Make sure you upload your EMC videos to here http://au.youtube.com/groups_videos?name=EMCCNC
117 vidoes now. Good job guys.
shouldn't you be sleeping?
sleeping? done that, time for another day of hassling people
have you uploaded your video, skunky? :)
* robin_sz blinks
My videos were done using mach, and I have since sold my machine, I cant wait to finish my new machine and test out EMC with it.
I have not looked at it yet. Maybe tomorrow
well dont forget. Its quite inspiring.
Anyone around know of a font editor that will edit .ttf files? I''ve found a font I like a lot using true-type-tracer to ngc but would like to edit out some of the smaller unnecessary detail in the font.
[00:53:20] <SkinnYPup> http://www.fonts4free.net/jetson-font.html
I think the one I used was called fontforge
it was super old and motif, so it might be a little challenging to get it to work
I've used it once or twice for simple things -- it has much functionality
thanks guys, I'll poke at fontforge. Everything I could google up was installing .ttf's into ubuntu...
Nice just what I was looking for, works great!
SWPadnos__ is now known as SWPadnos
Anyone set up a rig using HobbyCNC controller? I'm trying, but keep getting gotcha's ... lookig for suggestions or example configs that would work
yay, lathe oscillation is gone
or gone enough
I can't seem to get the TPI or steps/inch kind of settings right ...
the little bit remaining is in the position loop, which will be replaced
I know it is simple stuff,
servant74: are you using stepconf? if so what particular setting don't you know?
take them one at a time and you can get it.
cradek: great :)
It is on my machine out in get shop, not where I am ... I'll come back another time with my details... like I should have this time ... thanks for the offer of help... I'll be back
ok, sounds good
I tried using stepconf and it just didn't seem to work. i did use the step_inch.ini and it started getting there. I could'nt figure out what I did wrong
servant74: take note of the details of what's happening when it doesn't work, and someone can help
make sure you have the latest version of emc also..
"just didn't seem to work" - very hard to help troubleshoot from that
I'm in the process of re-installing the newer emc2 and ubuntu 8.04 on a faster machine. ... (1ghz rather than 700MHz)
I know, I will try again and get more detail ...
thanks for your time
skunkworks is right - in a couple older versions of emc2, stepconf had some problems
skunkworks: I ordered mesa stuff today... but, I think the existing control is fully working. believe it or not.
aww - you can just block that out..
are you going to use thier h-bridge?
maybe I can reimplement G76 in ttl logic :-)
the 2 axis servo board.
ah, nope, going to keep the amps I have
duh - I remember that.
also, I need the three encoder counters, so I would have to get the servo interface board anyway I think
well, not sure of that. I am not good at this mesa stuff yet.
darn, the piece of the puzzle I won't have soon is the resolver boards. that will be a few weeks.
I could work on the turret and stuff though.
(motion is the most fun to do first)
cradek: you're using the existing resolvers but you also also need 3 encoders?
jepler: the resolvers will be run through a quadrature converter board
well two of them at least. I haven't decided how to do the spindle sense yet.
cradek: oh, I misread "three encoder counters" as "three encoders"
Does the install emc to a stock hardy fix the boot priority so that rtai is the kernel booted?
yes, until the next stock kernel update
what chooses that order? this wasn't a problem before hardy - I wish I understood it
Good thanks Steven
I think whichever kernel gets installed last gets booted
ie, the latest update
so when you have multiple kernel packages installed, whichever gets changed last wins
why are updates installing? are both kernels on the cd?
but you should test that theory, it may not be correct
no, he said stock install then add EMC
previously I removed the standard kernel, so you never got useless updates to it
the CD has only the RT kernel
oh does this only happen when people do that? I get it then.
I think so
yes - I think it only happens when the stock ubuntu is installed - then the emc script.
Right. With the EMC2 CD there is only one kernel resident.
OT, where can I get a laptop with a serial port?
[14:20:34] <SWPadnos> http://www.mtechlaptops.com/
or usb-serial adapter
archivist_ub usb to serial is a pain with some plcs
thanks for the link SWPadnos
only the big ones have a serial port though
you may also try the Dell outlet - their used stuff
the dell D830 laptop I bought earlier this spring has a serial port. I've never used it, though. (it's no good for realtime, incidentally)
I'd rather get a new one if possible it is for programming PLC's and Drives etc.
you don't think a USB serial adapter will be OK for that purpose?
