who do I talk to about puting something in the EMC manual?
it is lacking information about the ~/emc2/configs directory
it took me a long time to figure that out
the _only_ mention of this directory is in the EMC manual was under the 8.7 Python Modules section
then I found clear documentation in http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Configuring_EMC2
here is some information about how to contribute to the documentation: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?BeyondWiki
we're always happy to get documentation contributions
especially right now, as we're preparingto release version 2.1
oh ok, I think I might do that then
that would be great
you're welcome, thanks for offering to fix it instead of just complaining :-)
alex_joni: I'm back
jmkasunich: did you automate the loading of rtai_smi ?
good :) care to try something?
promise I'll help you fix it .. if it breaks
that's so encouraging
ok, you still have the rtai sources in your home dir.. right?
which box do you want me to try on?
along with the newly built rtai_smi.ko
jmk-st: the one with problems
right - this one
ok, leave that there
download this package: http://dsplabs.cs.utt.ro/~juve/tempdebs/rtai-modules-2.6.15-magma_3.3-2+aj4_i386.deb
download as in apt-get install? or just copy the file?
then sudo dpkg -i rtai-modules-*.deb
it's not in the repo yet
I firefoxed it - so there!
on the desktop now
sudo dpkg -i foo.deb
jmkasunich@shoptask:~$ sudo dpkg -i /home/jmkasunich/Desktop/rtai-modules-2.6.15-magma_3.3-2+aj4_i386.deb
(Reading database ... 92875 files and directories currently installed.)
Preparing to replace rtai-modules-2.6.15-magma 3.3-1+aj4 (using .../rtai-modules-2.6.15-magma_3.3-2+aj4_i386.deb) ...
Unpacking replacement rtai-modules-2.6.15-magma ...
Setting up rtai-modules-2.6.15-magma (3.3-2+aj4) ...
ok, now it's installed
do a "modinfo rtai_smi"
do I need to give the path to the module?
jmkasunich@shoptask:~$ modinfo rtai_smi
modinfo: could not find module rtai_smi
just try it (depmod should have done it's thing)
maybe try "sudo depmod"
cradek: it should have worked .. right?
I was running the latency test (with my local copy of the module installed) during the dpkg
I shut it down when I realized what was going on
maybe that's the problem?
how do I un-do the dkpg -i?
did you get any info that there's a file in use or such?
you don't :)
no, I pasted the results here
try modinfo /usr/realtime-2.6.15-magma/modules/rtai_smi.ko
(path from memory, might be off)
modinfo with the path worked
jmkasunich@shoptask:~$ modinfo /usr/realtime-2.6.15-magma/modules/rtai_smi.ko
vermagic: 2.6.15-magma 586TSC gcc-4.0
try sudo insmod /usr/realtime-2.6.15-magma/modules/rtai_smi.ko smiDisableMask=1 smiReset=1
first make sure the old one isn't still loaded
its not - I rmmoded it right after I shut down the latency test
ok, insmod the new one as above
and run the test again
btw (while we wait for 64 seconds)
jmkasunich@shoptask:/$ sudo dd if=/dev/hda1 of=/dev/null
37608102+0 records in
37608102+0 records out
19255348224 bytes (19 GB) copied, 1020.87 seconds, 18.9 MB/s
doesn't 18.9M/sec seem slow for reading the disk
not that slow
what generation of drive?
if it were under 5M I'd be concearned
hmm - that's only ~2-3 years old?
made in Oct 2000
SWPadnos: ata 4 or 5
oh, 6YO - that's a fine transfer rate
well, latency max is 20uS
jmk-st: good, so it works
how is this one different than the one we made?
I added another insmod param
passing the mask as a param instead of #define?
so you can specify those #defines at insmod time
and packaged it :)
any clue why modinfo can't find it?
the problem with it is that you need to pass smiDiasableMask=1 to it
did you try depmod?
yeah, sudo depmod ran with no errors, but modinfo still didn't see the module
it can't find any of the rtai modules
try removing it, and see if modprobe finds it
(just tried rtai_lxrt)
maybe modinfo only looks in /lib/modules/...
this is outside
removing? you mean rmmod rtai_smi?
