how do you set the chip select bits in the channel descriptor? how are the CS bits connected to the slave device?
are there some dedicated CS pins on the 50-pin header?
Yes the CS bits are just a field in the descriptor. CS just special I/O like step/dir etc
Theres also a variant of BSPI that has decoded chip selects, in case you want to directly connect to chips
is there a datasheet for the 7i65 somewhere?
(7I65 uses non decoded CS to save isolation channels)
PWC: So if I want to send a frame using the second descriptor, I send the data to the ports base address+32bits?
No data sheet but the preliminary manual is on website
Geo: yes +04
PCW: so to send to the DACs you write the frames you want to the DACs' channel-descriptor's fifo, i see how that works
what should the GTag for SPI be?
how do you read from the ADCs? You put the right number of frames in the ADC's channel-descriptor's fifo, with dont-care data?
So transmit FIFO spans 0x40 byte address range (16 x 32 bit words one per descrriptor)
In other words in hardware when you do a write to the send FIFO, the top 4 bits of the
data pushed on the 36 bit wide send FIFO are the address bits 5,4,3,2
For the ADCs on the 7I65 You also send a read next channel address command
Geo BSPI tag is 0x0e DBSPI tag is 0x0F (config I sent has DBSPI = decoded BSPI)
So you have a gated FRAME/CS signal for direct connection to a SPI device
gotta go, see you guys later
For ADCs, I havent finished it yet but I'm adding a autosend buffer,
so a single write can trigger up to 16 pre-programmed frames to be sent
to avoid the overhead of sending multiple read requests
So the only things that need to be done to setup the DBSPI interface is to write the channel descriptors, and make sure the IO are the right direction?
That sounds great :)
(The auto send buffer)
Basically thats it (plus learning the internals of whatever device you are talking to)
So, I really could use the RAW mode if I don't have to read the device each time HM2 talks to the firmware.
I think raw mode would be OK for testing anyway, it would give
you a feel for the way the hardware works.
PCW: thanks for the help, I've got to run too.
good night all
anyone remember what size flash drive JMK used to boot his robot?
that's what I remembered, thanks
which distribution did you use?
things would have been much easier if I had a bigger flash drive
I had to put the installer and boot kernel on the flash, and transfer the rest over the network
1gig big enough to be easier?
much googling and trial-and-error
yes, the iso iso is ~700M
can I check my tach on my servo with the halscope without nuking my lpt port?
no. Tach is a analog signal - printer port it digital.
you could get an analog to pulse converter
Thank you guys, i'll see what I cant drum up on that analog to pulse
* is on ebay looking at scopes
$25 for a 20mhz analog worth getting or is there something I should know before I buy one?
UncleGemc: buying a scope is like buying a car
you have to be careful with buying scopes on ebay, lots of them have issues
if you don't know what you want you probaby shouldn't
hmm, do I want a sports car, a pickup, or a station wagon?
Well, I own a 1964 chrysler 300 and I like it pretty good
if you don't know, don't buy
I really just need to check the tach output to make sure it is "critically damped"
I don't need anything fancy I don't think max voltage on the tach 7v, it should look like this /````\ but it prolly looks like this $#^&% other than using a scope do anyone got any tips on how to get it damped?
I thought I was going to give away a scope but the timebase doesn't work for some reason. They must have capacitors in a timing circuit or something like that
rob__, your internet is borked
this is only a test
if this was a real chatroom, directions on conversation would immediately follow
i thought it wasnt working things were so quite
Did you hear something toastydeath?
i think i left the door open and it's the wind
toastydeath: Ah ok, I thought I was imagining things there for a sec.
it happens to the best of us
Guest865 is now known as skunkworks_
is the 9-number coordinate of each successful straight probe X Y Z A B C U V W?
hmm - dead here.
my string broke :/
afternoon all, i've finally got my 'test' encoder on the spindle, it has 32ppr and a seperate index pulse, is it possible to get some tips on integrating these signals into emc2 with the end goal of being able to do synchronised spindle movements
i've had a look at the lathe.hal demo but i think its over complicated for my needs?!?