I have only found one that works with most hardware
if you have to use the PLC programming software, and therefore Windows, then you can have issues with USB or ethernet to serial devices
or if you have too many serial port drivers installed
the Keyspan usb to serial is the only one that works with most devices
also I need a regular slot not an express as allen bradley DH+ communications device only fits the regular size slot
you mean PCMCIA?
yea, that one :)
I couldn't think of it
hmmm. that could be a bit harder in a new laptop
all the Mtech ones are expresscard
D830 has both styles of card slot but I've never used either
I think one of my customers has a Dell with a standard slot, I'll have to see what model it is...
'pc card' now... PCMCIA was so late 90's...
(as I listen to the REO Speedwagon song "Roll With The Changes" :) )
* BigJohnT listens to That Smell
as long as you don't have to smell it
7marythree at the moment
BigJohnT: want a tip?
yep, always random (on to U2, "Even Better Than The Real Thing")
BigJohnT: sunix uts2009b
great usb 2 rs232
that works for most plc's
all I've put it against so far :)
and dnc to the big mill at work
I program my avr's with it too..
yes, this great USB->RS232 thingie :)
hehe im not following
I'm sugesting (jokingly) that your magic USB->serial device is actually another PC ;)
with a serial port
im super happy with it..
so tell us what it is already. geez
thank you ;)
[16:38] <BigJohnT> always
[16:38] <anonimasu> BigJohnT: sunix uts2009b
oh. I missed that line
I should make coffee or something
make it pitch black :)
I'll just eat the beans
anonimasu: I have a usb to serial that works for me on one laptop but prefer to get a new laptop with a serial port as the AB PIC only works with a native serial port... thanks for the tip
BigJohnT: I'd tr that kind of port first.. it seems to work for most stuff that other ones dont
I have had really good luck with startech usb-serial
has done everything I have needed it to do.
we have to replace one laptop as the keyboard quit working....
keyboards are cheap...
my keyspan usb to serial works on everything except the PIC module
*laptop keyboards are cheap
what laptop is it?
call them and order a replacement
or ask for a quote
if you can easily remove the keyboard (usually a couple of tabs along the top will release it)(, then you can probably find a replacement
* BigJohnT goes looking
which model ;) http://search.ebay.com/search/search.dll?from=R40&_trksid=m37&satitle=laptop+compaq+keyboard&category0=
found a keyboard on notebookparts.com
how do u get it out?
that would require a internet search..
some are screwed from back.. some are clipped.. some have a bezel you need to pop off
I don't think I have had a compaq apart..
cool, I found a step by step takedown
isn't internet great? Someone has had the issue or done the proccess before.
you guys are great I don't care what Benny says :)
anyone know about /dev/lp0 emc setup for usb parallel port dev/usblp0 ???
one laptop in parts :)
you are trying to use a usb-parallel converter to run your machine?
yes usb-parallel converter
that will not work. you need to use a real parallel port.
I have heard emc2 uses low level realtime kernel drivers for direct parallel control, is this right or can I just recompile or set an ini file to use /dev/usblp0?
I blew my parallel port a few weeks ago
usb to paralell is a very bad idea
there is no way to make it work. those converters are only for printing.
* anonimasu nods
you can get a pci parallel port card and use that
cradek beat me to it
they're for bulk dataq transfer, not timing-sensitive pin control
I don't need timing sensitive control
yes you do, if you're controlling a machine
I don't have any limit switches or input sensors
do you have motors that are driven from step/dir signals from the parallel port?
as long as all parallel outputs occur at the same time surely it will be OK?
do you haev motors driven from parallel port signals?
yes step dir
mung: it's the output thats timing sensitive..
the step and dir pulses
ok, that is very timing sensitive
anyone tried experimenting with usb parallel converters?
nope. it's not worth the time, since we know with near 100% certainty that they can't work
at least not without extra hardware and a completely different driver
pci card is more than ten times the price of a usb converter
the experts here know.. This question gets asked on cnczone every few weeks. It will not work for mach also.
you can get a USB converter for $1.50?
pci cards are in the $15-$25 range, I thought?
mung: well, they wont do what you want to do even if they are free.
in UK pci cards seem >£25, I can get usb on ebay for £1
p+p is £2
thanks for info, I may try my own experiments, can anyone advise about altering parallel config?
mung: but they cant do what you want to do with them, due to the way they work, it's not a emc limitation but a limitation in how usb works
hmmm. well, sorry about that, but you really can't use a USB one. it just won't work
is there an ini file or where in the source are the relevant files?
