I've just sepnt a week in geneva,
I am gettin into that crap again too...too much...can't even get a good night's slepp haha
dang ... thats bad
anyway, being in geneva reminded me why I want to build machines ...
so I can accumulate enough cash not to have to actually be there building machines :)
I have not visited geneva in 5 years
geneva is *nice* ...
I go maybe 5 or 6 times a year
I always had to catch a train north though
2 10 day trips, a handful of 2 day ones
Our Gema division is in St Gallen
a bit north
right. Ive heard of StGallen ..
on the german border
nice little town
I spent most lunchimes sitting on the balcony, warm sun, shirt sleeves rolled up, watching the snow on the Jura
yup I remember snow up high even in warm weather
this was failry low I guess, perhaps down to 1000m
GEMA is our powder coating division
oh right, I remember.
picking up a powder oven tomorow as it happens
in fact ransburg gema invented powder coating
Ive got tired of sending stuff out to get painted
I'll try it in-house for a few weeks ..
powder coating is good....and I don't have the stuff to do it here
just as long as I can arrange to not breathe that crap in!
big convection oven will work
this is a 30kw infra red
Just testing today's sf and when I shut down get error messages.
no it is so sticky and charged that it does not float around in the air much
The message reads in part...
Removing module freqmod
generic.ini was not changed.
generic.var was not changed.
sed: -e expression #1, char 7: strings for y command are different lengths
sed: -e expression #1, char 7: strings for y command are different lengths
Got a clue?
les: going to build a wood and polythen tent, big box with a perf metal screen and an old cotton sheet as a back wall
paul_c: What's happening?
les: blower and a filter stack and bag to keep it sucked away ...
les: its not nice stuff to store in your lungs !!
we always used good face masks
positive pressure masks?
no, just filters
particle size is large
electret filters work very well
get a bit in your eyes and its unpleasant, like long-lasting grit
* robin_sz scribbles down 'electrostatic filter'
electret disposable filter masks
we have electret cleaning cloths on sale everywher here
I just pull them apart a bit and make filters for computers etc
what exactly is 'electret' about them?
permanently polarized polymer fibers
thats a 3m thing?
prob PP or PE varients
might be 3M
best order a decen tmask then I guess
I have 2 concerns
solidified under a huge electric field
1) breathing the crap in
2) factory covered in a film of the crap
all the boxes have warnings like ...
'causes heritable genetic defects'
I assume that means at a cellular level ie cancer, not fathering children with 33 heads
#2 happens pretty bad...as far as #1 amine curing agents and blocked polyisocyanates are not so good to suck in for sure
so vent to oven outdoors then :)
but on the good side the stuff really does not float aroud much
* robin_sz adds 'forget to check with enviromental health on this one' to his to-do list
more like beach sand
sticky light beach sand
like you want a factory full of beach sand ...
I already have 2 tonnes of the stuff in boxes
it gets all over the floor for sure...
could poder coat a city with that anount!
At Gema we had lots of mops
yeah ... but its not soluble, best you can do is move it about
I going to try the wood and poly tent, keep it tamed if I can
it will stick to the surface charge in poly very well
sticks to grounded metals too
that was the problem I worked on...BMW wanted a =?- 10 micron coating thickness tol
like +- 10 over a panel
(had caps lock on)
or even in the corners?
yeah...even in inside and outside corners
did you manage it?
or did they go water based?
that's why I buy their cars
not water on the continent at least when I was there
but conventioanl base?
I developed just the metering system
right, the sir/powder mix bit?
using "acoustic refractive index"
uh oh...but can't talk about it too much haha
to monitor the mix?
All I can say is that I got the idea at my shop in Chicago near the golf club's skeet range
in the summer it was "pop pop pop"
* robin_sz can probably fill in the gaps
during a snow shower it was "thock thock thock" even though the mass percentage of snow in air was tiny
same effect as foggy days ...
