[00:00:17] -!- logger[mah] has quit [Remote host closed the connection]
[00:00:29] -!- logger[mah] [logger[mah]!~email@example.com] has joined #linuxcnc
[00:05:16] -!- tronwizard has quit [Ping timeout: 265 seconds]
[00:42:04] -!- the_wench has quit [Ping timeout: 260 seconds]
[00:42:43] -!- archivist has quit [Ping timeout: 256 seconds]
mrsun: guess you did not have any latency problems to track down and lathe spindle encoders to add
not sure I could have done it with out the help of the guys here
so your cruising along now?
had to do some things with SMI as well as others
as far as EMC is concerned
one of the barring in the new motor needs replacing
make a hell of a racket
at least you know it needs attention
squeals every now and then
true, but a brand new motor should not need new bearings in just a few days
and I did not drop or over load the bearing
depends on the quality or lack there of
the only other thing is the spindle speed keeps jumping every now and then
the original one was
it lasted about 15 years
the new one is made in the USA
[00:55:14] -!- asdfasd has quit [Ping timeout: 255 seconds]
paid 2 times more then if I went to automation direct
but needed it now, so could not wait the week it would have taken for shipping
I scoped the spindle index input and it is solid, but when I use the EMC scope the spindle input has a spike every now and then.
noise on the line?
it is not the real input to the port as it is very solid
no, no noise at all
but yes noise when it is inside the computer
so port is messed up
[00:58:33] -!- tronwizard [firstname.lastname@example.org] has joined #linuxcnc
I am going to move to new input and try that
if that does not work then move to new port
I had to ground my motherboard to the drive cabinet on my BP cause the drives were noisy
will have to install one
it should be grounded through the Pport cable
should it not?
[01:00:07] -!- archivist [email@example.com] has joined #linuxcnc
yea, but I needed a bigger bonding path to clear up the noise
I can see no noise at all on the output of the opto encoder but sure do see it when I look at what EMC sees
I may try that
are all your wires shielded?
motor and signal?
the new drives are much higher switching speed so that may be a problem
it makes a big difference
but the noise is one spike even now and then
i rewired mine with all shielded and it works fine
the real scope looking at the input line going onto the computer is rock solid,
nothing at all for noise
I would put a .001 uf or so capacitor at the parallel port or breakout input to gnd
but when using the EMC scope
I see this small quick spike every not and then
and a pull down transistor
like I said the real world input to the Pport is rock solid
EMC's scope inherently pulse stretches so it could be a very short 1.5V noise spike that you might miss with a scope
I cranked the input voltage sensitivity way up and could not see that spike at all
Not necessarily true its not a sensitivity thing but a trigger thing
it is something to do with the port chip or the port is damaged or some sort of very weird ground problem
I guarantee it will go away if you unplug the cable
put schmitt-triggers into the line
RC followed by Schmitt trigger
and in noisy environments.. fibres or balanced lines and current loops.
thing is it worked for years with out any filter or transistor or anything at all
just a plain old opto
parts change characteristics over time.. so maybe it was working borderline to not working but stable.. now it aged to trip over to the bad side
[01:09:41] -!- rob_h has quit [Ping timeout: 265 seconds]
Other option is use the debounce comp in the base thread
(I think you can use it in the base thread)
I did that with mach to get it to work right
Well there you go
you need to do the same with LinuxCNC
First step is 'man debounce'
sw can sometimes fix symptoms generated by hw... but it should not. fix the hw. much simpler and reliable in the end
Then find a hal example that uses the debounce comp
Ordinarily I would agree but debounce is very effective and reliable against spike noise
john@zotac:~/emc-dev/configs$ grep -irl 'debounce' *
take your pick
I have 2 caps on the input
in fact our hardware encoders use the same technique on the A/B/Z inputs
pcw_home: true. but its usually a symptom of missing caps in input lines. even my chinese cnc driver boards have em
and a filter cap for power noise
The software ones also have the advantage of trivial time constant adjustments
I could not see this spike at all
Sometime you _dont_ want everything to be RC filtered
Well in any case. debounce should fix it
and with 2 filter caps connected to the Pport input for spindle index, that is all I have for the spindle right now
I can not see how I am getting any spike
could be ground bumping
its almost impossible to see what the parallel port sees unless you have a differential scope input
I know software debounce will work
Other options are cleaning up the noise source (common mode choke around the three spindle motor leads will help a lot)
the spindle motor is just a single phase 220 volt 1800 rpm motor
there should be no noise there
the spike is very erratic,
comes and goes
and no pattern at all that I can see
can get 2 or 3 jumps in rapid timing, and then nothing
most of the time it is just one every now and then
a bump here and a bump there
Oh I though it was driven by a VFD
I hope to be soon
will be putting new electronics in
So its just a fixed speed spindle motor?