25 < 1 + 25
there is no source to start from at the moment, you'd be in new territory
and you would need custom external hardware as well
I won't waste your time with further questions,
bummer about the blown port
I have not looked into it too deeply, but I will return once I have (if I find out anything useful)
sounds good to me
thanks for input anyway, im away
thanks guys we have a brand new keyboard on the way
mung I found some cheaper at a computer fair (UK)
about 12 quid
12 quid.. you uk people.. ;)
I was looking for ages and bought the only two the bloke had
is that a baby squid?
mung btw he was at the Northern computer markets fair at Wolverhampton racecourse
[15:33:31] <SWPadnos> http://www.byterunner.com/byterunner/category=Parallel+PCI-bus+cards/exact_match=exact
I don't know if they ship to the UK, if they do it should be <£15
ah, yes they do. $11-33 depending on method
just googled the ones I have and only get a greek site
ooh anders has some nice photos today. http://www.anderswallin.net/2008/08/partial-solar-eclipse/
Back agin, I have just found I was talking crap about pci parallel cards, I have found some on ebay and they are only three times times the price of usb parallel not ten times.
about £9 all in with p+p, and assuming they are compatable???. I cannot get any detailed specs so will have to take a gamble
if it's a standard parallel port, and the chipset is recognized by the kernel (so PCI addresses get set up correctly), then it should work
but it is a little bit of a gamble
But I can see a time when there will be very few parallel port options and as the volume of sales decrease the prices increase
sure. there are other I/O options that work with realtime
is there anything being done about future interfaces, I would like to run emc on my laptop
we've discussed USB, but there are feedback and I/O problems with it
there's a possibility of RTNet, which uses ethernet-connected hardware
there are expresscard or PCMCIA/PCCARD parallel interfaces
there are any number of generic I./O cards that can be made to work
there are also servo and FPGA cards that work (not for laptops though)
Also I was going try some experiments with usb and other stuff and am running emc from livecd. Was going to try some cat textfile > /dev/usblp0 experiments and have found the root password required, can anyone tell what defailt root pw is?
there isn't one
sudo and your login password
there isn't one on the liveCD, AFAIK
Ah live. I think the password is just <enter>
I'm probably doing something wrong, its a year or two since i used linux (I used to think i was quite good with linux??). I tried su, it asked for password, i pressed "enter", i was told wrong password, now I think about it maybe i should have typed "su root" ???
No. For su you have to create a root user account.
if you want a root shell.
sudo elevates your permissions to run stuff normally now allowed
mung: or you can just use sudo <command>
OK, thanks, i'll try "sudo sh"
anonimasu, is right about sudo su
[16:23:11] <rayh> https://help.ubuntu.com/community/RootSudo
for real root login
i'm away again, sorry to leach and disapear, thanks for your help
hello all im in need of help starting a proeject, anyone online available?
you expect us to drive over and help?
you have to swim some right archivist_ub
hows it going dude?
things are going well
SWPadnos: I'm currently working on making EMC work with my hexapod which basically works over USB (a USB - serial converter)
BigJohnT, so might you if he is in China
not bad, business is good
Gamma-X: is in the northeast USA I think
Hopefully I'll make it work pretty soon... It doesn't need to be realtime; actually the main reason for having realtime in emc is the direct motorcontrol; if this is not needed (which is the case if you have a serially controlled cnc) then emc could easily run on a laptop
ordered a Mesa card, going to have a play at setting up a CNC tube plasma
Another benefit is of course that you don't need to change your kernel
pmbdk, if you want accuracy you need good timing
pmbdk: so you'll just hope that emc can path plan for you in userspace.
pmbdk: and if the user launches firefox well crap.