I'm sure that will work - rmmod doesn't look in disk,it looks at loaded modules
I was curious about the next modprobe
not the rmmod :)
well, rmmod worked
and now I have bad latency
jmkasunich@shoptask:~$ modprobe rtai_smi
FATAL: Module rtai_smi not found.
jmkasunich@shoptask:~$ modprobe rtai_smi.ko
FATAL: Module rtai_smi.ko not found.
yeah ok, seems they don't look outside /lib/modules/..
[00:34:34] <alex_joni> https://mail.comedi.org/pipermail/comedi/2005-March/006925.html
I say don't bother more about it
he has the same problem, but unless I'm being dense I don't see an answer
ok, back to the issue at hand :)
no, no answer there
just ignoring the issue is best
I can do that just fine
we (emc) already use insmod and the full path
so it's a non-issue
another issue though
how are we gonna tell users to use this module?
hmm.. that won't work anymore :/
rtapi.conf way of doing things (at the moment) doesn't allow params to the modules
* jmk-st can't remember where the installed copy of rtapi.conf is
hmm.. me neither :D
somewhere in /etc
I found a bug!
from my installed rtapi.conf:
guess he packaged it :/
but that smells like a bug
same on this box
and I bet on every installation of 2.05
obviously that variable isn't actually used anywhere ;-)
might be used in RIP though
could be, but in RIP, its generated by configure, so it points to the right place
same here I think
(generated by configure)
and it is the path for emc2 and from where configure was run
regarding how to let people use the smi module...
suppose we add a variable "SMI_MASK=foo" to rtapi.conf
and change scripts/realtime, so if it sees that variable defined, it automatically loads the module, passing the variable as the parameter
if the var isn't defined, it does nothing
then all the user needs to do is edit rtapi.conf to add the variable
which can be described to them on the troubleshooting page of the wiki
sounds good enough
but it also means we need to detect rtai_smi.ko from configure
I just had a thought....
scripts/realtime is (maybe) in the wrong place
it shouldn't be a part of the emc2 package, it should be a part of the rtai package
I was thinking "why should emc2 need to figure out all the subtleties of the realtime installation?"
maybe a thing to add to modprobe.d to make it load by default
but if somebody does their own rtai installation, they sure won't cooperate with us by giving us a nice realtime script
jmk-st: does that module load without any rtai modules loaded?
do you mean "can I load the SMI module before I load any other RTAI modules"?
if so, the answer is yes
(thats what happens - I insmod the SMI module, then the latency test insmods the other stuff it needs to run
so loading the SMI module can happen at boot time.. right?
when I exit the latency test, it rmmods its stuff, and leaves the SMI module for me to manually rmmod
seems like you could load it a boot
then I say it should not be in emc2
or even triggered from there
btw, why is the module name (once loaded) "smi_rt" instead of "rtai_smi"?
RTAI sweet ways of fscking up things
'nuff said, forget I asked
although I have nfc why smi_rt is the name
* jmkasunich calls his ISP....
guess it was called that sometimes in the past smi_rt.ko
[01:04:27] <alex_joni> https://mail.rtai.org/pipermail/rtai/2006-September/015815.html
I'm driving a LED with the output of my pluto's "pwm" signal. with pwm-0-value = 0.1 the brightness is visibly different between pwm and pdm. should I suspect I have a bug, or that the LED is also a device with a time to turn on and off?
PWM is brighter than PDM?
how fast is the PDM clock
IOW,how long are the individual "chunks" of on time?
1ns I think
I swap all but the low 2 bytes and the fundamental clock is 40MHz
1GHz? I don't think so ;-)
1us I think
10MHz = 100nS
.1uS I think
I haven't had any spiked eggnog, I swear
.1us = 100ns
damned ISP.... every 30 seconds "please wait"
does the scope say you are getting 100nS on and about 900nS off (with some jitter I imagine, sometimes 800nS off, sometimes more, etc)
at .5uS/div, it's about 2.5 divs between "high"s
and a "high" is around 1/4 or 1/5 div
the high time seems right
100nS = 1/5 of 0.5uS
but 2.5 * 0.5 = 1.25uS = 1250nS, compared to the 900 or so I was expecting
of course, if you're not getting solid triggering (and on that waveform you probably won't) its hard to tell
how many bits is your master counter?
