i can't answer for sure, it's been off during the move. so i just ran update
no, not that error... its the one that rtai error
so i run rtai test and no overruns and tmax is well below my base period
is it this error RTAPI: ERROR: Unexpected realtime delay on task n
from the terminal window run latency-test
and let it run for a while
and what was the max jitter?
man, i spent days setting emc2 up, had it running good... ah jitter? ahhh max latency was like 19000
crap, mach3 reminds me of AOL.COM
BigJohnT_ is now known as BigJohnT
there are two different latency tests
one is latency-test, which runs a HAL-based latency/jitter tester
but the mach demo run dead nutz at 50ipm
the other is the RTAI one in /usr/realtime*/testsuite/kernel/latency
thanks SWPadnos I may drop off line I'm on generator power
which did you run?
hopefully I'll stick around :)
rtai test not hal
ok. I think the numbers are similar, but please run the HAL latency test
can I find that in axis?
you can't run both the latency test and EMC at the same time
so don't start emc while the latency test is running either :)
i no that
haha, i had this system (machine and emc2) running good 9 months ago.
[00:11:08] <BigJohnT> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TweakingSoftwareStepGeneration#Choose_your_BASE_PERIOD
twice2, usually that's a sign that you shouldn't change it ;)
took everything down and moved across country. only thing different
is this Ubuntu 6.06 or 8.04?
is that when I booted I did synaptic update
also, what version of EMC2, 2.0x, 2.1x, or 2.2x?
crap, sorry man, i'm not logged on from this machine
not a problem, just take notes ;)
let me do that when i can, so i don't waste your time
ok, come back any time
well talk to you later I'm shutting down in a bit
toastyde1th is now known as toastydeath
amp is coming togather... :)
I raised the gate resistance until the ringing subsided. I was fighting non symetrical current limiting.. It seemed to be the bybass capasitor I had between vcc and the comparator ground.
running quite a few amps and non gate drive failures. :)
no failures yet..
I really need to get the real supply togather.
cradek: how is the lathe working?
skunkworks: great except the cutoff slide
the solenoid air switch thingy quit working right
I guess, but I can't seem to figure out how it works
it can push either way - solenoid has a line to each end
surpluscenter often has a good assortment of air solenoids
including "double acting" ones
(or at least 2-valve manifolds that can do the same thing)
I bet SWPadnos would let you borrow his... (solenoid not lathe) ;O
all I have is a 7 or 8-port manifold set
(used for an effect in a play)
my lathe doesn't have a cutoff slide
hm, wonder if you have the solenoid for it...
I wouldn't expect to, unless all of them are plumbed for the cutoff
seems work is always replaceing air valves.
there's not a lot of tolerance for wear in something that has to be sealed against gases
I saw the link from Alex for building the RT kernel. Did you try it?
I just got home and started reading back
(snowstorm today, plus errands)
k, I just had a question, I tried to biuld it too.
I got through make menuconfig, but the make-kpkg does not seem to exist.
I was able to do a make all, but did not know how to translate the switches for applying rtai.
I just wanted to know if you had figured that out.
Also if the "core2" selection in menuconfig was the correct selection for smp.
For the Atom 330.
I have no clue with building kernels, especially RTAI ones
step 1 is to get this robot working on one core in time for the contest
step 2 is to try to use a second core to speed it up
2a would be to use the existing experimental smp core that doesn't have a working network (robot doesn't need network to run)
2b would be to build an smp RT kernel
I'll be astonished if I have enough time to finish step 1
k, np. I thought from the log last night that you might be looking into it.
what contest are you running a robot in?
an engineering week thing at work
[02:20:22] <skunkworks> http://imagebin.ca/img/703qoFX.jpg
that is a capacitor of the slobberknocker size class
medium at best ;-)
how about a 9x2 array of 7900uF 450V, charged to 700V
does that work?
does what work?
9 in parallel, 2 in series (with balancing resistors)
overcharging them would not work, loudly
the trick is to talk someone into sitting down on the array like the physicists do
the x2 hadn't sunk in
that would be murder
I think they have more energetic caps than you do
mine could easily kill anyone who sat on them
wouldn't it just burn a couple of holes in your behind?
8000 Joules worth of hole
that's gonna leave a mark
I forget who told me about that, but the first two pages of "sitting on capacitors" doesn't have any physics pranks
I need to remember to discharge my caps, grad students are like 4 year olds, they have to touch everything
skunkworks: neat - when are you taking orders?
100 0.00020 6.50410 6.50402 6.50423
over 100 "home"s, my X axis homing was always within this small range, .00020 inches
how did you measure that?