usb to serial is just anyones guess when the bits drop out of the end really
archivist_ub: only if you don't have external HW to do it for you. Which is the case for an external serially controller
so just use EMC for the gcode to xyz bit then?
anonimasu: It's true that if I don't have realtime I can't be 100% sure that I get everything... On the other hand if I have 2 secs buffer (or so) then it'll be very infrequent... In any case on a standalone machine I can simply uninstall any FF... :-)
pmbdk: why not just use the stuff that
that's already built for it ?
anonimasu: that's one way to look at it... if everyone looked at it that way we would never get any new functionality... :-)
pmbdk: no, we would never progress as everyone would be stuck reimplementing stuff other's already did
instead of implementu
ing new features
well, honestly, noone has never implemented my hexapod algorithms, so i don't really understand the issue...
i thought we had done hexapod?
with respect to the realtime issues, i have tested my current userspace "driver" under linux, winxp and win vista (yak!) and I have never got a buffer underrun in my controller. Obviously you can always force something to do it, but....
robin_sz: Well a generic hexapod, yes...
the problem is that this only works if ball-joints (is that the correct wording?) is used
in any case i still don't understand why we shouldn't allow serially controlled cnc's to be controlled by emc...
I don't think it's a question of allowing.
its just not practical
What your project comes down to is you writing a converter from what you can get from emc2 to what you need for your drive box.
I'd be interested in a simple and cheap dongle that removes the realtime requirements from the machine. I don't want this enough to work on it, but I think it would be a great option to have.
but if I had infinite time and energy, it's one of the many things I'd work on :-)
I can see a need to be able to use the serial servo types that sit on a network
rayh: yes, exactly... I want to use emc to drive something that emc currently does not support...
even a plain old step generator would be nice.
a stream of time and position commands
cradek: that's basically what I have... An N-bit step-generator and some I/O's... :)
pmbdk, your serial device is the controller, not the PC. your setup is like drip-feed on an old CNC control
for that setup, you're using EMC as a user interface, and letting all the work be done by the external device
that's not what mung was trying to do
Well I think he needs nontrivial kinematics.
SWPadnos: not at all... I'm using it as to control everythin just with a lightly different interface... Its just a matter of abstraction...
The only difference is really that i don't need step-generation in emc...
and that i need to add some other kinematics, which could obviously be done in emc as well
as long as you also don't need any feedback to EMC, and you don't need spoindle-synchronized motion, then you should be OK ;)
No it is not just abstraction
btw I have no idea what a drip-feed is so... :-)
It is time warping as well
PC sends G-code over a serial link to a CNC that doesn't have enough memory for the whole program :)
or PC acts like a tape reader ;)
I thought you had the external device do the hexapod kinematics
SWPadnos: No, the external device is only doing the pulse-generation and io handling (as well as the critical io handling)
i have a user-space driver for the kinematics
ah, ok. there are a number of devices of that sort on the market
SWPadnos: another advantage of making this kind of motion possible... :-)
yeah - it would be nice to have a "set it and forget it" mode
but it won't work for a lot of machines, which I think is one major reason none of us pursue it much
IMO the loss of "Sense, Model, Act" would make it completely uninteresting to me.
in a sense (no pun intended), that model is broken for any stepper machine
anyways... if I'll make it work its fine.... :-)
yep. more power to you (as long as you GPL the work ;) )
that could form the basis of a driver for the G-Rex, the SmoothStepper, and the ncPod
or the USB 7i43
hmmm. does that actually use USB for anything other than power?
yes, it can do serial communication over the USB bus.
