anyone recall that encoding method that only one bit at a time is changed?
used to be used in rotary encoders
or gray code
must be gray code :)
SWPadnos: hey you, get my email?
sounds plausible, no UPS should be necessary, at 10 seconds each, a day is enough
why 10s ?
button pressing, pausing for relays, detecting success ...
just a pessimistic guesstimate
that sounds awefully long
even so, a day is enough: 86400 seconds in a day
even faster if I could read in the eeprom =)
SWPadnos: Actually... the ds says it's a 3 wire eeprom
weeeeelll - if you attach the right scope to it at power-up, you just may see the data you want
that's what I was just thinking.... monitor the Rx line
clock and RX
nothing invasive on that
now the question is... how?
oh - it could be SPI with the chip select always on
I didn't look at the datasheet
I don't have a datalogger
my good scope could do it, it can decode SPI and I2C
I looked at it... and went uh huh
and even trigger on an address cycle
any way to turn a paraport into a nice lil data logger?
not for the speeds you'd need
parports are no better than 1 us or so per register read/write
(that's read or write, so an OUT followed immediately by an IN will take about 2 us)
it's a 1k eeprom
serial in, out, clk
I'm sure only a few bytes are read out
2MHz max, 1MHz typical
really paraport is that slow huh, never realized that
I have a BASIC Stamp
even worse, since it's interpreted
but for storage purpose, if it can just keep up
unless you can use the SPI port to receive characters, it's likely that it can't keep up
SWPadnos: Alright, half a tank of gas, 2 packs of camels, it's night and we're wearing sunglasses... how can I read the eeprom?
you can use an AVR or PIC with an SPI port
oh gawd... I have a atmel SDK and some avr 90's but don't know how to use them.
you could also do it via the parallel port, but not in-circuit, you'd have to remove the chip from the board (or at least lift the chip select, data and clock pins)
a lil too evasive I tink
if you know anyone local who has a nice scope, you can probably snoop the data as it's read out
if I do the manual thing with relays. the only part I haven't figured out is how to sense the LED is on/off
SWPadnos: yeah you
note the word "local" ;)
SWPadnos: note the word ONLY
note the word "probably"
unless I want something from 1930's
buy one from Halted and then return it ;)
would that scope I was interested in a ways back do it?
that cheapo LCD jobbie? I don't think so
you need something that either knows what the data is and logs it (like an AVR or something), or something that has a large buffer and captures the clock and data lines for a relatively long time
speed isn't the issue, it's record length
well, it could possibly work then, as long as you need <256 bits or so. you need at least 2 samples per clock edge, and there are two edges per bit
and all this assumes that the code is plaintext, which it may not be
well, I don't have a scope, so kinda scratches that
SWPadnos: are optos the way to go to iface with the keypad?
dunno, what kind of keypad is it?
so there are 7 wires coming out?
SWPadnos: the meypad, uC, and eeprom are all on the safe PCB
ok, is it a bunch of buttons that contact lands on the PCB?
yep, just like your tv RC
ok, and does it look like there are common wires going horizontally and vertically?
that's relatively good, you'll need to see which side is pulled high and which is pulled low
compare to GND?
it's a PITA to connect to that though, since you probably need to contact a row and column together, not to GND/VCC
yes, but with a scope, not a meter
ok, so I guess back to reed relays then
a meter could possiby tell you something, but it's pretty unlikely to be helpful
relays are OK, but you may need 12 of them
though I guess I could trace them back the uC and see what pins each connect with
12, one for each key
ok, 7 + a bus could do it
+1 for battery pack
no, you can't just connect to V+ or GND, you have to connect the correct row to the correct column
otherwise it looks like there's a stuck key
only one relay at a time would be on
well, ok two max
it's much easier to encode in SW than to add additioanl hw
what are you trying to do?
cradek: crack the safe
I'll bet that the way that keypad works is that the processor outputs a high (or low) on each of the 3 columns in sequence, and reads the 4 row bits when each column is energized
pretend to be a keypad?