I wrote a program
it repeatedly homes, and then records motor-pos-cmd
oh you mean within .0002 inches worth of motor steps
I get it
that's impressive for the switch style
this is one we took the lever off and poke the plunger directly?
yes, this is leverless
it's the one with leads that came without a lever
[02:49:53] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/rephome.py
nice. good homing is so very useful
# maxdiff last min max
99 0.00022 -5.14387 -5.14393 -5.14372
for iter in range(100):
that's "Y", the radio-shack switch I broke the lever off
iter += 1
what does this do? skip like in C for?
er, no, it does nothing
duh, must not if you got 100
(it changes 'iter' for the rest of the loop, which isn't very long)
your script is cool. I like emcmodule.
this is interesting -- I just reinstalled the Z switch tonight (unscrewed and rescrewed it to the machine). in the first run of 100, the max diff measured was .0063; in the second run of 100 it was .0031. this must represent the switch moving over time.
can you log the individual readings and see if the mean is moving one way or the other?
first run started at -2.34083 which was entirely outside the second run
jmkasunich: yeah, it prints each reading, let me pastebin
or is the std deviation just tightening up
it could also be any other part of the joint slipping/moving
yes. I jumped to conclusions since I know the switch mounting can slide a little bit and I had just changed that on the axis that showed this behavior
[02:58:06] <jepler> http://pastebin.ca/1321391
[03:01:19] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/homing-plotted.png
give it a couple nice whacks to settle things in their final position
"stress relieving" so-to-speak
here's "Y" axis data -- maybe it's trending too (just 100 samples, not 200 as earlier): http://emergent.unpy.net/index.cgi-files/sandbox/homing-plotted-2.png
jepler: linear regression!
math is hard! </barbie>
jepler: I'll help: it's a line. it slopes down a bit.
I, uh, calculated it for you
jepler: are the switch and motor on opposite ends of the machine?
well, actually, on X and Z it's the same side; on Y it's opposite side
well that throws a perfectly good theor out the window
screw, machine frame, anything
Y shows half-a-thou or so
if the axis is 8" long, that is 1 part in 32000, 31 ppm
aluminum is 22 ppm/deg C
so a degree and a half would do it
I bet plastic has a higher tempco
it's interesting to even be in the realm where one can talk about heating as a possible cause of change in home position
PVC has much higher tempco than al. NEC requires you account for it.
for conduit in wiring
wikipedia says PVC is 52 (vs 22 for Al)
jepler's machine isn't pvc, it is HDPE I think
no data here for that
So I am trying to decide where to mount my mother bd. Here is my machine http://www.machineability.com/Bridgeport_series_II.html
John_F: I think take out what you don't need, then see where there is room that's away from the big magnetics
Well I could just put the pc where the heidenhain was except I am woried about vibration etc
the heidenhane is built lik a tank
yeah, looks like a lot of room in that "pendant"
so my second coice is in the cabinet with all the big transformers
that seems worse to me
I think there would be less vibration
I bet you are right about that
in the cabinet. I could move the transformers then the I/O would be next to the servo amps
then use a long video and keybd cable to the pendant
I don't think it would work in the pendant
too many cables to run back to the servo drive cabinet
I'm envious of your way covers, mine were beaten to death with sticks
The way covers arn not as good as they look
there are lots of places for chips to get in
I could probably fix mine if I could get the wipers
I'm jealous of the saddle as wide as the table is long... that's a hefty machine compared to the series 1.
that part is really cool
the saddle is wider than the table
yes it also weighs more than 2x a series I
I can still move my by myself
is the travel still only 18" in X? looks like the same short table.
I moved this one by myself off the trailer and into my garage
travel is considerably more
30 X 14
table's bigger than it looks then
it's totally different than the series 1
It took nearly a day just to get it off the trailer
does your have a ballscrew on the knee?
36 X 15 table
the knee is just a lead screw
with air assist
my air assist doesn't
is there a scale on the knee or is it completely dumb about the knee motion?
there is dual scale inch and metric
It is kind of cool the way it works
do you mean graduated dial for the human, or glass scale to the computer?
dials for human
ah, got it
I hope to eventually put a scale for the computer on mine. it opens up some possibilities even if it's still moved by hand.
I have a scale for my knee, it's down the list a ways though
Yes that would be cool since the quil is only about 5"
I have a scale already, but unfortunately it's fairly low res (and metric)
win some, lose some (it was $20 or so)
aren't they all metric?