it counts 0 .. 2046
anyway, try sending 1.0/2046 (or 2047) to a PDM channel
wth does a module take it's name from?
so that I can load 2047 in the "compare to" register and get 100% duty
that should get you a single pulse every full cycle of the counter
the period should be 204.7uS (~KHz)
you should be able to trigger cleanly on a single pulse at 5KHz
jmkasunich: the RTAI build system is the problem, that's what gives that sh*tty name of smi_rt
then leave the trigger on that channel, and put the other channel of the scope on the channel you want to mess with
that's why during the 2.2 cycle somebody needs to make emc go directly on adeos, not on rtapi
jepler: Xenomai all the way :D
jepler: them as you increase the command, you should still have a stable 5KHz picture with more and more 100nS pulses appearing
guess I'm really waiting for LXRT to work properly
jepler: I'm all for HAL as a self contained thing over adeos
or whatever that user space RT stuff was called
then EMC is just another HAL application
* alex_joni looks away in distress
eliminate the parts of rtapi that we don't use, and tie rtapi/hal together
but right now, LEDs and scopes are more interesting
you mean not use a RT subsystem at all? go directly over ADEOS?
I'm not sure, jepler said it ;-)
that may be a lot more complex than it seems at first glance
jepler: sounds like a walk in the park, all you need to write is memory management, semaphore handling, scheduling
and a couple of other things I never heard about :D
alex_joni: not all of that
we're not looking for a generic RTOS here
we need shared memory, yes
no, but those things above we need
no semaphores I don't think
"it can't be that hard
I'm cocky after managing "sim" so easily
jepler: "others did this before, so it can't be that hard"
and scheduling - ideally you have a single interrupt source that triggers your fastest thread
how much spiked eggnog was that?
duh, I guess you need a scheduler to deal with slower threads
jmkasunich: I agree we need a simple scheduler
but that's not a trivial thing as it would appear
I worked on a RTOS once
but right now, LEDs and scopes are more interesting
for a simple DSP platform
and it was all but trivial
(so did I, but not for a modern pentium class CPU)
I did one for 8051
and for vintage x86
this one was truly hard RT
when you added a new task, it would analyze it and say if it can do it with the needed constraints or not
all wcet based
I didn't do any analysis
so it either worked as it should, or not at all (from the beginning)
still: right now, LEDs and scopes are more interesting
yeah, bed sounds good to me :D
jepler: what's happening with the scope?
jmkasunich: so I guess the rtai package is ok?
jmkasunich: I'm having trouble making it do what I want
thanks, merry christmas to you too
you mean the one you had me install on jmk-st?
seems that way
jepler: what kind of scope do you have?
jmkasunich: ok, then we might want it in the repo
I set it to 256/2047, or about .1250
the "high"s are around 100nS and the interval between rising edges is about 800nS. 1/8 = .125
IOW it's right
but still dimmer than PWM at 256/2047?
how much wire between the chip pin, and your led? and between the ground pin of the LED, and ground on the board?
it's all surface mount parts on the pluto board -- an inch tops
including the led?
(I thought maybe you had tacked that on)
the original board had the led
one perfectly good theory ruined by a dose of facts :-(
(I was speculating that the lead inductance was preventing the 100nS pulses from reaching full LED current)
could be junction capacitance instead
but that would take at least a sizable chunk of a microhenry, you probably have a few 10s of nanohenries
what color LED?
it's FPGA -> LED -> resistor -> GND
green or blue I could imagine, but red should be OK
what is the series resistor?
if I measure at the low side of the LED there's a negative-going spike at each turn-off
thats the junction C
(or your scope probe ringing ;-)
no there's not that bad a ring at the other side of the LED
it's a small resistor, I'm having trouble reading the marking
can you post a digi pic of the scope screen, showing the low side of the LED?
the marking seems to be 018
if the RC time constant of the LED C and the resistor R is a significant fraction of 100nS, that could have an effect
maybe but I thought I could see the hook on the "1"
roughly how long does the negative spike last?
err - how wide is it? ;)
3*.02uS before it is 90% gone
so ~60 ns?
that's a significant fraction of 100ns
but this happens at the start of the off-time which is 700ns or so
sure, but internally the diode junction needs to charge up before the LED will light
it shouldn't take the same amount of time (I think), but still ...
