High all.... what an awsome Hang gliding day... ; )
para gliding.. not so good...
you take a camera with ya, or have someone shoot ya!
sorry.. i have none.. chk the gallery at www.uflyontario.com
how long are you up for?
both.. ive flown withthe video camera.. but that show where you screwed up to others... so i stopped.. lOL
today.. avg flight was 20-25 minute... 2700' max alt for the day..
always wanted to go hang gliding
it was blowin' 30 mph though... a VERY technical day...
helps to be an instructor..
not in front of me...
for the one(s) that have died... what were the reaosns?
or on my shift as i say...
flying too close to the planet..
equipment? weather? operator?
MOST are pilot related... remainder are usually wind and obstructions
object fixation is a killer...
power lines, trees, etc?
power lines and wind is a nasty combo...
yeah... dont hit the tree .. dont hit the tree... dont.. ahhh the tree...
what, ppl dont know better to go to the big empty field?
we had a field with 20 acres on 1 side of the tree line and 20 on the other.... 1 stray tree out front.. near the clubhouse... i wahtche 10 ppl in 5 yrs wrap gliders in that tree
it was a joke.. im on radio.. yeallin AWAY from the tree... turn away from the tree....
and i acnt even call them an idiot WHILE they are haning from the tree
why not?! lol
we had a russian Mig pil;ot come learn to HG
I'd have camera in hand and offer up 8x10 glossies
what a piece of damage control required there
he ALMOST got the wills wing 195 thu afore mentioned tree line...LOL
hot air baloon $190 per person =(
HG tandem for 150 canadian
or i can use some pull and get ya a free one if your friend pays..
Guess all the ones around here are up in Napa (wine country)
but you pay for the scenery...
SkunkWorks: ok but I don't know what I'm doing
I even went to Jon E's tuning class but I have nfc
want some help?
not sure how you could help without seeing my graphs?
you could post em
besides I can offer general comments
first question: what are you using as a stimulus?
jogs in emc, or siggen, or?
I tried jogging and then I tried step changes (sets 0, sets 0.1)
I read using step changes is best (but jon E used jogs in his class)
I used to believe in steps, but not anymore
speed steps maybe, but position steps just tend to saturate things
i dont understand how this can be such a black art
its not really a black art
its just that the main input is visual, so its hard to write down
maybe it's just an undocumented art?
that was the thing with the class
`from looking at this graph I see I need to increase "I"'
of course he was right
but that didn't help me
but he probably couldn't tell you why
yes I even interrupted once and asked him to explain why he said that - what features was it on the graph?
I guess I left just as clueless
for me, step one is to get the scaling right
step two is to set limits
step three is to start messing with it
iterate until done
maybe I should wait because I don't even know my scales yet
I don't even know my encoder ppr
good, lets start from the beginning then
you have your motor hooked up?
have the encoder counter hooked up?
ok, note the encoder count, turn shaft one rev, note count again, subtract
if it doesn't count, you got a problem
if it does, you know ppr
well, sort of
I definitely know it's counting
it might be hard to tell the difference between 500 and 512 counts per rev unless you are carefull to turn it exactly one rev
put a halmeter on encoder.X.counts
no index pulse?
I think it's 124? which means the encoder before the /16 is 1984?
but not impossible I guess
I turned it 10 turns and got 1242
what is your belt ratio going to be?
and your turns per inch on the leadscrew?
I think I'll end up with 3:1 (I think I will have parts for 2.5, 3, 4)
the leadscrew is 1mm/rev
do you intend the ini file to be in inches or mm?
the leadscrew is 1/25.4 in/rev haha
probably mm will be most natural? I don't really care but I was thinking I would have a mm ini and run g20 programs most often
I wouldn't let the screw pitch decide the ini
what units do you think in?
for cnc I seem to use inches, for manual stuff, especially small, I tend to use mm
I'm moderately comfortable with either
well pick one
if you want a mm ini, lets do that
let's use mm since I want to still have the handwheels
that seems most natural
with lines you could probably open it up and count the number of lines
124 counts/rev * 3 motor revs/screw rev * 1 turn/mm = 372
so that will be your input scale
I'm guessing the actual number is 128 counts/turn, but you can always fix that later
or are you using some kind of divider circuit?
he is, /16
ok, you got your input scale set in the ini file?
you can also setp it into the encoder module scale parameter
yes I'm changing the other things for mm sanity
maxvel 25 maxaccel 600
those motors are rated 19V, right?
what is your supply voltage?
yes, I'm using 12v now
when I build a supply I'll shoot for 19
ok, we need to figure out the motor's speed constant
at 5v I got ~ 1300rpm
maybe 1400, hard to tell exactly with my "tools"
does that tell you what you need?
I was gonna suggest a ddt block on the encoder position output, feeding a halmeter
and then spin it with a fixed 12V or 5V or whatever is convenient
ok I'm firmly in the mm realm now
ok, lets assume that you got 1300 RPM with 5V
ack, I don't think it's 384 or 372
I hope my circuit isn't fubar
the scale I mean
what are you seeing?
I put a pointer on the shaft
I guess I have a tuning that works somewhat
when I move a integral number of mm, the pointer moves
IOW, not an integral number of revs?
hmm, when I move back, it's still off
so I must have a hardware problem
can you disable the divider and route the encoder signals straight to the counter?
(straight to the parport)
not very easily
so it could be either an encoder issue or a divider issue
cradek: what about with the etch-a-sketch board?
it's a noise problem - by playing with the motor (it's fighting with me to stay in place) I can get it to move
turn off the motor power supply (or disconnect a motor lead, if the same supply runs the encoder)
try turning it back and forth, see if it loses counts
it seems to count right
I turned it 6 turns, the screen said 2mm, I turned it back 6 turns, it says 0
if it works ok with the motor disconnected, time for a sprinkling of 0.1uF caps and prayer beads
I already sprinkled them...
can you post a schematic and board layout?
[02:07:39] <cradek> http://timeguy.com/cradek-files/emc/brd.png
[02:07:39] <cradek> http://timeguy.com/cradek-files/emc/sch.png
hmm, translation required
whats a 4516?
duh, up down counter?
yeah it has u/d and "clock" inputs
it's presettable but ignore that
so JP1 isn't really a jumper, its the encoder input connector
right it's a 4-screw terminal
clk on the 4516 is edge triggered?
bzzzzt - you lose ;-)
edge triggered circuits are very noise sensitive
can you put a few uS of filtering on the clk input?
say 1K in series and 1000pf to ground
is the 4516 a cmos part?
good, ttl input currents would make a 1K resistor a bad choice
for now, tack the 1000pF to the board as close as possible to the clk input, and install the 1K off board (unscrew the wire, screw one end of the resistor in its place, tack solder the wire to the other end)
whats the 4030 all about?
looks like a funny and gate symbol
software selectable divide ratio
* cradek kicks eagle
oh, binary to quadrature conversion
fenn: no, it's wiring-selectable
I wonder if you can get glitches on the xor output
when both counter bits change on the same clock edge
you get a race
yah get rid of the xor gate
it doesnt do anything
it looked ok on the scope, but hard to tell.
the encoder counter is sampling, not edge triggered, so it most likely wouldn't see the glitches
fenn: sure it does - it regenerates the quadrature
jmkasunich: that's what I was thinking too
except once in a billion counts, probably when you are machining something out of unobtainium
oh I don't have that around here - it's hard to find
ok, then don't worry about it ;-)
seriously, I'm probably being anal, the xor isn't likely to be a problem
ok, what powers the 4516s and the 4030s and the encoders?
didn't see a vcc pin on them
that's why I put caps everywhere
* jmkasunich kicks eagle
the schematic hides it
it's the blob on the left
ok, C2 is the bulk cap on the motor supply
it hides the 7805?
C7 is a high freq cap on the motor supply
C1,C3,C4 are for the 7516's
is ic6 the 7805?
the ground plane is pretty chopped up
yeah it's hard to get a good ground plane on a single-sided board
whats the shortest path from the ground pin of JP1 to the ground pin (8) of IC5?
looks like left from Jp1 gnd between pins 14-15 of IC5, up to middle pin of IC6. around the right of JP1 and down to pin 8 of IC5
03jepler 07HEAD * 10emc2/src/hal/classicladder/files.c: remove accidental assignment
or similarly around to the right
03jepler 07HEAD * 10emc2/src/hal/utils/m5i20cfg.c: fix warning: prototype for getline() is only available for _GNU_SOURCE
not horrible (not as bad as I thought at first)
around the right?
