03cradek 07HEAD * 10emc2/src/emc/motion/control.c: this may be triggering when there's not a problem. wish there was a better way.
03cradek 07v2_1_branch * 10emc2/src/emc/motion/control.c: this may be triggering when there's not a problem. wish there was a better way.
I sorted that printer, more digging in the service manual (that they only supply to service engineers) ..
I mean, just because you've bought the thing, they seem to think you cant have the info to maintain it
it seems like the only defense for that crap is to find out before you buy.
"recurring revenue" is thought to be the key to a successful business these days, the "razorblade" business model
yet another reason to go open source, or at least people that are trying to do the right thing
same as the people at Comau, whose robot I seem to have
bought the thing, (new machine, but in an auction) it had dumped its memory and the boot discs are corrupt
so, i ring them up.
try and get them to email the duff files
so, this is their first interaction with me.
do they a) bend over backwards to help
or b) try and screw £60 an hour, minimum 8 hours to send a service tech over ...
c) try to figure out the best way to make money off you
they agree I could do it myself if I had the files and it will take only 30 minutes for the tech, but
8 hour minimum visit
well, no .. actually its NOT the best way to make money off me
well, to be fair, they don't owe you anything...
heh. kind of like $40 printer or serial cables at Office Depot. they're looking to make money off those that don't know any better
because now, I am tempted to hack the servos off it, sell them on eBay and make aprofit, putting htr cash towards an ABB
cradek, true, but they could have been helpful for less time/effort than they spent arguing with me
sure, I understand
anyway, its left a bad taste in my mouth
and thats not a good first impression
ah christmas is over
but vacation is not ;-)
vacation? whats that?
not being at work
for me, it's back to work tomorrow
I only went to the office once today :)
[02:28:32] <jepler> http://gcode.blogspot.com/2006/12/pf-tracing-results_116705865715026523.html
wow, you guys really grind
to get a some paper for this poxy printer
jmkasunich: how do you tweak dyndns automatically from linux?
I glanced at some of this guy's code, but it was way over my head. From these graphs, it looks like he is varying the interrupt rate based on the next time a step is required...
seems like that won't easily extend to movement in more than one axis at a time
you mean update it when my IP changes?
jepler: what guy?
cradek: there are multiple clients, I'm using ddclient
cradek: this "Dmitry" guy with his own new CNC program and realtime layer for linux
I believe there is an ubuntu package for ut
does dyndns still have a free service?
yes, thats what I'm using
cradek: what has you messing with dyndys? don't you have a static IP?
not at home
timeguy is at your isp?
timeguy is a sort-of colocation
I think I'll look into this for ingrid's office
she has dynamic ip but often I want to ssh in to do sysadmin stuff
Hey guys. I just finished the ubuntu thing. Wow. I'm impressed. Axis is really pretty neat. I haven't really looked at anything real with halscope, but it seems very nice.
thats initially why I did it - so cradek could ssh into cvs2
I'm not sure how ddclient can work since my inside machine doesn't know the real IP address
lerman: thank you! good to hear it
it goes out to a website to get it
(I'm using NAT, so I have the same situation)
I'm running remotely using x0rfbserver -- couldn't get tightvncserver to show a decent desktop.
ok - it just worked with your nat?
the package I mean?
