my machine does something that makes me uncomfortable (pluto_step). my stepper drivers take an active-low signal as amplifier enable. between the time that the pluto_step firmware is successfully loaded and the first execution of pluto-write, the amplifier enable pins are at the startup value, which is LOW. you can easily notice the amps coming on for a second or two during startup if you listen
maybe I should change the power-on value to be HIGH, because that's the unpowered state as well due to weak pull-ups in the FPGA...
that sounds like the right fix to me
on the other hand, right now you can override the pull-up with a stronger pull-down if the safe state on your machine is low, and this would put those people in the same boat I am now
* jepler continues pondering
HIGH is more normal for a disconnected pin, so it's better as a default
unless it's a low inpedance current-sourcing kind of HIGH :)
ideally, there would be (at least) an "enable outputs" bit in a control register, and the pins would be tristated until that's tickled at least once
yeah that'd be nice but I don't have the protocol or the gates for it
(and to even try I'd have to dig out the vmware image with the altera tools installed :-P)
hi,is there anyone who are interested in small linux based emc2?
such as cncuser 's coolcnc
I have an interest. I would like to see it embedded and very small.
do you have it done yet? :)
i found cncuser website failed
Sebastian, you there?
i also just begin this job,and i got ubuntu8.04+emc2 in a cf cad,it is very easy,but it is too heavy for embedded system
so some light linux distribution maybe be better,somebody has done some job on puppy
unfortunately, he just left (the devel channel)
puppy was a nice demo, but I'm not sure it's good for something that you want to keep up to date
if you want to install and then leave it for eternity, it may be fin
SWPadnos: did you see that board I linked you to?
hmmm - no, I think I didn't look at it
and more worse,the early puppy linux distribution cannt be found any more.
how small a flash and memory footprint are you looking for?
woohoo - he's back :)
and buggier than ever
ehj meet seb_kuzminsky, seb_kuzminsky meet ehj
maybe 128M ram ,and 586 cpu is also ok.
hi eric :-)
how big a CF/flash disk?
i looked at the dmesg output you pastbinned, and it ends before hostmot2 is loaded
you two may want to go to #emc-devel, if yer gonna get all technical and stuff :)
i think it is not a big problem,you see,the cf card is very cheap now
sure, sounds good
See you there.
scutsxg, I was thinking the same thing. even an 8G SSD is relatively reasonable
and CF is almost free these days
are you thinking of a PC-type controller, or something like an ARM/PPC/something else
so embedded is a relative term - today's desktop is tomorrow's embedded system
actually, the PC is nearly free these days too: http://www.newegg.com/Product/Product.aspx?Item=N82E16813135091
see above ;)
hahah,but you also have to know how to intergrate a linux system yourself
the way a lot of people use the term "embedded" is "specific-purpose"
I want < 3 second boot
I pointed out to some of the guys at ESC that Google is an embedded system
but it's far from small ;)
database systems are also often considered embedded
the big thing is that the ideal system is one that just manages to meet its performance requirements (and will continue to do so for whatever lifespan is desired)
so embedded is always about fitting as much functionality as possible into as little resources as possible
also, I like discreet components rather than all-in-one boards.
I may be able to get most of what I want with the Linux bios.
I'm a component stereo kind of guy, myself ;)
Linux Bios is cool, but last I saw they didn't support a lot of chipsets
one thing dies - the whole system does not have to be replaced
yes - it should get better though - and one chipset is enough if it is the correct chipset
Linux does that for you in many ways, since more or less all drivers get installed with every system
last I knew, it was mostly server-type chipsets
it's used in cluster computing a lot
I haven't looked for a while either
I do like the idea of serving the image remotely
where can find cncuser,his website also down
I think the european mirror has the puppy ISO on it
ok,could you give me a website?
uh -one sec
i search all the puppy site,it seems that they dont support early puppy distribution
[02:46:31] <SWPadnos> http://dsplabs.utt.ro/%7Ejuve/emc/
[02:46:35] <SWPadnos> http://dsplabs.utt.ro/%7Ejuve/emc/get.php?file=coolcncb05.iso
I should stick that on linuxcnc.org
I'm just going to make sure it's "backed up" - you can download it from dsplabs.ro right now
err - dsplabs.utt.ro
i have a friend who named his son heron, i think it's a pretty cool name
I have some other heron pics in that directory, but this one was taken this evening
went sailing with a friend - the heron was hanging out along the river that we take from dock to lake
anyone in here have a few minutes of emc2 expertise to lend?