03jepler 07HEAD * 10emc2/src/ (Makefile configure configure.in): first stab at making MATHLIB and MATHSTUB work again
under ic1 then around the right and bottom edge of the board
03jepler 07HEAD * 10emc2/src/Makefile: stupid typo
that doesn't get you to pin 8 of IC1
you have to go through the jumper wire
I guess its pointless for me to sit here and dissect the board layout
its done and milled
yes save your energy for dissecting my board layout :)
cradek: do you have the caps and resistors to put those filters in?
still working on the mechanical parts of that task
so, 24VAC is 34VDC..
* fenn bonks himself
rolling caps from aluminum foil and oil soaked toilet paper?
pros use saran wrap
fenn: sqrt(2) and all that
03jepler 07HEAD * 10emc2/src/Makefile: fix placement of MATHSTUB
* jmkasunich us confused by "mechanical parts"
yeah, i knew that, it just didnt quite sink in until i measured the power supply i made
I could send a photo, but trust me instead
the wiring from the encoder is that flat ribbon stuff with crimped-on ends
now I'm even mode curious
I'm replacing it with wire
]$ nm siggen.o
000000000000048c T fputs
cant you just split out one of the wires and solder a resistor inline?
no, not without replacing the wire
jepler, that means its using glibc libm I think?
03compile-farm 07BDI-2.18 (2.2.18-rtl3.0) * 10emc2head/: build FAILED ; see http://linuxcnc.org/compile_farm/emc2head_slot2_log.txt
cradek, when you get to the adding caps part.. make sure the ground end of the cap goes to the 4516 ground pin, _not_ the header ground pin
ok the wax dip on my cap is almost dry
03jepler 07HEAD * 10emc2/src/rtapi/mathstubs.c: use the rtapi_math.h header
ok here we go
03compile-farm 07BDI-TNG (2.4.18-rtai) * 10emc2head/: build FAILED ; see http://linuxcnc.org/compile_farm/emc2head_slot3_log.txt
they make 1000pf caps that are polarized?
I doubt it
oh i misread i think
polarized caps under 1uF are rare
under 0.1uF pretty much unheard of
I'd like two please... http://www.atcsd.com/lrad.html
sigh, it's still counting wrong
03compile-farm 07BDI-2.18 (2.2.18-rtl3.0) * 10emc2head/: build FAILED ; see http://linuxcnc.org/compile_farm/emc2head_slot2_log.txt
"LRAD can communicate in various languages via the built in MP3 player or Phraselator."
I have encoder -> 10k -> 100pF to ground -> CLK
but it still wanders when I mess with the motor shaft
wanders as in when you let go it doesn't go back to where it was?
yeah if I turn it gently one way it ends up rotating the other way and stays there
what position does it report?
bad tuning could explain what you are seeing
the position on-screen stays the same
we need to distinguish between bad feedback and bad control
is that commanded or feedback
it damn sure ought to change when you force the shaft, even if it does come back later
yeah it does
as soon as I let go it jumps right to the commanded value
but the shaft may have turned
but the shaft jumps back to someplace else
let me see what looks noisy here
maybe it's the parport pins
could be anywhere - not necessarily the counter input
true, but edge triggered stuff like the counter is the most likely
I don't think the approach to quadrature dividing is really going to work. http://emergent.unpy.net/index.cgi/sandbox
(excuse the ascii art)
you're even riht
thing like that are why I like to sample quadrature with a state machine instead
argh here too, I've worked with quadrature and thought about that exact problem many times in my career, I should have spotted it right away
ideally, someone (me?) would have spotted it before I built this whole thing
I guess it would have had to be you, since the rest of us didn't see the circuit until now
cradek and I had talked about it
jepler: I shouldn't have said that, sorry
I don't suppose the 4516
* fenn highly recommends solderless breadborards
4516's are socketed?
03compile-farm 07BDI-2.18 (2.2.18-rtl3.0) * 10emc2head/: build PASSED
they are? great
unplug them and jumper the signals straight acriss
oops, forgot about the xor
fenn: I tested it, but only with the motors rotating nicely
03compile-farm 07BDI-TNG (2.4.18-rtai) * 10emc2head/: build PASSED
unsocket that bastard too
hehe this is always a necessary feature on any device you build: "Attention-getting and highly irritating warning tone for behavior modification"
jmkasunich: now let me know if they *work*!
about to give it a try
ok divider defeated
now I get nice oscillation
make the scale number 16x bigger
i wonder if i should be following along.. i have all the parts
* fenn goes to get more chocolate
ok this works much better now
I think the encoder is 500 line (scale is 4*3*500)
edges at 127kHz?
jepler: if you say so...?
that's what units said
127kHz at what speed?
base period is?
that's a lot of counts
I can do at least 20usec
so 50000 counts/sec absolute top speed
25000 with margin
not going to cut it
need a better divider
you got 1300 RPM at 5V?
so about 5200RPM at 20V
86 revs/sec * 500 lines * 4 = 173KHz
you want 25KHz
so divide by 8
I had assumed 1000 line so I did /16
do you do PIC'ish stuff
or AVR, or similar?
yeah and here's where jeff is going to point and laugh
using a micro to divide?
yeah he wanted me to do it that way, but I thought it was stupid
oh, hes not laughing at the idea, hes laughing because he suggested it and you said no
how many pins do your favorite uCs have?
we need 4 per channel
I have a wide variety
the fastest implementation is a LUT based statemachine
do those chips let you implement an array of constant bytes?
6 bits of current state + 2 bits of input, index into 256 byte array, get 6 bits of new state and 2 bits of output
this sounds like what's in encoder.c
use the same LUT for all channels, just keep independent "current_state" vars
I could just copy that code
yeah, except you make the table bigger, cause you want it to repeat after 8 quadrature cycles not one
I didn't like the whole microcontroller approach because you'd have to worry about it being fast enough
if you really hate using a micro you could implement it with a 256x8 rom and one 74HC374 per channel, plust a little counter
of course that would take 10x the board space
re speed: you'd be insane to code it in anything but assy, the code itself is so simple
is there a better way to get position input into emc other than the encoder module?
how tight a loop can you write that merges two input bits with 6 bits of state and does the lookup
thru a parport? I doubt it
you could count in the uC, but then you have to communicate a binary number to EMC
that is probably more demanding of speed
those little uCs run at 20MHz or so, right?
16 I think
are they riscy, one instruction per clock? or more like the 8051, 12 clocks per instruction?
hmm, the cheapies I have are 10MHz
so 5-10 mips
assume the loop is 20 instructions per channel
at 5 mips thats 83KHz
not gonna hack it
at 10 mips it would be ok
(or if you could do it in 10 instructions)
found some 16MHz parts
do they have 6 input and 6 output pins?
does the arch have bit access instructions, or do you have to mask and shift?
I'm not sure, I write C...
let me look
you could try it in C and see what you get
state &= 0xFC
state |= phase A shifted to bit 1
state |= phase B shifted to bit 0
state = LUT[state]
outputA = state & 0x80
it has setbit/clearbit
outputB = state & 0x40
what about getbit to read the inputs?
skip if bit set/clear
state &= 0x0fc
skip if phase A clear
set bit 1
skip if phase B clear
set bit 0
state = lut[state]
skip if state.7 clear
skip if state.6 clear
repeat for two other chans, using two other state vars
then write output to a port
(gotta clear output at the start)
looks like 10 instructions per chan, plus a couple for writing the output and clearing the output byte at the start
say 40 clocks total
at 16MHz, you can loop every 2.5uS, 400KHz
the division is built into the lut somehow?
thats one where a pic (state diagram) is worth a thousand words
visualise a circle of 32 states
whichever one you are in, based on the inputs you either stay there or go forward one or back one
the state numbers are arbitrary, they don't have to be sequential
ok I see what you mean
so make them in gray code
and output the two highest ones - presto quadrature
seems you didn't mince words in your recent reply
Can't stay long. I'm much to wound up.
jmkasunich: Mystery Guy
Now you see why I must not be leadership.
just realized the state &= 0xfc is unneeded, since the value in the lut can have zeros in the low bits
rayh: wanted to ask you who mystery guy was at fest, but I forgot the url for the pic
rayh: I think you're fine, don't worry
Oh. Got a call or two about him today.