grid of optos?
so you can't just pull a row high, or it will think there are 3 keys pressed at the same time
to do it in software you'd have to read the scan and output at the right time
with a grid of optos you'd just turn one on and wait a while for it to be read
you can do it with 7 relays, by having 3 column relays and 4 row relays, and connecting the "other end" of all the relays together
SWPadnos: that's not much different than just holding in a button
well, you need the opto to be energized at the right time, and an opto could cause problems depending on the scan rate
I'll probably have a harder time on the mechanical timing than anything else.
if you had an opto for each button, you'd just leave it on long enough to be scanned a few times
SWPadnos: ok, reed relays
yeah, opto or relay
optos or relays don't really matter, with the slight advantage to relays that you don't care about the polarity
and the slight advantage to toptos that they're quiet, faster, and have no contact bounce ;)
relays are harder to drive
that's true in general, but not really an issue for small reed relays
contact bounce is no worse than a keypad I'm sure
got a reply from Greg Jackson
carbon pill keypads are surprisingly clean (not perfect, just surprising)
he said there was a $5 handling fee, and he has lowered it to $2
an ebay auction that seemed to be playing the "cheap item, rape them on shipping" trick
[02:22:24] <jmkasunich> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=310074742697&ssPageName=STRK:MEWA:IT&ih=021
These eight I/O lines comprise Port A. PA6 and PA7 are open-drained pins with
pull-up devices whereas PA0 to PA5 are push-pull pins with pull-down devices.
PA4 to PA7 are also capable of sinking 8mA.
These six I/O lines comprise Port B. PB0, PB3 to PB5 are push-pull I/O lines with
pull-down resistor. PB1 and PB2 are open-drain I/O lines with pull-up resistor.
are those the ports the keypad is connected to?
in respect to your scanning method
at least some of them, others to the EEPROM I'd suspect. would have to trace them out first.
hopefully hte keypad connects to 7 port pins and that's all (except maybe some caps and pull-up/downs)
ok, lets assume that is the case.
They only thing I can't figure out is how to tell if the GRN or RED led is on/off. Use a phorodiode maybe?
you should be able to connect directly to the driven pin on the LED
is that a "good idea"?
or even better, find the processor pin that drives the transistor to turn it on, and directly connect there
sure, it shouldn't hurt anything - it's very likely connected to a 5V supply
but you should make sure of that before connecting to it, if it runs on 4 batteries, it could be a problem
(if it runs on 3, that's a different problem)
err - 2
then use a resistor ;)
ok, lets say a non evasive route.... would a photodiode do the trick?
probably even, but it may be a pain to tune, since you can't test sensitivity levels until tjhe light turns on
is there such a thing as a SMT "chip clip" ?
they're not exactly cheap, but they're readily available
< $20 ?
* SWPadnos looks at digikey
its either a 16 or 20p SOIC
in the $15-$40 range, it seems
uh - hold on a sec, I have an unusable search URL at the moment
[02:39:45] <SWPadnos> http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=923665-20-ND
OH, okey... I think I can get one of those from frys
that make it MUCh easier to physically interface to.
you should find that the keypad is 7 pins, the LEDs are 2 pins, and you can probably tweak the reset pin instead of removing power
the question is, what does it send down those 3 wires into the safe>
Yeah... nfc there... but as long as I can bypass the lockout timers from the external electronics. I'm good
oh - the lockout isn't on the keypad board - well, you may need another relay there :)
and removing the power and replacing it makes it instant
It probably is on the ext PCB. the battery connects to the kypb PCB
be careful there, repeated power cycles may have bad effects on the EEPROM. I've seen them get corrupted before
SWPadnos: Well, I'll have 100K tries to find out =)
the ds says 1M write cycles =)
at least it's usually the first byte only that gets screwed up. if there are 3 codes, then 2 of them should still work
SWPadnos: Ok, if I use the pins from the uC to determine which LEDs are on, how should I iFace that to paraport pin?