5i something and the 7i something to go with it
yes, for now
but what I get the 7i something board in I can add the spindle speed controller
I wonder if you have noise form some other obscure source (Static is one possibility)
that will be very sweet
5i25 + 7i76
I believe that is what I should get
and the 7i76 should do motor speed control as well
started on a shoe string and have had to do things on the cheap
Aero-Tec: ah.. its a motor with coals?
had to build up tooling and metal inventory
not that old...LOL
well.. for contacts
graphite blocks in slides with springs
no, just a AC 1800 RPM motor
has 2 startup caps
one for each direction
it is a nice motor
big and heavy
but want to get a 3 phase one and the vfd to go with it
was going to get it all from automation direct
get a GS2 and you can control it with LinuxCNC
via the serial port
is that better then a 0 to 10 volt control system?
I think so you can get feedback like load and other things
you just plug it in set a few parameters and load the gs2 component and hook up the hal and your good to go
I use the on my BP knee mill
[01:38:27] -!- bedah has quit [Quit: bye]
I was going to cheap out and use the GS1 and a 0 to 10 volt control
but it looks like a GS2 would be the way to go
[01:40:07] <jthornton> http://linuxcnc.org/docs/html/drivers/GS2.html
[01:40:18] <jthornton> http://linuxcnc.org/docs/html/examples/gs2_example.html
I've done all the hard work for you
is there a page like that for setting up debounce?
not that I know of but the example configs I listed before will have them
with info on how to and why and what does what?
dunno, you would have to look at the configs
doesn't look all that complicated with only a couple of pins http://linuxcnc.org/docs/html/man/man9/debounce.9.html
[01:48:26] zz_satyag is now known as satyag
[01:51:44] -!- acdha has quit [Quit: Leaving...]
so if I make a change to hal and or the INI
is there a way to get EMC to load it with out having to restart EMC?
got it all zeroed and ready to go
but would like to get the debounce thing going as well
jthornton: any way of loading in a new ini and hal with out having to restart EMC?
or a way to lock in every thing so when EMC restarts it is zeroed and ready to go
I have backlash turned on and so that needs to be in the right dir so the zero does not move the amount of the backlash
can anyone help with this?
I think you are overthinking it.
do you have home switches?
I load tool 1
and zero from that
all other tools are offset from tool 1
[02:17:33] -!- tronwizard has quit [Ping timeout: 256 seconds]
but that is a pain to do
perhasp I misunderstood what you want to do.
would be nice to load in new ini and hal with out restarting EMC
you want to preserve your current zero, or just in general?
this isnt 1997 brah
write down the values ;)
its the offset in the variable display
but I have backlash turned on
you could restart and make sure your first move is in the same direction it moved last.
how would I make sure the backlash is right by entering in the numbers?
so say move t1 to z0x0
and then restart
then make sure z and x get moved in the same dir then load t1 and zero it?
guess that would work
[02:22:25] -!- tbone_atl [firstname.lastname@example.org] has joined #linuxcnc
as long as your first move after restart is the same direction as the last move before exiting, it should work.
it is the x I am the most wanting to restore
so you could just move +.1 on each axis, exit, restart, move +.1 then -.2
the z get moved each time I rerun the program
or, just refind the spot and don't worry about it.
will try it
but need to find the debounce sample and try to get it to work
tbone_atl: It is many of the same people, but rather more of them.
I would love for the spindle speeds to read right
I'm the proud owner of a new chinese laser that came with a moshi controller :) I do plan to convert it - am I in the right place to ask some questions?
Aero: I have a debounce hal if you want
[02:25:22] -!- servos4ever has quit [Quit: ChatZilla 0.9.85 [SeaMonkey 2.0.11/20101206162726]]
tbone_atl: congrats. if you havent learned about cutting corners, you'll do now :)
BTW any help on setting debounce numbers?
would love to see it
i've built a couple of 3d printers - i bought the cheapo laser with expectations of hacking it to make it do what I want :)
but I'm pretty new to this side of the CNC world
it's for Z
jdh: cool, thanks
Aero-Tec: I think my limits were set to 15 when I had a base thread. so a whole 300uS of travel past the switches.
not sure how applicable it is.
tbone_atl: i guess you know by now that you need a new controller board?