lathe.hal is simulated, it's no good for you
look at nist-lathe.hal
except use encoder instead of counter
you could also look at etch-servo/etch.hal for encoder hookup example
you won't need any of the PID stuff
er, at least i'm assuming your VFD has that built in
okey dokey thanks i will examine that nist-lathe. eventually i want to hook the encoder up to a pid so i can have closed loop spindle speed control
pjm__: if you can get/wait for 2.3 the wizard does a pretty good job of getting you goint
ah ok thanks perhaps i'll download that and compile it on another box then run the wizard to make the config
the config might not work with 2.2.x
so you have to use it with 2.3
ok, well perhaps i can look at the file it creates and cut/paste it into my existing configs
i'd like to try and understand how to make it work thoough
[16:03:23] <BJT-Work> http://www.linuxcnc.org/docview/devel/html//examples_spindle.html#r2
BJT-Work thanks that stuff at the bottmom of the devel page looks exactly what i need!
great i will try it in a few mins and report back! brilliant
i just finished making the encoder electroncs, so its outputting nice clean edges
Does anyone have or know of a working hm2-servo 5i20 .ini and hal file ????
run emc from the menu and select a config from there and tell it to save a copy for you
they are basic configs with no limit switches
I have pre-2.3 (like the day before release) and it has a 7i43 config but no 5i20
can you give me a path to the hm2-servo-5i20 config?
dave_1: the 5i20 was in there for quite a while as I have been playing with them for some time now... you may be looking in your configs not the ones in usr/share or something like that
the top of the list is what is in your ~emc2/configs directory
apparently I'm looking in the wrong place
will try again
[16:41:52] <BJT-Work> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/configs/hm2-servo/5i20.ini
* BJT-Work is off to take my crack head to the machine shop to be repaired
crack heads are so '80s, nowadays it's meth heads
BJT-Work, but but you are the machine shop
archivist: but I don't weld cracked tractor heads and rebuild them :/
eric_unterhausen: actually it is a 1956 head
BJT-Work, doing it properly does require cast iron cooking I know :)
I can weld cast iron but don't have the equipment to do the valve job...
I last did a casting in a forge, was entertaining
that sounds like fun
be back after while
Ni rod and pre-heat is supposed to work on cast ... I've never been very successful
braze may work as well
it does work but care is needed
if care is needed, it's no good for me
also cast rod and a muffle furnace with a hole to work through
just mig weld it
and a gas torch
Jepler/BJT having the right file certainly makes it easy. THANKS! Now off to wire limit and home switches (ugh)
unfortunately we haven't been able to find a way to allow "net" commands in the hal file to actually do new wire runs
that would sure be nice.
I'd pay twice as much for EMC if it did that
make new "wires" using net in a hal file???
like "rewire this net from that output to this output instead"?
jepler was talking about wires, not "wires"
oh - that makes all the difference :)
maybe I'll drink more of this here coffee
need to link EMC to Eagle to route the wires
that's been though about before
and tomp even mad some eagle components representing HAL components
so you could make diagrams and stiff. I don't think he ever got to the point of exporting a schematic like that as a HAL file though
so that's not a joke?
SWPadnos: are you typing with your toes?
I have no idea
eric_unterhausen: he's been on vacation ;)
just woke up from about 14 hours of sleep, after being up for ~36 hours getting home
SWP just needs more coffee to calm him down ;-)
getting to it :)
that's the opposite of coffee :)
[17:22:34] <alex_joni> http://www.reghardware.co.uk/2009/03/03/asus_lamborghini_vx5/
I'm betting there's an extra zero in the price tag
just spoiled my day
nothing that lots of $$ won't cure
too bad about the limited ram
other than the styling, thbe quad-core CPU, the video card, and the 1TB SSD, it's more or less the same as my laptop (excepat mine has a bigger screen)
I like that "more or less"
I like the other than part :D
mine has dual-core, 17" 1920x1080 screen, 4GB, 200GB disk, and 8400M graphics, so it's in the same league at least, if not in the same ballpark
probably much more bang for the buck!
heh, yeah. I bought it over a year ago, and it's still usable
(about $2k at the time)
haha .. all in one: http://www.reghardware.co.uk/2009/03/03/review_inkjet_printer_brother_mfc_990cw/
my brother's Lamborghini http://envirofuel.files.wordpress.com/2007/10/r6-100-fs.jpg
runs circles around Asus
alex_joni, that's probably a very good machine actually. I have an older one (B/W only, not a flatbed) and it's excellent
brother has some of the best paper handling in the FAX world
SWPadnos: I found it funny that it includes a phone and an answering machine
but it's actually pretty smart, as you need the phoneline for the fax function anyways
well, so does mine :) (at least the phone, maybe not the answering machine)
gotta go wire .... see ya later.