So I took some laplace transforms and stuff...
sure, ... 2 days later you had the maths all wrong, so you just got some piezo sensors and tried it anyway?
haha...YOU ARE TOO SMART...i AIN'T SAYIN A WORD ABOUT PIEZO STUFF
so ... if you had been real smart ..
youd have booked out 2 weeks on the golf range and a few weeks of skeet as "R&D"
heh...oh i did...I built there system in my garage
hah. so you know exactly how much mess it makes then :)
yeah pretty much
They cut me a lot of slack I guess
i think these guns im getting have some basic mix control based on current flow into the particles
I forget what brand they are ...
yeah tribo stuff
box on a vibro stand, lower a feed arm into it, spray.
it's a fluidized bed basically
basically, yes. the air feed to fluidize it is just on the end of the pickup arm
'locally fluidized' would be more accurate
tribo is unreliable to determine mass flow rate...depends to much on particle size and aspect ratio distribution
yeah, but for manual application ...
the human has a lot of impact on the coat thickness
yeah might be ok...or just watching while you spray
watching the part, and thinking .. 'done a lot on that bit already' seems to work
bmw was unusually nasty and tough as usual
experience of manual cellulaose application puts you well ahead of the game
I can imagine ...
+- 10 is not much at all!!
how they measure the coating, laser and a reflection?
it's +/- 5% or so..really tough for powder in corners etc
inside corners are the hard bit
I can imagine the curvilinear sqaures right now
only servo mass control could do it
that's where I came in I guess
and a robot
yup we have plenty of those
faraday cages and bmw specs=work
how was the application startegy developed? trial and error, or hard sums and 3d models?
haha can't say
lots of it
right .. hard sums then :)
wrote some serious code
at least when youve done it that way, you can stick another panel in and get it right first time
but hey...we can mess with e field vectors right?
well, now theres a thing .. I guess you can launch the particles in just the right place then you kow where their flight will take them ...
anyway as you will find...powder tends to be self limiting...try to spray on too much and the charges are neutrilised
right. I had noticed that a bit
so it does not stick past a point
that helps a lot
right. almost idiot proof.
not enough for the likes of BMW...but it does help a bunch
I guess the base and colour coats get a hard time ..
a trip through a powder oven cant be much fun for them
well mica and metal flake don't work well
Ive read theres people doing crazy things with infra red these days too
like on MDF
powder coated MDF is widely used
we have tried poeder on it
they are talking 30S total cure time ...
of course my information is 4 or 5 years old
yes with a quick IR blast I could see that
apparently LARGE anounts of just right lambda IR can do the same as 20 mins at 210 ...
hit the methyl groups and such
still suprised its so quick though
theyre using IR pre-heat on conventional ovens now
ha...and I am using old uralkyd varnishes
20 seconds of IR melts the paint, once molten its much better at conductiong ambient heat into the base metal
so they IR flash it as it goes in, get 2~5 times speed up
and some of it reacts strongly exothermic
right, thats news.
spekaing of which ...
my mate was playing with soem BIG xenon lamps last week
in a laser cavity
for a laugh, took the lid off, stood a big fat electrolytic on the edge of the cavity (like 100mm from the lamp)
and fired it.
the xenon flash totlaly removed the plastic on one side
like .. gone.
can he still see?
he had the laser googles on, eyes shut and well out of the direct light
hes not that daft:)
amazing the amoutn of energy in a big xenon pulse
this has like a 6U 19" rack full of elelctrolytics ...
several 1000 joules at a guess
yip ... or a very instant tan :)
25 W/cm^2 on the retina
with a point source in focus
[00:58:58] <robin_sz> http://students.washington.edu/mdang/Humor/tanlines.jpg
...toast cone and rod cells
* robin_sz nods
assuming the cornea doesnt exlode with the UV
haw funny link
uv does not reach it
thats the outer bit, right?
* robin_sz thought it absorbed UV like a sponge
as far as I can tell...remeber I did that welding helmet for Miller
so doed aqueous humor
AFAIK, the soreness you get from too much sun/weld flash is the cornea burning the outer layer, then sloughing off
it regenrates overnight htough
cornra and crystalline lens get nailed by uv
48 hours and the 'sand in your eye' feeling should go away ...
"sandy eyes" we call it
the cornea is the outer outer layer right? the very front bit?
im pretty sure that takes most of the UV hit
I think the uv causes cataracts there
probably if you do it too often I gues it does
and NEVER contact lenses!!
I hired nasa's space welding group to answer that question but they really didn't
Iguess their lawyers told em not to commit to anyting
One of our government/industry cooperation incentive programs
it was a waste
they werent accustommed to answering 'industrial' questions?
well the lady NASA engineer doen in Huntsville was fun anyway
but thats another story, right?