I forgot his name.
him = mystery guy? or paul?
mystery guy then
Mystery guy set up the mazak and helped DaveE with speeds and feeds and such.
yeah, he seemed very knowledgable about the machine
lol nobody knows his name at all
several people say the pic and wanted to put a name on the face
I'll have to ask Roland to go through his list and find the name.
we should put a captioned copy of that pic somewhere
Ah, HERE we go... http://www.electronicsam.com/images/KandT/cncworkshop/cncworkshopemc.jpg
jmkasunich: thanks for your help - looks like I'll be making a new board and stuff
Dopey, Grumpy, Doc, Happy, Bashful, Sneezy, Sleepy,
oh here comes snow white
jmkasunich: should always plan to throw one away, no big deal
sniff, sniff. Someone said snow white?
Skunk Works == SW. Snow White == SW.
on the plus side, there will be fewer chips, so less board stock, and maybe a better ground plane
seems like a waste of a good micro
if you are gonna lay out a new board, lemme take a look before you start milling
might have some suggestions about routing and noise
fenn: they just cost a couple bucks, cheaper than glue logic
cradek: problems with your board?
skunkwrks: my quadrature divider circuit is bogus
but the driver part is fine
what was the problem - 74192 would do a cheap divide by 10 :)
looking at the existing schematic (which will change I know) the power circuit to the right of the DB-25 and the quad ckt to the left should have independent grounds, tied together only at the connector
the regulator is the only other tie point
regulator ground should go to the quadrature side
and regulator input might (being anal here) go thru a small resistor or ferrite bead, with a cap on the regulator side and going to the regulator ground
a pi filter
I had a successfull milled board tonight. found out though the machine is quite a bit out of square. .001 or more an inch. so you really see it when the board is flipped. It willl work thoug.
only with independent grounds
SkunkWorks: flip it the other way then :)
:) would have to think about that
main problem is getting the cam scripts to output rotated by 90 degrees
just swap X and Y in your output
or fix the problem :)
fix what problem? flip and rotate is the proper way to do it, i wot
keeps the same three points in contact with your fixture
er, two points i mean
if the machine is square - it isn't a problem. the holes at the bottom of the board lined up perfectly - as the hole got further away from the registerd edge - they where off.
(can't believe it was that far off) but we had never really checked it.
kinda amazing you can do circuit boards with a huge machine like that
is it something that can be adjusted back into alignment?
here is the scary part - it is only driven on one side of the x and y axis. that will be the next upgrade - duel lead screws.
later guys. keep up the good work.
how big is the machine? is there a photo somewhere?
skunkwrks: belt linked or two motors?
[04:01:27] <skunkwrks> http://www.electronicsam.com/images/KandT/DSCgantry.JPG
belt linked - easiest
Here's an interesting question: when patching and configuring a realtime kernel, just how much stuff do you *really* need to leave in?
lol - all tables get covered in crap, even machine tables
dont remind me
asdfqwega: when in doubt, rip it out!
when something breaks, put it back
I'm wondering just how much really need to be there
repeat until satisified
probably not much
will compile a lot faster without all the extra crap
Who's going to run a RAID array on a cnc machine?
more importantly, are _you_ going to run raid on you cnc machine
Well, there is that
Still, I'm curious about the range of computer hardware people use for emc
from massive overkill to massive underkill
i'm one of the underkill people
"rapid rising from affluence to poverty"
"Yeah, we got one of the spare Compaq Proliant's with SCSI raid and token ring networking running our Sherline."
Heh, I've just made a .deb of kernel 18.104.22.168 patched with rtai, and I started with the 'defconfig' option, and re-enabled the stuff I wanted - and the .deb is almost a third of the size of a regular kernel .deb
03jmkasunich 07HEAD * 10emc2/src/module_helper/module_helper.c: added a couple more modules to the whitelist - they're used by an older version of RTAI
sorry - just noticed it was 2mb http://www.electronicsam.com/images/KandT/cbottom.JPG
the holes look pretty well centered to me
looks really nice
a couple spots in the upper left look a bit weird
here is the bottom ;) http://www.electronicsam.com/images/KandT/ctop.JPG
I mean top
the bottom was milled last and drilled. so that looks really nice
I've had some turn out like this too
I notice you can't put traces between dip pads - my traces are MUCH finer
notice the bottom edge of the top of the board lines up nice.
skunkwrks: whoa, what *IS* that monstrosity of a machine?
yes it does - definitely looks like you have a rotation problem of some kind
it is a mizomex film step and repeat machine - converted to have a z axis. :)
if I set a square on the outside lines - it is visbly out of square.
for trace size comparison: http://timeguy.com/cradek-files/01144545838/stepdriver2.jpg
I wanted fatty pads.
(I didn't lay out this board, only milled it from someone else's design)
what do you use to mill the big clear areas?
not the 60 degree cutter
often I don't bother
no broken cutters tonight though - thank goodness.
and eagle automagically generates g-code for both the skinny and fat cutters?
or is there some postprocessor?
jepler and I wrote the output program, eagle has a scripting language
I am using some script someone wrote.
skunkwrks: are you using ours?
skunkwrks: using eagle right?
not yet :) it was something I found with a google search.
oh well - time to go to bed. maybe do some soldering tomorrow. might need help setting up freqgen if I get stuck.
skunkwrks: I've got some encoder dividers to sell if you want them
cradek: I thought the divide by was tested?
skunkwrks: for some value of tested, they were
just not fast enough?
skunkwrks: the scheme doesn't work when the motors reverse
ah - oops.
I'm an idiot for not noticing it
I never have issues like that :)
could be worse - it was cheap parts and a homemade board
ordering $100 worth of boards and waiting a week would have sucked
now the hard part will be not giving up
just happy I didn't break any bits - and the register pins worked great. just need to square the machine.
I don't see the holes for registration
sould be able to do a 13X13 circuit board pretty easy.
3 1/8" pins registerd the edge of the board
I will take a picture tomorrow.
not that I will ever make a 13x13 board.
But I like the option :)
* fenn goes hunting for a discrete gate drive circuit
oo cool i can use my micro's charge pump both to generate the high side voltage and as a safety feature
fenn what you workin on?
all n-channel mosfet h bridge
need to generate a voltage above the main power rail to drive the mosfets
whats the voltage/current rating?
(assuming your making a stepper driver)
at least 100W peak 24V
DC brushed motor
pseudo servo driver?
no its a real servo
the loop is closed in the computer
i figure i will implement step/dir if there is interest
but i find that really distasteful
hogs cpu cycles and give crappy results
I can't comment on that... I'm running steppers
and haven't looke/dfound enough old tredmill machines to scrape out
it makes sense for a cheap stepper interface
but all this PLL stuff is ridiculous
just send a velocity command out and get position feedback
or whatever combo you want
Yeah, sounds cool. It's a bitch if my steppers stall and have to relocate it's position
But, I think I've figured that portion out (at least till I get some geckos)
[07:00:08] <alex_joni> http://www.bash.org/?564197
mornin mr Joni
nice lesson learned there :)
oh.. btw, did I tell you I got a drafting order?
to join them military?
you did? good for you!
not really.. :)
alex_joni: WE can take care of that!
Jymmm: it's done
alex_joni: Nah, we'll get you enlisted yet!
if you insist..
gonna go die for US oil eh?
well you're probably safe, i hear they aim for the civilian contractors
[07:15:16] <Jymmm> http://www.bash.org/?639861
some people are soo stupid
I remember seeing an auction for wireless lan cable once ;)
some people actually believe the stuff on bash.org
Some ppl actually buy air guitars off ebay too
probably the same people
or pay $20 for a cd of EMC
well to be fair it was $3
we should prolly send him an updated version
fenn: it was an Ubuntu LiveCD
Oh I found the guy, he's 40 minutes from me
with emc packages? did you ever even release that?
alex_joni your release is for sale on ebay too
Jymmm: I know.. and I'm ok with that
is MPSA06/MSPA56 a 'normal' transistor? i've never heard the name before
alex_joni for $85 USD ?
