but the link to the pdf is 404 now. Though I do have a copy of the pdf locally.
CHEAP-ENC V1.0 GNU/GPL (C) 2005 Lloyd Godsey. 2.00",
8bit PWM, Tach, 45/90/180 Sync. Print at 2450 DPI or 25400 DPI.
Still $14K USD, and only 65W... http://www.emissiontechnologies.com/xy.htm
24x48 are VERY VERY hard to come by for this price.
that device is not for sale is it?
robin_sz Yeah, I'd expect at least 100W
it is, and it's a KIT too
it looks like a lawsuit waiting to happen
high voltages on the tube ...
He's been selling laser tubes and PS for a few years now
exposed class 4 laser beam ...
hes an idiot then
that is just so dangerous ...
oh you should have seen his other 4x8 table... now THAT's scarry... even the 48" tube was FULLY exposed
you might do it at home, inyour shed
but to sell it?
[00:20:02] <Jymmmm> http://www.emissiontechnologies.com/XY/RasterVector/PIC00001.JPG
thought that is his 24' x 48"
robin_sz that's why it's sold as a KIT
semi releases him from a lot of liability
I doubt that will save him from product liability suits
but not all ...
direct exposure to class IV co2 like that .. well, goodby retina
But, it does make for easier maintance being a kit (i would think) instead of a closed propritary system
they added a SS enclsure to the kit
see lower in the page I linked
thank $deity for that
[00:23:01] <Jymmmm> http://www.emissiontechnologies.com/XY/Top%20Cover/PIC00001.JPG
I would have thought propreiatary was easier maint ...
looks sharp being SS
just phone the service technician
Well, epilog charges $1200 for a refilled tube, doens't include the matched PS either.
Universla charges $900 for a matched refilled tube and PS
I have a nice spare tube and PSU here ... too big for you though ;)
Yeah, need 120VAC @ 15A
I think the blower motor for the main device in the PSU draws more than that
WEll, that would be on a seperate circuit
along with the PC
mine draws about 30kw .. a bit big for you ;)
you mean the exhaust blower, or air assist?
the coolign blower for tha PSU is more than 15a on mine ;)
ah, this one is water cooled
ah right ...
I have a watercooled YAG laser
but again ... its not small ;)
I really have no REAL desire to cut metal (maybe thin brass,etc) with a laser, so ~100 W I'd be happy with
bras is really really difficult to cut with a laser
just artistic metal work, not functional
whats the main metal in brass?
is what I'm saying.
hint: begins with C
mmm .... close
Well, I meant with ANY laser CO2/Yag, etc
and guess what metal they use for the mirrors in really really big CO2 lasers?
and guess what metal they use for the final output nozzel in industrial lasers?
copper is pretty much a total reflector for 10um
very hard to make any impression on it ... and you may well damage your output coupler trying ...
Do you know why a laser is REALLY appealing to me?
mmmm .... the smell of buring wood?
the prep/finish work needed.
for acrylic... it'll even polish the edges for you
that's a HUGE timesaver
it is very very good
the 450w ferranti I had did GREAT acrylic
one word of warning ... the fumes
even wood too... no deburing
oh yeah, already looked into activated charcoal
acrylic is horrible fumes
it stinks and it coats everything ...
slides .. screws ...
a few 5 gal buckets daisy chained together filled with activated charcoal and water
just vent it out of the roof :)
but ... you'll find it soon coats the inside of the enclosure
Na, I already saw and smelt that setup in person.
wait for a few hundred hours of use
blew a bulb onthe YAG yesterday ...
that was a mess
Yeah, I really have been doing my home work, even the math on the exhaust system, plumbing etc.
broken glass in the water systems etc
now if I can only get the pocket book to where the homework is, we would be in business.
it was only 4 weeks old
less than 300 hours on it
whats the avg life on that?
sounds like a defect - dirt
I never knew that could happen.
