how do you call a connection with 70% packet loss?
you're beeing generous
or was that a technical question? :)
anyone tried a cardbus parport?
I think I've seen them mentioned on either CCED or the Gecko group, but I haven't used one myself
jmkasunich__ is now known as jmkasunich
[19:09:55] <JymmmEMC> http://news.yahoo.com/s/ap/20070921/ap_on_fe_st/odd_bungling_burglars;_ylt=Ai5kaLYjF2gYnrZ0JA2OD6ntiBIF
any adc's been used with emc? i use m5i20 and am looking for a solution that doesnt require buying another controller ( stg motenc ) speed and accuracy not high ( 10Khz and 8bit is fine ) I'm looking into usb as r/t isnt needed
tomp: SWPadnos is working on a high end AD/DA card
alex_joni: wireless, rate shaping, throttled,
tomp, if you have $800 sitting around, I can sell you a fast accurate multichannel AD/DA card for the 5i20/5i22 :)
I have ambitions for something in the 12 bit area
my plans are much cheaper, but I'm very slow
JymmmEMC: it's not wireless.. cable-link
if it's ethernet, change cables
thanks, & SWPadnos good luck ( no, not 800$ ;)
alex_joni: someone needs to learn how to crimp coax
alex_joni: hard line baby!!!
i went broke buying 7i33's 7i37's and breakouts ( no, not bad prices on those )
I think I can make a less accurate version for only $300 or so :)
especially if I don't need them delivered via Einstein Express
random number generator ( low accuracy buy fast ; )
JymmmEMC: nevermind, it's at 100% loss right now
alex_joni: somebody needs to learn how to connect coax together.
fenn_ is now known as fenn
tomp: you're using the mesa stuff for an RNG?
no, stupid joke... a random number generator is a inaccurate cheap solution for an adc
tomp: ah, ok. heh
Many I hear so much of copper theft these days - wire and plumbing (mostly in new construction)
the new high speed train system in taiwan is electric, except kilometers of it were missing every morning
tomp: if you don't mind doing the analog section design yourself, USB HID devices can report analog values up to at least 16 bits .. you could adapt obdev.at AVR USB firmwares for your purposes. http://axis.unpy.net/01187535387 http://www.obdev.at/products/avrusb/hidkeys.html
thanks i was browsing the avr at obdev
my board design exposes two of the AVR ADC pins on the header and includes the recommended LC filter on AVCC input, but I haven't used it yet -- and I wouldn't trust me to do analog design really
I'd only use the onboard ADC if accuracy is unimportant, especially near zero or full scale
SWPadnos on the other hand I would trust to do analog design
the 10-bit A/D is more like an 8-bit ADC, and even less outside the 10-90% of scale range
because I sound like I know what I'm talking about?
or why the ADC sucks?
why the adc sucks
oh, they just do. I know this after using the internal ADC on AVRs and some Motorola DSPs specifically
were you sampling a high-ish frequency signal?
they basically tell you that they're not going to be accurate below a few tenths of a volt
1000 samples/second per channel, and there was a 200 Hz or so filter on the input
what if you bias it?
you can do that, and it improves things a little, but you lose resolution
since you're compressing full scale of the input signal to something smaller than full scale on the A/D
as well as doing things like putting most of the mcu into sleep mode while you do the adc sample
tomp: Here ya go.... http://www.fourmilab.ch/hotbits/hardware3.html
strangely, sleep mode seemed to make things worse
I was oversampling
I had a 4-sample ring buffer
but it only helped (somewhat) away from 0 or FS
the problem with gaussian noise and a unipolar A/D is that you can never read below zero
so you only get half the noise
and it's always the positive half (near 0)
[19:26:32] <jepler> http://www.atmel.com/dyn/resources/prod_documents/doc2486.pdf
page 248 gives claimed ADC characteristics
oh, maybe that was contributing to your error then
single-ended conversion error @ 1MHz typ. 3LSB
btw you can introduce 'artificial noise' which wouldnt hav ethat problem
a sawtooth signal, basically
so these usb based scopes are not likely to have valid output near 0Volts?