I guess I don't know
metric would bo OK the whole machine is mostly metric
the ball screws are metric pitch
they are, 10 or 20 micron is the usual scale
ok then, guess it's normal
since there are exactly 25400 microns per inch, it works out
there are some strange older scales that were in inches
took a while for my glass scale knowlege to kick in, I used to buy lots of encoders on ebay, it's a sickness
is that a power drawbar on top?
It is nice for tool changes!
nice. not QC30 then.
(looks like the same 2HP head as the series 1 with QC30)
same spindle then, just drilled through
yes I think so
mine is solid so a drawbar can't be used even if you want
does it have a nut at the tool end?
eric_unterhausen: QC yes, NMTB no
yes NMTB I can never get that right
John_F: was that a yes, there is a nut?
sorry no nut I was talking about saying MNBT before
I think the QC stuff will fit on an NMTB machine
Yes I know somone that has the QC and it looks the same
yes it will, if the drawbar matches
I think most QC30 has 1/2-13 thread, but some is metric
I didn't think the QC used a drawbar
it doesn't, but all the holders have the threads anyway
all of mine do anyway. they are not used.
I thought Joe Vickers (sp) used to sell a drawbar for QC30
sure could be that some spindles are hollow - mine's not
he used to proselytize the series II machines
if it was, it would sure be tempting to switch to a power drawbar
yeah, the nut is problematic
can you drill a hole through the spindel
I had my face mill come loose once - now I always use a wrench for anything remotely taxing
John_F: can *I*? doubtful :-)
well maybe I should ask how hard is the material
I don't know...
my spline is worn anyway - I would just try to find a drilled one to replace it with, if I messed with it
they used to show up on ebay occasionally
as did the ballscrew knee drive
but I couldn't see spending 3-400$ on a ballscrew for a $1000 machine
probably false economy
$1000 machine that costs $1000 to move out and replace with another $1000 machine...
the $1000s add up
moving it costs about $500
and then the replacement, another $500
but you get to play with a big forklift
I let my daughter drive the forklift up and down the street
I have a gantry on big casters and a 3 ton hoist
I envy you the space to store such a thing almost as much as I envy the gantry
I envy the space more than the gantry
The best part is my wife made me expand my garage and add the heated room!
se didn't like the smell of machines in the basement
evil plots come to mind ...
time for me to go. .. nice chatting.
comcast and thier dns again ...
anyone want to try this and see if it lives up to the hype? http://www.designnews.com/article/160365-A_Better_Belt_Drive.php?nid=2337&rid=1696394
fenn: I want to dismiss that, but it looks like a good idea
Lerman_______ is now known as Lerman
jepler: what about sane_period2.patch ? will be in TRUNK ?
micges_mill: thanks for the reminder; I thought I already had checked it in
btw shall i add a page on making pci parports work under ubuntu/emc2 to the wiki ?
pjm_: there already is one, you should add to it if you have new information
confusingly, the page is called "netmos", I think
use the search feature
i'll add stuff to that page then
is gets rtapi error on startup... how to run the other test that shows the threads an jitter
ok, it's been a while, i found it
when i run latency-test: insmod: error inserting '/usr/realtime-2.6.24-16-rtai/modules/rtai_hal.ko': -1 File exists?
is EMC running?
if u do an lsmod, look to see if that is loaded
EMC not running, yes there are some realtime modules loaded
so rmmod and retry the latency test
ok, so those m odules do not get unloaded when it crashes
anyway, so max jitter BT is 17739, good no?
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-01-29.txt
yep, the other latency test looks good too, but i get rtapi error on startup
twice2: what is the error?
RTAPI ERROR: Unexpected realtime delay on task 1
twice2, how long did you run the latency test for
and did you run glxgears etc and other apps to work the system hard
30 minutes or so, with glxgears and browser and irc...
it's like as soon as axis opens i get the RTAPI ERROR
dmesg has these ones - threads: Unknown symbol hal_create_thread
yah, that fixes things :/
my uptime for this box is 71 days at the moment
our novell 4.11 server has been up 230 days.. :)
this look correct: rtapi: no version for "nano2count" found: kernel tainted?
twice2 i see "[ 4226.944307] rtapi: no version for "nano2count" found: kernel tainted." in my dmesg also
pjm_, just been reading old CQ-TV mags didnte realise how long ago my 10ghz was 1988
wow that is a while ago!
i was about 15 then!
ok, if i start emc do F1,F2 then i can jog. but when i move the mouse i get RTAPI ERROR
i've been playing PCB milling with a dremel add-on, although i'm not overly convinced that the dremel output shaft is as slop-free as it could be
a lot of people have had trouble with dremel runout
front bearing is rubber mounted iirc
is there a fix apart from replace it?
not that I know of
ok perhaps I can be the first!! I'll have a look at what can be done
I wear dremels out from daily use
and a few kill modes :)
33k rpm and a commutator fell apart
i thought about building a proper spindle and using the dremel motor etc, i'll have to see if its worth the effort
iron filings cutting the towbar of under the car got in the vents and killed the speed controller
u could make a replacement controller
with a variac and a bridge
I got that one replaced under guarantee
i found in my crap a variac with motor control and feedback pot, no idea what it was from or what to do with it, more crap!!!