[02:35:23] <jmkasunich> http://checkip.dyndns.org/
I had to put my info into the ddclientl.conf file
I'll pastebin mine, less the private stuff
[02:37:14] <jmkasunich> http://pastebin.ca/292054
not much to it
jepler: i've been following his (Dmitry) stuff, and it seems all the pieces arent there, (I dont think you could make anything move). he came to this irc once i think, and was using 'iso' as the extent for his gcode files ( if that rings a bell ). it looks impressive but i think pf-gauntlet has nothing to do with his nc work
no, he posted on the mailing list with an announcement of his cnc project
I wish him luck
it's a very ambitious project, since he's writing his own realtime layer as well
03jepler 07HEAD * 10emc2/src/hal/drivers/pluto_servo_firmware/ (pluto_servo.rbf quad.v): re-fix index pulse detection based on talk with swpadnos
03jepler 07HEAD * 10emc2/tests/counter-encoder.0/ (README expected test.hal): add new test
03jepler 07HEAD * 10emc2/tests/abs.0/ (expected test.hal): test of the abs component
03jepler 07HEAD * 10emc2/docs/src/hal/rtcomps.lyx: encoder now implements 'velocity' pin
03jepler 07HEAD * 10emc2/docs/src/hal/pluto_servo.html: make docs match index pulse handling
03jepler 07v2_1_branch * 10emc2/docs/src/hal/rtcomps.lyx: merge rev 1.6: encoder now implements 'velocity pin
03jepler 07HEAD * 10emc2/src/hal/utils/comp.g: improve error handling. introduce 'unsigned' and 'signed' as names for the HAL types, and deprecate 's32' and 'u32'.
ho ho ho
03jepler 07v2_1_branch * 10emc2/tests/abs.0/ (expected test.hal): merge new tests from HEAD
03jepler 07v2_1_branch * 10emc2/tests/counter-encoder.0/ (README expected test.hal): merge new tests from HEAD
wow, jepler is giving us christmas presents
oops looks like the comp.g change was botched :-P
I deprecated s32/u32 but didn't make signed/unsigned work
dropping s32/u32 is only in comp right now?
since there's only one integral type I want users to call them 'signed' and 'unsigned' instead of referring to their width
that makes sense
I'm open to other opinions about whether that's a good idea or not
but the true change is a lot more widespread
(well, two integral types, but not the wealth of widths)
every "newsig foo s32" in every config for example
(fortunately there aren't that many integer signals)
plus halcmd, halmeter, halscope, etc
maybe I'll back that part out for now, and put it off for the great mythical hal reaming
or maybe I'll start accepting 'signed' and 'unsigned' but not print a message if 's32'/'u32' are encountered
accepting both names is a good way to ease into it
this is HEAD so we can decide how far we want to go on this release
and I have no problem with changing comp to accept only the new names in the .comp files
changing the names in hal.h is the step that has wider repercussions
edge.N.out-width-ns signed rw
Time in nanoseconds of the output pulse
* jepler debates about showing "signed" in the manual page
that doesn't help people type the thing that will be accepted by halcmd..
a purely cosmetic but annoying issue will be the output of hal show
today its float, bit, s32, u32
nothing more than 5 chars
(how come out-width-ns is signed??)
(so it shows in decimal in halcmd show?)
I wrote it, so I should be able to answer the question
the hex display may also be something that should change
03jepler 07HEAD * 10emc2/src/hal/utils/comp.g: actually accept signed/unsigned. don't warn about s32/u32 yet
03jepler 07HEAD * 10emc2/src/hal/components/ (edge.comp threadtest.comp): use 'signed' instead of 's32'
03jepler 07HEAD * 10emc2/docs/src/hal/pluto_servo.html: additional rewording of index pulse handling
03jepler 07HEAD * 10emc2/docs/src/hal/ (comp.lyx out8.comp): use and recommend 'signed' and 'unsighed' over 'u32' and 's32'
yay typos in commit messages
03jepler 07v2_1_branch * 10emc2/src/hal/utils/comp.g: merge rev 1.12: signed, unsigned types and improved error reporting
03jepler 07HEAD * 10emc2/docs/src/ (Submakefile index.tmpl): make the index say the version of emc the docs were generated from
03jepler 07v2_1_branch * 10emc2/docs/src/ (Submakefile index.tmpl): merge from HEAD: make the index say the version of emc the docs were generated from
says: This documentation refers to emc2 pre-2.1.0.
it's a slow day so I've made it to #1: http://cia.navi.cx/
events today: 16
but it's waaaaay past my bedtime on a school^Wwork night
hrmmm... are you supposed to put the load before or after the relay controlling it?
ejholmgren: usually after..
though there's nothing that prevents you from controlling earth to your load..