Jymmm: you sure?
I would be glad if it was worth so much.. lol
alex_joni Yeah, I bet you care now @ $85 huh?!
I love this one every time I read it.... http://www.bash.org/?405221
jymmm i dont know why you think jameco is expensive, they're at least 2x cheaper on everything i've looked at so far
than everywhere except futurlec (which takes 1 month to ship from thailand)
fenn: because doing it yourself could be cheaper if you had equipment for it
er, i guess
hard to beat a couple cents for a transistor
One example... $11 for a 12VDC@500mA walwort http://jameco.com/webapp/wcs/stores/servlet/ProductDisplay?langId=-1&storeId=10001&catalogId=10001&productId=162996
hey thats 6W man
* fenn cackles
who buys those things anyway
$50 for a 200W PC PS http://jameco.com/webapp/wcs/stores/servlet/ProductDisplay?langId=-1&storeId=10001&catalogId=10001&productId=172822
will a pc even start with 200W nowdays?
more like with 500W
COme on... $2.49 for an orange LED http://jameco.com/webapp/wcs/stores/servlet/ProductDisplay?langId=-1&storeId=10001&catalogId=10001&productId=131895
fenn maybe your just lucky on the IC sode of it
if you bought 1000 it wouldnt be that expensive
$1.79 EACH @ 1000 <---- that's expensive
more like a $0.0x something
No, it's $1.79 for EACH LED why you buy 1000 of them
then your dealer is shitty..
That's why I buy my LED's from HK =)
the leds at my dealer is about
$20 icluding shipping gtets me 50
those guys are real ripoff artists
if you buy 100 they are 0.72$
$20 icluding shipping gtets me 50 ($0.40/ea)
and they're white leds too
yah well what if you want an orange led? :)
$0.20 from radio shack
well, radio shack puts elfa to shame when it comes to ripping people off
usually, but once in a while you'll find good stuff there.
i'm not listening to you anymore jymmm
fenn What? about RS ?
Hey, there's a RS Surplus store here =)
is that thoose guys selling the cheap servos before?
hm, I'd never buy leds from thoose guys.
anonimasu what RS you mean?
RS == Radio Shack
surplus center sells the motors, not RS
RS is a retail chain of electronic stores operated by the clueless sales ppl.
ah thoose guys
Yeah, owned and operated by Tandy Corporation, Texes
Yes, most of their stuff is crap, but once in a while if you dig thru the shitpile you'll find somethign good.
shitpile being theri whole product line.
a quick question, anybody know which axis tcl file to start for remote GUI?
EvertL is now known as Evertlammerts
bash: cls: command not found
is anybody here?
hi there :)
Usually there is always someone answering stuff you might need to find out
I was asking if anybody knows which axis tcl file I need to start for remote GUI
it's been awfully quiet :)
So you kan just write your questions anytime and somebody might wake up
I'll have some more patience the next time ;)
That one I don't know...
alright, doesn't matter... I'll keep digging and checking this IRC
do that, have you downloaded the emc2 documentation?
yep, most of it. I still need to get a good checkout from the new docs on cvs though
good you mention it, i'll get on it right away
* alex_joni is on and off
alex_joni: which axis tcl file I need to start for remote GUI?
* Evertlammerts keeps on checking back here every now and then :)
alex_joni: It is Evertlammerts question...
Evertlammerts: Is that the answer to your question?
any video of a cnc working with emc2 around ?
nobody use a camera to shot a video ??
Evertlammerts: have you tried to start the axis file?
chinamill: no, I lost my connection, did anybody suggest anything? this morning I did try axis.tcl but that gave me an error
alex_joni: sudgested the file axis
ehm... just axis? I think I need a tcl file. but where can I find this file?
I guess the axis executable in the bin folder? seems a little strange, but I'll try to run it
03jepler 07HEAD * 10emc2/src/rtapi/ (rtai_rtapi.c rtai_ulapi.c rtapi.h rtl_rtapi.c rtl_ulapi.c): new API function: rtapi_vsnprintf. behaves like C library vsnprintf.
03jepler 07HEAD * 10emc2/src/hal/ (hal.h hal_lib.c): new functions to create pins and params with printf-style formats
03jepler 07HEAD * 10emc2/src/hal/drivers/hal_parport.c: use hal_XXX_newf
03compile-farm 07BDI-2.18 (2.2.18-rtl3.0) * 10emc2head/: build FAILED ; see http://linuxcnc.org/compile_farm/emc2head_slot2_log.txt
Just another remote GUI question to digest. Axis executable is the file that starts axis. specifiying our ini file, it sais the statusbuffer is not valid (emc.stat() & emc.poll()). I see what happens there, but don't understand why. Is it because there is no connection between server and client? Off to a meeting now, but hope somebody can help me :-)
btw, this all happens on the client while the server is running, of course
woo, my lathe conversion is finished (or rather, the X and Z axes are done, spindle on/off and encoder is not done yet)
Lerneaen_Hydra: Congratulations! Do you have any pictures?
yep, several pictures and a video
a few minutes and it will be done
How does amc handle encoding on a lathe?
currently you need a special version to do G33 (from cvs)
ok... with some extra HAL modules?
afaik you can also either have a normal encoder or a reading fork (?)
I'm not sure of that, I think it goes through the servo encoder system, but I'm not sure
Do you know who is behind the lathe encoder code?
has 2.0.1 been released?
hey cradek - read the conversation between you and jmk last night. That sucked.
can't believe there isn't a simpler solution than a micro
oh about my circuit being bogus? I was bummed a bit too, but that's how it goes
yeah that's my thinking too, but I trust jmk if he says there isn't
yeh - he is the man.
99% of the stuff he talks about goes over my head ;)
he does sure know his motors'n'stuff
he knows what works
I should be soldering today - if I can find enough parts.
ther, compression finished
the images are of the lathe and the part done, the movie (mpeg4 simple profile) is the entire run
jmk you in
Lerneaen_Hydra: at workshop someone had one of these lathes for sale
it had a very cool tool changer on it
I got two of those for free (no toolchanger)
I have seen them go on ebay every so often
I was really interested until I saw the $3500 price tag
yeah let's just say it didn't sell.
with the intergrated plc-ish system, I'm not surprized
that system is a *real* PITA
no the controller was replaced with an old PC
hence the EMC conversion ;)
the spindle sensor was not hooked up so it couldn't even thread
that counter module I wrote for this is in cvs but it's entirely untested
cradek: 1 sensor?
1 pulse per rev?
I think his is one pulse per degree or so plus a separate index
nearly one pulse/degree
there are 200 holes in the disc
and two sensors, one index and one 200 holes/rev
if anybody is looking for lathe i have 3 for sale
the feedrate varies a bit when running as I was adjusting it while it was running
alex_jon1 is now known as alex_joni
when it goes close to the chuck for the first time for example
oh, there was something I noticed that was strange when performing a G92
when I set up the position of the tool in Z axis, I do a face turn, and when done I don't move the tool away, and enter G92 Z0. This works fine, however when doing for X, I remove some material from the cylindrical portion, keep the tool at that X-position, move it out in Z axis, and measure the diameter. Then I do a G92 X <value that I measured/2>. However when running I get diameters that...
very incorrect, up to maybe 0.5-1mm
* alex_joni is
but I never used G92 ;)
what do you use?
how do you measure/set up the part?
why the second G92 ?
one to align the Z axis, one for X
or well, not align so much as offset/adjust
sorry.. can't help you here.. not quite understanding the problem
oh, the things is that the diameter is incorrect when using G92 with a non-zero value (AFAIKT)
it shouldn't.. do you have some backlash in the axis?
nope, at least nothing easily seen
it has ballscrews
did you properly divide by 2 ? ;-)
you do know that m2 unapplies g92, right?
how's your free day?
fine so far
nice to hear that..