I would appreciate it if someone could confirm that I have this right: my machine came with a cheap software package (MoshiDraw) that is mostly garbage, but it's good for engraving. If I convert the controller to use Mach or LinuxCNC then I will be pretty limited to cutting only. Is that right?
roh - i pretty much knew that when I ordered it :)
Do you mean raster engraving or vector engraving?
well, maybe I don't know the difference - is vector engraving the same as "pocket"?
(learning as I go here)
tbone_atl: moshi seems to be horrible... we use newlydraw on our cutter (for now) which does vector stuff better
I've played with mach3 and SheetCam for another project - Sheetcam has a pocket operation that removes material
moshi seems to be mostly for rastering (back and forth, line by line modulating the beam power)
roh - is NewlyDraw like Moshi in the sense that it has it's own control board that only works with that package?
in other words, is it possible to run NewlyDraw if I have the Moshi controller?
tbone_atl: yes. but it seems to be a different board to the moshi ones
atleast i have seen no picture of the same board we have anywhere yet
the seller asked us what we want to do and recommended to get newlydraw instead of moshi and the fitting controller for no extra money
if you already got the machine you see that its basically a drop in space for the controller, regarless what kind of. 4 screws and some plugs
yeah, that would have been nice - I actually made that request after I ordered it (ebay) but the um...language differences...were comical
yeah - mine is USB. LinuxCNC requires a parallel port, is that true?
we want to retrofit our machine to emc also
ours is usb as well
emc = LinuxCNC, correct?
seems to use some arm7 and 2 pentawatt case stepper drivers
what exact machine have you got?
[02:35:18] * tbone_atl looks for ebay listing
can CO2 laser do deep engraving of aluminum?
ours is about 40x40cm cutting area.. space for maximum 45x45cm material.
nope. no metal removal at all
one can use elox. aluminium and make nice markings
saw a laser doing deep engraving, but it had to go over it like 20+ times
[02:36:33] <tbone_atl> http://www.ebay.com/itm/40-Watt-C02-Laser-Cnc-Router-Cutter-Engraver-Engraving-Cutting-Machine-Usb-K40-/290817166731?pt=LH_DefaultDomain_0&hash=item43b60dcd8b
thats what co can do
^^it's basically that machine but I paid $600
plus $180 shipping
roh - is elox. aluminum the same as cermark/thermark/
Aero-tec -that might be a YAG laser
that is just burning away the black dye from anodizing is it not?
thermark is some special stuff to mark metal i think
eloxiert means anodized
yeah, you spray it on - and the laser burns it
sorry.. missed to translate that
how did you get it for $600?
heh, guess with the $180 shipping it is about the same though.
these machines exist in basically the same visible appearance and building concept in different sizes
jdh - it was a different listing. There are a ton of different sellers more or less selling the same machine
from 20x20cm worksize up to 40x40
K40 is one of the common ways they are descbribed
could one keep dyeing it and relaserring it to get a nice cool interplay of colors ?
ours is about 100cm wide, 70cm deep and 30cm high
ecpur.com - this is the website of the people I bought from
which seems to be the biggest variant in that case design
yeah roh - that's a decent size
the tube is 30-40W
my cut area is about 9X12"
depending on who you ask ;)
but I'm going to be modding the thing
thats about 22x30?
thanks for the debounce sample
going to try it
to me it seems like the machine could do more area, just the mechanics to hold a workpiece are smaller
may be back with questions
yes, roh - that's why I'm going to rebuild
EMC requires parallel port, yes?
also.. i recommend getting some honeycomb bed
do you have plans for a driver stage yet?
where do you get it from? I read some people don't like honeycomb because it reflects and marks the backside
tbone_atl: Not-USB is correct, but there are a bunch of PCI alternatives (including PCI-ext) as well as parallel port
roh - not sure what you mean?
[02:45:32] -!- andypugh has quit [Quit: andypugh]
tbone_atl: well.. what you want to use as new 'controller board' .. since linuxcnc would do the controlling.. something would need to interface from parallel port pins to your steppers
tbone_atl: no honeycomb also means that whatever you vapourize needs to get sucked upwards and cannot escape in all directions
we do not have air assist (yet) but on stuff like acryllic one can clearly see the fumes move whereever possible
what would be an alternative to honeycomb for cutting if i may ask? just a horizontal sheet would reflect even more
and clamping from the sides means everything falls into the machine
tbone_atl: i will be building my own driver card for such lasers and share the design soon.
tbone_atl: it will be just a few ttl parts for deadlocks/chargepump and some drivers from pololu. mail me if you're interrested
the steppers on our machine are faily small, so the drivers should have enough power
stepped away for a sec
what did you mean by "plans for a driver stage"?
is that a controller?