I need a variac to to test a psu
u could have this one then
chop off the motor and add a control knob
I came half way down last night
i still have this linear slide
Oakley near Bicester
archivist: did you find a mosfet?
a cheap HP toy for spares
skunkworks, not yet
I have to pony up and pay Agilent the £19
an HP badged device dont knoe what it really is
i cant play with http://www.collection.archivist.info/archive/DJCTM/EQ/2009_01_01_a/P3222889_3562A.jpg
till I get one
interesting, what it is ;-)
whats the HP part # on the fet too?
two chan dynamic system analyzer, just the sort of thing for mechanical measurements like resonance etc
bah cant remember and its at home :(
did u see http://users.qld.chariot.net.au/~friends/HP%20Parts-1990.pdf
HP to 'normal' part number cross ref
might be in there
archivist: did you see http://imagebin.ca/img/703qoFX.jpg
I think it is pretty
hehh that is a big cap!
poor pcb under that
we have a ton of them.. Mainly why that is what I am using them. (free) they were out of some large ac drives (parajust)SP
that will get bracketed at some point.
pjm_, a lot more of bench briefs are here http://www.hparchive.com/bench_briefs.htm
ah good site, i need similar for anritsu for a schematic for my SA
pjm_, if you ever need old manuals ask me i have a...few
thats handy to know
thanks ;-) I'm after one for my MS612A Anritsu, I've emailed them and no positive response
I just checked and I only have a dipole manual
[15:08:35] <archivist> http://www.collection.archivist.info/searchv10.php?searchv4page=1&errlev=0&searchstr=anritsu
hmm.. crap ..
my wifi doesn't work on the latest hardy kernel
(the latest I have installed: 2.6.24-23.46)
it's using iwl3945, works just fine on 2.6.24-22.45
did you get l-u-m for the new kernel?
whee.. no I didn't :D
* alex_joni hides
.. in shame even
I see new packages (which it's installing now..)
I still prefer the 2.6.24-16-rtai :)
YEA! the electricity came back on :)
oh - you were part of the ice storm
yep been without juice since tues evening
energy managment you know like Green Acres
heh, when will yank utilities learn to put cables underground.. we dont get those problems
everything from the highway back is underground but a lot of it is real old
the above ground stuff
food in the fridge survived?
I have a small generator that I run to power the coolers and the heater blower
I just run it a couple of hours at a time
jepler: that's fun
I'm not sure whether I'm solving them in the intended way -- sometimes it looks like my design should have toppled, but it doesn't quite
what level are you at?
cool.. I just reached 8 or so
(realized I can rotate the parts too :D)
I'm new to EMC. Could you please give me some help?
hi starter, you don't need to ask permission, just ask your questions
just ask your questions, we will try
I need 2 steppers to work on the same axis
in this case, I need a A to be a slave axis o X
a gantry machine?
and A should be inverted so that the axis moves in the right direction.
any reason to call it A?
you can wire their step/dir drivers together
you do not want to call it A because A is a rotary axis
further, you do not want it to be a separate axis because they need to move together
I meant it solely for the purpose of understandig
i read on emc2 documentation that i could wire it out in hal file
starter: indeed you can
you can wire the two drivers together with actual wire outside the computer, or with virtual wires (a "net") at the HAL level
the actual problem, you will quickly find, is homing the two motors/screws ("joints") so they are synchronized
I had this: net xstep stepgen.0.out => parport.0.pin-03-out parport.0.pin-09-out
that looks perfect
but it's not working...
more details about "not working" please
emc gives an error while opening
and the error is?