OK you're not saying that the -ve spike is the reason it's darker, but it's a way to gauge charging the C the other direction when it turns on
but at anything <50%, I think you'll get only single on-times, and multiple off-times
so the LED should probably be ~30% dimmer for anything below 50%, and should get progressively closer to the PWM brightness as you approach 100%
(the 30% is a guesstimate)
sounds similar to what I'm seeing
at 50% there's still a difference, and it get smaller as i go higher
I think LEDs are a little weird anyway, at least in terms of perceived brightness
when doing PWM on high brightness LEDs, I found that 1/256 on time was a significant jump from off, but I could hardly tell the difference as I added 1/256-ths to the duty cycle
I think eyes are log devices
the difference from 0 to 1 is a lot more than from 1 to 2 (on a log scale)
* jmkasunich curses AT&T
I suppose I would be better off calling some other time than 8pm on christmas eve
that might improve your chances of speaking to a human
in your time zone +- 3
I spoke to a couple humans, in india
in your time zone +- 3
none of which could actually do anything anway
anyway... I gotta find something productive to do this evening
hmmm - productive. I guess opening new DVDs wouldn't count, huh?
tore into the christmas presents early eh?
I know - I can look for connectors
I want to be able to disconnect things without unscrewing or unsoldering individual wires
for instance to remove the millhead or table from the machine
or to remove the electrical box from the bench
hmmm - what kinds of connections?
servo/stepper drivers, VFD, ...
steppers, limit switches, jogwheels, etc
IOW, both power and signal
do you have a bin(s) of circular connectors?
yes, but probably not big enough
I probalby have lots of unmatched paris
I wonder if HD power connectors would work
I suppose only if they're protected from swarf/chips
I want something more rugged than that
for signals D-shells would work
trailer connectors are usually inline, not bulkhead
do you plan on panel-mounting one end?
for steppers I could probably parallel pins on d-shells (kluge, but would work)
for the spindle, I need 240V
* jmkasunich digs thru the box of circular connectors
maybe the spindle would be betrer served with an arrangement of contactors
you have two motors, right (lathe and spindle)?
contactors switch power, they don't get it in/out of a box
I'd hard wire the input, and possibly the outputs to the two motors
use a reversing contactor, and parallel the VFD output to both inputs, then run one the FWD contacts to one motor and the REV contacts to the other
I don't want to hardwire anything unless I absolutely have to
I'm definitely gonna have one VFD feeding two contactors, with one motor wired to each contactor
but I still want to be able to inplug the motors from the box
I found two matched pairs of 17 pin MS connectors
you probably know that VFDs really don't like being run with no motor attached ... (hence the reversing contactor scheme, which keeps one motor connected all the time)
no it doesn't
except during a switch ...
reversing contactors have three states, fwd, rev, and off
thats not an issue - I know how to do interlocking
I was thinking of the mechanically liknked pairs
of course :)
even so, the interlock simply prevents fwd and rev being ON at the same time, they can both be off
SWPadnos: in regards to shipping and insurance - sounds good to me.
ok. I'd say that's close enough to a quorum to make it decided ;)
hmm - the spindle motor has the same 17 pin ms connector for the encoder signals
except the motor has a chassis mounted male
I have chassis mount female and cable mount male
an MS connector on a ShopTask - weird
you may be able to swap the innards around
an MS connector on the big-ass AC servomotor that I intend to replace the POS chinese single phase motor with
hmmm - nevermind. not with the chassis mount units
the power connection for the servomotor is also a big MS
with 7 heavy pins - three are the windings, two are a winding thermostat, and two I forget
(got it written down somewhere)
back soon - dog is whining
heh - see you
found some more goodies
two 9 pin cable/connector matched sets
and two 6 pin
the 6 pinners seem like signal or moderate power, the 9's seem pretty hefty
I actually have 5 of the 9 pin cables, but only 2 of the chassis mount mating connectors
wish I had more of those
8 conductors (one colored green) plus a separate ground/shield conductor
all 16 gage
sounds like two steppers to me
got to break them out somewhere though, so each motor can be removed individually
SWPadnos: does your bport have motors, or are you gonna be adding them?
