anybody know if LINUX_KERNEL_2_2 is ever defined? I don't see anything in emc that would define it, and a google search fails to imply that the kernel or C library would define it...
(in emc2, that is)
i dont see anything in 1 or 2
why isnt there a cvs grep?
in any case, kernel 2.2 is awfully old to still be supporting
it hasnt been changed at all since 2004
I've changed emc2 to use a 32-entry ring buffer when sending commands from userspace to the kernel, and also modified 'task' so that it doesn't sleep when it's putting stuff on the queue. Here's a graph of the queue length seen from kernel space: http://emergent.unpy.net/index.cgi-files/sandbox/qlen2.png
before, the kernel queue could grow by at most 1 per trajectory period, now it can grow by up to 32, and often grows by 5 to 20.
the first samples in the graph are: 0, 17, 45, 73, 103, successive differences of 17, 27, 28, 30
on its own this doesn't do anything to improve contouring, but it means that it would be possible to adopt algorithms in the kernel that is able to retire more than one segment per cycle and not have starvation
I assume that task still sleeps at some point - what now causes a sleep?
when it failed for some reason to queue the command, it sleeps
for instance, because it reached the maximum in-kernel queue size, or because it has to wait for a toolchange or a programmed pause
('top' still shows milltask taking <1% CPU time)
jepler: still around? or sleeping (if you are sleeping, don't bother answering this ;-)
* jmkasunich goes to bed
[Global Notice] Hi all. Please check out the news on http://freenode.net/news.shtml
.... contents: Tor access and cloak formats are changing.
[Global Notice] Thank you for using freenode, and have a great day!
jmkasunich_: I was, in fact, sleeping.
steves_logging is now known as steve_stallings
steve_stallings is now known as steves_logging
SWPadnos: good day. how 'bout a quick opinion of this opto isolator board: http://www.industrologic.com/uob8desc.htm
or anyone else?
personally i'd look for one that had a parallel port connector on it
thanks for the reply fenn, i'm going thru the m5i20 and 7i33 from mesa.
and don't want to risk anything to my less than perfect wiring!
limits, homes, enables, etc, etc.
Roguish: that board looks very overpriced to me
overpriced? got a less expensive alternative?
if you can manage to wire your own (on a protoboard or something) you could get the same functionality for.. uh.. bananas pretty much
if all you want is optocouplers
i'm much better at buying than making. i'm mechanical, not electrical or electronic, which is why i need the 'protection'.
i like OPTO22, but they are $$$$
if you know of anyone else that can do it the optocouplers cost around 70 cents each
[21:36:37] <Lerneaen_Hydra> http://www.fairchildsemi.com/pf/CN/CNY173.html
i realize there's not much in parts there, except the board and the time.
it all depends on how much your time costs
that's a fact.
i have a protoboard with one opto circuit working.
would it be hard to duplicate that?
it's the whole board thing that i don't really want to handle.
i've never made a board or anything close ..........
"made a board" what do you mean by that?
a circuit board
oh, why not leave it on the protoboard forever?
i'm trying to put together a sellable package.
oh! sellable! that's something else. right. now I get what you mean
and with as little work between the stuff you get in and the stuff you send to customers?
hmm. what is this kit intended to do?
depending upon the customer, they can have this $49 item or a $500 OPTO22 setup.
i do custom machines. retro fits, etc.
oh, the entire thing?
from manual machine to cnc?
after the optocouplers, how much do you do?
opto22's are expensive because they incorporate a solid state relay
fenn: so they output high-current output
yeah and i think it can handle AC too
uh SSR, is that mosfet or something?
so you can do PWM with them?
i dont think so
oh, pity. otherwise a servo system would be nearly trivial to wire up
ah wikpedia has failed me
they didn't have SSR's...
Hmm, if it's all electronical then what wouldn't like PWM?
lol "pwm ssr servo" the first hit is the EMC wiki
a SSR is a mosfet with an opto-isolator built in
the opto-isolators are to keep my 24v control away from the mesa boards and the computer.
or, two mosfets i thikn
oh. so pwm should be a non-issue
Roguish: that's a similar setup to what I have
the optos are not for the pwm, but they could be with the right ic.
says it takes 0.5ms to turn on/off which is nowhere near fast enough
fenn: that bad?
here I was thinking it was in the 10µS range
there are some pretty fast optos out there. i've seen mega hz numbers.
good enough for some sorts of speed control but not servos
fast opto's are expensive
i mean the optoisolator part
$4 instead of $.40
uh, are you sure of those times?
according to this datasheet: http://www.ortodoxism.ro/datasheets2/d/0jedoh4ph3rpg669tyx5u9dy7lky.pdf
times are in the 5µs range
and these are 70 cent optocouplers
err, 10µs I mean
SSRs are a faster alternative to electromechanical relays because their switching time is dependent on the time required to power the LED on and off - approximately 1 ms and 0.5 ms respectively.
it shows a diagram of an optoisolator directly driving the gate of a mosfet
so i guess it takes a while to charge up the gate capacitance
some SSRs (AC output ones) use SCRs or Triacs instead of mosfets, and are even slower
nothing that can be fixed with a newer, faster, mosfet though
what do most servo drivers look like?
SSRs are not suitable for PWM unless you are talking about _extremely_ slow PWM like you might use for a heater
servo drivers are usually H bridges
anyway, it anyone knows of a less expensive pre-made optoisolaotor board, please let me know. that at $49 is the lowest i've found so far.
servo drives usually use mosfets unless you get into really high voltages
for 8 i/os
Roguish: have you looked on cnczone?
yeah, a bit
Roguish: also keep in mind that the opto's are probably too slow for quadrature encoder signals
on that board
you will have to do some math to see if they'll work for your application
sure, i'm just going for limits, homes, enables and such, buttons and switches.................
encoder signals are inherently low voltage, so i'm not to concerned about them. they come out of my yaskawa drives.
i thought the reason for all this was in case you get a short or crossed wires somewhere
you can convert 24v to 5v pretty easy with a couple resistors
I find it nice to seperate machine ground from controller ground when at all possible, except for the main grounding cable
avoids nasty problems later on
yeah that too
why "except for the main ground cable" btw?
do you mean safety ground? (the one with pipes going into the dirt)
so signal ground is never connected
but "power" ground is
gotta watch the ground loops
this is all probably more suited to #emc though..
ok, thanks for the assistance.
I'll take this opprotunity to say goodnight
It's late here in borkborkbork land
tanstaafl or whatever you say
[22:09:04] <Lerneaen_Hydra> http://video.google.com/videoplay?docid=3569590578701395201&q=bork+bork+bork
goodnight in swedish? that would be godnatt
hmm, I can't say that I can figure out what that's supposed to mean
what was it supposed to mean?
hmm. I can't seem to connect that to any of the common terms for that...
but i digress, goodnight to you all