that's ADC clock 1MHz, not 1msps
jepler, right - 3 lsb means a 10-bit converter is actually an 8-bit converter (or 8.1 or so)
JymmmEMC: still reading hotbits trying to understand it
tomp: It uses radiatic decay to produce True Random Number Generator
tomp: just the DIY version we are talking about
tomp: i'd hope something bought'n would have a better ADC in it
SWPadnos: pessimist! I get log(1024/3., 2) ~ 8.4 bits
JymmmEMC: gotcha RNG
that's theoretical ;)
* fenn squints at 0.4 bits
tomp: Close TRNG
fenn: a lot of are avr inside ( or pic )
SWPadnos: was that to me?
no, to jepler
tomp: is that all?
fenn: just ones i looked at (bitscope et al )
What's all this A/D about?
What's all this A/D talk about?
i considered using a usb scope design as basis for simpler adc
[15:10:37]<tomp>any adc's been used with emc? ...
re adc: i wanted to read a real world value as input to hal
tomp: what bandwidth you looking at?
doing the conversion isn't the issue, it's getting it into the PC that's the problem
USB has timing granularity (and driver) problems
10khz 8 bit bipolar (7 with sign )
USB has problems, period.
damn #@^$%$&*%^$^ polling
the 5i20 should get rid of lots of that, and you can even make a config that talks to a micro over SPI
but it's not trivial
SWPadnos: solved easily enough with a buffer on the adc (since realtime isnt an issue)
in fact, I should get back to that :)
a bit buffer
tomp also stated at the outset that it doesn't have to be realtime
it's just for a readout or something
fenn, if all you want is a sample every once in a while, then you don't even need a buffer (at least, not larger than one sample)
it's nice to know when the sample was taken though
for me, r/t adc is not important, but for SWPadnos it may be desirable for a product
only when you need 10 K servo updates/second on 6 inputs and 8 outputs ... :)
who 'needs' 10kHz servo updates?
at 16 bits resolution (real required resoution >= 14 bits)
depends on what you're updating now, doesn't it
I should have said PID update rates, it's not servos
it's not controlling something mechanical
actually, I think it's the power supply for an ECM machine
why are you doing that with a pc anyway?
"engine control module, a device which monitors engine operations"?
basically a big-ass current source
its for the velocity of the EDM, begins with average voltage, vesrus desired voltage , error generates new velocity
a variance of 0.5 of 100 is needed ( low accuracy)
it's a 36V 20000A ECM supply
oh your gen , yes
oops, i meant electrochemical machining
JymmmEMC: fourmilab / fermilab thats confusing
so we have a 4th axis post that seems to be right but axis doesnt show it correctly... wow it didnt take long for us to catch up did it..
axis doesnt show rotations
at least, not last time i checked...
shows something looking rotated.. but its wrong
alex_joni: is sudo the command to become root,im trying to get the usc diagnostics program to work.
jepler: did Jon ever catch up to you?
jlmjvm: 'sudo su' will make you be the root user, but you can run most commands just by preceding them with sudo
fenn: I always wondered about that, thanks
fenn: Now to figure out how to becoem root in GUI to edit files
another sudo trick is 'sudo tee' for redirecting files to a non-writable location
I think I'll learn screen before tee =)
tee is easy
cat foo | tee bar
prints foo on the console and logs a copy to the file bar
* JymmmEMC dusts off the dot-matrix printer.... the ULTIMATE in "secure logs" ;)
you mean the granite-carving machine
I loved my dot-matrix printer.... full color
had a feed thru for printing labels, envelopes, and would even feed the fan fold paper to the cut line, then reposition it back to top of page for printing.
you could even set it up for cutting half-sheets, like for a receipt or video store rental.
thanks fenn,i didnt have the su in the command
now its saying no such file or directory
JymmmEMC: no, was he looking for me?
03fenn 07TRUNK * 10emc2/configs/hexapod-sim/minitetra.ini: RPW to ABC
has anyone ever used the remote desktop in ubuntu before?
jlmjvm: I use it to connect to a doze box
jlmjvm: what are you trying to do?