mine has a handset on the side, and sometimes you need to call with voice, then send the fax after someone on the receiving endhits a button
we stopped using fax a few years ago
email has mostly replaced it
but not for some signature-related things
btw, i know this might be a thick question but could someone give me an example of g-code for M6X1 tapping say 10mm deep? I've just hooked up the spindle encoder and want to do a fresh air tapping cycle first to test
I think K is 1.0. does that help?
one rev per mm movement of Z yeah that is right
it may be 1mm per rev ...
so M6x0.75 sould use K=0.75, not 1/0.75
g33.1 z-10 k1
ok that simple. thanks ;-)
i will test
hmm... what will cut ferrite magnets good without shattering them ?
grinding or ceramic cutting diamond wheel
thanks, think it will end up being cost prohibitive. :(
SWPadnos_ is now known as SWPadnos
nah a tile cutting wheel should be cheap enough, see also wht lapidary people do (we diamond cutting) all on the cheap
well trying to come up with a way to solve a issue we have with several thousand round magnets. thay used to work just fine but now we are craping over 10%.
also a green grit silicon carbide wheel may cut ok
mm.. well i'm just the unlucky guy that's forced to assemble them before thay get shiped out for welding. it's basically a stainless hollow hockey puck that we put a magnet in (prevented from moveing by locktight and a oring on the inner diameter).
drill hole larger
actually ferrites have a loose size spec due to sintering process
yes, thats why thay use the o-ring
whats the fitting problem ? length ?
the problem is lately 10%+ of the magnets have come back lose after welding, the puck is welded on one site to enclose the magnet compleatly.
lose/cracked thay rattle when shaked.
what's the O-ring material?
I can imagine the ring getting a bit warm during welding
donno the specs, but it has worked more or less flawlessly for 3 years im told
and locktight cannot handle temperature
if the O-rings changed from rubberto something else, or the loctite compound was changed, then that could be the cause of the problem
which has nothing to do with cutting though
well thay intend to try a oring meteral that suposedly will handle a higher temp.
well about the cutting, i was thinking about makeing 2-4 slits to keep the magnet from spining inside the puck.
I would use a better potting compound
and most important educate the welder to heatsink the puck during welding
eg put in a vice
thay are spun on some rotary setup with a tig welder, very nice welds.
we machine them after welding
thay are used to transfer power from a motor to some form or drive unit. to prevent oil/contamination.
eddy current drive
another more reliable trick would be a spring
well there cant be physical contact
well thanks for the input, im off for a bit.
Kohlswa: i'd say use black silicone gasket maker instead of the o-ring
it can handle 350C
(and heatsink the puck while welding of course)
if any goo is soft and expands or out gases it will damage the weld
there would be some air space inside
say, does anyone know if 3 x 12 oz.in steppers would be sufficient to build this introduction-to-CNC-toy machine: http://www.instructables.com/id/Easy-to-Build-Desk-Top-3-Axis-CNC-Milling-Machine/
dpy: try it and find out
i'd say yes, but that thing is known to be a piece of crap no matter what you do
fenn: no problem with that, it's an introduction to the subject
ive seen better in the scrap yard
these steppers are also 37.5 Ohm per coil, so this is going to be slow anyway
ask yourself what you want to make first
with the machine you mean?
yes, perhaps use a real machine to learn how stiff and solid they are
I'd be happy if it got to mill PCBs and maybe some plastic/nylon/delrin parts
PCB requires some accuracy to your building of it eg the squareness
also, I hope it will be able to drill holes in kind of accurate locations
yes, low backlash, high squareness and table flatness, and low spindle runout are all important for circuit board milling. for etching+drilling, things can be a bit sloppier.
so maybe use this as a bootstrap machine to make a better one with it
I would not start with something so flimsy
well, I like the agile development adagium: small baby steps
better to have something that works from start to end, than have something that is for 1/3th best-of-the-best and 2/3 not finished, ergo: unusable
that design has no inherent squareness, will require some setting up
archivist: that's fine, I see it as a stepping stone
I dont want to see people tripped up by a stone
I have seen too many people who buy the best microstepping driver boards, NEMA 32 (or better) steppers
and end up not finishing the mill
or whatever it is they intended to build with it
they need more round tuits
archivist: failure is an option, if it only is capable of engraving my name, or even writing my name with a pencil, then I have a starting point for improvement
first-time-right is not one of my beliefs
ok, go for it
tnx for your approval:)
no its your approval not mine
I have an advantage, Ive used all sorts of machines before I built my cnc, so I have an idea what not to do
dpy: that thing won't be milling plastic unless you use tiny tiny tiny bits (and even then..)