I think I fscked my back again
moving too many robots
lifted some stuff again..
you're too young to have those problems
wearing an elastic belt now
well.. gotta be a bit more carefull
I keep telling myself that.. but to no avail
wait wait - can I say it? Don't use g92
axis has a new "touch off" feature where it lets you set g54 so that the current position is a certain value
I recommend using that instead of g92
however in LH's case I suspect lost steps or a moving tool or something that might be found with a dial indicator
unapplies g92 offsets
which version of axis has that "touch off" feature?
only the very latest (cvs head probably)
it's a few days old only
I will package it with 2.0.1 probably today sometime
released in a week or so?
spiffy? what kind of word is that?
it's like.. uh.. cool, neat, nice and so on
hmm, the counter module won't be in 2.0.1 so you will need to use the cvs version
[16:32:25] <Lerneaen_Hydra> http://en.wiktionary.org/wiki/spiffy
cradek: I don't think that there are lost steps or a moving tool
once when doing: G92 Z0.5, the X value also changed (the row of three values shown in the axis window in the upper left of the opengl window)
fenn: you still at the keyboard?
there seem to be many people interested in that clip :|
Lerneaen_Hydra: yep... 3 minutes left to download!
Lerneaen_Hydra: sri to flatten your upload for 10 minutes or so
lucky I don't have a bandwidth/month cap
the thing is, my max upload is 100, and it's maxed now. (5 people or so)
not that it really matters though
Lerneaen_Hydra: not bad... I am seeing about 60k bytes/sec here
we got the remote GUI to work
great to hear that
don't know what the comments in client.nml look like in the current release, in my release it states that remote GUI can be started on the client with tcl/tkemc.tcl -ini <infile>
that should be changed
we needed to make a custom scriptfile that exports some variables, and then starts the tcl file
how fast is the remote gui?
what did you need?
EMCSH I presume
pretty fast. we didn't run any extensive tests though
yes, EMCSH and TCLDIR
I say by hard, don't have it here
fast enough to control it (feedrate, manual jog, etc etc)?
yes, that went fine
we used a simple gcode script which ran very nicely
I'm off for dinner! see you guys later
Lerneaen_Hydra: Nice pictures!
EvertL: Nice to hear some progress :)
Lerneaen_Hydra: great movie even..
mind if we publish it on linuxcnc.org ?
a bit slow but good nonetheless
sure, I can encode a more high-res version if you want
nah, it's a bit big as a file right now
the current one is only 320x240 @ 500kbit/s
maybe encode it with 2x speed?
can you do that?
keep it around 5-6MB should be great
2x speed, I'll raise the audio pitch to match and lower the framerate so it's the same as it is now
one of these days I'll start a page with user apps & examples & machines
we need a bit of marketing ;)
that's true, some images at linuxcnc.org wouldn't hurt
Uh, dinnertime now, I'll send a link to the movie later
ok, no problem
cradek: How did you solv the lathe enc. functionality? A new HAL module?
chinamill: how do you mean that?
there is an input to the motion controller, that receives the spindle position
alex_joni: what codec/framerate/resolution do you want?
Lerneaen_Hydra: no idea.. what you think is ok
something like 320x240 or maybe a bit bigger
as long as it's around 5-6 mb?
keep it relatively small to be able to download
is 6.7mb ok?
if so then the file cnc1.avi (in the ftp root dir, ftp://temp:firstname.lastname@example.org ) is what you would want
alex_joni: you there?
on and off
have you gotten the movie/checked it out?
getting it now
slow connection at your end?
other stuff going on at the same time ;)
what type of connection is the linuxcnc.org site hosted on?
pretty good one ;)
I kept the pitch the same becuase it sounded strange at 2x the pitch
yes a bit more exciting than the earlier one
can't your lathe go faster?
or would that cause problems with the cut?
it can go faster, but I just guessed for a feedrate that would work guaranteed
that's the first part I machine in it
I ran at 100mm/min, it can go up to 700mm/min
how about spindle speed?
from 48 to 3200 rpm
there are 6 belt positions
one of them goes from 300-1200rpm
with a dial on the front of the machine
so it's partly full control via the dial
03jmkasunich 07HEAD * 10emc2/src/rtapi/rtl_ulapi.c: typo fix
nice, you could make that come from emc2
* alex_joni goes away for a while
I'll be back in an hour or so
yep, that's why I'm nagging about CSS support ;)
and feedrate as mm/turn
hmm, I just realized something
jmkasunich: what is that?
for machines with gearboxes we need a g- or m-code to specify the gear range to be used
but M101..M105 might do the trick
you mean if the gearchange is automatic?
or whatever custom Mxx
originally I thought it could just be a hal comparator, if greater than some threshold use high gear otherwise use low gear
but if you have css it might cross the threshold during a cut
that could cause issues if it hovers around the threshold
and you have css
that could get nasty...
before I was thinking only of mills, where you do S<something> and it would set the gearbox at that time
an m-code to select a gear is probably the easiest
yes, mills certainly are simpler/easier sometimes
on the mazak we have auto-gearchange, but manually requested
so mXXX changes to a certain gear?
(IOW, the ladder logic handles the actual change sequence: slow down, shift to neutral, shift to new gear, speed up)
no, right now you have to set a hal signal to force it to use low gear
although it would be pretty trivial to make a custom M code that does that, it would just do "halcmd sets use-low-gear 1" or "halcmd sets use-low-gear 0"
03compile-farm 07BDI-2.18 (2.2.18-rtl3.0) * 10emc2head/: build FAILED ; see http://linuxcnc.org/compile_farm/emc2head_slot2_log.txt
failed! how dare it!
jmkasunich: you will be happy to know I have started a bidding war on your raratan box :-D
duh, same typo in two places, I only fixed one
what utter foolishness, it's time to smite the compilefarm
LawrenceG, so you are ve7it?
Raritan MasterConsoleII 4-port KVM with 4 cables
thanks.... too many computers here and not enough desk space
oh, where are you located jmkasunich?
oh, a bit far for shipping for my taste
03jmkasunich 07HEAD * 10emc2/src/rtapi/rtl_rtapi.c: typo fix
Hola good ppl of mars!
Hola, nos ablas como les marsianos ;)
03compile-farm 07BDI-2.18 (2.2.18-rtl3.0) * 10emc2head/: build PASSED
jmkasunich: when a build is done, is there any testing to see if the program works? (I assume there isn't as that would explain the BDI issues you had... fun... with yesterday)
automated testing would be nice, but is pretty hard to do
the last couple days is probably the first time its been "tested" in 6 months, maybe even a year or more
(on those platforms that is)
"tested" is in quotes because the testing I did in the last couple days is still very limited
those boxes don't have X installed, so I can't run any of the GUIs, for example
they are 200MHz Pentium 1 with 128M ram
except the ubuntu one, which is just running on my own PC
is the cvs compiled and tested every time someone makes a change?
or well not tested, but it checks that the compile succeeds
yes, compiled every time theres a commit
if it fails, messages are posted here and on the commit mailing list
which all developers should be subscribed to, so they see that they broke it and can fix it right away
the nasty ones to fix are the ones that don't get noticed right away
errors that compile correctly but give bad behavior when running?
of course the vast majority of those do the same bad behavior on all systems, so they get spotted by people testing on ubuntu or BDI-4
the ones that break only on systems that pretty much nobody uses are the ones that get missed
those errors sound nast
yeah, fortunately they don't happen often
because the vast majority of the code is the same regardless of the system you are running
03jmkasunich 07v2_0_branch * 10emc2/src/module_helper/module_helper.c: merge a fix from head - rtai modules needed by older systems, added to whitelist
except for those inline's you were working with yesterday?
yeah, things like that
that file has a lot of conditional compilation that depends on compiler versions and such
p200.... do those still exist?!
no dount needed though
jmk you there
you were talking yesterday about copyright laws
do you know where emc came from
NIST originally wrote it
your (if you are an American) tax dollars at work
kearney and trecker programming manuals and allen bradly
they also did a fanuc
there's nothing wrong with reading a manual to learn about the control
however, making a copy of your manual to send to me to read is a violation
and i am sure hal came about that way to
did NIST write the c/c++/whatever code?
or "just" the RS274 spec
would that be EMC1 then?
code yes therre is always a better way to skin the cat
the original emc1 code was written by nist
and emc2 is very much of a complete rewrite?