Ionnieh: what's the problem?
when I'm jogging my steppers, the movement is very inconsistent
I can't seem to figure out the timings to get everything running smooth
[06:44:21] <micges> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Stepper_Drive_Timing
you have board or use software step generators ?
[06:47:22] <micges> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TweakingSoftwareStepGeneration
look at those two pages and if you didn't find any help you must wait for others to wake up to help you
I think I have exhausted those pages :-( . Thank you for the help, I'll hang around for a while
and this too ?? http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Stepper_Formulas
not yet, I'll give that a read
lonnieh: can you rule out an electrical problem?
yeah, I think that the wiring is OK, not 100% sure about the driver but I get the same results from all 3 axises
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2008-09-12.txt
I think the news was supriesed when the weather service used "face certain death," in warnings of ike.
[12:18:07] <archivist_ub> http://news.bbc.co.uk/1/hi/world/americas/default.stm
seems UK hasnt suppressed it
I think most americans need the 'shock' of it to get thier a$$'s moving.
SWPLinux: I bought 4 of those http://www.newegg.com/Product/Product.aspx?Item=N82E16813135091
for work. these will fill some empty cases we have.
I just hope ecs has solved thier exploding capasitor issues ;)
yeah, they solved it by dumping boards with the bad caps at a real good price :-P
well, time to get going to the office..
see you later
* archivist_emc pokes the doc keepers, http://www.linuxcnc.org/docview/html//gcode_main.html#sub:G17,-G18,-G19:
empty section links
got a nice ballscrew yesterday, new with its test cert
ohh nice ;-)
i made a scalar feed on my mill last night: http://pjm.dyndns.org/feed1.jpg
its for 8.4GHz
each of the rings is 0.5mm in width
epay item number 200248881852
trying to knibble dinner and a customer has turned up!!!
yeah that is a nice leadscrew!
heh cnc make the microwave easier
I must make another 10ghz discone
pjm_: I'm not familiar with a "scalar feed"; is the distance from one ring to the next related to the wavelength?
jepler yep thats it, i scaled common Ku and C band 'chaparral feeds' and made an excel spreadsheet to calculate them for other frequencies
[13:49:53] <alex_joni> http://www.quinstar.com/antenna_scalar_horn_33ghz.html
the EMC2 is really excellent with the on-screen rendering of the gcode, its great to be able to see what the program will do
ah i have a question,,, this might be silly but
last night when i was machining that feed, i dropped a file on the floor and as i went to pick it up my head hit the estop button which is at waist height
i didnt know if it was possible to resume the program
so i started it again from the start
beginning i mean
so is there a simple way to resume after an estop?
you should at least be able to set the enxt run line
it's in the menu..
ok i need to rtfm again
select the line where you want to start from, select run from line from the menu
pjm_: there is "run from line", but it has a fair number of gotchas
probably the most practical way is to edit the g-code
then when you start the program it will run that line
for instance, the spindle won't actually start turning if you skip over the M3 line
best thing is to position yourself at the end of the previous move
and check spindle, mist, lube, whatnot
yeah i thought about editing the program, but i was concerned about the offsets in the circles etc, since cam-bam seems to generate 4 arcs per circle
but yep i'll try editing the program next time
restarting on an arc has its own set of gotchas
yeah this is what i thought... and not being a gcode expert it was simpler to just redo from start
might want to move the estop button
yes that is on my list ;-)
or the head
although i put it there so if i'm getting mangled by the spindle i can 'knee' the button
you need a scream-activated-estop
wireless of course
hah I saw this and thought "dalek!"
fenn: unfortunately, microsoft holds a patent on that..
alex_joni that feed for 33GHz is pretty neat. I have to build a similar one for 25GHz for my 'lunar recon orbiter' receive project
that looks like an advanced machining project..
yeah its pretty neat..
i'm well pleased with the 8.4GHz feed i made... the previous one was made with bits of copper 1mm flat bar bent into circles and soldered down to a backplate
but one milled out of solid is a lot nicer
how precise do these have to be?
