SWPadnos_ is now known as SWPadnos
solexious1 is now known as solexious
I'm working on the "final" bug in a clock I built. How frustrating I have to test it on a leap year on 2/29 and after 2/29, and also on a non-leap year
also, I wrote it 6? years ago so the chances of me getting the fix right in one try are pretty slim...
(at least I found the source)
at least I think this is what's in it...
my electronic clock has no date shitzu in it
pic chip counts mains pulses
eeeeeeasy! (and pretty decent timekeeping)
slow debugging process there....
jmkasunich: yeah it's been going on since about '02 (intermittently of course!)
like in the olden days of synchronous clocks, cheap and chearful, and deleberately blinks if the shop keeper doesnt use it properly
is this a recently built clock, or just taking advantage of the once every four years debug opportunity>?
I found 3 things that look wrong but I can't quite convince myself they would cause the wrongness I see
no I built it in '02, never made a case, found this bug in '04, never fixed it ... etc etc
you need more round tuits
I ordered a truckload, but something came up and they never got delivered
we had a clock here a couple of weeks ago that had an evil bug
200 year old and the strike goes out
I didn't think you worked on the kind of clocks that can have bugs?
jmkasunich: just because there are no electronics doesn't mean there are no bugs
turns out that there is a cyclic phase error to the locking and has to be built to hide the error
we work from 1600's to new
do any atmos?
wound by air pressure
jmkasunich: a clock powered by only atmospheric pressure changes
pseudo perpetual motion
people think they're finicky but I suspect the trouble is usually operator error
yes, free power, just not much of it
enough for a very carefully made clock
[02:40:03] <archivist> http://www.atmosclocks.com/history.html
the one I have: http://www.atmosclocks.com/atmos540.html
(what a terrible picture though)
ew eyes bleed picture
from the description it sounds as if its driven as much by temp changes as by pressure changes
"Inside a sealed capsule, a mixture of gas and liquid expands as the temperature rises and contracts as it falls"
one cycle of the pendulum per minute...
must be a torsion pendulum
I've had people look at mine and think it's not running
you have one? then you must show us photos!
it looks like that photo, but not so grainy
looks like they have four for sale...
you could have a nice collection
less blurry photo: http://www.atmosclocks.com/images/Anniversy.bmp
(they have lots of photos, most are in more boxy cases)
that one's very neat
so is it gonna be 4 years before you know if your bug is fixed?
seems like I should simulate it somehow, but in reality, yes
what kind of clock is it?
sets from wwvb, shows "hh:mm:ss Ww mm/dd/yy" on some neat 'vintage' 5x7 LED displays
oh, so the "sets from wwvb" is what makes it tricky
showing Sa 03/01/08 currently
thats the bug I assume?
it only gets time from wwv, not date?
you need a TZ offset in it?
wwvb transmits "day of year" and year and a leap bit
and the leap bit isn't getting processed right....
so today is day 60 which on some years is 03/01 and others 02/29
does that mean it would be off for the rest of the year? (what happened in 2002?)
yes I think it's off for the rest of the year ('04)
[03:00:18] <cradek> http://tf.nist.gov/stations/wwvbtimecode.htm
when you consider that you have a utc offset, the leap problem, and the various rollovers required, it's a bit complex
yup we saw a mysql bug of that type
it seems like you don't need that many rollovers - you get shiny new data every minute, you only have to fill in the seconds
or do you want it to run for long periods with no signal?
it should be able to run without a signal, but stay perfectly set when there is one
and its probably running some miniscule CPU too
my 'older' model sets once a day and if the signal isn't there, it keeps trying (the current set is lost when it starts) - not enough ram for anything else
bug/not seeing leap bit
this one does everything I want (but slightly wrong)
if you look close you'll see there's something obvious missing from the wwvb signal...
(no space for it I assume)
any kind of checksum, parity, etc
so this clock reads for two minutes and compares
probably not needed - if you look at a few consecutive minutes you can check
the old one will *occasionally* show a wildly wrong time for a day
(or until I unplug it)
cradek: Well if you live to be 100, you'll have 25 chances to get it right =)
no, he's already missed at least 8
I wasn't born yesterday (ha)
"an efficient antenna system requires a radiating element that is at least one-quarter wavelength long. At 60 kHz, this becomes difficult"
just a bit
Come on, you dont have 7800 ft of copper available?!
the copper is easy
the tower is the tricky part
I probably do, but dont have the land
ground spacing is a problem
Electrical poles... copper included =)
wwvb is one of my favorite radio signals :-)
what do you use to recieve it?
it's neat to see looooongwave still used for something
antennas ripped out of drugstore wwvb clocks
I sometimes use 198 khz for frequency
I mean, do you have a receiver that tunes 60KHz?
or just the one in the clock?
just the clocks
I guess you'd need a receiver with a BFO anyway, since the only modulation is the timecode
I guess you could hear it like CW -- right
(if you wanted to hear it)
this clock has an LED so I can "hear" it
helps you orient the antenna
I should build something around the MSF modules I have here
actually wwvb is so strong now that you can get it in the basement and everything
before '99 it was a bit of a trick to receive it, even here
"When properly implemented, the use of bit and frame averaging can be more effective than increasing
the sensitivity of the receiver, or increasing the field strength presented to the RCC by many decibels"
hello, hello, is this thing on?
we can't comment for national security reasons.