I wanted data on High Light flux weapons research
did not get a peep
heh .. well quite. "we have no such programme"
I know they have a program
sure ... sure ...
blind the enemy and all that
and they dont use laser dazzle weapons either
no one does
against the geneva convention
several companies sell them ...
its just no one uses them ... riiiighhht.
theyre UV by the way.
I just wanted the info to establish a baselinr for eye damage with auto darkening welding helmets
cause the aqueous humour to flouresce
depending on wavelength
the laser dazzel weapons do use a apecific wavelength to do just that
must be uv?
seems so ... from what i read
you could blast em with IR and do the job permanently I guess
I didn't think we gave much concern for Geneva conventions these days.
well I was just concerned because my mig welding helmet was transparent except during the arc event
I remember ...
did it go on sale?
NO...I gave a presentation and said it was an unacceptable risk
I did not like the idea
right. one out of synch moment and ... zappo.
A VP asked me if it was possible
I said yes...he gave me some money...and I did it
but REALLY dangerousd
it did work
you can see the work with no arc light
you can see glowing metal though
much too dangerous for a product
neat though .. you have one at hoem I bet ;)
funny thing is how much money I got to spend on it
yeah I have one...do not use it
dang .. 1am
go to bed
* jmkasunich lurks
good morning paul_c
03paul_c * 10emc/remove_realtime_base: Rework the sed to try to make it a little more portable across versions
I just got your reply to zwisk's tracker
I must say I agree...
however some statistics could get read from proc
for some HAL components (which are registered) cat /proc/hal/stepgen.0/max_time
or something similar
* alex_joni goes away for a while...
There is a limit to the amount of data that can be normally passed via the proc interface
for x86, this is one page of 4096 bytes.
Hi Ray - Committed a fix...
Can you update from SF and test.
Yes I can. Will do in just a few minutes.
Did the download complete?
I should be able to just move remove_realtime_base into the compile I did yesterday?
* paul_c is back from lunch...
Download finished around 05:00
* alex_joni is back
paul: I use proc mainly for some limited output from it (e.g. cat /proc/meminfo)
so I would like cat /proc/hal/sysinfo or smthg (with some information on a page)
alex_joni: can you access sco.com
and without www. (http://sco.com)
paul_c: You 'da man! remove_realtime_base works as promised.
Hi Alex. Did the candidate you wanted win?
there will be a second tour for president election
Neither did the one I wanted;(
yeah.. I agree on that
did you manage to sort the iptables stuff?
Not yet. I had a sawmill breakdown.
* paul_c is on the phone to ntl about their spamcop listing...
My ISP has been dropping stuff all over the place.
I've read the logs from yesterday... about GEMA ;)
oh...haha we had quite a discussion about powder coating
familiar with GEMA?
we're kinda competitors :P
we sell Wagner here
also from CH
oh haha yes
I worked mostly on car finishes for them
Hi Les. Did you leave your business manager hat on the hook this morning?
well I have to do a digi-key order...and buy some teak...but I just fired up the shop heat
temp 50 dewpoint 40
so I can only go up to 60/dew point 50 without cast iron sweating
the unvented gas is comfortable but dumps a lot of moisture in to the air
so dew point goes up as temperature does
Should really switch to direct vent heat
I had some of the same problems here until I vented the burner.
yup...I want some moisture..but not as much as I get with unvented
want 40-60% rh
I control it with dehumidifiers in the summer
paul_c: The download was from the guy in town. Should have a disk in three or four hours. Gotta blow some snow before I can get there.
That gives me memories
* paul_c wonders what snow is ...
we have hd some at the higher elevations even down here
paul: if my memory serves me right it's something white & cold
how's the BDI going?
emc running on latest kernel I hear
does the latest linux disto you are using handle usb, easy samba setup, and things like that?
USB drivers are compiled, but I've not tested them...
and samba is a black art even with a webadmin interface
samba's cool ;) and easy to set up...
hmm well usb memory sticks would be handy for those big gcode files
* alex_joni sets his samba up ... (it's been 4 years since I started)
I'm almost there
You don't need samba
paul: I agree that on a BDI you don't need samba
samba is only needed if you work with the devil
smbclient should be there...