Jepler: two comments - the restart has an 'unfair' amount of gotchas - and - no one in his right mind will try to restart in the middle of an arc - :)
stuste1: oh, you haven't met them all yet
if you are using tool diameter compensation the restart is much more complicated also - the restart is a special case and extra steps to restart are warranted
alex_joni: probably not
fenn the positions of the scalar rings have to be within something like 0.1 wavelength
fenn so fairly precise
stuste 1 some people are in their left mind
stuste1, or restart and rotate A axis :)
up till today I have not needed arcs
BigJohnT, , http://www.linuxcnc.org/docview/html//gcode_main.html#sub:G17,-G18,-G19:
empty section links
pjm_, 25ghz I can generate that but not measure :((
BigJohnT, trying to learn arcs today, so might find a few other comments
ok that seems to be something to do with the building of the html's the links work but don't have the numbers in the brackets
heh - here's a great quote "He told the paper he could walk or kayak out if necessary. "It's just water, man.''"
from a guy who plans to stay in Galveston during hurricane Ike :)
sounds like they are going to get hammered
yep. my sister said that the water was already over the seawall this morning
luckily, she was in the car, on her way to Austin
That is good :)
I wonder if her house will survive
yeck - where abouts is it?
a few blocks from the water. near the corner of Ball and 12th (street I think)
to sym umoren i sled malko si lqgam
[19:22:53] <skunkworks> http://www.harborfreight.com/cpi/ctaf/displayitem.taf?Itemnumber=66052
[19:23:12] <skunkworks> http://www.harborfreight.com/cpi/ctaf/displayitem.taf?Itemnumber=66051
IT'S A LIE
I am sure it is ;)
I'm sure it's every yuan it costs to produce
[19:26:30] <SWPadnos> http://www.siegind.com/CNC_MACHINE/KX1_MACH.html
heh - they'll even provide a PCMCIA parallel port card if you don't have one
that machine weighs about the same amount as my rotary table
what is a direct drive stepper listed there? I'm not into steppers :)
dushantch: it means there is no gearing or pulley between the stepper and the leadscrew
and what's a hybrid stepper?
I'm not sure what "hybrid" means in that sense
jepler: probably same as those "multimedia" PC cases :) just adding words for nothing :)
"Modern steppers are of hybrid design, having both permanent magnets and soft iron cores." -- wikipedia
so in this case it just means they're the common type used today, not some older worse type
pffft, like saying some modern AC motors are of hybrid design, having both copper wires and soft iron core :)
btw don't laugh at chinas CNC's just look at all the plastic you bought, those tools make them by milions :)
and those cnc's make those tools :)
my friend tried once to buy some china CNC's but they sell on minimum of 10 units directly from factory :)
sounds like he only needs you and 8 other friends for that to work out nicely
well 10 same lathes :), but he got some nice guaranteed planparallelism machine vices i think 120mm's for 200eu
he said that plane transport and import taxes costed as much as vice :)
[19:48:32] <skunkworks> http://www.cnczone.com/forums/showthread.php?t=64357
good night all
heh that kx1 is £2195.00 over here
hey all. is anyone using a CNC4PC pendant? if so, what's the opinion? good, bad, or ho hum?
Roguish: I think that is the one that Jon Elson had at CNC Workshop. It seemed solid and heavy, but I don't have any experience with real commercial pendants so take that with a grain of salt.
thanks. i am going to replace a Flashcut setup and need a pendant. it's a big gantry style machine. about 30' x 15' x 8'
and i don't want to make something that i can purchase for less.
what's Elson's irc nick?
he isn't here very often .. when he is, he's jelson or elson, I think
ok, thanks. i'll ask via email list.
I think BigJohnT or someone like him got a pendant and created a config for it
[22:17:57] <SWPadnos> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Hooking_Up_A_MPG_Pendant
yeah, already been there.