I have servos for it, but I need to figure out the mounting on my own
so it wasn't built as a CNC
except for Z - I snagged an Anilam quill drive mount off eBay
it had apower-feed, and that's it
what do your motors have for connections? just wires? or a little junction box? or plugs of some sort?
they're Baldor servos
do you have the mating halves?
24-7, I think
I got some of them at Fest
there was a guy who had loads of MS connectors - was selling them for $1-2 each, or thereabouts
I remember that
yep - 16-pin, configuration 24-7
he actually had the right ones - lucky!
heh - with backshells even
if you need that type, I have others that aren't the particular type I want to use
no, my motors have 7 and 19 pins
ah - ok
the lathe spindle motor is gonna be interesting
why is that?
[02:43:15] <jmkasunich> http://jmkasunich.dyndns.org/shoptask/lathe-motor-new-back.jpg
if I point the connetors up, they hit the table
if I point them toward the machine, that _might_ work, but would be ugly - a heavy chip area
down means holes in the bench
and out (as in the pic) means right at the back of the electrical box
its tempting to point them back and figure out how to get them directly into the box without a cable
are right-angle backshells too tall to go under the table?
I have about 1" from top of motor fins to bottom of table
lemme turn the motor connectors up and see what that leaves
oh - that probably is too little
the connectors look like they go to ~1/4" below the fins
3/4" from connector to table
facing the machine, but with a shield over them, is probably the best solution
but maybe not, from a chip/oil/fluid perspective
I have this vision of a box/frame that goes around both connectors, and is gasketed to the box at one end and the motor on the other
then just access the connectors from inside the box
it'll have to be a big box - those connectors don't just slide on and off
it's a real pain to get the alignment correct, and then to screw the thing on for the inch or so of fine thread
if inside the box, I don't really need backshells
and maybe not even housing/threads
(just brainstorming right now)
I don't think I'd use those connectors without some threads, especially in a vibration-prone location
you're probably right
heh... txt msgd everyone on my phone a Merry Christmas... getting replies!
another possibility - the back cover plate of the motor - I could move the connectors there
yet another - use the box/frame idea (a tunnel of sorts) from the main electrical box, to the space between that red stripe and the end of the motor
which is a combination wiring space and encoder housing
then just run wires thru the tunnel and into the main box
the control cabinet will stick out from the end of the cabinet, no?
ie, it's not going to be under the bench
[02:53:56] <jmkasunich> http://jmkasunich.dyndns.org/shoptask/lathe-motor-new-back.jpg
(sorry - meant bench when I wrote cabinet)
that pic is just temporarily places, but its mounted now, in just about the same spot
the top mounting ears are gone of course - they would hit the table
that was the motor pic you just pasted
[02:55:17] <jmkasunich> http://jmkasunich.dyndns.org/shoptask/big-box-open-side.jpg
but in the motor pic, the control box would be in front of the motor and bench, right?
the motor pic is taken from the back of the bench, so "front" is ambiguous
like loading a truck "put it in front of the sofa"
top of box is a bit (1/4" maybe) higher than top of motor
exact front/back location of motor is undefined
one approach would leave space between motor and box for cooling, but crap would get in there
ok. I didn't realize the cabinet was going to be above the bench top
thats why my talk of a tunnel didn't make sense
yeah - a 4x4 box on the back of the cabinet would be fine, with appropriate sealing
as you pointed out it might be hard to mate the connectors
but not out of the question
do a 6x6, though you'd need a pretty big cutout from the enclosure (and backplane)
I've been thinking about the backplane
since I don't already have one, I have no investment there
I', thinking of maybe using several smaller ones
it's a space issue more than anything
one for the PC, for example, and one for the geckos plus power supply
are there many studs in the box?
ok, so mounting of backplane(s) is an issue
doing sub-modules makes a lot of sense though
I was thinking of perhaps 1/2" square aluminum bar, an inch or two long, whereever I need panel support
ideally, with only connectors or barrier strips/DIN blocks in between modules
screwed in from the back, with a stud in the front for the panel
heck, maybe just run screws in from the back, then double nut them to make studs
I have some freedom in back, because there is a 3/4" x 3" wide strip at the level of the benchtop
the rest of the box back has 3/4" of air behind it, so screw heads won't hurt
some gasket goop or RTV or something when I screw the blocks in to keep it sealed....
through bolts as studs would require pretty wide aluminum bar ...