I think you've said too much
maybe i'm using misdirection.
so you're saying that if I think that you think that I know what you know, but you think that I don't know if you know what I know, that it doesn't matter if you tell me what you know?
i propose we use the term 'vector' instead of 'axis' so that there is no confusion with joints
fenn: i can dump the opengl to indiv frames with xvidcap, and then make movies later. so far i've dupmed to xwd at 6 fps with no dropouts. dunno how fast or what formats can be saved.
tomp: do any wedm controls use A and B to specify the rotation of the wire? instead of UV
not that i know of, xyuvz is std axis
yeh, not ab cuz not rotation,
but it is rotation
why do you say rot>
the wire rotates relative to the workpiece
the upper wire guide translates realtive to the path
yes i know
if there's no rotation then where does the taper angle come from?
from trig, dist from upper to lower and offset
and that's not a rotation..?
the angle is not used, the offset is, the angle is a product, not an input
so, lets say there's a drawing of an extrusion die. do they typically mark the angle on the drawing or the two different profiles?
some systems let you specify the angle, but the resulting code is 4 linear dimesnions
normally the print is draft angle,correct
the machine tool does not work at the top and bot of the die
it works above and below
how far? depends on machine
so designer doesnt know machie setup, so oper must figger a lot of this out
depending on clamping & section
the wire guide is fixed at a certain Z height, yes, that's what i'd consider part of the machine kinematics
lower guide is fixed, upper not ( typically)
do you change the guide height often?
yes, for me every cut, but i dont do production, each is custom
what is the advantage?
neccesity, not advantage, differnt thx
er.. why do you need to do it then
like on a bandsaw, it's to keep from cutting your arm off, and reduces deflection of the blade
does the wire deflect that much on an edm?
to get the angle the print request, (dies angle varies with matherial and thickness), you must have the correct spacing and offset. the spacing varies with setup (material thx and clamping), the offset dimension must vary
wire deflects ( action,reaction) causing 'bicycling' around corners, guide path and wire path are not same
so dynmaic speed control is common, also dynamic power control to make it straighter ( will alwys bicylce some) bicycle is the path your bike makes thru a puddle, watch the tire tacks of front tire vs back tire
they dont match
i dont see how that relates to edm
what causes the wire to deflect?
the upper guide is advanced of the wire center
hydrodynamic pressure between the wire and workpiece?
gas pressure from the sparks?
yes gas pressure from teh gas bubbles
does the wire bend as it enters/exits the workpiece?
when going around a corner
the upper guide is advanced of the center of the tensed wire length
the wire bows anytime there is lateral pressure exerted along its length, more at ctr, l;ess at supports ( guides )
since the normal units of measure are tenths ( .0001") or um (.001mm), this effect is significant ( approaches or exceeds tolearance)
btw: imagemagick 'animate' just made flip movies faster than i capture with xvidcap
you're saying xwd works better than xvidcap?
now, still experimenting finding the tools. i used xvdicap to grap indiv frames at 6fps. now gotta join them into a vid
but xvidcap didnt drop any frames using this 'to frames' trick, while 'to vid' drops a lot
its what you noticed, it'
s too busy making the vid to keep up
right, it wastes a lot of CPU encoding compressed video
so i let it do that later
right, your tip got me on the right track
i'll look for a wedm tutor that shows the axis relations
i didnt have much luck using xvidcap in single frames mode
tomp: i'm more concerned about how EMC can accomodate many different geometrical paradigms (naming schemes, axis vs joints vs vectors vs bodies vs relations)
and still seem somewhat coherent and not hacky
extra linear axes doesn't make much sense when you're concerned with the position and rotation of a milling tool
damn edm wire doesn't have a tip to use as a center of rotation
the extra linear axis can be used to shift the work envelope, like cradek's mill knee W idea.