the arc gradually eats away the glass wall of the tube ...
maybe shitty tempering of the envelope, but...
goes chocolae brown near the cathode
eventually it gets thin and pops
good to know... note to self: Enclose tube in explsion prrof cabinet
so .. lamps cost £125 ...
oh thats a pair, 8kw
so it will use £800 in electricity in its life ...
whats that? $1200?
plus ... the cooling
thats about 50% efficient
so $1800 to run the device for 1000 hrs
so the cost of the lamp is minimal compared to the cost of the power to run it
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2006-11-17.txt
so running 80v into the big servos gives me a non load rpm of about 500rpm.
it needs to be bolted down for testing :) - thats all I have to say about that ;)
can't deney the laws of physics
these things are monsters - at 10% of 80v you can't stop the shaft. (around 50rpm)
these where direct coupled to 2tpi ball screws - now they will be coupled to 3tpi ball screws
skunkworks, if I did the math right, 20A in those motors = 22 foot lbs
= 4325 oz-in
the tag says 17ft-lbs at 16.2 amps stalled - how did you calculate it? close
did you calculate it from the tag?
80V * 20A = 1600 watts
1600 watts / 746 watts per hp = 2.144 horsepower
2.144 HP * 5252 (magic number) / 500 rpm = 22.52 ft lbs
if you repeat the calcs with 16.2A, you should get 17 ftlbs
cool. They have a lot of rotational inertia. but they are going to be hooked to a beast anyways. not worried.
I got 18.24, but that's in the ballpark
the calculations assume 100% effeciency, etc
and the price was right ;)
500 rpm / 3 tpi = 166 inches/minute
the machine used to have a max speed of 150 ipm
so a pretty good match
around 1000 rpm would be great - the spec on the servo is 1200 rpm max
just use more voltage
did I say I was excited?
the encoder should be mounted tomorrow night.
these things have 6 brushes
we figured that is why the machine was scrapped - 2 of the sevos had hung up brushes.
6 brushes means a high pole count motor - low speed and high torque
and the numbers prove it
gotta reboot this vm, just installed the RT kernel...
how to pass from joint to axis ?
how do you mean?
i need help
what do you need help with ?
i have 3 motors on my desk
i want to do XXY with gantryskin.c
on axis gui, i see 0 1 2, i home all this axis
i not change to world view
I'm afraid I can't be of much help, when I set up my gantry I used a single axis and just slaved to outputs to it
err two outputs
axis doesn't do that
tkemc can allow that
does not matter which UI, you can do it in HAL
just hook to pins up to the same signal
and invert one of them so both motors drive in same direction
why not wire one motor inverse?
then axis could'nt be use for gantry
well that would depend on what kind of motors you are using, steppers or servos or such but you can do that aswell
i have servo
i haven't found an example
dallur-thc is my config, I think it is in the 2.1 samples tree at the moment
Dallur: you are building a plasma table right?
how fast does it need to go?
[13:43:35] <Dallur> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Torch_Height_Control
is where I put all my current config, im still working on improvements so it's not a "production" setup but you can look at the config for examples
hm, so how fast are you running?
Dallur: he didn't mean that
he is using special kins
I have gone up to about 15meters per minute
and when the emc2 GUI starts up it is in joint view
how fast do you need to go? was rather what I were wondering about
after homing he can switch to world view (XYZ coordinates)
but afaik only tkemc knows how to switch
well received alex : only tkemc
AXIS should be extended to do that..
but I don't know anyone with non-trivkins machines around
it's a pity, axis is more easy...
so there's not much drivingforce behind this
i could do that with a little help
I don't know any python, else I would have tried already
the emc-related stuff is pretty trivial
Dallur: so, how fast are you cutting with it? lets take when cutting 10mm sheet..
I'm not sure how the backplot should react to this change
anonimasu: I only did a couple of test before dis-assembling everything to coat it and prepair
it could be possible for me to become developper ?