I guess it does - now I remember asking if the aprport could be used as I/O (no)
I believe the design incorporates a USB-serial chip
I *think* FT245R has an 8-bit parallel bus a bit like EPP or ECP to the connected device..
could be - I was thinking about whether the DB-25 could be used as I/O when the 7i43 is used USB connected only
jepler: it's a straight 8-bit bus
btw, it's a really nice chip; no-fuzz 8 Mbps serial driver
i guess EPP/ECP gives slightly more functionality, but still
the 245R or the 245B? (there are a bunch of chips in that series)
the 245R is (afair) simply the ROHS version of B
"A FTDI FT245R USB interface chip is used for the USB interface"
probably with minor differences
the -R has some extra features, it isn't a direct replacement (the 245B/BM have "L" versions that are lead-free)
SWPadnos: ok, I only used the B...
err - a -Q package that is
I've nev er used either, I'm looking at the FTDI website ;)
in any case it is still nice to be able to have such a high-speed interface with virtually no problems
it's nice that they've gotten rid of the external EEPROM and clock
pmbdk: so the linux userspace interface for ft245b is a /dev/ttyUSB?
jepler: well, the eeprom was not really needed...
but still; there was at least one chip (non-ftdi) which required virtually no other components, so... :)
you know - I just realized that it's Friday. I had thought it was Saturday
must be nice
I was about to start getting rude with the people making business calls on a weekend
well, it was, but now I have to get work done in half the time ;)
sorry just testing, I have finally got the net working in the workshop
nice - what distance?
not far, only abour 250meters
but did not seem to work without the cantenna
250m is quitea ways for wifi
i havent even got it aligned yet, was going to use a laser pointer but it just worked with a rough point in the generally direction
ah - wikipedia says ~32m indoors, or ~95m outdoors, with stock equipment
soon have enough UK users for a brit user group
I was here a few weeks back an someone suggest cat5, but wifi I think works out cheraper and less hassle
not as reliable, but almost definitely less hassle, and probably cheaper, depending on what the 250m is like
I still need to get a parallel card, then maybe I can get on with testing my cnc
250m is longer than the max allowed cable length, so you need a differnt solution anyway
I use a little driectional antenna on one end, and a little omnidirectional on the other, pretty reliable about 100M or so.
the directional is on a USR game adapter (ethernet to wifi bridge), the omnidirectional is on a 'FON' router/gateway/bridge/whatever
I got to go, dinner time!
SWPadnos, said "that model is broken for any stepper machine"
IMO it is only broken at the last link to the position.
It is not broken for any other thing like feedrate override or trajectory change based on velocity.
true - that's more of a "model - act" thing :)
no real sensing going on
No it still senses the computation of pulses at the far end of things.
heh - ok, fair enough
And an outboard step generator can't do that.
not via high-latency buses like USB, no
If there were a way to return total pulses along with some time at which they should be compared it might be better.
That is one of the issues I had left over from g10x
and SS and ncpod
what's "interesting" is that I think non-feedback systems may make parts of the wrong size if the clock isn't perfectly synchronized with the PC
though I could be wrong about that - I need to think about it a bit more
if you output velocity instead of position, yes
it's like assuming that because your servo amp says it is "velocity mode input" you don't need PID to actually ensure you end up where you want
I certainly agree that something would need to be done to ensure that the outboard generator was working from the same page as the pc.
Or free run the outboard and drip feed it data as needed.
isn't this already solved in the USC scheme?
it is velocity command with position feedback right?
cradek: yes, but I think we're talking about situations like what pmbdk is working on where the time delay between command and feedback might be on the order of 2000 servo cycles rather than less than one servo cycle ..
oh I see
so you need position control of coordinated motion in the device, unlike usc
I think that the position (steps) data that pmbdk sends also has a time to completion variable.
So within the block of data the coordination is assumed to be linear.
But it was a while ago that I asked those questions.
I'd hate to speculate about what code that I haven't seen (that may not yet be written) does :-P
I hate to speculate about what code in front of me does
I hate to speculate about what code I've written does
I've only written a few lines of lightweight stuff and it's a crap shoot!
[19:24:52] <SkullWorks_PGAB> http://cgi.ebay.com/RENCO-Low-Profile-Encoder-RCML15-500-0_W0QQitemZ200084103247QQihZ010QQcategoryZ67003QQtcZphotoQQcmdZViewItemQQ_trksidZp1742.m153.l1262
Cheap - with index!
I am assuming 500 line?
"Resolution of 500 pulses per revolution with no commutation" would seem to indicate that ;)
yes - thats line count
heh - missed that
I went to renco.com and pulled the RCML product sheet pdf.
what does "commutation" mean in the case of encoders?
i'd guess an optical counterpart of the hall output on a brushless dc servo
aha -- a "commutating encoder" has U, V, W commutation signals in addition to A, B, Z signals.
