appear out of order, and then notice that the tests pass!
EMC: 03cradek 07concave_comp2 * 10emc2/tests/ccomp/mill-zchanges/expected: reinstate interp's debug comments, restore comment handling in sai, sync regression tests with trunk, enqueue comments so they don't appear out of order, and then notice that the tests pass!
EMC: 03cradek 07concave_comp2 * 10emc2/tests/ccomp/mill-line-arc-entry/expected: reinstate interp's debug comments, restore comment handling in sai, sync regression tests with trunk, enqueue comments so they don't appear out of order, and then notice that the tests pass!
EMC: 03cradek 07concave_comp2 * 10emc2/tests/ccomp/mill-g90g91g92/expected: reinstate interp's debug comments, restore comment handling in sai, sync regression tests with trunk, enqueue comments so they don't appear out of order, and then notice that the tests pass!
EMC: 03cradek 07concave_comp2 * 10emc2/tests/ccomp/lathe-comp/expected: reinstate interp's debug comments, restore comment handling in sai, sync regression tests with trunk, enqueue comments so they don't appear out of order, and then notice that the tests pass!
EMC: 03cradek 07concave_comp2 * 10emc2/src/emc/sai/saicanon.cc: reinstate interp's debug comments, restore comment handling in sai, sync regression tests with trunk, enqueue comments so they don't appear out of order, and then notice that the tests pass!
EMC: 03cradek 07concave_comp2 * 10emc2/src/emc/rs274ngc/interp_convert.cc: reinstate interp's debug comments, restore comment handling in sai, sync regression tests with trunk, enqueue comments so they don't appear out of order, and then notice that the tests pass!
yeow - that's a lotta commit right there!
much less than some others, but in more directories
"updated comments in all files" ... ;)
but this is exciting - trunk's regression tests pass
I notice that none of the tests use inverse time mode :-/
I haven't read the code or commit messages too closely, but something is tickling my "questionable code" detector
it seems like there are a lot of special cases
I don't know why that is, but that's usually a sign that some design somewhere is sub-optimal
can you be more specific?
not really :)
I just noticed notes in the commits like "haldme perpendicular ...", "handle concave A to B..." etc
the only thing that's made me uncomfortable is detecting and handling the corner cases that happen when you have tangency (circle-circle, line-circle, circle-line)
and you mentioning inverse time mode also made me think - at what level do inverse time and normal moves become the same?
the canon level (output of the interp)
in IT mode, the interp sends a new feed for every motion
those 'special' feeds are not yet queued
ah, ok. it's not an issue of the comp handling the modes differntly, it's that IT probably doesn't work as expected because the feed commands won't be in sync (or won't be executed)
yes, right now those feeds will each come out one move too early
N move(s) too early
yes, where N is usually 1
I think what you're seeing is many checkpoints while I'm working on a relatively multifaceted problem
yep, could be. and as I said, I haven't looked at the code or the commits too closely - some things just jump out at me after all these years of debugging
do look at it and give me something more specific if you can
and, in case I've forgotten to mention it, great work - I'm looking forward to being able to use this stuff :)
I hope it's not a big ball of mud that nobody will ever touch again but me
yeah, that would be good
ok, back to getting this 5i22 to work with an opto-22 module rack
did you ever update the eeprom?
speaking of "questionable code detector": http://rafb.net/p/Zwgkqs54.html
I missed the punchline at first
seb_kuzminsky, no, I deferred porting the utility until avfter my massive work crunch this week
ok, the module numbers are reversed, and an optot22 rack works nicely
I wonder if it'll trigger a camera :)
EMC: 03cradek 07concave_comp2 * 10emc2/src/emc/rs274ngc/ (Submakefile interp_convert.cc): move canon queue to its own file, which I should have done first
man. alias is sure useful when using a mesa board
what you dont like my names!?
alias pin hm2_5i22.0.gpio.023.out camera
setp camera 1
setp camera 0
click - aaaahhhh
your fingers will atrophy
so, what do I do if I want to use a 5i22 and a 5i20 in the same machine? how do I specify the firmware for the two cards?
or two 5i22 for that matter
(with different firmware)
the config modparam to hm2_pci is a string array
config="blah blah","blah blah" ...