I need someting other that burning CDR to get the code to the shop
I've run a couple of usb devices here.
smbmount perhaps too
yeah...get the machine on internet and just ftp the file sfrom the office
but I am a little uncomfortable with that
or local LAN
well need samba...office boxes have to be windows
you can smbclient to a windows machine
just set a share on the doze box, and browse it from linux
or run a web server on the office boxen, and use wget
if there was a place I could read up on that I will
les: just say when you need help
just: smbclient \\\\WINMACHINE\\SHARE
it's not urgent...but burning programs to get them out there is a little silly
those usb memory stick drives are a nice modern-day replacement for sneakernet (floppies)
but Les is running BDI-2.12
No USB support.
Might be a reason to change
1GHz boxes are cheap now..
* alex_joni thinks IBM had a point when they said: never touch a running system
I only upgrade if I HAVE to...and that might justify it
well I am in production now so I can't touch it for a while
cradek: firing up the shop today and will test the abort program
jepler was pretty sure he figured out the pattern.
do you know who would be the one to fix it?
it's assigned to jmk but that was a default assignment
well I can't make much of the code
some things I understand
too much global stuff for me
me too. I think I'm missing the big picture. I don't understand the flow of the messages.
but I just want to see if this is an older problem or something that has broken recently
yeah, that would be nice to nkow
although I am using a 6 month old build
Ah! "the big picture" That was the subtitle of my fest presentation.
wish I could have made it
I was at NAMES but had to leave before Fest
I wish that we had a clear statement of all of the EMC1 global variables.
Globals have a distinct advantage as we saw in the recent mars landers.
When the rover started doing geology it created a priority inversion in the real-time code.
Some routines took to long to complete so the system nearly locked up.
By submitting a change to CYCLE_TIME it sorted it self out on the next rt loop.
heh I nominate ray to do a write up on the big picture
or a transcript of your talk
durn...on the phone and can't find any teak
There are some real issues with globals in the EMC@ stuff.
well the big issue with me is I can't read it
granted...I am not a programmer...but I have written a bit of c
That is where a glossary of globals would help.
But I don't know more than a handfull of them.
how about fred p and some sodium pentothal?
Wonder if I could twist Fred's arm enough to get a list.
It would help if the essential variables were seperate from the debug vars.
Chemistry might help;)
debug is a critical problem.
well I run full debug
slows the gui and stuff badly it seems
It does, but when you need it you really need it.
You do need to be carefull with one of the debug options
that is ...running a terminal and the gui gets kinda slow on an old machine
paul_c: Do you think it is possible to separate the debug vars from run vars?
but I do it
Any debug info that has a timestamp from realtime has a flaw
Unless the kernel is compiled with TSC, any get-time calls are not thread safe
most of the time, the call works, but there is a chance that the system could lock up.
rayh: One thing I didn't mention in my last post to emc-dev
was the addition of a /dev entry for EMC
well...I can't ever recall having a lock-up with emc1
* rayh reads through the fest outline.
except segmentqueue testing
Using a /dev/emc node gives us the scope to separate essential shmem realtime data from debug and status
and the proc interface also provides a simple debugging window on the internals.
What would we need in order to track stuff through /dev and /proc.
The /dev/emc node could also provide a means to reconfigure on the fly.
Something like halscope?
Right. I can see that.
/proc/emc/* can be read with any text reader, even cat
proc entries are (generally) human readable
I was thinking about the time critical relationships down there in rtai.
You can easily cat /proc but you get a time slice rather that a look at the process over time.]
There is a ring buffer used by the logging interface (emcplot uses it)
but it hogs huge amounts of memory.
Right. But if properly configured and triggered it would look like a digital recording scope.
Then we could strip all the rt debug stuff.
Space for the ring buffer is allocated at insmod time, which (in my opinion) sucks...
a lot of stuff is decided at insmod time
which in my opinion sucks too
Using a dev entry, it could be allocated dynamically
A pile of bones left over from 09J?
e.g. nr. of axes
A legacy of the NIST testing program..
Some of the data recorded is meaningless.
ok guys... I'm outta here
I'll catch up with the logs tomorrow
See you alex.
With a little bit of work, halscope (or something similar) could be made to work.