with some series resistors, since it's a 6V board
I shouldn't isolate it?
you may not be able to pull up with the PC
is it connected to mains power or big nasty motors?
the PC is
sure, but the safe isn't
not yet =)
stick it on a wooden bench and don't bother with isolation
what, the safe?
just don't drop any bare AC wires on it
it's battery operated, isn't it?
so, leave it on the concrete floor - no big deal
well, lets pretend I might fubar and set a pin as output
if you'r eusing series resistors, then you won't have a problem
all your outputs will be through relays
this would be input from the pins on the uC to the Q's to the LED's
isolating the safe from the PC won't help you if a wire falls on the parport connector itself (and an output is shorted to another output at the PC)
what's been fun lately in this chan... I haven't had much change to hang out in here these days.
those will swing from 0 to ~6V, so you use a - (picks number our of the air) 4.7k resistor
SWPadnos: ok, ok
SWPadnos: btw.. come pick up your UPS damnit!
all I want is my ferrule crimpers and ferrules ;)
SWPadnos: do you really thinkg hitting the RESET line would do in place of cycling power?
if the timeout is controlled by the micro then yes. if it's controlled by the internal electronics, then no
a-l-p-h-a, no clue, I've been traveling for a couple of weeks
SWPadnos, fun, where to?
Steven's Point Wisconsin and Dallas (Texas)
austin was kick ass, when I went to sxsw.
Austin is a lot nicer than Dallas
Austin was beautiful. Friendly city.
some very cool recumbent bikes here: http://www.velokraft.com/+products.htm
yep, not like Texas at all ;)
they're generally friendly, but not so much if you don't like country music and republicans
if I didn't live in Toronto, I would move like to live in Austin.
Austin has all sorts of music.
especially as sxsw
one of my sisters and one of my cousins used to live there, it's a nice place
south by south west.
I met my wife there actually
sxsw is the short form.
it's time for me to hit the hay, night everyone
I'm off as well.
Jym is now known as JymmmEMC
hi guys.... added a new gcode generator to http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Simple_EMC_G-Code_Generators
it engraves volume control bezels
thanks to BigJohnT's work
alex_joni: there are many changes in 2.3 version, when it will be available ?
ra3vat is now known as dimas_
micges: I guess the answer is .. when it's ready
but that means we need to stop adding new features, and start the testing phase
there are a couple of things I wanted to add, but I think they won't make it for 2.3
somewhere I readed that in about 6 months
could you list what you want to add ?
[10:28:19] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Alex_Joni
micges: then there are things from here: http://sourceforge.net/tracker/?atid=356744&group_id=6744&func=browse
micges: and here: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?EmcFeatures
alex_joni: and you want to do it alone ?
micges: when I get around to do it.. which lately didn't look very good
alex_jon1 is now known as alex_joni
[13:16:09] <skunkworks_> http://www.cnczone.com/forums/showthread.php?t=63013
"When i push stop button Z goto zero automatic" ???
that's a big safety violation
(but of course possible with HAL ;) )
bad idea bad idea
back from your trip?
just creep past my jig and awkward shapes as well
I saw one of these in Stevens Point: http://www.velokraft.com/-vk2.htm
now if he means automatic-quill-up-at-M2 I'm all for that kind of option
but if he means go up at abort/estop he's just wrong
hm, he explicitly says "When i push stop button"
yeah, but who knows what kind of stop button it is: optional stop, pause, estop ...
I guess someone could ask :)
SWPadnos: not me!
I'm doing it - I'm a glutton for punishment
WOO WOOOO (train wreck sound)
Keith Rumley did some work on this kind of stuff in his version of the interp.
chugga chugga chugga chugga
Machine logic. What happens when ???
that subject sprung to mind when cradek mentioned M2
_Poincare is now known as Poincare
There are a fixed set of assumptions made in the interpreter.
His notion was that these assumed behaviors should be variable and read from the ini file.