the thing that we now call EMC1 is the sum of the NIST code and outside contributions over several years
i was talking about buying a programming manual for you
you don't have to copy it
if you want to buy a programming manual and contribute it to the cause that would be most appreciated
the ideas where not threres
I might not be the best person to have it, but it can be passed around as needed
when you read somebody elses work you can always look for points to improve upon
yesterday when that conversation was going on I was up to my eyeballs in a bug
that is almost on how aldesign is done
I should have just stayed out of it and concentrated on what I was doing
i am only tring to help to get emc to be close to a comercial control
like I said, yesterday I just wasn't in the mood to talk about new features
I'm afraid I was rather rude to several people....
you have done a very good job
emc2 is way farter ahead then is was 1 year ago
lots of people to thank for that
have to go now have family to take care of
"farter" ahead ;-)
(this is when you say that it's all thanks to yourself ;) )
jmkasunich: as a sidenote.. there is keystick in emc2 now
so you should be able to run on non-X platforms too
the only thing that's strange is that it uses xterm to run it .. no idea why
but changing that, and it should run (I ran it over ssh.. although I had issues using the Fx keys)
jmkasunich: I think the new board is ready for cutting
cradek: with the micro on it?
wow - that was fast
alex_joni: that's why I changed it to use xterm - it hasn't kept up with some console termcap changes I think, it needs to be fixed to work on the console
also emc was spewing stuff on stdout that messed it up - I think I've removed it all now though
SkunkWorks: I'm afraid for jmk to look at it - he'll say the grounds are all wrong
:) I know - I know.
right now I have a ground plane on the top and bottom of the board - with through connections at about 20 points. Not good I am sure :)
[20:18:00] <cradek> http://timeguy.com/cradek-files/emc/brd.png
[20:18:00] <cradek> http://timeguy.com/cradek-files/emc/sch.png
I want it single-sided so the ground plane suffers
although it has better coverage this time I think
I'm kind of jealous
how much was auto-routed?
I wish I could design electronics like that ;)
autorouting doesn't work very well at all for singlesided boards
cradek: hmm.. you still have a round pad on that big cap
I think the diodes are round pads too
on electrolytics they make the + side round I think
you can also make pin 1 on ic sockets different (square?)
yeah, only mentioned it because you said octogons are better for cutting
can't you put a trace between the F25 pins?
round takes a lot more gcode lines, that's the only problem
yes notice pin 14
not like that
vertical I mean
that would get rid of that connection on the second layer
the conn from pin1 on the Mega8
the F25 pin 14 trace is in the way
yeah, but you can route that outside
around the pin 10 from the L298
move the F25 pin 16 & 17 a bit to make room
get what I mean?
yes but I don't want to do it
ok.. just thought I'd mention it ;)
that's the reset line for the uc, I don't want it any longer
I appreciate it, you're right I could get rid of the jumper
not sure how tight the space in the F25 is though..
pretty tight, the spacing is a little less than .1in
if you care about length.. why have it so far away from pin1 then?
I have run two traces between ic pins before, but it doesn't seem to help the autorouter much
on the Mega8 I mean..
it has to go up to the programming connector, and the rest of its signals are on the other end of the chip
yeah, I know.. but you go pretty far the other way
you could put the via just next to pin 28
(if I can count to 28 ;)
it's a wire on the back.. no reason to have it vertical ;)
since I wire the jumpers by hand I like them to be straight so it's harder to mess up
was flushing gunk out of the airconditioning condensate drain, yuck
looking at board now
you only have 3 :)
what's the PCB going to do?
jmkasunich: sounds fun
Lerneaen_Hydra: it's a 2-channel servo driver and 3-channel encoder divider for my lathe
a crystal... bet that costs near as much as the uP?
oh, that sounds nice
it's free if I already have it
why go for PCB rather than protoboard?
JP7 is the programming connector?
Lerneaen_Hydra: I want to make two, and this is actually less time-consuming for me.
with debugging and all
cradek: before I forget it.. nice job ;)
heh thanks, but let's wait until jmk looks at it
I only routed a couple of boards.. have a friend who does it usually..
whats the bright green color of the +24 line mean (schematic)
oops nothing, it's selected
cradek: there are a few areas which aren't connected to GND
I would jumper those too
I wish eagle could tell me that
especially the ones around the connectors
it will say if any ground pins get isolated
I think gnd plane which isn't connected is worse then none at all
alex_joni: I think those are all connected...?
they connect under the ENC pins
oh.. missed that from the picture ;)
yeah, you can't zoom like I can
hmm.. hang on a bit
time for us in other timezones to dissapear ;)
so there is copper between the enc pins?
I see the Gnd from Senc is connected to the copper around it
but I don't see where it's connected to the Gnd from Zenc
[20:35:19] <cradek> http://timeguy.com/cradek-files/emc/brd2.png
sorry.. now I see
ok, so the copper around the enc connectors is tied to the main ground "bus" that runs up the right side of the board (by the jumpers)
what about the SENC?
ah, I see
it weaves thru the ZENC connector
jmkasunich: I didn't see it a first..
I know all the pins are connected - eagle will tell me if not
I notice that the same weave didn't work at the XENC, you needed another jumper
I just fixed that
had to move that trace left a bit
bet if you come straight left from the 2nd pin before angling down its better
I think I'll take out the top jumper now
ok, now what about the ground region under the uC and crystal
where is it tied to the rest? by the DB-25?
so the encoder connectors are grounded to the right, and the uC to the left
I'd try to make both go the same way
I could jumper over the three traces to the right of the xtal
maybe remove _both_ of the jumpers by XENC and go left instead
ideally the tie point between power ground (the driver) and signal ground (the uP and ENC) would be near the D-shell
ok I removed that, need to jumper to the left now, where should I put it?
the criss-cross under the lower end of the uP is unavoidable?
unless you rotate the uP
alex_joni: then they all cross each other
but then you have to move all the rest
or could you swap the inputs from the encoders and the outputs to the PC (dunno what the flexibility of the uP ports is )
cradek: not if you put the IC on the back (mirror it ;)
I don't want to start over on that unless I really have to
but I'm only kidding
cradek isn't anal enough
* cradek already hand-routed this once
this is probably too much trouble for this simple board
of course I don't know how much of a pain the tool is
where should I jumper the ground to the left?
jmkasunich: it's manual to get it this good
I do everything manual
if the tool is nice, its not hard, I actually enjoy routing
jepler enjoys it too, I don't really
if that criss-cross wasn't there, you'd have solid ground down the middle of the uC, and paths from there between pins to all the ENC connectors, and to the d-shell
same here.. although I lose interest after the 3rd version
jmkasunich: only way I see is to mirror it (put the IC on the bottom)
it will look funny, and it will be hard to solder
alex: the pins that are used are arbitrary I think, both the inputs from the encoders and the outputs to the PC are I/O ports on the uC
hmm maybe I can switch them without too much trouble
so you go back to the schematic and swap them
good plan ;-)
always a good plan ;)
cradek: see.. no fun asking us.. we'll just nag till you go nuts
yeah, that is a problem with things like this
jmkasunich: did you catch the stuff about keystick?
if I was doing it myself, I'd do all kinds of little tweaks here and there
you should be able to run emc2 on BDI2.x with that
alex: just skimmed it as I was catching up
ok.. just that you don't lose it out of sight
right now the farm does a normal build (not run-in-place)
I've been manually going in, re-running configure for rip, and doing make again
then verifying that realtime start works (first hurdle), then that loadrt siggen works (second, found the math problem), then loading threads, and verifying that siggen can make a sinewave
only once tried loadrt motmod
never tried starting emc itself
would need a config just for that I think, and explicitly invoke that config since pickconfig won't work
there is a keystick config already
so only scripts/emc configs/sim/keystick
so only scripts/emc configs/sim/keystick.ini
or similar ;)
would it work over ssh? or do I have to login at the console?
it shoudl work over ssh
I ran it yesterday over ssh.. couldn't do anything (F1, F2 didn't work)
but at least I saw that it started :D
that's a test too
how do you exit from keystick?