I'm just trying to add this line to the standard stepper_mm file that comes with emc2
alex_joni: like this solution (don't look if you don't want a spoiler for that game): http://emergent.unpy.net/files/sandbox/cannot-be-intended-solution.jpg
can i name it another name other than stepgen.0.out?
for example stepgen.0.step?
the error is "EMC terminated with an error... etc etc"
I thought that this was also possible:
[18:12:06] <starter> http://emergent.unpy.net/files/sandbox/cannot-be-intended-solution.jpg
sorry I mean
stepgen.0.out (or whatever) is the name of the pin you need to connect to a parallel prt pin
net xdir stepgen.0.dir => parport.0.pin-02-out parport.0.pin-08-out
you can name the signal anything you want
like net nondescriptive_name stepgen.0.dir parport.0.pin-02-out
so why is this giving an error when opening emc with this on the hal file?
because I actually want to wire port 3 with 9
to figure that out, I think we'd need to see the actual error, and maybe even the HAL file itself
and port 2 to 8
the problem is that I also need to invert the signal on port 8
you should run emc from a terminal and paste the actual error
so that each motor works "inverted
that's no problem, just setp parport.0.pin-08-out-invert 1
ok, so the signal on port 2 will be inverted on port 8
everything works ok if i take this net statements from the hal file
jepler: if you place objects ontop (where the whiter stripe is, you can put more than one object ontop of each other :/)
if i insert them, emc closes
starter: open a terminal
there will be an actual error printed on the terminal
starter: from Applications->Accessories->Terminal
then type 'emc'
to start emc, and you'll get a better error
it's likely that you either (a) mis-spelled something or (b) tried to connect the same pin to two signals
but it's hard to tell without the error
i'm checking the error out
parport 9 already linked
I saw it now O_O
so much easier :)
just to be clear
is it important to reset pins at opening?
I have a line with setp parport.0.pin-09-out-reset 1
that doesn't reset the pin
so can i erase it from the file?
there's a feature of the parallel port driver that allows it to change the state of the pin twice in the same interrupt
it gets set, then after a time it will get reset
if you intend to use pin 9 for stepping, and you want high step rates, then leave it in there
that isn't the cause of the problem you saw
I was just cheking out
the problem is that files created with stepconf utility
always insert an A axis
although I just choose XYZ
I just erased the a axis part on the file
you need to make sure you set the pinout right
and it's ok
using stepconf, you should be able to set more than one pin to X step and X dir
which will do what you want without having to fiddle with the hal files
use the drop-downs on the pin configuration page
so all this configuration can be done in stepconf?
net xdir => parport.0.pin-02-out
net xdir => parport.0.pin-08-out
this is what stepconf generated
starter: probably so
for the xaxis
there's no wiring involved
in this case
what i assume here is that it will send the same ammount of pulses for both pins
the outputs will be identical
which could cause some problems around homing, as cradek said
for homing I'll have to use a plc with a pulse generator
I guess it's the easiest way...
not necessarily - how will you combine or switch betwen the PLC and EMC2 step signals?
I guess the PLC could be in standby untill the job is finished
when it's finished, it can be signaled using the estop out function
I think someone had a suggestion of using an AND (or OR depending on the logic sense) gate with the step signal and the home switch as inputs, and feed that to the motor drive
then, it could handle the steppers by itself and return the machine to homing position
so the first one to hi its switch stops moving while the other catches up
you might even be able to do that in HAL, if you use the proper AND/OR hal component and put it in the base thread
it could be a little bit enhanced
if, instead of switches one uses a distance sensor
there are cheap ones already
the idea is to move the machine at maximum speed and ramp it down when it reaches 20cm from the end
well, you can do that if you choose the home switches and accel paramaters well
then it would hit the home (0) swith
but that may not work well with multiple slaved motors and an outboard AND gate
the concern is to align both sides of the gantry, at least at the end of each job
or the beginning :
well I'll try this configuration ow
I'll come back with news regarding auto-homing
sure, come in and bounce suggestions/experimental results around
Since emc2 also has embedded ladder
incidentally, you could also experiment with gantrykins
i think it would be possible to do this in emc without a plc
maybe you can help me with this in the future
but you'll have to get down & dirty with HAL to do that, it's not part of stepconf
and if not me then someone else should be around
isn't there a way to program ladder in a graphical way?
i mean, for emc?
if you pay airfare, I'll even come help out physically :)
classicladder has a GUI editor and status monitor
i have just installed emc/ubuntu latest iso
and i can't find classiclader
jepler: darn. I'm stuck at 29
it should be in the Machine menu in AXIS
alex_joni, heh, me too :)
yay, made it
i cant find it
it's not in the machine menu
it is in the file menu
alex_joni, I think it was half luck
itg's grayed out
you may need to load a configuration that uses ladder
use one of the sample ones (I don't recall which at the moment)
then I would change it?
ok, I'll try
So, if I want to start a fresh one myself
I always have to start with an already made one, right?
starter: you need to read a LOT fo stuff to start fresh :)
I'll try to stick around with a few lines
have a nice weekend thern
and thanks for your help
you bet.. have fun
SWPadnos: I hate that it's inconsistent
alex_joni: aww darn
alex_joni: I just finished the last one
yay - I have just unlocked inferno
ugh inferno is ugly
but it has much more infernal sound effects
aaargh.. I hate 29
did you guys start like this? http://imagebin.org/37066
mOizt: welcome. please ask your questions in this channel, not in private messages.
no. I used a hypoteneuse on the gray for the lower right block
I figured more surface area might mean more "grip"
jepler - Sent You a mail yesterday about the hexapod kinematics impl.. Did You get it?