(with regards to sudo/su) ?
trying to get this usc working
there is a diagnostic program to run
alex_joni: remote desktop doze box?
alex_joni: you can use nix to RDP to a M$ box?
the "Terminal Server Client" in Ubuntu connects very nicely to a remote desktop on a M$ box
it knows RDP and RDPv5
although I don't know the difference
alex_joni: but it's not secure at all, is it?
it's as secure as the protocol :)
i was wondering if it worked,a guy from synergy configured my software thru terminal 1 day
alex_joni: that's not what I asked =)
jlmjvm: sudo /path/to/univstepdiags bus
JymmmEMC: I have nfc
I use it all the time
it's certainly lots more secure than VNC
alex_joni: I'll be damn
alex_joni: colors are a bit funky, but it works
glad it does
Though, I think I need to check the security (or lack there of)
alex_joni: this is the command im trying to use # ./univstepdiags bus
heres where i got it http://pico-systems.com/codes/univstepdiag.html
jlmjvm: then use 'sudo ./univstepdiags bus'
or use 'sudo su' to become root, then use './univstepdiags bus'
without the '' of course
when i do it says no such file or directory
which one of them?
you must be in the directory where univstepdiags is located
that's what the ./ means
my file unpacked at /home/cncmill/usc/univstepdiag
jlmjvm: cd /home/cncmill/usc/univstepdiag
jlmjvm: sudo ./univstepdiags bus
or is univstepdiag the file?
03jepler 07TRUNK * 10emc2/tests/stepgen.1/checkresult: these scripts require bash features, change #!-line accordingly
03jepler 07TRUNK * 10emc2/tests/stepgen.2/checkresult: these scripts require bash features, change #!-line accordingly
03jepler 07TRUNK * 10emc2/src/hal/utils/halcmd_main.c: error and warning should go to stderr; otherwise, the setexact message fubars all the testsuite
03jepler 07TRUNK * 10emc2/scripts/latency-test: having gone to the trouble to create a temporary directory, latency-test should go ahead and use it rather than dropping temporary files in the current directory.
had to be in that directory
jlmjvm: echo echo (13:43:06) SWPadnos: you must be in the directory where univstepdiags is located
but its not seeing the board
what is it... lspci or somethign like that
JymmmEMC: nope, it's a parport connected board
jlmjvm: make sure your bios is set to EPP or soemthing like that
alex_joni: I thought he was playing with USC
alex_joni: and USC is a PCI board aint it?
rebooting with bios change
JymmmEMC: nope, all boards from jone are parport based
jlmjvm: there's also a program called pcisetup to make the parport be EPP
(for some PCs where the BIOS is not enough)
hooray,it sees it
03jepler 07TRUNK * 10emc2/debian/extras-Ubuntu-7.10/emc2.files: extras for 7.10
03jepler 07TRUNK * 10emc2/debian/extras-Ubuntu-7.10/etc/modprobe.d/emc2: extras for 7.10
03jepler 07TRUNK * 10emc2/debian/extras-Ubuntu-7.10/etc/udev/rules.d/emc2.rules: extras for 7.10
03jepler 07TRUNK * 10emc2/debian/extras-Ubuntu-7.10/etc/xdg/menus/applications-merged/cnc.menu: extras for 7.10
03jepler 07TRUNK * 10emc2/debian/extras-Ubuntu-7.10/usr/share/applications/ (5 files): extras for 7.10
03jepler 07TRUNK * 10emc2/debian/extras-Ubuntu-7.10/usr/share/desktop-directories/cnc.directory: extras for 7.10
03jepler 07TRUNK * 10emc2/debian/ (drivers.files.in emc2.files.in): new files to install
03jepler 07TRUNK * 10emc2/src/hal/drivers/probe_parport.c: linux/config.h is unneeded and caused a problem on 184.108.40.206-magma kernels
03jepler 07TRUNK * 10emc2/src/ (configure configure.in):
* Fix kernel version checking for 220.127.116.11-magma
* Fix lyx detection for 7.10 -- binary renamed from lyx-qt to lyx; support both
03jepler 07TRUNK * 10emc2/src/Makefile: extra files to install
03jepler 07TRUNK * 10emc2/debian/ (configure control.in): 7.10 uses python 2.5; make debian/control adopt to the major version of python installed
03jepler 07TRUNK * 10emc2/debian/emc2.postinst: 7.10 severely limits per-user locked memory; increase this across the board for systems with emc installed
... and with that, emc2 builds on the upcoming Ubuntu 7.10 "Gutsy Gibbon" release; experimental packages including realtime kernel at: http://linuxcnc.org/experimental/gutsy/
cool. I'll try to test that this week
is that an SMP kernel build?