Dallur: so you have no clue at all?
anonimasu: but the speed just depends on how powerfull the plasma machine is, the top speed for the table is way higher than the capacity of even 300A machines
anonimasu: probably around 2m per minute (6ft per minute)
I have a 300A machine I'll be building a plasma table with..
sorry.. lost my irssi somewhere
anonimasu: Thermal Dynamics or Hypertherm ?
going to cut thick stuff..
hypertherm rulez ;)
oerlikon is ok too
Dallur: I like SPT ;)
anonimasu; Nice, only used welders from them, never plasma eqiupment
Dallur: from sweden :P
Alex: nope Iceland
[13:51:42] <alex_joni> http://www.spt.se/
I meant SPT is from sweden
Alex : i try with tkemc
I dont need speed..
it's for cutting thick stuff..
anonimasu: what's thick for you?
stuff between 1-3mm..
you need about 100A for 30mm
Alex: Never used them, I think there is a store around which sells them and I noticed they use alot of torch stuff from Thermal Dynamics
Dallur: we distribute them in romania.. very good machines
mostly the 9C http://www.spt.se/sparcin5c_9c_eng.htm
wait a sec..
it's not oerlokon..
anonimasu: yuck ;)
40mm.. is doable apparently..
I might have messed up my amp rating..
It's all Air Liquide in any case
i have tried with tkemc, it's ok / Alex
though just a stepper machine..
Didier: you can run both (tkemc & AXIS) :)
I will code Axis to do this
anonimasu: 300A is a _huge_ plasma
I've seen a few.. scary
both TKEMC & AXIS ????
Didier: why not ;)
numbers dosent stick in my head..
you can run as many GUIs as you like
at the same time ????
Didier: yes, even on different computers
this one is pretty alrge.. but not that large
not on the same division though :D
[13:59:35] <anonimasu> http://www.weldersworld.com.au/plasma3.jpg
the only bad thing is that it isnt a machine torch..
how to do ? DISPLAY = tkemc and next line DISPLAY = axis ?
Didier: no, you need to manually run one of them
DISPLAY = tkemc
then run emc
then open a terminal: scripts/emc_environment
i have opened emc_environment, and what to do with bin/axis ?
I think you need to tell it the ini
bin/axis -ini configs/whatever.ini
axis works with kins now
i need to import the last emc to see it, cradek ?
it's not in emc 2.0
so you need to run the cvs code, or wait for emc 2.1
well i will the last cvs code
thanks Alex, thanks Cradek
talk you later
didn't remember that
it's been there for a bit now
off to breakfast...
wow - nice work
<troll>too bad there are so many issues with emc according to paul</troll>
hmm.. emc_environment is also part of HEAD only
so I wonder if he's running 2.0.x afterall
hm, that reminds me of when I ran bdi..
mm the issues.
wonder what I did earlier to lose irssi.. it was still active, but I couldn't get to it
might be . shouldn't fg bring it back then?
I have a scary part to machine
the .igs file takes 1.1mb
anonimasu: alibre is great btw
the free version is all I need so far
* anonimasu nods
I found it a bit annoynig when you break a sketch,..
that it dosent show you where the error are
I have installed it - but have not played with it.
[14:36:09] <anonimasu> http://www.almaskin.se/wh1.jpg
that's the scary part..
skunkworks: it's a strange and twisted man who has time to always check emc2 for compile errors, yet never wishes to contribute anything useful.
jepler: sounds like part of the compile farm to me
oh.. missed an i there ;)
jepler: it makes me wonder why he even bothers to check it for errors, if it's sure he doesn't plan to use it
probably tuxcnc is comig along too easily so he has too much spare time
let's change the subject...
anonimasu: what's that part?
seems like a cutting tool
got it.. big top secret project
cradek: it's for cutting special undercover agents
into subatomic particles that can't be traced
heh let's be sure to show this to LH: http://blog.modernmechanix.com/2006/11/16/model-airplane-motor-drives-scooter/
I'd rather use a turbine engine ;)
thoose tiny ones :D
and a huydralic motor..
anonimasu: nuclear power
hmm how about anti-matter powered but you use cold fusion to make the antimatter ?