it uses the stepper as a thrust bearing so you get lots of slop from the motor (it wasn't designed to be used that way)
slop = broken tiny bits
machine is flimsy and that means frame spring chatter and broken bits
and I note the cheeky bar steward wants money to download the pdf
one more vote for "don't do it"
dpy: i'd suggest getting some 5/16 threaded rod and using the skate bearings for thrust bearings, and then make a frame and add ball bearing drawer slides for the ways (if you're cheap and dirty)
make a frame -> out of concrete reinforced with threaded rod
or MDF if you are a wimp and enjoy failure
the mori in the background is making the most awful sounds, it's finish cutting roughed out carriages, and the rough cut left huge fins of metal sticking up. it sounds like electric arcing.
howdy, i have a g33.1 question. I have the encoder / index pulse linked to EMC2 properly, when i start a rigid tapping cycle, the Z descends properly, but when it gets to the bottom and reverses spindle direction, Z does not retract, and the MDI input wont take any further commands such as M05 to stop the motor, any idea what I'm doing wrong?
does the spindle reverse?
it just doesnt retract
watch motion.spindle-revs with halmeter. rotate the spindle by hand and see if it goes up by 1.0 for each turn clockwise and down by 1.0 for each turn counterclockwise
now i'm not using a quadrature encoder just a 64 ppr + 1ppr index and i wasnt sure if emc2 would look for a change of direction via the spindle encoder
you need quad for rigid tapping - yours will work for threading though
if you don't have quadrature, it's impossible to sense the reversal
you need a real encoder
ah yes ok i will mill another encoder disk out with quadrature
i used the config as per http://www.linuxcnc.org/docview/devel/html//examples_spindle.html#r2
at the bottom
great, well i'm a) getting somewhere and b) learning
it's true that you can cut threads without quadrature because the spindle doesn't reverse
ah yes of course, on a lathe that would make perfect sence
it's possible you can get quadrature out of your homemade disc by adding a third sensor
yeah i'm just on the cad now gonna add in another set of holes for a 90deg phase shift then add another opto switch for the 'b' channel
you don't need a separate set of holes
just offset another sensor by an integer number of holes + 1/4 hole
just move the opto a bit
ah yes of course
you will want to tune the exact position for good quadrature signal with a scope
yeah i can do that
err 1/2 hole, sorry
you know what I mean
yeah, i got it
well i'll test this on my home made encoder, if it works nicely, i'll find a 'proper' 100ppr quad + index encoder
i found a 1250ppr heidenhain encoder but think its too fast for my parport inputs
depends how fast you want to tap
at a few hundred rpm might be fine - you'd have to do the math
yeah 500rpm gives just over 10KHz
* archivist contemplates 1 mm tap and sensible surface speed
* cradek contemplates getting extra taps
I have not broken a teeny tap in a while
* fenn contemplates testing the tap in some sharp cheddar
that's actually a brilliant idea
i'll be making a business of custom cnc'd party snacks
hey that joke was from next door a few hours ago
i just need to find someone to handle the marketing end
joke? what joke?
us and cheese in #cam
good evening.because in the m3 s1000 or m4 s1000 commad emc not visualization the red line on m3 command partprogram?
when i start the partprogram the emc not evidence the m3 line.i setp false motion-spindle-at-speed , but the problem not change.when emc read the m3 line the red line on actual line cnc read not see
motioncontrol: I think that will be better in version 2.3
the line highlighted is usually the one currently causing motion -- m3 does not cause any motion so in some cases it is not highlighted
ok if i can help the comunity i have happy
today i have install linux debian 5.0 lenny with rtai 3.7 test 1 .very good
hi all i just updated my CVS and it is now telling me i need to use kernel 2.6.24-16-rtai
robh_: what's the exact error message?
1 sec see if i paste it
EMC2 requires the real-time kernel 2.6.24-16-rtai to run.
Before running EMC2, reboot and choose this kernel at the boot menu.