I didn't get smp to work
then I may not test this week :)
also I have no idea why my kernel-image package is so much larger than the official ubuntu one..
192M did seem a bit excessive
did you look at what's inside/
it has about the same number of files as an ubuntu kernel
but the first modules I looked at were about 3x bigger on disk
they aren't multi-architecture, are they?
maybe you enabled some debug? or verbose stuff?
try modinfo on one
modinfo looks about the same
'objdump -h' shows a bunch of debug sections
I'm sure that's the reason for the difference
jepler: hmm, your kernel has APM and ACPI enabled
I suppose you did that on purpose?
alex_joni: CONFIG_ACPI=y is deliberate, "# CONFIG_APM is not set"
I have one piece of hardware that requires CONFIG_ACPI=y for the onboard network to work .. seems to be just fine for realtime as long as the power manangement parts of ACPI are not used
I think the base ACPI is also needed for multi-core CPUs
though that could be APIC - don't remember
that too, but ACPI may be needed to turn the other core on
jepler: the CONFIG_PM fooled me
# Power management options (ACPI, APM)
on this machine of mine, the ACPI has something to do with IRQ management
just out of curiosity, does 2.6.15-magma work on it?
my recollection is that breezy and dapper both didn't work with the onboard NIC -- detected but didn't work right
so if 2.6.15-magma is the realtime kernel for dapper, no, it doesn't work
probably because of the same issue
* alex_joni goes to bed
good night all
great name they chose
I'm trying to see how it gets set for the ubuntu kernel, haven't found it yet
alex_joni:thanks for the help earlier
"mumble" is my own interjection
jepler: I figured that :)
jepler: might want to check /boot/config-* ?
alex_joni: yeah, but INSTALL_MOD_STRIP is not anything from .config
tristate "STRIP (Metricom starmode radio IP)"
there's this but it's a wireless NIC of some sort
anyway .. bbl -- and I mean it this time
btw, you can easily update your info on cia.vc now
I just did it for http://cia.vc/stats/project/emc
[21:56:39] <JymmmEMC> http://www.arduino.cc/
Site Stats Summary
100k visitors/year for linuxcnc.org isn't that bad
good night all
JymmmEMC: maybe of interest (diy usb i/o) http://electronics.wikia.com/wiki/PHOENIX http://semifluid.com/?p=24
tomp: Thanks, someone just mentioned the other one to me, I thought it was novel idea as it's an open source HW project, they even give you the eagle files for the PCB's.
watching the intro vid now :)
heh, I didn't know it had a video! lol
its based on older thingy i looked at 'wiring', thanks again http://www.youtube.com/watch?v=PeScmRwzQho
arduino not hid :(
ow that arrduino video is painful to watch
get one of these and write some software: http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=AT90USBKEY-ND
i just bought parts for 25 usb programmers, based on attiny2313
usb avr programmers *
so why would i mess with that
for the analog USB I/O thing people were discussing earlier
that's actually pretty decent
the headers look stupid
no buffering or anything - it's like the Pluto
they arent even 0.1" though
2mm then, most likely
ah - hm. the pitch is different from the row spacing. that is weird
probably fits some header that they cheaped out on
1.27mm dual header
yeah, I see they're standard items
still weird though