1/2" bar would be fine for 1/4" bolts
oh right - you did say 1/2: bar, didn't you? ;)
I think the existing studs are 3/8, but if I'm using smaller panels thats overkill
actually, if I'm going all the way thru with a bolt to make a stud, I'd just use doubled nuts to space off the panel
maybe not so good in a vibration-prone environment ...
that could work
or I could go back to the original idea, with the screws from the back into a bar, and then either screws thru the panel into the bar, or studs in the bar
(panel fasteners offset from bar-to-box fasteners)
for input power: http://www.mcmaster.com/catalog/112/gfx/small/7162kp4s.gif
nice - twist-lock?
unfortunately $40 for a matched pair,panel mount male and cord mount female
that's about the going price for cord-mount anyway (depending on the NEMA rating)
this would be three pole 4 wire (120/240VAC, with neutral and ground)
L16-20 or something
16-20 is good for 480V
ah - going for the exact voltage rating?
yeah - since I'm running off a phase converter, I figured I'd get a rating or two over
if I see a 16-20 in the dumpster, I'll use it
if you are buying, you should buy the right thing
those things are a form of keying to keep you from plugging a 240V load into a 480V receptacle
sure, you don't have any 480V receptacles NOW, but you never know
I got ones that are the same, they may have been L15
the keying was different across the current ratings, but the same across voltages
[03:17:20] <jmkasunich> http://www.mcmaster.com/catalog/112/gfx/large/d1bnema1l.gif
master decoder ring
they're not all the same for different currents
hmmm - maybe they just looked the same - but the pin opposite the keyed pin is a different width across the voltages
we had some neat ones in our old lab
600V 60A four wire
I can imagine
hate to think what they cost, and we abandoned them when we moved
must have been 50 of them in the lab
way too big for this tho - about 4" diameter
probably $2k in connectors (double if it was 50 pair)
it was 50 pair
hmmm - actually, probably closer to $3-4k, or $7-8k for pairs
I bet it was easily $100 per mated pair
the receptacles were mounted in the ceiling, and wired back to the distributuon panels
(we had 480 and 240V receptacles)
werll, I did get L16-20 - the L15 are 250VAC only, and I wanted some margin over that
then there were drop cables to each bench
your machine is 240v though?
yep, but the phase converter isn't perfect
those things have plenty of margin
I have outlets on the phase converter
most industrial (and even residential) wiring devices are insulated to handle spikes to several Kv without arcing over
I think I was measuring 256 or so, not during surge events
yeah, they're certainly big enough ;)
voltage stress is a non-issue, I bet even the 120V ones are safe at 600V
the reason they have all those ratings is for keying purposes only
yeah, I'd be surprised if I could convince one to go through dielectric breakdown
which you''ve neatly defeated ;-)
since I'm rewiring all the machines I get, it's a moot point ;)
just don't sell any
(added electronic motor controls and a washdown motor to the bandsaw, haven't had the time to do the same to the bench grinder ...)
true - I can always put the old plugs back on (you don't think I threw them away, do you?) ;)
didn't know you had old ones
sometimes old and icky - I've ended up replacing the cords as well on all the machines
yeah, I don't like cords in general
for 120V stuff the IEC plug is a great invention
SOOW is so much nicer than the old paper-wrapped oily stuff
when I scrap out stuff I tend to save the IEC power inlet modules
then when I have something with a cord (like a bench supply, or whatever) if I can replace the cord with an IEC thingie I do
and for any project, I use the IEC thing
sure makes it easy to replace the cords
some are really nice - built in RFI filters and such
but less useful for 240V/3ph stuff ...