I tend more to favor making them variable as a part of task planning.
yep, something like the card stack idea we've discussed, with the cards coming from a config file
But then we did quite a bit of thinking about that some time ago.
I would make them variable by having integrator/user defined g/m codes.
G/M codes would also be defined by a sequence of primitives
with configurable preconditions/postconditions
We probably already have most of the primitives we need. Turn an output signal on/off, sense an input signal. Coordiated move to a location in X, Y, Z at a specified rate.
At two levels. First, I need to add the user defined g/m code stuff (and some more introspection). Then the integrator can "simply" program what he wants in gcode.
sure, with the appropriate safeguards against circular definitions
My "plan" is that within the user defined codes, the original ones can be accessed -- say by using G-1 to mean use the original defn of G1.
IMO the original definition ought to be default if G! is used without any parameter
Hmmm. The disk seems to have died (at least partially) on my machine running EMC. I guess it's someones way of telling me to upgrade to EMC2.
Lerman: sounds like I can simply change my system so M6 <- (G30, then M6)
Or, redefine G1 X.. Y.. to be original G1 xsin[theta] ycos[theta] to use rotated coordinates.
I'd say that's a kinematics function
you wouldn't want the DRO to look funny ;)
yeah, seems like everything would work goofily if you did that
Well, if it were a small rotation that amounted to a few thousandths, it wouldn't look to funny.
(ie, I program G1X1Y0 and it moves to X0Y1 because it's rotated 90 degrees ...)
to -> too
The 90 degree case is easy to (mentally) deal with. The 45 degree case would be awkward.
I would be very concerned when my X and Y axes appeared to swap with each other ...
(more or less)
SWPadnos: add some Z or A in there
...another topic... If I run from the CD, can I access my hard drive so that I can look at my old .ini file?
Does networking generally work?
yes and yes
I would like to start with my old .ini file when I do the latest EMC install.
(I would hate to have to retune my servos.)
you can also install to a different hard disk, then plug the old one into the same machine and read files from it
Hmm. I might just do that. Since the problem is that my old drive didn't fsck too well and I can't boot it.
try the live cd -- I think you get icons for detected drives. if that works you can copy the config to a usb flash drive or whatever
Thanks guys. I'm downloading now. 1 hour 29 minutes remaining.
Lerman: whart's it mean when (debug,#xxx) prints ###### instead of a number?
maybe xxx is ##### ?
I believe the number is out of range. I forget what format the sprintf uses. Is it #xxx or #<xxx>? If the later, it could be that the named parameter can't be found. I'll take a look at the code.
it's a number
lerman: that was exactly right. thanks.
Which was exactly right? Looking at the code, it puts in the ##### when it can't find the paramter. For whatever reason.
that my number was out of range
xxx was out of range?
If the param number is negative or greater than RS274NGC_MAX_PARAMETERS.
greater than or equal to.
alex_joni: is that off the top of your head or did you look that up.
Good news is that I've now downloaded 9%. Bad news is that the new remaining time is 2 hours 20 minutes.
As I make progress, it keeps giving me longer estimates. Does that mean that the last .0001 percent will take almost forever? :-)
it's like lightspeed
Lerman: depends how fast you are moving away from the download source
* alex_joni grins
Too fast, it seems.
cradek: re contactor with nc contact. i buy 'aux' contacts that snap onto the sides of the contactor. common on std frame sizes. They are mechanically driven by an extension of the solenoid. Usually a single aux has both nc and no contacts. like <15$ for the Aux.