EMC2 requires the real-time kernel 2.6.24-16-rtai to run.
unless its an update machine decided to install
robh_: you probably got a kernel update installed
which then is selected by default at boot
you need to reboot, and at the grub menu (just after BIOS) you need to press Escape
then you'll see a menu with different kernels
uname tells me i still have
emc 2.6.24rtai-3.6.1-smp #1 SMP PREEMPT Sun Feb 8 00:57:57 GMT 2009 i686 GNU/Linux
one of them is 2.6.24-16-rtai
ah, so you are using the smp kernel
i had coustom kernel, CVS was working fine on it till i updated few hours ago
that means you probably are running the wrong version of emc2
(the installed emc2, not the run-in-place one)
remember to ". scripts/emc-environment"
i uninstalled old one
type "which emc"
/usr/local/bin/emc - thats only as i installed cvs
aaah, i see some rtai-modules-2.6.24-16-rtai was downloaded in an update
yep that sorted it
with a tool change do you need to tell EMC the new tool number? as i see there is a iocontrol.0.tool-number (out) no in
i would think there is a Txx M06 or vise versa
im thinking throught my PLC for the ATC
robh_, are you planning on changing tools manually?
no bidirectional ATC
EMC2 assumes that once a tool change has completed, the requested tool is in the changer
the only place you might need an input is to determine what tool is initially there (at startup)
yea, i need away of always knowing whats in spindle or where atc is
from power on
is iocontrol.0.tool-number in paramiter file?
or just taken from the program at the time
I think the tool number is now maintained across runs of EMC2, but I don't recall the method (could be a parameter or a file similar to POSITION_FILE)
I think tool number used to default to 0, which wasn't too great if it was wrong
no pocket zero
zero ok if u have arm changer i guess
the Mazak has a 0, but that's not why it defaulted that way
has more to do with C than anything else :)
shame u cant tell EMC max tools, no biggie ill just determ in ladder if tool >16 or <1
hmm. yeah, that would be a good thing
as long as u could still read past set tmax in TBL
I imagine the max would just be so the interp could throw an error if you do T<too high>
correct, as other wise if dont check it, changer just goes round and round looking for it
is there already a PLC on the ATC?
no, well old control had a PLC yes, but was built into control it self
you can likely use HAL/classicladder if there isn't already one there
im rewritting it all so
i did take a print out before i took it all apart so can see how they did things etc
apart from magic box's they dont want you to know about, guess they need you to call a fanuc engineer out some times lo
thx SWPadnos ill take look see if tool-number retains its value
add M61 Qxx
* allows changing the current tool number (the Q-word defines the loaded tool number from now on)
* for now it's not allowed to call M61 from a program, only from MDI (or while in manual mode)
aah ri cool, that will let me set the current tool in spindle if like
check for KEEP RELAYS... on old fanuc's
hi dmess, yea all iv dont is looked at these for concept
untill like i said u get a box they dont want you to know about
well this machine had mitsubishi, but very much like the fanuc
could it be a virtual/satalite drive..?? how old was the fanuc??
machine is 1985
i have fanuc manuals .. but i dont believe thats where the problem lies
you need machine tool builder manuals
i got both
for all our machines
this is the machine, http://cnczone.com/forums/showthread.php?t=73943
there has gotto be SOMEthing??
mykro is now known as notranc
1st.. it a random t/c you are chasing 16+ 1 in the spindle keep relays i believe... i seem to know that machine
no radom, the dahli is random with arm yes
very sweet hardware BTW... accurate and reliiable once you knew her... finicky like a woman
machine im doing it carosel
so non random.. called pot IS spindle pot
so curret pot position equals spindle tool
yes, pot facing spindle always has to be empty
so in with empty pot, release tool and down with atc, find out wich way to go, how many upin with new tool, back out
no encoder on atc to make it more fun, so all counters in plc
should unless you have mucho $$$$ we had a 24 t/c umbrella that could hold 5 tools at 1 place i played
iv ATC with atached changer, so when run out of 24 or what ever in machine, theres a 2nd one changing the pockets too, that gets fun
^iv seen atc
timers will be your friend...its a t/c macro you build out of g codes that drive the ladder
emc has no Gcode macro calls, M19 etc
but old one does yes
call a sub
if in MDI can do M19 - M24 and does motions of atc
ok so you need to look at their atc sub-programs as im telling you to build... maybe aquire them instead...
yes i do have them, all they did was call M code into the PLC section
so you will need to build Mcodes to match them with t/c motions and appropriate timers to allow NOT pulling the machine apart as ther probably arent switches for all limits of things
there are switch for ATC postions, but yes like u said need to delay when going in out etc and wait for ATC to get there
i b off now, gd night all
arm in.open.switch. colletopen .timer.airblow.timer.pullout..switch....rotate to position.. swich. in and reverse
say goodnight Gracie