Such a interface need not be real time. It just needs to display that data.
Agreed - The only part that would need to be realtime would be the data logging routine.
But we already have that in the EMC code.
Could we do some graphing of that like we do for the backplotter?
Or is that already there in stripchart.
a 3D spatial plot ?
I think I'd be content with laying several vars on a timeline.
[15:40:51] <paul_c> http://www.kitware.com/case/vtkinuse.html
[15:42:07] <paul_c> http://www.kitware.com/images/VTKInUsepvwave75A.jpg
Set it up right and the area under a slice would be the cpu time committed to a specific thread or process.
With toolkits like Vtk, the only limit is your imagination.
Yes it is and the other kit systems complement it by extension.
bleh sent a man out to pick up teak...$13.75/ bd ft
and now off to the shop...enough of this office stuff
"ParaView runs on distributed and shared memory parallel as well as single processor systems"
You'll be wanting to run a Beowolf cluster next ;)
Did you look at the dart stuff. I've see that with laptop and rt someplace.
Yeah, see, what you need is a beowulf cluster of machines with a USB to parallel converter, so that it's *fast enough* to drive a stepper system.
There you go.
but at least you can do it with python.
Or tickle. Those features would be very nice.
I was also looking at the cable stuff because that is intended to be a high level connector
between c c++ and the scripting language.
It does not look like it will make it to the generalized level we need.
methinks Jon E. isn't going to like some of the changes imposed by the 2.6 kernel build system.
Does it break his parport comm?
Not the comms, but it will break any modules that have different variants built from a singe source file.
Ouch. Don't we build a lot of variations upon emcmot.
Yes - But I've already split up emcmot to get round that.
Oh. I see.
This is a part of your cvs rather than sf.
At the moment, yes - Just got to figure out the best way of merging in to SF
You are also using your librcs?
At the moment, yes - It makes life sooo much easier to produce a deb.
I'd support simply creating a new module emc1 that is your repository.
We would essentially freeze emc where it is.
With that we have essentially parallel 1 and 2 modules.
A fork in the emc2 tree may be a better route to go - librcs and the configure script are both the same...
as is much of the make structure
I think I'd find that really confusing.
If it were possible to split out the 1 and 2 specific source while holding the common source in one place...
Could add a couple of lines to the makefile...
and hal_rs232.c ;)
not that easy
some minimal serial thingy
no kernel support
you can do some other thing
but i have a Heindenhain IK220 card here it supports two counter inputs
mybe i write a driver for it and buy another one, but it is a bit expensive
about 800EUR :-)
the counter chips i have are 16EUR and they can also speak EnDat
but they don't speak with my PCI Decoder
maybe i have talk to them
ok i will go
paul: still around?
I was serious about those boards...
checking Farnell's pricing
drop me an email and I'll ship you the cards
gotta check if I have 2-3 at work, but if not it'll take a few days to make them
email sent to robcon
pity you didn't use LS7266 counters
* paul_c has a box of LS7266 chips.
what's the difference?
* alex_joni is checking footprints
24 pin DIL package
also have the core code to read the 7266 in dro.c
well the code is pretty much the same
some bits here and there different ;)
can't seem to find a pdf
[23:50:32] <paul_c> http://www.usdigital.com/data-sheets/pdf.shtml
[23:50:52] <alex_joni> http://www.usdigital.com/data-sheets/ls7266r1%20Data%20Sheet.pdf
[23:50:55] <paul_c> http://www.usdigital.com/data-sheets/LS7266R1
Manufacturer's Data Sheet.pdf
hmm... a little bit different than LS7166
internal is pretty much the same (only double)
let me check what I can do..
we'll talk about this in a few days .. ok?
Can I twist your arm for IO ?
what kind? how many?
16 in 16 out would suffice
ooh, meep meep alex_joni
today I acquired ...
paul: don't think that will fit on a standard PC104
an electrostatic powder coat gun
at least not with optocouplers
robin: those are nice to play with
yeah, or 85kv possibly
paul: is there a problem if the board outline is bigger than standard?
a couple om mm
I'll try to see
or 8 channels?
IO or encoder ?
e.g. 2 LS7266 or 4 ?
four channels are enough for most apps.
I don't need more myself
Just checked my processor card