halcmd loadrt hm2_pci config=firmware=hm2/4i68/SVST4_8.BIT,firmware=hm2/5i22/SVST824S.BIT,firmware=hm2/5i20/SVST8_4.BIT
but it does the trick better than anything else i could think of
ok, with quotes if I need spaces, for instance to set the num_xxx options
no, it isn't
yeah, it's not a simple problem
as we know from long discussions about it :)
strange. inverse time mode will never command a rate less than 0.1 units/minute
rate = MAX(0.1, (length * block->f_number));
it makes sense to insure that it doesn't round to 0, but 0.1 seems like a high threshold
BOSS's minimum speed is 0.1 inches/minute. that bit me recently.
well 0.1 units, no matter what the units, is a silly kind of threshold
on BOSS if you set F0.1 and turn the feed override below 100%, it stops
not at all slow enough for grinding etc.
not that bridgeports are usually used for grinding ...
you could probably convert the boss and sell the control in working condition :)
skunkworks_: heh, good idea
I am sure more than one person is still using the old control..
skunkworks_: it's hobbled together a bit - I would have to take phots from just the right angles.
oh certainly. it's quite adequate.
just leave the cabinet doors closed
SWPadnos: one doesn't close anymore :-)
"action shots" are the key here
EMC: 03cradek 07concave_comp2 * 10emc2/src/emc/rs274ngc/interp_inverse.cc: inverse time mode queues up feeds here - make them work too
[21:57:52] <skunkworks_> http://desc.shop.ebay.com/items/?_nkw=bridgeport+control+boss&_sacat=0&LH_TitleDesc=1&_fromfsb=&_trksid=m270.l1313&_odkw=bridgeport+control&_osacat=0
[22:00:13] <cradek> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=7518726408
what's funny is this really is the "EMC" board
what does that stand for?
I don't know
mumble motion control?
EMC: 03cmorley 07TRUNK * 10emc2/src/emc/usr_intf/stepconf/ (stepconf.py stepconf.glade): start of adding prebuilt pycvp samples to the advanced page
cool, inverse time even works
You are on a roll !!
well, I don't see what EMC means in that context, but this could be useful to you cradek: http://www.cnczone.com/forums/archive/index.php/t-34347.html
sure - looked useful
EMC: 03cmorley 07TRUNK * 10emc2/configs/common/configurable_options/pyvcp/ (thc_vcp.xml xyzjog.xml spindle.xml): add sample pyvcp files
about every tenth boot, mine loses its mind (offsets, tool table, and program)
it would be nice if ti didn't do that
(I replaced all the RAM and checked the batteries)
come on... You know the solution... ;)
ok. I think I saw a battery reference somewhere
hmm novocaine is wearing off...
mine always reports "AXIS LIMIT" on boot, which doesn't mean AXIS LIMIT, it's just that AXIS LIMIT is also detected by that board, and it only knows how to signal one error, and that's the most common problem
EMC: 03cradek 07concave_comp2 * 10emc2/src/emc/rs274ngc/interp_convert.cc: new comp also on ZX plane / lathe
I am so so tempted to put this on TRUNK now
if it appears to work, and it can be reverted, go for it :)
(both true, from what I can tell
I will play with it some more first. But soon I will need help to find the problems - one only tests things one thought to code
EMC: 03cradek 07TRUNK * 10emc2/src/emc/iotask/ioControl.cc: I don't remember writing this doc clarification but it appears right
ANY LAST WORDS!?
I think my IDC modules don't work
EMC: 03cradek 07TRUNK * 10emc2/src/emc/nml_intf/ (canon.hh interpl.cc interpl.hh): merge concave_comp2 branch: new cutter compensation algorithm that handles inside corners
EMC: 03cradek 07TRUNK * 10emc2/src/emc/canterp/canterp.cc: merge concave_comp2 branch: new cutter compensation algorithm that handles inside corners
EMC: 03cradek 07TRUNK * 10emc2/src/emc/rs274ngc/ (13 files): merge concave_comp2 branch: new cutter compensation algorithm that handles inside corners
EMC: 03cradek 07TRUNK * 10emc2/src/emc/sai/saicanon.cc: merge concave_comp2 branch: new cutter compensation algorithm that handles inside corners
EMC: 03cradek 07TRUNK * 10emc2/src/emc/task/emccanon.cc: merge concave_comp2 branch: new cutter compensation algorithm that handles inside corners
and there was (will be) much rejoicing
Runtest: 27 tests run, 26 successful, 0 failed + 1 expected
and good news - trunk's tests pass (I did not change them)
I think I stand a very good chance of not breaking any old programs
can the new mode be enabled/disabled?
(not that anyone in their right mind would be likely to want the old behavior, given a choice)