tom1: thanks for the hint
re: integrator defined g/m codes. the system i'm working on now has all m codes as text files. but it also preparses all program to be run (except in drip-feed mode for long progs), The longest prog i'
ve used parses faster than i can press enter
cradek, you usually remove the side stickers to see the actuator hole and tapered dove tails the aux slips into
bbl ( ubu-update requires reboot:(
BigJohnT, hi... added another gcode generator http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Simple_EMC_G-Code_Generators
oooh pretty picture http://antwrp.gsfc.nasa.gov/apod/ap080818.html
lewin1 is now known as lewing
parport b/o board with & w/o optos http://www.sunrom.com/index.php?main_page=index&cPath=96&sunid=jajosb7bbuut2u3hm2jcibh9n4
hmm.. 60$ seems a bit much
yeh, but then its done too, so depends on how handy/blind you are ;)
I'd rather pay 120$ on a gecko, than 60$ on this :P
eh? oh yeh better quality item
use any 'linear actuators'? any prefered vendors (ina/skf/thk/rexroth...)?
any of the above :)
usually skf and rexroth
thx ( leaning towards rexroth, customer already uses it so it likely will work :)
good night all
hm, this guy ended up going down to a paltry 2in/min for cutting out circuit boards with his wolfgang engineering spindle: http://www.cnczone.com/forums/archive/index.php/t-41190.html
I'd hate to go that low -- cutting out the board could take more time than cutting traces!
still sounds like an good trace cutting spindle ( and a PC router can be used to cut the board out )
porter cable router
I'd rather do it all on one cnc machine though
right, it just seemed easy to chg spindles on those
when it's cutting out well, the results are very good -- nice clean edge. but it's true that the speed dips
chg the motor? move the drive belt to the big motor?
changing the motor might improve things some
we dont got no stinking gear shift, we got motor swap ;)
I try to get a pcmcia to parallel card to work with the live cd but i do not
right, that will not work
you need a real parallel port, either on the motherboard, or on a pci card
what is the problem with pcmcia ?
for EMC to run stepper motors it must access the parallel port directly and in realtime. pcmcia does not allow this.
thank you for the answer
I have found some mispellings on the french documentations pages how I can report them ?
the primary editor of the french documentation is Francis TISSERANT <tissf AT free DOT fr>
ok i will contact him :)
that's great, we always want to hear how to improve the project
so I have something that my mill cut which should be rectangular, but I measure the two diagonals as 4415 and 4428 mils respectively. is that enough data to figure out how skew my X and Y axes are?
Your help the other day got me going with the joypad.
oh great -- glad to hear it
jepler: not really?
what's next to configure on your machine then?
i mean if you do the trig you might be able to fudge it pretty close.
as long as you know how it was sitting on the table before you took it off
oh the cuts were entirely axis-aligned .. two edges cut by pure X move, two edges cut by pure Y move
do the trig, make the adjustment and see what happens
it's not the canonical test, but it should get you close enough for government work
how stable are the measurements across the diagonal?
:-D Well, I'm not 100% done with the joypad yet, but the basics are working. Trying to get fancy now with the variable jog speed buttons.
toastydeath: it's a thin piece so I can't measure the diagonal at different depths
well i meant like, in the same spot
if you take the same measurement 5 times, do you get the exact same answer each time
within about 2 mils
which is pretty good compared to the total amount, but not very big compared to the differenec
do you have an angle plate
or any other 90 degree angle reference
then do eet
hm I don't trust my trig
also be sure to measure the leg lengths
like, the bottom and top, and left and right sides
because you might have taper as well
and that will screw you up if you don't take it into account
then you'll have a good idea of what adjustment you have to make, and once you line up a parallel and bolt it down
make your adjustment
beats me how I would just skewness on this plastic monstrosity anyway. maybe try to correct it inside emc with kinematics.
there's no way to loosen some bolts?
is it a bed mill?
[23:09:10] <jepler> http://axis.unpy.net/files/01188441458/img_7045-medium.jpg
a router I guess you'd call it
look at how the wayshafts attach to the machine and see if there's any way to get play in either the X or Y
that's all I can think of
but i'd be a little careful with the X axis, i would be concerned about screwing up the Z axis alignment
but that seems like any adjustment might make a difference to Z
I'll try not worrying about it until I see how double-sided boards come out
a toe clamp should adjust that once you nest the rest ov the mahine base
or just shim in/ shim out... with the end sections