I put one hypotenuse on the grey, then built up in blocks until I got to the top
you may have a different mail address than what I found..
yep, that was my method too
mOizt: I generally don't respond to private e-mails about emc -- questions about how to set it up and so on should go in public venues like irc and the mailing lists.
jepler - question was not on basic set-up, but on the possibility to control servo/arm actuated hexapod creation.
still, I'm not sure why you're contacting me privately instead of participating in the emc community.
jepler - simply because of reading genhexkins.c .. with Your name in it :)
YAY.. made it.. ty jepler
mOizt: I have made some trivial changes to genhexkins.c so that it continues to compile as other parts of the emc code change
I don't understand the math or have a hexapod machine
jepler - was it a origninal name in the code then? Who implemented it?
It cant be just the nist code? it seems to be more generalized..
I think that may have started at NIST actually
you can tell by looking at the history at cvs.linuxcnc.org
I think it's as old as the NIST EMC
the history before 2006 can be seen here: http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc/src/emcmot/genhexkins.c?graph=1
jepler - do You know who in the community could be brainy enough to go into the kinetics of this, and maby address my questions?
what are your questions?
if you ask on the list, then the people who might be able to help you will see the questions
also, the answers are available for everyone to see then.
you will probably not get a commitment from someone to help you in private. I hope you stop looking for that, and work in public on the problems you are having instead.
No problem about public, but this is a quite narrow topic, thats why
there are lots of smart people here and on the lists.
its not strange behaviour to look at code and contact author.
true, but often on old projects the author isn't involved anymore. that's certainly the case with most of the original EMC authors.
and with genhexkins
should be stated in the code that jepler is not author, but modified to compile with newer emc by jepler
"Derived from a work by R. Brian Register"
I don't see jepler's name in the file at all
mOizt: you might have an _really_ old file maybe?
where did you get it from?
in 2.2.8 it has $Author$ CVS keyword, and I had touched it last, so it has my name
he changed it now then, my file: (newest archive) says: * $Author: jepler $
ah. those keywords are deceptive. we have now removed them.
anyway, to the point: I would like to play with hexapod using a el-cheapo setup like the following: http://www.x-simulator.de/forum/download/file.php?id=1562&mode=view
they are automatically generated by the revision control system and "Author" means "who touched it last" which is not useful information
mOizt: looks interesting
..but then the genhexkins would need some overhaul..
it's only approximately that the legs change length and their endpoints stay fixed
the nice thing is alot of people could manage/afford building this toy
..and could probably even fit a dremel to it :)
jepler: the rest were easy
wonder if there is a better, but still simple, leg setup
linear will, for a while atleast, be expensive
using rc linkage-rods/servos, You could do this for 30$
hmm.. it looks a bit travel limited
the more travel, the worse the unwanted endpoint motion
its not going to have a huge travel, no..
I'd use much longer levers and keep them nearly horizontal
there's also the following video (I don't think it moves until about 1 minute in)
if you consider horizontal "0" the unwanted endpoint motion varies only by cosine
mOizt: emc can generate pwm signals on the parport, which I believe is how RC servos are generally performed
jepler - servo is ppc
wrong one, here it is: http://www.youtube.com/watch?v=ve1ydnKpJ70
I'm not familiar with ppc
its the pulse-width that is important..
pwm gives variable pulse width with fixed period
could probably use pwm
isn't that like pdm ?
ppm is the name, not ppc
here's what I read about RC servos: "To control the servo, you command it to a certain angle, measured from 0 degrees. You send it a series of pulses. The ON time of the pulse indicates the angle to turn to; 1ms = 0 degrees, 2ms=max degrees(about 120) and anything in between gives a proportional output angle. 1.5ms is generally considered to be the "center"."
[19:43:29] <jepler> http://www.geocities.com/bourbonstreet/3220/servobasics.html
..or I dont know anymore.. when searching now, I find alot of pwm stuff.