Dallur: not interestng unless it also generates an anti-gravitational field, so you don't need the wheels anymore
alex_joni: Everyone knows anti-gravity impossible, now that's just crazy talk :D
alex_joni: ahh I saw that a while back, and they did try to verify it,, they had some problems
alex_joni: wait, not I thougth it was the anti-gravity stuff done by that Finn, sorry
Dallur: did you see my recent comments on the "gantry" feature request on sourceforge? http://sourceforge.net/tracker/index.php?func=detail&aid=1075625&group_id=6744&atid=356744
sorry about the spelling everyone, one of my mem chips failed, I removed it and now my work laptop can't keep up with my typing rate and keeps dropping key entries :(
jepler: sure did, can't wait to test it :D
jepler: My table is currently in pieces, some parts getting zink coated, others getting nickel plated, and such
Dallur: I see
I can't wait to see how the Electroless Nickel plating works for all the surfaces, it's the same stuff they plate HDD platters with
The hardness is way up there and the plating is uniform unlike hard chrome or other active plating methods
Dallur: so the homing sequence I describe in my message sounds like it would work for your machine?
not everything I described is implemented yet
jepler: it looks like a very good solution for my problem
Dallur: do you currently have a home switch for both sides of the gantry axis?
jepler: yup, I have seperate home switches for all axis
jepler: although im replacing all of them due to EM problems, High Freq. really causes problems unless you have PNP/NPN so im switching to inductive PNP
jepler: Both EM and HF that is, even with shielded cables normal magnetic switches go crazy when the torch comes close
Dallur: luckily I don't have to worry about those details in the software -- I just assume the HAL signals from the switches are reliable.
jepler: I wonder if using combo limit/home switches would cause any problems with your seq. ?
Dallur: I believe homing already handles that, and I don't see that my changes would affect it
jepler: ok, just a thought :D
is your final homing move onto the switch a second time? I kinda assumed that in what I wrote
jepler: I have not done any homing work, I was homing manually because of the gantry problem
has anyone here heard of any EMC user that goes to chalmers?
becuase there were a few there that I met today that ran emc :D
the college I go to in Sweden
oh, rayh, were you looking for something to do with emc and lathes?
That is an upcoming project.
I only know of and have met one EMC user from Sweeden.
I also met two of his kids.
Anders Wallin is in Sweden, isn't he?
He engraves some sort of rollers.
not gothenburg though
[16:57:05] <jepler> http://axis.unpy.net/index.cgi/01163779208
"truth table" compiler for emc2
from the description, can I (correctly) infer that a pin that appears on both the left and right side of the vertical bar is an output, but its state is considered in the evealuation?
yes, that's how it's supposed to work
ok - neato
how hard do you suppose it would be to make the pin names multi-character (separated by commas or something)?
it probably wouldn't be hard
I guess the lhs would need to be an array of string instead of array of char, and an additional split would be needed (if a comma is found) ...
or use csv from a spreadsheet
(I really need to learn python
what does the zip function do?
does that just allow you to iterate over multiple arrays at once?
yes, something like that
>>> zip("abc", "def")
[('a', 'd'), ('b', 'e'), ('c', 'f')]
ok, essentialls splits multiple arrays into a series of ordinates (tuples in programmer-speak, I guess)
it's such a pain to be able to see the syntax easily, but not know what libraries are available
yeah there are a *lot* of useful functions available, but you only learn over time what they are
I'd end up writing my own version of lots of standard stuff, just because I don't know the name of the available library version
so in this loop, "k" gets the names of each of the left-hand-side items, while v gets the corresponding letter on this rule line: for k, v in zip(self.lhs, lhs):
sure - I got that, just wasn't sure if the parallel iterators were part of the language or done by the zip function
it's funny - I've been working on microcontrollers so long that the first thing that pops into my head is "what can be done to automatically optimize the resulting code?" :)
"I dunno, but I leave it up to gcc"
I was thinking along the lines of sorting the input cases and doing nested ifs instead of a chain of else if clauses
or build a table of input values and use case statements ...