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2006-12-26.txt
that was suspiciously slow
do you think logger_emc is holding out on you?
* Lerneaen_Hydra is starting to think that logger_emc is a small chinese person alex has imprisoned and is feeding only a bowl of rice per day
* aip_tom offers some tofu chunks to logger_emc
no doubt he transcripts it by hand, and hasn't learnt how to pipe the output from IRC directly to a file
g'night (or g'morning, whatever is most appropriate for you)
'morning for me
Happy Christmas for all
i have emc 2.1 and i want creat a external buton for cicle start
IIRC that should be doable in 2.1.x
the devs would know
* Lerneaen_Hydra pokes alex_joni
how install emc in ubuntu 6.1?
breezy-install or dapper-install?
my ubuntu is edgy eft
I don't think there is a version for 6.10
Guest375: we don't have packages for Edgy. Only dapper.
* Lerneaen_Hydra what cradek said
we chose to support Dapper because it is a LTS (Long Term Support) release.
it's definitely possible to use Edgy but we don't plan to do the work to support it - you would have to build the realtime kernel and EMC2 yourself.
I have to go get ready for work, bye
that mync project by that russian guy looked interesting
didn;t see any clues as to how he was planning on doing backing-up the path
robin_sz: custom to perhaps
a few days or so
* alex_joni says at 6pm
seems that cold didn't pass around me :/
sorry to hear that
not that bad though
I get to spend the rest of the day in bed :(
vq_ is now known as Vq^
03cradek 07HEAD * 10emc2/src/emc/motion/control.c: spello
03cradek 07v2_1_branch * 10emc2/src/emc/motion/control.c: spello
03cradek 07v2_0_branch * 10emc2/src/emc/motion/control.c: spello
they call me spello yello
SWPadnos: thought you don't like Donovan
I hate Donovan (almost as much as LabView)
but I still know the words to those damned songs
yellow is the colour of the ..
in the morning, when we die :D
anyone know what the standard thread pitch is for M8?
1mm or 1.25mm
1mm for fine-pitched
eholmgren: http://www.efunda.com/DesignStandards/screws/screwm_coarse.cfm http://www.efunda.com/DesignStandards/screws/screwm_fine.cfm
(agrees with what alex_joni said)
Rugludallur: stay a bit away, or update your antivirus..
* alex_joni has a bad cold
:( bahh over the holidays, :(
I just got news that I need to go to Boston on Friday,, hope to make it back before New Years
that will be a very short trip then.
24 hours hopefully
This is starting to become a habit, I was in NY on the 30th/31st last year :P
delaying the new year?
Anyone here in MA that could give tips on something interesting to do on a Friday night in Boston ?
alex_joni: Starting next year :P
Rugludallur: does it help?
ah - VMWare 6 beta finally finished downloading
sorry - sometimes I don't know my own strength
I'm finished playing for today
03jepler 07HEAD * 10emc2/docs/src/hal/pluto_servo.html: fix contradiction noticed by cradek
any christmas emc hacking going on ?
I added a few more items to the hal testsuite yesterday
ok. I'll probably get my servo motors and amps in a week or so... then I'll be busy trying to put all components together
what do you think about re-winding a transformer with a 110VAC secondary to produce a bit lower output voltage ?
etla: what kind of transformer?
it's a 220/110 VAC step down transformer. 3kva
I bought it for 100VAC servos and it was ideal for that
but now I wan't to use it with 90VDC servos and need about a 100-110 VDC +rail
is the secondary on the outside?
is it E laminations?