(never used it at all)
or kill from another ssh if all fails ;)
[20:53:24] <cradek> http://timeguy.com/cradek-files/emc/brd.png
really, it's END
ok fixed, much better
cradek: I know.. but it didn't work for me
MUCH better (the board)
now... if you move one more trace
then you'll get rid of all jumpers
[20:54:49] <cradek> http://timeguy.com/cradek-files/emc/sch.png
the trace below the S25 pin 1
just a bit for the gnd plane to close
the trace up the center of the uC still cuts the ground plane in half
I think I'd be tempted to keep the jumper for that one, to preserve ground integrity
the uP main ground is pin 8, right?
its cut off from the main ground plane by the trace going to pin 11
I guess you only have 5 IO in the pin-2 to 6 block
so the last one is gonna be a bastard child no matter what
11 should be Vcc
no.. sorry.. 7
just looked at the new schematic, lol
cradek: think one atmega is fast enough to do 3 quadrature dividers?
jmkasunich: should I do something with that pin 11?
jepler: if the code is simple I think it'll be fine... jmk did the math and said it should be ok
ok .. I haven't scrolled back that far
what I like to do is draw a line between the uC ground and the encoder grounds (for example) and see how long and convoluted it is
you'll have to scroll back to last night
cradek: you could run the trace to pin11 through C1 & C3 and use a jumper over the Vcc line
anyway, right now (unless I'm missing something) that line goes from pin 8 up and right past the crystal, around jp7, between the pins or R1, down the left of the uC, between pins and then down to the bottom, and finally back under the uC and out the right side to the encoders
[21:03:17] <cradek> http://timeguy.com/cradek-files/emc/brd.png
even better: the trace to pin11 through C1 and C3 to the XENC, then take the Vcc for the uP from the same connector
ahh, much nicer
solid ground plane under the uC, nice direct connections to each encoder ground
cradek: take Vcc from pin 2 from XENC, then it won't cross
and you'll lose the jumper
good point alex
it was right there ;)
now time to nitpick the power section
pins 1 and 15 are the power grounds I think
[21:05:50] <cradek> http://timeguy.com/cradek-files/emc/brd.png
1 and 15 of what?
cradek: I think the mega8 has to have the analog vcc and gnd connected even if unused
jepler: ok, just to vcc/gnd?
how about aref?
agnd is easy
avcc needs to be done carefully to avoid splitting the nice ground plane we just got
cradek: the datasheet says avcc, I don't see if it says agnd must be connected
what pin is that?
agnd is simple - avcc makes a split again
ah, I see it... from pin 20 (avcc) up and left, between 11 and 12, then up to the vcc trace
I guess they call them both 'gnd' in the datasheet
avcc = pin.. ?
sorry, up and _right_ from pin 20
they label 8 and 22 both gnd in the datasheet; I think eagle calls one agnd
yup, throught the C's again
I think aref can be unconnected
worst case, the xtal and C's need to move down a tiny bit
ah shit, that cuts off the gnd to XENC
unless you go between XENC pins 3 and 4
do it like I said..
oh, I see
from pin 20, between 10 & 11, through the C's
and finish at XENC pin 2
it cuts less
or even the trace that is close to C1
heh, this is funny... collaborative board routing
probably less funny for cradek who is madly mousing trying to keep up with us
[21:13:11] <cradek> http://timeguy.com/cradek-files/emc/brd.png
cradek: just tell us to buzz off when you had enough ;)
there is copper between XENC pins 2 and 3?
one more thing ;)
a few pieces of floating copper (betwen the caps, and between C3 and the uC) but not big, and not important I don't think
I'd move the reset trace on the other side of the uP
I don't like how far the gnd for the crystal caps has to go to get to the uc
I'll put the caps under the xtal
kills two birds with one stone
duh, I was full of crap talking about floating copper
didn't realize the caps were grounded, and those are thermals
[21:17:00] <cradek> http://timeguy.com/cradek-files/emc/brd.png
can I cut it now??
I usually do it otherwise ;)
put the XTAL vertical
and the caps to the right
keep the XTAL lines as short as possible
if you don't want to do that, at least move the vcc line (below right of caps) further to the right and down (make the diagonal part shorter)
that way you'll have more copper around the grounded end of C1
now I want to look at the power section
pins 1 and 15 of the L298 have PWMed current flowing in them
they need a low inductance path back to the bottom of the cap
as do the bottoms of the low side freewheel diodes
pin 4 needs a low inductance path to the top of the cap
as do the high side freewheel diodes
the high side diodes are pretty good
I could move the big trace on pin 4 up under the lower diodes
I'd move the trace going from pin4 to the right of the cap
alex's suggestion seems easier
I'm not sure what you intended with the trace from pin2 either..
move the trace from pin 2 to the top of D5 so its under the diode
drat, that still leaves pin 1 with a long path
yup.. either way..
that chip really seems like its indended to have the outputs head out from the two sides
dunno how radical you want to get... I'd be tempted to turn the chip 90 degrees, and put one set of diodes and one connector at each end
try it with pencil and paper first, see how it lays
[21:25:18] <cradek> http://timeguy.com/cradek-files/emc/brd.png
I see a ground buss on the bottom, connected to pins 1, 15, and the low diodes
I think this is a big improvement for the lower diodes
a power bus on top, connected to 4 and high diodes
diodes are good now
* alex_joni grabs some food..
1 and 15 are the trouble spots I think
bet this will take a while ;)
I'll just add gnd jumpers then
15 to 1 and 1 to gnd near low diodes
cradek: when in doubt if it's worth the trouble.. take a look at http://emergent.unpy.net/files/projects/01142347802/crw_3713.jpg
(which I heard works pretty well :)
one final touch - a 0.1uf cap from the pin 4 trace to the outside ground (the low side diode ground)
the cap combined with the ground jumpers mean you have a local bypass with total leadlength not much over an inch or so
I wish I had eagle
* jmkasunich is an obsessive-compulsive layout freak, especially for power circuits
is eagle free for a board this size?
jmkasunich: the free goes to 80x100mm I think
I don't know how big this board is
this board is ~ 120x85
[21:33:15] <cradek> http://timeguy.com/cradek-files/emc/brd.png
ok, high side diodes and chip main power are nicely bypassed to 1 and 15
how are 1 and 15 connected to each other and the main ground?
through all that meandering ground plane
I can't convince it to allow me to put in jumpers
it says "it's already routed, stupid"
use a resistor in the schematic
I think you can WIRE 'GND
or a fuse
* jmkasunich bets cradek he can lay this ckt out in 80x100
ooh a bet
a bet about "mine's smaller"
* jmkasunich downloads the free version of eagle
I'm sure it can be much smaller - but if it's near 4x6" it takes up a whole board anyway, so I don't care
jmkasunich: want another bet?
by the time you have it done, chris will post a pic of the already cut board..
I won't take that bet
4x6 = 100 x 150, right?
so I should be aiming for 100x75, then you can make two in one blank
the free version of eagle is a source tarball, right?
I think it comes as a tarball though
duh, of course not
ok I got jumpers
there are rpms and tgzs
I assume I want the tgz
I've gotten used to either apt-get install or tarball/untar/configure/make
forgot there were other ways of getting software
final version, I hope: http://timeguy.com/cradek-files/emc/brd.png
that will work
can you send me the eagle schematic file?
I'll send you my version of the board later ;-)
[21:45:45] <cradek> http://timeguy.com/cradek-files/emc/lathe1.sch
[21:45:45] <cradek> http://timeguy.com/cradek-files/emc/lathe1.brd
btw, what exactly are you using for all the connectors? they just look like 2 pin headers
nice little screw terminals
little must be the key word
they're 0.1" centers, right?
screw head diameter about 0.075" or something?
probably something like that, the screws are pretty small
jewelers screwdriver I guess
how big of a wire do they take?
cradek: looking great (the final board)
12 I think
I guess you don't need anything bigger than 18AWG for this level of current
14? I dunno, surprisingly big
a 12AWG wire is darn near 0.1" in diameter
more than 0.1" with insulation
well you don't put the insulation in the screw terminal...
really? .. hmm... maybe that's why my last board didn't work :D
no, but two wires side by side need to fit into two holes on 0.1" centers
moot point anyway, this board doesn't need anywhere near 12 guage
I'm sure 14 fits
what's a 12 wire btw?