The OFF time is varying, as you can see. This has no adverse effects as long as it is between about 10~30ms. The ON time determines the position of the output arm.
so pwm would be fine
slooooow pwm (so you would have plenty of resolution)
anyway, the problem is that emc's software pwm doesn't have very many distinct levels of control
at 30msec period it would have plenty wouldn't it?
jepler - what about usb then.. even tho not rt, could it be used?
and it doesn't really generate RC servo waveforms
if base-period is 40uS then there are only 25 levels of control
.. for widths between 1ms and 2ms
I think the idea is that the re is a pulse of varying width delivered every 20 ms or so
oh I didn't read enough, sorry
here is abit about rc servo (bottom of page) http://en.wikipedia.org/wiki/Servomechanism
same width pulse gives the same degree of rotation
so You have to repeat it to keep servo rotated
seems like you would need some external circuitry to generate this kind of signal
or a custom component
You get 10ch usb interfaces for $15
but I thought we were talking about kinematics...
cradek - we are :)
emc does not support any usb hardware for position control.
is it not possible for a "toy" either?
even tho not rt?
there's no support for sending motion commands to a Linux device
like writing to a file, which is what you'd need for USB
you can try.. but it will probably suck
but with parport it should be possible to control them directly, with no if required.
you'd have to write it in userland, but alex is right - it won't work very well
yes I agree that's possible
you can probably write a userspace component which polls the motion's controller current position
you would have limited resolution like jepler said
and then send the positions through USB /whatnot
but I doubt it will work very well
so a kernel thingie to make usb look like a parport might work?
just don't use usb - you are on the wrong track by thinking about usb
I understand it will cause problems if You have webcam or harddrive on same hub..
only thinking usb because my laptop has it :)
mOizt: you can't control USB from realtime
and even if you went and implemented a driver
alex_joni - I know, but it will be more realtime anyway
you could only change something once every 1 msec
there is no "more realtime"
it's either userspace or it's realtime
userspace means it can pause for half an hour when the planets are aligned strangely
but back to basic, the servo if will be easy anyway: What do You think about implementing the kinematics for this type?
"half an hour" is probably a slight exageration
mOizt: you need to know your math really well :)
but it surely has been done before, and could be done
who would be the person that will accept beer-bribes for implementing this? ;)
to a decent approximation, the existing kins should just work, neglecting the unwanted endpoint motion, which you can minimize by keeping the servo lever around the horizontal point
* mOizt is homebrewer (allgrain)
I'd suggest a visit to your local/closeby math university
alex_joni - hehe-.
cradek - but looking at the defines for the machine-types in genhexkins.h, it seems to be needing fixed end-points for base, and servo-type is not.
a live-cd with a small gui to configure the servo/platform placement parameters (Arm length).. would be really nice.. ;)
mOizt: did you investigate the various hexapod types from genhexkins?
I understand that - do you see what I mean about the approximation above?
alex_joni - I have searched my *#¤& off for couple days now, but not much (any?) info on the different types.
MINI_TETRA, UF_HEXAPOD, Ingersoll, OCTA_TETRA, GEN_TETRA?
yeah, sure. I have searched and searched.
mOizt: ok.. not much more info here..
maybe you can get hold of fenn (in here sometimes)
he investigated hexapods more lengthly
but I think all of these types are linear (from the defines)
just a bit different configs.
[20:04:02] <mOizt> http://www.youtube.com/watch?v=IuzknDc1VPs
good night all
What do You guys think? interesting?
six people talked about it for 45 minutes - you can tell it's interesting
please let us know how it works out - I look forward to seeing your videos etc.
[20:12:27] <jepler> http://www.youtube.com/watch?v=jHjFxJVeCQs
what? you get through 80 levels and all that happens is there isn't another one!
I need help, thats it. Find one guy who are capable of implementing.
jepler - allready made one ;)
nice machine: http://www.youtube.com/watch?v=DGNc1g7MBI8
if it plays that theme music while running, nobody will buy it
this is nice also.. crazy people http://www.youtube.com/watch?v=KW18GGGKJQM
Is this fenn guy here often?
he doesn't talk often
jepler - do You know his email address, could send him some info, and see if he would be interested in looking at it.
it would be much better to mention the project on the emc-users list
why guess who might be interested - just let them decide
But You guys know who has the brains to be able to.. no need to start a big deal before anything is able to work..
mOizt: you could start by stating what part of the project you are willing and able to do yourself, and also where you need help.