hmm, yeh, no tool tip, got a tool edge almost
ditto for grinding
well, a toroidal cutter doesn't have a tip either, but you can pretend there is one
i guess you can just use one of the wire guides as the center of rotation
anyway that's not what edm people want
it occurs to me that there isn't enough information in two profiles to reconstruct a three dimensional shape
if you have two circles as the profiles, the relative phase of the angle around the circle will affect the shape, going from a cylinder at 0 through some hyperboloid hourglass shape around 90, and a pure cone at 180
[08:15:27] <fenn> http://en.wikipedia.org/wiki/Image:Ruled_hyperboloid.jpg
yes, theres the 2 paths, and a synchronization between the 2, so knowing the guide position on 1 tells you the other
lots of 'wedm art' like that ( toolmakes with too much free time )
i think the ever popular nudie girl on mudflaps is from the early wedm days, tho we cut 'BJ's which are illegal to display ;)
huh. the reason they use hyperboloids in cooling towers (like at nuke plants) is because the reinforcing beams are straight
[08:23:43] <fenn> http://math.arizona.edu/~models/Ruled_Surfaces/
i imagine they're more impressive in 3D
damn closed information services, pay for info... "Simulation of Complex Curved Surface WEDM System Based on OpenGLFile Format"<font size="3"> http://www.ttp.ch/0-87849-470-7/364/</font>
maybe they think adding more frames will help protect their data
gnite fenn, did it agin , 3am
ala Homiere Simpsonne
god reading old emails is scary
dif är fina grejjer :9
ooops wrong window...
i would like to test the sampler hal module. I wrote a simple hal file ( http://pastebin.ca/924476
) and display the vaues on the screan using halsampler, but all i can put on screan is one signal at the time.
how can i display al of them at the same time
"halsampler -t -c0 -c1" doesnt work
to show what values?
kwaj you can probably do something like: watch -n 0.5 'halcmd show sig mysignals'
[#emc] "This is the #emc channel - talk related to the Enhanced Machine Controller and general machining. Website: http://www.linuxcnc.org/,
wiki at http://wiki.linuxcnc.org/"
alex_joni: yes please
fenn: that was an ironical remark..
alex_joni: bout time! lol
alex_joni: Every time I sneak in here, you've been sneaking to bed.
well.. you sneak in too late
alex_joni: Nah, you sneak out too early =)
but but but it's 7am! brunch maybe =)
I'm looking to create clickable building/cubical maps (from original CAD drawings, then converted to SVG maybe) that would show me where people, network drops, printers, etc are in the building, based on a db backend, and various overlays (electrical, network, telco, plumbing, etc). Zoomable would be cool, and being able to import/convert the CAD drawings with the least amount of pain would be nice. Any suggestions on what/where to start? This is all with
truncated: last that appeared here was "This is all wit"
This is all within a web browser btw.
I dont need a full GIS system, but something a little more elegant than just using IMGMAPS. just not sure of what that happy medium might be. Since CAM proggys "basically" transform CAD drawings into XY points, I thought this might be one potential way.
JymmmEMC: look at X3D
I saw a couple websites which have 3D rooms, where you can walk through them, look at things, click stuff, etc
alex_joni: 2D is all that's really needed, plus overlays
oh, I wouldn't know then :)
See, I "could" setup and use server side image maps to do what I need, then on the backend pre-generate the overlays and export as bitmaps, just not sure how to maintian it all
google maps uses AJAX to do the overlays, not sure if I want to go that route.
i was trying to animate a bunch of gifs, and the default viewer was ImageMagick. the anims were dead slow, even when delay parms were minimized. then i opened the same gif in ffox, and it files. 'd'eaux'!
now i can save opengl renderings to animations w/o dropouts :) like emc vismach anims
damn, like 3 different techniques all worked but i couldnt tell becuz IM bodger'ed 'em, all fine in ffox
anyone got an answer for this question http://www.cnczone.com/forums/showthread.php?t=53595
alex_joni: You know... since the CAD drawings are to scale, maybe I could just map the four corners of the bldg to lat/lon and use mapping application (or a subset) for the internal building layout and such.
alex_joni: Google Maps are really just on-the-fly genreated GIFs... http://maps.google.com/mapdata?latitude_e6=33945610&longitude_e6=4176576196&zm=4000&w=1250&h=1250&cc=&min_priority=2
Though, I'm a bit surprised they used GIF over PNG's... I'll have to look into the reasonign behind that. Probably lighter weight in processing.
4 axis vismach http://imagebin.org/14469
toolchain is vismach 2 xvidcap 2 series of gifs 2 gifsicle
thx, gotta up the capture rate
JymmmEMC: theres some java tools for auto-animating network maps
Heh... if emc can compensate for great circle, we can CNC THE WORLD!!! LOL
tomp: java == scarry shit (to me)
tomp: I can either 1) reinvent the wheel, or 2) plagerize all the PhD's that google has hired to create google maps (to some extent) for this project. I think I might do the later =)
I'd MUCH rather use SVG's, but it looks like even IE7 doesn't natively support them, damn M$
tomp: what are you working on?
a axis on end of Z, the motion is not what i will be doing
BigJohnT: emc doesn't presently have any hardware drivers for Firewire devices.