I need to check, but I think the primary and secondary are separate windings
oh side by side? that's nice
core probably looks like a box viewed from the side
you don't want to rewind it, you just want to UNwind some turns, right?
yeah, guess so, from 110VAC to about 70VAC or so
that's a third or so?
that might not be bad, especially if you can do it without taking the laminations apart
or you could add more windings in the opposite direction. maybe - maybe not - that is quite a bit of voltage drop
I was thinking that, but that's a lot of windings
ok, need to try taking some windings off next weekend
do you need isolation? you could use it as an autotransformer and get 220/3
yeah, isolation is needed
yeah that's a terrible idea anyway :-)
hopefully it's not all embedded in epoxy or similar
I've modified a few transformers - it's always worse than you think it will be
a new one with 2x35 VAC in series would probably be good
yeah a new transformer will be easier
but a 1-2kva one will cost something
do you have an idea for sizing the psu for a three axis mill
probably do not need to size for max simultaneous current on all axes
no idea, sorry
steves_logging was talking about that at fest - I bet he can help you
he had some sample transformers
the appropriate toroid transformers are smaller than you would think
03cradek 07v2_0_branch * 10emc2/src/emc/motion/control.c: tweak detection threshold
Rugludallur: go to the Italian village, goto Cafe Vesuvio, drink Grappa & strong coffee, try to understand the game the old guys play on a pool table with cards, dice, money & a lot of yelling
tomp: sounds fun
its really a 'carom' table with the little 4 sided markers (tetrahedral?) and 1 white and several red balls
is "grappa & strong coffee" two drinks or just one?
etla, remeber, toroids are rated for continuous, they will deliver 200% for short times, so you probably dont need anywhere near as big a txfmr as you think
it's just that I have the 220/110 3kva transformer sitting on the shelf
and how many amps do you need?
the motors are 90vdc
and how many amps do you need?
and I think max continuous current about 7A
let me check the datasheet
right, so when would you actually pull 7A?
continuous 7.8A and peak 38 A
etla -- that must be one strong shelf to have that big a xformer sitting on it. :-)
it's not _that_ big...
you can get them on ebay
Save the transformer for another project. Does your boat need an anchor. -- Seriously, though, you might need an isolation transformer, some day.
trouble is, building big linear supplies is not simple
It is if they are unregulated.
no, its not
What is the hard part?
something like this http://item.express.ebay.com/2000-Watts-Step-Up-Down-Voltage-Converter-Transformer_W0QQitemZ5853683932QQihZ008QQtrZexpQQcmdZExpressItem
for example, how would you switch the power to that xfmr?
are you worried about a current surge at poweron ?
I used a dual pole switch/breaker that I bought from McMaster(I think) -- although I'm only running around 1KVA.
thats one worry
well, inrush current will be quite something on a 3kva txfmr
I think the mazak charges the big caps through a resistor at first
You could wire it to a panel type breaker.
Three times that of a 1kva xformer.
cradek, that will make zero difference to the inrush in the mains side
although that will protect the rectifier
well you could do the same trick on the input side if that's the problem
so there's a relay that switches out the charging resistor once the voltage rises to a certain level ?
cradek, inrush on big txfmrs is significant, and not related to the secondary load.
they tend to fry mains switches eventually
It does take a fairly large resistor, though. I really don't like the idea of using a mechanical relay to do this type of thing. Anyone have a good circuit for a solid state approach to this problem?
or you use a PTC inrush limiter
etla, id recommend ptc;s on the input side
um, what's a PTC ?
Positive Temperature Coefficient resistor
and it gets hot really quickly at poweron
high resistance when it's cold
we used to put 2 or 3 in // on the bigger amplifiers
PTCs (1) very hot, (2) don't solve the inrush problem if power cycles off and on rapidly -- like we get when a car hits a power pole.
would it help to turn it on at zero crossing or does that not matter?
well, we used PTCs extensively when we made power amplifiers. it solved the exploding mains switch issues and we never had a problem after that
I suspect one could come up with a nice circuit that put a pair of back to back transistors in series with the transformer and gradually turned them on over a cycle or two.
lerman: with solid state relays doing the switching ?
zero corss helps a little, but you still have a major inrush when initially charging the cap
I'm not sure how you would do this.