make it easy to store on the shelf too - I remove the cord
that's where I like the twist-lock plugs and good NEMA-12 cable entries
and/or liquid-tight conduit
ligui-tite doesn't work for cords tho
(where one end is a plug)
I like being able to point the garden hose at the bandsaw now ;)
great for starter box to motor, etc
I can't point a hose at the shoptask ( because of the wooden bench if nothing else), but that level of sealedness is my goal
because of cutting oil and chips
so... lathe spindle motor may go direct into the box
yep. that or move the connectors to the back of the motor
mill spindle motor will need cords that plug at the motor end, may be permanetly connected at the box end
steppers, I might use the 9 conductor plugs, run those cables to a junction box on the machine somewhere, then connect the individual motors there
16AWG should be good enough for 6.1A steppers
why the junction box? do the steppers not have connectors on them?
I don't think so
hmmm. then I'd probably use separate small junction boxes for each motor
[03:34:25] <jmkasunich> http://www.kelinginc.net/SMotorstock.html
item 3, first one
is the motors I'm planning to get
are those the ones that need a special tweak to Geckos to get them to run well
or was that some servo of theirs?
the only gecko tweaks I'm aware of are for nema 42 motors
there was a special version made for some keling motor, I don't remember which one
hmm, thats news to me
not just for branding, but because the standard one didn't work well
ok - it was a servo, not a stepper
from Mariss: "This turns out to be an unexpectedly good servomotor when a G320 is modified to counteract the effects of this motor's pronounced cogging. See the report's scope pictures for details."
was starting to get worried - I already have my gecko drives
I wonder what kind of nice 4 conductor connecters are available for steppers
(the kl34-150-90 motor, in case you're wondering)
nema 34 frame - interesting
the shoptask is built to accept nema 34 frame steppers
yeah, and pretty cheap for a servo
no encoder included
$30 from USDigital
no driver purchased would be the bigger problem
skunkworks style H bridge ;-)
SWPadnos: I saw skunkworks mention shipping and insurance - have you sent out an email?
they are relatively low voltage
not yet - was asking for opinions on inbound insurance, rounding - that sort of thing
I should get that out before I go to bed tonight
round up - don't sweat the small stuff.
yep - rounding up to the nearest dollar. I've gotten rates for outbound shipping (also with insurance)
works for me. A big thanks for the trouble! Hmm - as usual, getting called away.
no problem. have fun :)
interseting family of connectors: http://dkc3.digikey.com/PDF/T071/0318.pdf
6 conductors plus ground, 10A each 250V
mated pair about $15-16
yeah, well, can't have everything
keep them away from the really hot chips
they are nice
maybe put pigtails on the motors, long enough to go all the way to the box, with these connectors at the box end
yeah, or mount a small box with a panel-mount connector on /at each motor
that would be nice
probably tricky though
won't have the motors until after the first of the year, then I can look into it
man MS connectors are pricy
yes, I know
in some cases, one item number comes with both male and female ends though
I don't remember which cases
I just sent you an email. let me know if there's anything missing (I intend to send a similar one to every buyer)
looks pretty complete
* jmkasunich debates whether to add another LED board
I think they have pass-through connectors, so they can even be useful in a machine
a couple things about the email:
and test points, though I'm not positive about that
you address (for people sending a check) and your paypal address (for people doing paypal)(
oh, duh ;)
so you haven't actually placed the order with mesa yet?
I might think about the LED board a little longer
no. they weren't going to ship until the 27th anyway
I may want to make sure it happens on the 28th, so it's past my credit card billing cycle
may as well pay in Feb instead of Jan, if I can
I don't see test points on the LED board
I think I'll just stick with one
most of the stuff that I'm gonna be messing with will be stuff that doesn't LED well
pwm, serial a/d converters, step/dir, encoders, etc
I figured it would be good to have a complete set for experimentation - 3 boards for one card
I guess it's a little easier to grab an IC pin than a connector wire ...
its only $25....
ok - how's that?
go ahead and add one more 7i31 to mine (total 2) when you send out the official email
heh - ok.