that's the funny part - you get lots of ugly shit that may run a little faster, but it's mostly unnecessary on modern PCs
check if an output is used on the LHS; if not, eliminate the corresponding local
sure - that would be good
check if a pin is ever tested for an edge; if not, eliminate the corresponding "o_" variable from the instance data
but any sort of analysis like that kind means that you ahve to deal with the "whole thing" as well as individual lines - right now it's basically a filter
* jepler just implemented multi-letter pin names
everything's easy in python
yeah - if you speak hiss :)
A truth-table file consists of a title line followed by multiple rule lines. Each line is divided into a "left-hand side" and "right-hand side" by the vertical bar character "|". If a side contains whitespace, then it is broken into items separated by whitespace characters. Otherwise, each letter is an item.
On the title line, each item that appears on the right-hand side is an output pin, and each name that appears only on the left-hand side is an input pin. (If it appears on both sides, it is an output pin but its prior value can be used in a rule). All created pins are bits. When multi-letter names are used, "_" should be used in place of "-".
sounds good to me
</new text on the tt2comp page>
is "unchanged" a useful condition to test?
possibly - use an = sign?
yeah - unchanged would give shorthand for !(^ || v)
while I'm at it I'll add !otherpin and =otherpin for the LHS
it's unfortunate that in the column for 'x', '!x' and '!' have different meanings
(one tests the prior value of x and triggers on edges; the other tests the current value of x and never triggers)
hmmm - how do you intend to use/allow =something or !something?
not sure, maybe it's not a good idea
jepler: nice work.
just got kfc. weak moment
the ! gives you "either edge" (and = gives the opposite, "no edge"), ^ and v give you specific edges, and 1 / 0 give you "present value comparisons" - I thikn that's a complete set
heh - my kfc moment was last night ;)
those crispy chickens are pretty darned tasty about once every few years
yes - more like once every 2 months
look at this
[17:50:54] <skunkworks> http://www.cnczone.com/forums/showthread.php?t=4849
we should get him to update this tread with his experience with emc2
Holy crap. He wire wrapped the laser control board
wire wrap... tasty
I just don't understand wire wraping
before your time?
Seems like a pain in the a$$
before mine too
Lerneaen_Hydra_ is now known as Lerneaen_Hydra
it's an odd technique, but it worked pretty well once you got good at it (which I never did)
a friend of mine built an entire computer (TRS-80 compatible) using wire wrap
we were happy to have etchant available at Radio Shack back in those days :)
the big K&T has I don't know how many cards in it maybe 500.. The back plane is all wirewrapped
* skunkworks shivers
hmmm - I can't imagine using WW as a production technique, unless you need the ability to customize the products
I think these where pretty low producion - it is like a GE model 1 or something like that ;)
hmm, any idea why doing an SCP from two fast machines only goes at 1mib/s?
because SCP is doubly inefficient
cpu use is low on both
oh, you mean that it's so inefficient becuase it sends lots of "uneeded" packets and waits and so on?
if you're using ssh as the connection, you end up with some weird encrypt / encode / re-encrypt / re-encode <-> transfer <-> then undo it all at the other end
at least there was a problem of that sort with puTTY and PSCP - I'm not sure it's a generic problem
the cpu is not even close to being the bottleneck
and this is both win-lin and lin-lin
no, but the amount of data transferred over the network is much higher than the amount of actual information
it may do base64 encoding (possibly more than once), so you send 8 bits on the wire for every 6 bits of source data ...