"revolution without commutation"... didnt that start the Revolutionary War?
yo ho ho
and a bottle of rum
apparently there's irc for cellphones
my review: it works
theres a thing called SMS too
its even better
u can txt grls on it
wots grls can you machine them
* robin_sz nods
where can I find more about your amazing product
im going to call it "real world" .. it might catch on
it sounds very web 3.0
I took a bunch of photos at work
so I will now have something to put up on my website
my boss offered to help out if I needed extra hands for the machining tutorials, which was cool
I thought so!
although I need to get more ideas for howto articles
hm.. how to cut random metal and not screw up
hahha I don't think I am qualified to do that one yet
I will probably do dumb stuff like edgefinding and tramming/checking a vise
"I'm sorry pmbdk, I'm afraid I can't do that"
in sampler i can use something like "net Xpos => sampler.0.pin.0"
now the last part i easy... I can get pin names using show pin
I hope you aren't going to ask how to get signal names
But how am I supposed to know what Xpos is?
it starts by knowing what the pin names mean
if you look at the manpage for motion, it will tell you that axis.0.motor-pos-cmd is the X position command
if you show pin axis.*pos-cmd, that will tell you all the signals that are connected to those pins
the stock configs use nice names like Xpos, Ypos, Zpos. The names can be anything though
man motion is clearly a good place to start... :-) Thanks!
heh - sure
you can man most any HAL component (though YMMV as to how useful the information will be)
let me see if I understand this: "net bla1bla2 => bla3 bla4" is equivalent to "net bla1bla2 bla3 bla4" which basically creates blaX if it does not yet exist and connects them all together?
ok! tanks... :)
"arrows" are optional, and are only visual aids for hte operator
bla1 may not be the name of a pin, I think
oh wait - was blah1blah2 meant to be one word?
no, bla1 and bla2 are two seperate words... Does that matter?
the arrow isn't needed, but it implied to me that you had both the net name and the source (writer) pin name before it
net signalname pin pin
the net command goes like this: net signame pin1name pin2name ...
there are only two restrictions that I can think of (plus some connection restrictions that don't have to do with "net":
1) the signal name may not be the same as any pin name
2) you must specify at least one pin name, since net uses the data type from the first pin as the data type of the signal
you may specify the name of a signal that already exists, and as long as all pins specified meet all the connection criteria, they will be added to the existing net
net == signal, in case you didn't notice :)
halcmd help net will tell you all this, I think :)
this hal-thing is _really_ cool... :)
no sh!t :)
its an idea from connecting signals in electronic circuits. A network of many wires, made of individual wires that are common to one circuit, called a net. Many nets make a complex circuit. A single net is identified by a single signal that goes between some connecting point on some devices.
I used it for a waveform generator in a power supply
ooh - hoe many points do I get?
from the freethrow line or halfcourt?
hmmm. the far endzone
oh - they actually use end zone in basketball, don't they?
a question about configs/sim/core_sim.hal: two hypot components are used to generate the XYZvel signal. One should have been enough shouldn't it?
ahh, but XYvel may be used also somewhere...
I think hypot only has 2 inputs
oh - it has 3 now. it may not have when that config was first made
ok... leftover from the old days... :-) But on the other hand its nice to also know the XYvel... :)
well, it looks like it had 3 from day one - dunno
hmmm. looks like there's an error in core_sim9.hal
? what are most people using for encoder hookup wire?
Anyone know whats up with dxf's made in qcad having errors less than .001" that gives gcode that emc can't use? Particularly entities that have been rotated to make a whole part like constructing a gear or synchronous pulley?
sounds like an open chains issue.
Yeah I've looked from the gcode back to the part feature , arc centers and endpoints will be mathmatically off by .000x" but its a product of qcads trim , round or bevel function...
I can make plenty of things that do make good parts using gcnccam, but so far no very useable things where entities had to be broken at constructon lines or radiused
does the cad have things like "trim to intersection"?
you'd think that if you trim two entities to their intersection point, the ends would match
can you increase the resolution of the output? 0.001 seems pretty coarse
I'd increased the line length and angular res up to 6 places with no avail
still same errors totaling less than .001 "