SWPLinux, but thats on the secondary side
if the current is limited on the secondary side, isn't it limited on the primary as well? what am I missing?
SWPLinux, unrelated to primary inrush
cradek, inrush on big txfmrs is significant, and not related to the secondary load.
the core has to mumblemumble saturate
even with no load on the secondary?
so, ptcs on the input side
may I never have to deal with such a large transformer...
When the transistors were on, they would dissipate a diode drop times the average current -- say 15 amps -> about 10 watts. That's a reasonable sized transistor.
well, hes taking about 3kva ...
lerman, a complicated solution to an easily solvewd problem with a couple of 10c PTCs
just have a really crappy service that drops the voltage while inrushing ;)
don't they need to be pretty high power resistors ?
more than 10c ?
yeah adding active components (three legged fuses) in a high power app where they're not needed seems bad
they are about 10c, and 15mm disc, 2mm thick
do they solve the cap charging problem on the secondary side too ?
well, no thats another problem
Just don't stick your hand near them. I believe they get up to a few hundred degrees.
hot to touch, sure, but not burning hot
like lerman says, the problem of not cooling down on a short interruption is still there.
and they are just in series with the mains ? both poles ?
Either pole is enough -- they are in series.
we'll there will be a 120x120mm fan in the case so they will cool down reasonably quickly
these will save the switch long term ...
but switching the power on off might not be a good idea
The short interruption problem is solved by the main breaker. And the fan stops running when the power drops.
ok, on our 3kw amplifier, we used the best rated panel mount switch we could find and had multiple failures, we added PTCs and didn't have another switch failure in 5 years.
is that good enough?
If you use a contactor with a push button switch to turn the power on, if the power fails, you will have to manually reset it, anyway. For a machine tool, that's a very good idea.
thats also a good plan .. and probably a legal requirement in an industrial situation
Let's see... a microprocessor with a thermistor to measure the ptc temperature. If it is high, inhibit turning the power on :-}
let's not make it too complicated...
I think he was kidding...
secondary side might be OK, but rate that bridge big enough
It has to be complicated. If there is a computer in it, we get to charge more.
you might be better off with 3 smaller supplies, its just a lot easier
get a 3-phase transformer
Big (high current) diodes are pretty inexpensive.
and start it with triangle, then star
If you have three phase power, you can get away with a smaller capacitor (but you have more diodes).
90V at 15A .. gonna need a fairly decent capacitor
I've gotta run. See you later. Ken
see you later ken
I bought some 400V caps earlier, don't remember how big they were now...
ESR will be too large
as the voltage goes up, the ESR goes up, thus the heating for a goven current goes up
always choose caps with some headroom on voltage .. but not too much
etla: re the bf20, what type of stability/repeatability were you able to get with the acme screws?
and dovetail slides
L_H: something like 0,01-2 mm I think
but with about 0,05mm backlash comp on X and Y
I assume that's after a finishing pass?
coo. 0.05 backlash. not bad at alll for acme
the nuts are adjustable
I think the cross slide on my lathe has about 0.5mm
that's quite a lot of backlash, was that becuase of springyness in the screw and the rest of the assembly (becuase the dovetails have to be tightened)
etla: oh, so there are pre-loading nuts included?
I think the backlash comes from the nut
yeah, the nuts can be tightened. but that ofcourse increases friction
still 0.01-0.02 isn't bad
what about the Z axis?
with the counterweight the Z can be kept to one side of the backlash
so I think we don't use as much backlash comp on Z
oh, preloaded for that one too?
I think so
nice. btw, do you know someone from chalmers, gothenburg working there (I'm not sure of his exact profession) whose first name is göran?