I'm only using around 15% of the potential bandwidth (100mbit)
basically I can't seem to find the bottleneck
well, I dunno, but when I did the same thing on my 100mbit LAN, I found that things were significantly faster using FTP :)
do you know if there is an intentional max bandwidth limit in openssh server?
I don't think so
well, I know that I don't know :)
which ftp server did you use?
I don't remember
I may have done it as http transfers, come to think of it
SWPadnos: I almost flipped one of those huge servos onto the floor... oops.
* skunkworks needs to bolt it down for testing
yes, they do tend to jump around a bit
SWPadnos: for instance, you can get o=a xor b with this: ab|o / - =a|1
sort of - that won't set to 0 for b != !a (without another line anyway)
then I guess it's two lines with or without =<name>
as shorthand, it can reduce the number of lines needed by half, since it allows you to do the cases of both=0 and both=1 at the same time (or !=, as the case may be)
(both inputs, that is)
hmm - the order of the rules is important. consider this:
ab|0 / 10|1 / 01|1 / --|0
that's xor, assuming that wither of the two earlier rules would prevent the last rule from being evaluated
oops, and assuming that I learn to spell 'o' :)
When an output value is set multiple times, the assignment in the last matching rule takes precedence,
so you'd put the -- rule first
I'm thinking like C "break", but you can't do that, since several outputs may be changed - OK
err - C case with break ...
a b |o
- - |0
so here's the version using !<var>
yep - saves one line
I'm not sure it's clearer, and I'd rather have clear than compact
yeah - I was trying to figure out whether the added complexity of the syntax helps much
I suppose it would if you had a non-trivial truth table, with more than 2 inputs
- v |d
is this the truth table for a D flip-flop?
actually, thinking about it at the HAL level, the = or ! syntax would be useful for xx-invert pins
I think so
since by default q will retain its previous state
(if you coded it in VHDL, you'd need to explicitly reassign q to its current value :) )
do you think that output values should not be "held" by default?
SWPadnos_ is now known as SWPadnos
what a strange error: Closing Link: 127.0.0.1 (Connection Timed Out)
apparently Windows only provides the loopback address when a network card has a link
(that error came up when I accidentally unplugged my switch)
sounds like something windows would do
Rugludall is now known as Rugludallur
Anonimasu, I checked some of my logs for the cutting speed and for 10mm steel the cutting speed is 1.3m per minute
Anonimasu: I max out at 25mm @ 0.4m per second
Anonimasu: and at the other end I can cut 3mm plate at 7m per second
Anonimasu: replace second with minute (bonks self on head)
Jymmmm is now known as Jymmm
only 1 more hour
Tonight I _bolt_ the servo down and play with it closed loop.
alex_joni: big plans for the weekend?
about 20h worth of sleep
I hope at least
sounds like a plan
yeah, sure does
my gf has plans for tomorrow morning (yay)
so I can sleep late
wife is having the girls over tonight - so after cooking I am probably free.
that's nice ;)
think I'll start implementing that plan
night all :)
LawrenceG: any progress on the 3 phase servo drive?
rugdallur: Thanks, if you do read the logs! :)
robin_sz: are you awake?
naw... working on the dc drive at the moment.... am getting spurious resets... driving me up the wall
I thought it was a software problem in the brush servo drive, but it looks like hardware... I see the 5v supply spiking down enough to reset the micro
yeck. What do you use for current limiting?
the current limit is done in the output driver chip... supply 0-5v and you get 0 to 7 amp current limit
aww that is cheating
yup... I actually use the direction signal to turn the drive full + or full - drive and use the current limit to set the output amplitude, so it always runs in controlled current mode
rats... print is confused after power failures here.... got to power down....
LawrenceG: would you want he free wheeling diodes to bypass the sense resister - so that back emf doesn't spike accross the resister?
jepler: the truth table thing is very cool
like a minimalist version of ladder logic