Anybody know how long a spindle pulse EMC2 needs for syncing?
threading on a lathe is the intended purpose, and constant surface speed machining if that's possible
it depends on your BASE_PERIOD and latency
to be sure the transition is seen, the pulse duration should be BASE_PERIOD + 2*max_latency (I think)
maybe only 1*max_latency
you need to decide how accurate you want the spindle orientation to be, because your spindle speed will be limited by that as well
so fir example, if you want the orientation to be accurate to 1 degree, then the spindle can't be turning faster than 1 degree per (BASE_PERIOD+max_latency)
if your BASE_PERIOD is 25000 and max_latency 10000, then 1 degree / 35 us is the highest spindle speed you can use
that makes sense, but i hadn't thought of it
not that way
my circuit has a rise time of about 35us
that's probably an upper limit too
but that should be constant for each cycle, so it may not matter
or maybe wouldn't affect accuracy as much
what should be constant, and what cycle are you talking about?
are you talking about revolutions of the spindle?
I haven't mentioned slot size either :)
of course, if you want the index accuracy to be 1 degree, then you can't have a slot any larger than 1 degree
actually, I guess it's a minimum - if you have a very fast base_period and very low latency, you can have a slot as big as you want
it shouldn't matter what size slot you have as long as it triggers at the same point each time
but you need a slot big enough to be seen in one base period+latency
there's no such thing as "the same point" - there will always be some variation, which is what the max_latency figure is
for sampling by emc2
yeah i understand that
but as long as the signal gets to emc2 at the same point in spindle rotation, then it would be governed by max_latency right
that's why I thought of the timing aspects first I guess
if the circuit has a consistent rise time of .010 seconds then it really wouldn't matter as long as it was consistent
it could have changed just barely after the last sample was taken, or just before this one
yeah i see what you mean
i'm trying to run latency-test now but it's giving me an error and wanting to see a library
which latency test?
the HAL one that pops up a window?
is this a compiled version of EMC2?
i probably forgot to load that variable list like you told me to before :)
emc-environment is your friend ;)
yep that did it
base thread 10,752ns
for max jitter
latency, that is
max interval 35000 ns
yep - it's set for 25000, so 35000 is around the max jitter plus the requested period
well, set for something close to 25000 anyway
so to get 1 degree of accuracy (assuming constant rotation speed) then i can't have the spindle going any faster than 1 degree per 35u
if you set your BASE_PERIOD to 25000, yes
I think stepconf chooses a number that's on the large end of the range, to reduce CPU load
i can check my config hang on
when you pick the resolutions and speeds for the axes, it calculates the BASE_PERIOD you need
35500 is what it says the "minimum base period" is
thats what I guess it set it as, but there's nowhere to modify it. I just type in the max jitter and it sets that up
ok, so you'll have 35000+10752, which is very close to 46000
so 1 degree per 45us
you can change it later, but you have to edit the ini file and any of those edits will be lost if you re-run stepconf
so that's about 60 rpm
so you get about 23 degrees/sec
err - 23000 maybe ;)
so if i do any threading then i'll need to slow 'er down
that's 60 RPS, not 60 RPM
damn thats fast for threading
my machine tops out at 1818 rpm
(as i just found with my newly constructed beer-can-interrupter-circuit)
in that case, you'll have better than 1 degree accuracy in detecting the index pulse ;)
so i can thread at full speed!
closer to 0.5 degree
EMC can thread at full speed, but who knows about the axis drives or the spindle ;)
i wonder how acceleration works then
of the spindle or the axes?
the z axis
to cut a 1mm pitch thread at 1818rpm it would have to be movin pretty good
i guess about 30mm/sec
well, you'll just have to experiment to see what the bottleneck is
1mm pitch at 30mm/sec limits you to 30 RPM, which may be a bit slow to regulate
how do you figure it limits?
you mean if you try to thread faster than Z can move?
30mm/sec at 1mm pitch should limit it to 30 rps right?
not 30 rpm
or 1800 rpm
oh, mm/sec - no problem :)
that darned min/sec thing again
i puzzled over my oscope for about 15 minutes trying to figure out why it was telling me my lathe was only moving 30rpm when it was supposedly at 1800rpm
checked every time setting, recalibrated it, set the trigger over and over
finally figured out that a second is not the same as a minute, and then it all made a lot more sense
well i got another question for you... can emc2 do constant surface speed machining with only an index pulse?
or is that something i'll have to look at adding
(attempting to add)