I went to ICBP2004 which was at chalmers
hmm, ok. he knows about you in any case. :p
but otherwise I don't know anyone there
or at least your website
from model yachting ?
a conference on biophysics
I'm not sure of how he found your site
ruglu: you have a satellite connection in iceland ?
elta: nope, the reverse dns is just to mess with ppl
etla: ADSL at home
etla: fiber at work
nice lathe: http://www.youtube.com/watch?v=oGq-9NNmr3o
how the hell did it cut that hex??
the spindle sunced to the rotating blade was nice
I know, but I can't picture the geometry
[21:48:31] <SWPadnos> http://www4.profilator.de/dienstleistungen/videos_animation.html/
yeah, does the rotating tool move in x/y also ?
no, it doesn't need to
look at the second one in the "Polygon and slot machining" section
uh, what language is he speaking?
I think he's reading an engrish script ;)
it's very funny that he has no accent
sound like an english translation of chinese or japanese
why would an english speaker do that?
I don't know
"excuse me, there are some errors here .."
because that's what the borchure/script says?
cool stuff though, huh?
yeah extremely cool
you'd have to be a special sort of idiot to do that, or be following the instruction of special sort of idiot
how is it cuttinghtose hexs?
look at the second link
if you have a six pointed star and spin in sync with a rod and feed you'll get a hexagod
that was sweet
do you need a different diamter star for a different diamter hexagon?
and the two sided flat?
I'm not sure. I'd assume so
I don't know how they did the tang on that big taper
kewl stuff though
well - we have a new goal for the NIST-lathe at Fest ;)
I can bring a small servo to use as a secondary spindle ...
only 50 oz-in though :(
it's interesting to think about the minimum hardware required to do that
obviously? one tool is enough
it depends on the sync. I'm not sure if you need to have the curring tool and the workpiece rotate at the same rate
which video shows the hexagon stuff ? I looked at the polygon/slot machining already...
haha, apparently my lathe video on youtube has got 2800 views! O.o Don't people have better stuff to do? http://www.youtube.com/watch?v=MX4SWe8WFA4
(man - I can't type today. maybe I hould have some Bailey's in my coffee)
Lerneaen_Hydra: your tool is mounted too low
the second one in the poly/slot section (the 1.94Mb RealVideo or 7.64Mb mpeg one)
cradek: I'm lazy
I couldn't be arsed to get it right and just eyeballed it
the carbide tool was meant for steel or other hard stuff so the wear shouldn't be that bad
L_H: do you have steppers or servos on the lathe ?
on that lathe? haha, steppers from the mid 1980's
how come yours isn't making 'gons ?;)
the engirish is hilarious
single-winding 72 steps/rev
robin_sz: I played the whole thing again just to hear the talk
robin_sz: I don't even understand what the part about hex stock is about
robin_sz: the hex stock is too loud? or something?
it's a PITA to handle with auto-feeders, so why not just make it in the lathe
I think he is saying a conventional machine feedin hex bar makes a lot of noise at high speed
wtf. that text is completely incomprehensible
they use round bar and can make hexs
and the accent didn't seem to be awful either
he says "lakes" means "lengths"
fwiw I have to agree with "special kind of idiot"
oh, lakes hahaha
Lerneaen_Hydra: I'm pretty sure he's a native english speaker
almost 100% sure
cradek: then why! why!?
but I guess he wouldnt know one end of a lathe from a screwdriver
I thought it seemed to be good too
oh, that's a thought
so, he didnt question the text
could have been a marketing drone
like ai said ..
"a special kind of idiot" ;)
someone from marketing probably
I do not have any sterotypes at all!
at least not any boring ones
the motor that rotates the tool needs to have plenty of power I guess ?
it cuts the hex VERY slowly
cradek: what do you mean ?
the feed is slow, I bet the cutting force is high
so could you do it faster with a stationary main spindle and a moving live tool with an end-mill ?
a hex maybe, but I bet not all those shapes
and end mill would be more versatile
Fenn: How goes it? :)
* fenn grunts noncommittally