AWG = american wire guage
another stupid american measure
* alex_joni looks away ;)
an strange way of specifying wire size
we have ours in mm
or square mm
mm^2 I think
both ways .. but mm^2 is more common
the catalog says 30-14 AWG
is 14 bigger then 12 ?
small number, big wire
just like drill bits
but drill bits have different numbers, of course
all the guages are thay way
sheet metal is the same way
but screws get bigger as numbers increase
now I know why you said another _stupid_ american measure
like shotgun gauges....
over here they are all in mm
small numbers, small size
big numbers big size
guess what size hole a #4 screw goes in?
a 12ga shotgun is one where 12 lead balls the diameter of the barrel weigh 1 pound
the answer is (I think) #33
I think wire and sheet metal gages are similar
this is too funny
alex_joni: we're kind of used to it
alex_joni: it's still frustrating though
I imagine that too
heh, the big wires are a little more sane
actually we have fractional inch, number, and letter drill sizes
500MCM is exactly twice as much copper as 250MCM and half as much as 1000MCM
how do you calc. how much current a wire can handle?
look it up in the national electric code book
I said calc. ;)
looking up is easy
(current ratings don't vary linearly with area anyway, so even if the wire size explicitly tells you the area, the calc is non-trivial)
ok ok .. I give up ;)
a wire with twice the area will _not_ carry twice the current
I know.. but I still find it more usefull to use mm ;)
probably cause I'm used to that
at least you know which one is bigger
and I know how big it is
our wire numbers are kind of a log scale
if it's a 5 (drill bit, screw, hole, wire diam)
I didn't know that
10AWG is 10 times the diameter and 100 times the area of a 30AWG
they all are the same size
10AWG is 10 times the area (and sqrt(10) the diameter) of a 20AWG
I guess if you compare 30, 14 and 0, the jump from 14 to 0 is relatively huge
there is a 0 wire???
yeah, that's the biggest wire ever
0 is often called 1/0
* alex_joni runs away
then you have 00, 000, 0000 etc
jmkasunich: I was joking
there are 2/0. 3/0, 4/0
how do you spell that - aut
after 4/0 you start using MCM, 250MCM, 350, 500, etc
jmkasunich: so similarly, as drills get bigger, you have #2, #1, #0, ... then what? (I know the answer btw)
pronounced like "ought"
jmkasunich: these are bigger wires then the initial standard had?
sorry, I get no end of entertainment out of this
I think a machinist over here would laugh his ass off if I'd say to him, hand me the #0 drill
actually I'm not entirely sure there is a #0 drill
it may go 2 1 A B
or wire ;)
send a bunch of eu machines to the usa, and they'd probably have no asses from laughing them off
notice how it switches from decreasing numbers to increasing letters
cradek, yea, there is no 0 drill
jmkasunich: I see nothing funny/irregular in the usage of mm
[22:02:31] <jepler> http://www.yarnfwd.com/needleconv.html
alex: I know (that came out bass-ackwards)
lol at jeplers needle chart
jmkasunich: bet they'd still laugh ;)
bbl, I've got a board to make
cradek: you were right about 0, 00, and 000
but it's for canadians
jmkasunich: familiar with IEC-61131-3 ?
its got IEC in front, I don't look at anything like that unless somebody forces me to ;-)
it's a standard for ladder programming & other stuff
I'm interested in the other stuff now
darn, I thought it was for wire
[22:05:55] <alex_joni> http://en.wikipedia.org/wiki/IEC_61131-3
i have full ascii drill charts...
[22:06:22] <alex_joni> http://en.wikipedia.org/wiki/Sequential_function_chart
lots of heavy reading there
it's some sort of visual programming for PLC's
the sequential function chart looks like a flowchart
and other sequential machines
it's basicly that
or maybe state machines
but with extentions for inputs/outputs conditions, macros, procedures, etc
ok.. nm if you don't know it
interesting concepts there
have to come back to it when I'm not so interested in doing a board layout ;-)
lol.. still interested?
I intend to fit that ckt into 75x100mm or less
I want to re-enable probing for emc2
anything against that?
I don't think so
not sure how much work is needed
don't think very much
and there are some german users interested in it
i am too...
some german & a canadian user
i can supply the mothion commands for many macros... alter variables and test... ; )
laser probe macros for tool lenght and dia. measurment
"some " custom tools
any thought about a READ/ type command to let you read another file name to read/end
nope.. no thought ;)
not this late .. :)
ok i' forget it then... ; (
some other day.. sure
you mean a G-code instruction for READ-ing a new file?
like O-READ whatever?
its a beautifull day here... 6:21..
01:26 < dmessier> its a beautifull day here... 6:21..
go to sleep alex
that's 1.5am ;)
kinda to load macros and the likes... keep the vars local..
my back hurts
I already am on coke
did yoi did something???
dmessier: heavy lifting..
not the best position
i know ;)
ice and immobilization has worked for me since 1988
and a friend called mary jane..
heh.. don't know her yet ;-)
strained and in spasm BITES....
I think I'm just looking for excuses right now :)
what ever you do DONT try Tyzanadine....
no drugs for me yet..
my eyes WERE still open cause OTHERwise i would have slipped into a comma...
using one of those elastic belts..
for support, and it's pretty ok
they arent bad... in the beginning..
yup.. after a day or so they don't help anymore..
gut if you can at nite in bed HEAT it.. even nthe deep heat smell bad stuff
bbl kid run.. ; ) story to follow...
my board is done - all I know so far is the 15 volt regulator works :)
what's your board going to be?
2 inputs from emc psudo pwm - hopefully high power output
kinda a test to see if I can make it work. then go from there.
ok guys.. off to bed
SkunkWorks: sounds similar to what I'm trying to do, but probably higher power
my website seems to be down :(
right - but no inputs yet - I have a buffer board I will use for that to play
I hope this board isn't a bust - I'm almost out of solder wick
poor jmk is going to be routing for hours...
I finally found some solderwick - milled boards seem to bridge over easier. Not bad though
or it might be the club of a soldering iron I use.
this is pretty close to the circuit
[22:50:19] <SkunkWorks> http://www.electronicsam.com/images/KandT/eaglesch.JPG
crap - I forgot to take a picture of the vaccum table. Dad was tramming it today.
those are 8A, 500V MOSFETs?
damn it - I need to redo that with the correct componets - don't go by the values. I am using irfz44 to play with. (left over from about 10 years ago when I was playing around with stepper drives.
This is my first eagle board. - I was just grabbing stuff for the pad size I wanted. sorry.
oh just a wimpy 55v device
60v 50a under perfect conditions :)
will be cool to see if it switches the n-channel mosfets with motor voltages above the control voltage (15v)
suposidly it work :) - has a boot stap circuit.
[23:05:16] <jepler> https://www.mwave.com/mwave/images/checkout-step3move.gif
SkunkWorks: Will you PLEASE stop using a branding iron to solder... you're giving us all a bad name
I can solder the whole board at once :)
SkunkWorks: Then at least use a blow torch!
Ug Ug Ug
jepler: I don't get it.
it is a 30 watt pen soldering iron - does ok
I like it because the tip doesn't burn up so fast.
maybe it is 35 watts - some where around there
when I leave it on.
I am a slow solderer
we should have held a "worst solderer" contest at fest
sorry but im back... ; )
worst solderer??? i got my sons new 1 Gb mp3 playeres cause he yanked the earplug socket off the board... i listened to him call in for the replacement gift cert and had it working fine in 15 minutes,,, that the 2nd one ive scammed off him that way...
he doesnt like me much cause mine work before his new ones come in.. lOL
how much heat can a small mp3 player mic take??
its a tight top solde fit.. and im sure one isnt right ... ; (
the whole 1 Gb unit is on a .75" x 1.75" board.. WITH a 1.25" x .375" display
and has a built in usb2 plug
a few weeks ago - I had to solder a power plug back into a toshiba laptop. boy that was a pain to get to - had to remove the mainboard.
i peeled this TINY thing... but the microphone is RIGHT on TOP of 1 lead...
i went fast and strategic... but NOT for quite long enuf i guess... 1 chanell works fine and both work if you wiggle it
eeeew, I hate that... sound popping in/out all the time