JymmmEMC: VisFlowConnect ask google to show you the cached version of http://citeseer.ist.psu.edu/738059.html
jepler: I didn't think so but didn't want to open mouth and insert foot without asking the experts
BigJohnT: that was 'A axis' not 'a axis' as in the thingy what rotates about X ( dang english )
tomp: I get na "out of order" at that url
ask for cache
tiny type at bottom of google hit
I don't think ormec is going to release the command structure either
(in reference to the question on cnczone)
tomp: Got it, but more of a ref page... http://22.214.171.124/search?q=cache:uxuNhA52Z28J:citeseer.ist.psu.edu/738059.html+http://citeseer.ist.psu.edu/738059.html&hl=en&ct=clnk&cd=1&gl=us&client=firefox
I asked ormec for info about one of their motors, it's like a state secret
right, i fiiger you'd like the animated network links
tomp: Ah, well all the cached links are "out of order" too =)
always wondered how citeseer ended up at Penn State
I believe I'm overthinking this proj a bit... KISS
JymmmEMC: mmm, hope thats temporary, but there are studies and code available for animated network traffic
tomp: ah, gotcha. One of the NetAdmins might be interested in that.
[16:45:48] <eric_U> http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.60.1822
I like the old format a lot better
but at least you can download the paper
Yeah, I think that's a bit of overkill for this. The only thing I considered (and quickly dropped) was an automated "You are here" based upon their network connection =)
Though, that may not be a bad idea for the VoIP system.
man I hope they have a good estop system: http://www.youtube.com/watch?v=_WBjijWwBI0&feature=related
[17:10:46] <eric_U> http://www.youtube.com/watch?v=oBWv7UoyPCc
we got too much stuff, when there's industries based on the crumbs we leave behind
wonder why the don't dump a box of hard drives in at once, just for doc that it was destroyed?
tomp: no kidding huh
funny narrator: http://www.youtube.com/watch?v=cwDSiXLrIf0&feature=related
whoa, where can I get one of these? http://www.youtube.com/watch?v=-oZd_P_jk7M&feature=related
any links to shaft couplers that are mm to inch? ( 10mm leadscrew shaft to 3/8 stpper shaft ) i only find inch/inch or m/m :(
hmm, i got 1.54" between shaft end, and at least .75 exposed, will make a 10mm-.375 hard bush female to male, then a stock in/in disk coupler
some couplings come in three pieces - hub, disk or rubber thing, other hub
you might be able to mix and match the hubs
tom1: see pg ~14 http://www.bsc.com.au/PDF-bsc04/6couplng.pdf
jmk, Jymmm thx!
wohoo! i got a 10mm to 14 mm disk coupler already, now to find a lathe for the adapter 3/8 to 14mm
tom1: take a look at huco.com they have alot of couplings from mm > inch
Duc050 is now known as Duc05
Duc05 is now known as _
Hello, anybody about?
Do you know anything about installing the NVIDIA drivers?
I just did a powerdown to insssstall another parport card, and while x is running ok, emc claims there is no GLX extensions.
Its been working for about 2 months before the powerdown
I have an ATI card in my laptop so I'm not so sure about that
that's pretty strange though
hopefully someone who knows more than me will be around soon
gene: what kind of paraport card is it?
Ahh, ok, thanks. I'm installing the 4th axis on my micromill, hence the powerdown. I'm still plugged into the mobo's own port, haven't been able to run emc & try the Rosewill dual port I just installed at all.
gene: if you haven't installed any sw/drivers for the new paraport card, power down and remove the card and try it again. If the problem goes away, there's a conflict between the paraport and video base address I'd suspect. Does your BIOS have PnP OS enabled or disabled?
Lemme see if dmesg has any enlightenhment to offer. Bear in mind the problem is the nvidia video card and GLX extensions. Dunno about pnp, back after I check that.
pnp=no. I rebooted in single mode and re-installed several times, installing what the NVIDIA installer bitched about, which was about 40 megs worth of gnome and x stufff, but its working.
Now I can make up a test cable & see if it all works. BEFORE I put the last 75 pounds of UPS up on that same shelf :)
It seems the fans I put in this box might not be happy with their supply voltage, the old 3 axis xylotex boxes fans were also 12 volt fans and they have been happy as a clam running on 18 volts for 2 years now, but I can smell these, or something when I turn on my power supply.
Tisn't chips, they are unloaded and cold ATM.
whats all this 'straight probe' then? as opposed to 'bent' ??
as opposed to some other path, like maybe an arc
G1 is called a "straight feed" in ngc lingo
tomp: you can probe with a rotary :P
my >guess< is a rotary swinging a probe would be using 'straight feed' even if the path were G91 X1 Z1 A1 , even tho an arc in motion is not in interpolation
i mean, even tho the motion is an arc, it is not circular interpolation
* alex_joni is off to bed
good night all