I need help on the hard math
I could probably do a gui for configuring the setup and documentation
I also have some information about how to do the math, I'm just not able to process it :)
this seems to be of interest: http://www.newton.dep.anl.gov/askasci/math99/math99075.htm
cradek - You able to program this?
mOizt: sorry, I'm not interested in doing a programming project for you.
guess more than me would benefit
mOizt, are you wondering if anyone here can modify genhexkins to support that RC servo-based stewart platform?
I want a real stewart platform..
got any ships to work on?
* skunkworks_ searches ebay ;)
skunkworks_: shouldn't you get your PUMA working first?
[21:33:59] <skunkworks_> http://cgi.ebay.com/Hexapod-Stories-Book-Edith-M-Patch-1922_W0QQitemZ170151788717QQcmdZViewItemQQptZLH_DefaultDomain_0?_trksid=p3286.m20.l1116
jepler: I should get a lot of things working first. :)
Getting there though.
jepler: btw - raising the gate resistance also made the pluto not hang. (I must have been radiating a lot of noise)
skunkworks_: I recently tried my pluto on a different machine, but it didn't work right
oh - different computer?
(would program, then go back into unprogrammed mode at the first EPP cycle)
skunkworks_: yeah, different computer
Like I say - the few computers I have hooked it into have worked great./
piece of junk. people should buy mesa.
yes - now.
Do you know the printer port hardware that was in the computer?
could it have been netmos?
the one that didn't work? whatever was onboard. the motherboard is via chipset.
skunkworks - yes
was for SWPadnos
my take on it is that it's a very hard problem, with the only benefit that you can make a relatively inexpensive demo platform
there are several things with the math that are hard. one is that the strut endpoints aren't fixed
another is that the servo position has a non-linear relationship to the effective strut length
cradek: was this something you had fixed? http://www.cnczone.com/forums/showpost.php?p=558921&postcount=12
(timing between pyvcp and emc)
SWPadnos - noone said it was simple, but I think doable?
skunkworks_: yes I fixed that back in August. it will be in 2.3.
mOizt, it's probably possible
fix problem that required you hold down pyvcp buttons for them to work
is it true that vcp can't show an s32? that seems unlikely to me
I thought I'd seen it display a tool number.
cradek: thanks :)
SWPadnos - so a challenge to a knowledged scientist ;)
mOizt, I think the problem with it (for me anyway) is that it isn't interesting enough to be fun, and it isn't useful enough for me to do it if it's too much like work
that's aside from whether or not I actually have the mathematical skill to solve the problems :)
I would believe it would be a nice experimental platform
desktop candy :)
it may be possible to help the math problem with mechanics
like using a wheel with a slot in it to push the rods, which might give you a fixed point to work with (you'd constrain the rod by running it through an eye of some sort, and the location of the eye would be an effectively fixed endpoint)
SWPadnos: look here: http://pastebin.com/m6a865280
the simplicity is part of the nice thing about it 6 servos and 6 linkage rods and You should be mol done :)
well, it would be cool to see, but not cool enough for me to spend the time necessary to get it to work
I'd be happy to look at it once you're done though :)
with one of those bldc rc motors it could be a pretty nice tool
Uh oh, swp is on the road again
SWPLinux: SWPadnos The S10 is on sale at PCConnections for $299
SWPadnos: Lenovo IdeaPad S10
oh right, the smallish thinkpad
SWPadnos: Ubuntu does work, just pwr management aint there yet
hibernate,sleep,wake,etc lil funky
yeah. that's been my experience in general, though I've seen reports of Ubuntu (or other Linux) outperforming Windows
there was a recent patch that fixed a long standing bug which caused the hard disk to get accessed every couple of seconds (journal or something)
which was a bad thing for power usage
SWPadnos: With as much as your on the road lately and not needing a paraport, For $299, I think you should check it out.
I'm not on the road actualy, I just have the laptop connected also :_)
you know what I mean =)
I think you would like it, and cheap enough if lost/stolen not to worry about it.
have you heard anything about these? http://www.newegg.com/Product/Product.aspx?Item=N82E16883220004
it's the data that I'd have to worry about
SWPadnos: rdiff-backup and/or rsynv
it's funny - the company that makes my laptop has a bigger/heavier one, and on the configuration page you can add a small notebook computer
No I ahven't I'll chck them it out in a bit.
you for when you just want to check email or something :)
you know ...
well, rsync only works when you have a network connection, and unfortunately thieves don't always wait for you to back up your data before taking your stuff
oh, on another note, the calufornia tax code is a royal pain in the ass, especially for non-residents
or California eve
(gotta turn up the heat, fingers are more frozen than normal)
mmmm - nothing better then new batteries in the volt meter