(I thought that was the official mail :) )
no, that was the proofreading mail
got the official one
I'll mail a check tomorrow
thanks for all the work
ok. Tuesday would be fine ;)
no problem. I'm glad it seems to be working out
tuesday we'll be out of town
I'll drop it in the mailbox tomorrow, but the mailman won't pick it up till tues - how's that?
that's OK by me
hey, I may have found good (cheap) stepper connectors
[04:39:15] <jmkasunich> http://dkc3.digikey.com/PDF/T071/0319.pdf
bottom of the page
4 pin housings, cable or chassis mount, about $4
contacts $0.18 each
doesn't spell out the current rating, but they have contacts that accept 14ga wire
yeah - I was looking for that
also wants a crimp tool :-(
the crimp tool, as usual, makes up for the connector/contact cost
yeah, but tools you get to keep ;-)
so true ;)
I might order one connector and some pins, and see if they would lend themselves to soldering instead of crimping
or crimp with some cruder tool, then solder for reliability
it looks like the bulkhead connectors have pins (they don't say "contacts sold separately")
they call them headers
intended to mount to boards?
how about pin/socket extractors? (drive-by question)
what about them?
the extractor is cheap - only $21.47
the crimper is not cheap, only $343.44
prototype crimp tool is $41
(or $41.34 for the prototyping version)
missed the extractor. wondered if they were available.
they don't appear to have a backshell, I wonder if they're intended for discrete wire (bundles) only, instead of jacketed cable
another possibility: http://dkc3.digikey.com/PDF/T071/0321.pdf
bottom of the page again
4 pin mated pair $12
I think so - the usual automotive stuff. BTW - the bulkhead connectors are labeled headers - may be PC mount.
jtr: yeah, I noticed that
gotta run. goodnight and Merry Christmas!
have a safe and happy holiday
see ya. enjoy
ok, now its official
heh. thanks. same to you and yours
ok. time for me to go beddy-bye. good night
I didn't completely follow the smi module discussion. are we going to advise users to insert the module at boot time through /etc/modules or the like?
or are we going to put stuff about it in rtapi.conf or .hal files?
I think the first version is better
it can be loaded at boot time (it doesn't require any other RT stuff to be loaded already)
and it needs some hackery to get it done properly by rtapi.conf
that's an odd thing reported by Mario
the 1.1x threshold may be a little too low
you mean BASE_PERIOD ?
no it's done in the servo thread
it keeps track of some history of the elapsed time between calls. if the elapsed this time is > 1.1x anything in the history it gives the error
I'm not sure I know what you're talking about
ahh.. ok, now I see it
you mean the RT error notifying ?
maybe it should be 1.2 or 1.5 or 2 or some other number
ok, that might be one thing
I think that's what he's talking about, although it's hard to be entirely sure
I don't know what that 1ms might be about
I don't even understand what takes 1ms
I wish he'd just give a bug report instead of trying to figure out the cause, because he's on several wrong tracks at once
and the problem is probably his hardware
I think jmkasunich agrees :-)
not my fault - I was playing doom, must have hit the caps lock by accident
doom is the perfect christmas game - peace on earth and good will to men, and all that...
several times I've said to myself, I wonder if he's found a real bug
then the urge to try to replicate it goes away
jmkasunich: that's the spirit
"Whenever I feel ambitious I sit down in a chair and wait for it to pass" /some person I can't recall
bet it wasn't you
oh, definetly not
that seems to be the PHB's raison d'être
Lerneaen_Hydra_ is now known as Lerneaen_Hydra
doom? cool. I played the original wolfenstien on the computer in collage. Then a year or 2 later I came across a game called doom - I thought - hey that is made by the same company that did wolfenstien (id) .. That is the only game I have ever played all the way through. iirc
speaking of the company that made doom (John Carmak)
do you know what he's doing these days?
[21:07:13] <jmkasunich> http://en.wikipedia.org/wiki/Armadillo_Aerospace
ah I think I actually heard about that
there was a artical in popular science/mechanics
they have a website (which seems to be having problems at the moment)
they post updates monthly
including test videos, etc
he flies their rocket with a joystick attached to a laptop
I would think he would have the software prowess to do it.