together with linuxcnc
moshi or newly seem to do what would be the 'make gcode equivalent and send to machine' part
but the controlling the steppers and stuff is done on the arm on our machines currently. with linuxcnc that job moves onto the pc
thats why it needs an rtai kernel
also the power amplifiers for the steppers are on there (stepper drivers)
which is just a fancy word for '2 h-bridges' per stepper
this looks interesting: http://chriscircuits.blogspot.com/
he has the same laser and made a 'drop in' solution for a new control board - no re-wiring needed
if you only do isolation and stepping, yes.
but i#d recommend adding a chargepump and deadlock for safety.
a parallelport can have lots of states before the machine runs and linuxcnc is started
his design is missing those
besides that its very similar to what i just proposed
also the ideas for adding pump monitoring are nice. we just built a new cooling system for our machine and still need proper monitoring there
atm we use a old digital rpm meter on the flowmeter and some pc panel with temperature display for the sensors
but the pc doesnt know about these.. and it would be very nice. thanks for the link
my ideal setup would have X Y and Z, plus A - interlock on the door, water flow sensor/kill switch, E-stop button, water temp readout
So EMC can engrave or not?
it can, but its not that easy ;)
the guys from a hacklab.to did that afaik
check this out.. http://www.buildlog.net/forum/viewtopic.php?f=5&t=1078
they retrofitted a old air cooled laser some while ago and raster with a special hack
it is based on the hacklab work
ah. nice. i wasnt that far to check up on recent sw developments. nice to see it progressed that far :)
[03:11:08] -!- r00t4rd3d has quit [Quit: Leaving]
we just rebuilt the machine (new tube, new cooling, cleanup, etc) and when i am finished with my current project i'll build the controller/driver board
i find it quite spooky to see these new china-co2-tube based designs recently.. one even had the tube moving on the Y axis
any suggestions for a homemade chiller option?
use oversized pc tuner stuff
the piping is quite small diameter so use a high flow pump and some chiller for 2 or 3 12cm fans
pc hardware tuning stuff.. you know.. idiotic blinking fans and watercooling for gamer systems? ;)
made all that stuff dead cheap and availible.. we used a flowmeter from the same line of equipment
what about fume extraction? those systems are expensive. I can vent to the outside, but the little air mover they gave me is weak
remove the fan that came with it and make your own adapter for a round 10 or even 12.5 or 15cm diameter flextube from aluminium
run that to a fan and then outside with directional lid
our fan is something like 400cubic meters per hour or so.. much more than the chinese thing, but it could be even more
check out the recommendations for sizing the fan from the epilog helix laser manuals. they are online as pdf somewhere
our tubing for venting is 10cm dia, with adapters to 125mm for the fan on both sides.. but the total length is less than 5 meters, so thats fine.
if you are in a residential area you propably should think about a charcoal filter system
yes, looking into that
in total our fume extraction system cost something less than 200Euro. the fan about 80, and some small parts
right now I can't seem to cut more than one time per hour without the software crashing so my neighbors are safe :)
heh. thats evil
but it sounds like that will change
looks like I messe3d up trying to do the debounce
[03:21:00] <Aero-Tec> http://pastebin.com/6qV22G26
error message trying to start emc
hopefully i will be able to consistently control this thing...soon
can anyone help decode it?
tbone_atl: i'll go back to completing that last contract of mine.. so i can get back to hacking the laser ;) bbl
is gcode used by Mach3 the same as EMC? or are there different "flavors"?
thanks for our help
your help :)
I can post the hal file as well if that will help
jdh: you still around?
is there anyone here that can help with programming the hal file?
[03:32:44] -!- kwallace has quit [Ping timeout: 255 seconds]
[03:35:43] -!- skunkworks2 [email@example.com] has joined #linuxcnc
got time to help me out of a jam?
need some help with linuxemc
ask away.. but I am fading
trying to do some hal programming
get error mess
[03:38:27] <Aero-Tec> http://pastebin.com/6qV22G26
can post hal if you need
trying to add debounce to spindle index
old-lathe.hal:91: Pin 'parport.0.pin-11-in' was already linked to signal 'spindle-index'
will get the hal file for you
I have no idea how to rework it all so it works right
will post the new hal I messed up
maybe you can fix it
plus are you editing the file on a windows machine?
I sure hope so
this is windows
I copy and past using linux
so I save a temp file using win
and open it in linux
and copy info over
[03:42:13] -!- Valen [Valen!~Valen@c122-108-45-139.blktn6.nsw.optusnet.com.au] has joined #linuxcnc
anyone here use MoshiDraw on a cheap chinese laser? :)
[03:45:46] <Aero-Tec> http://pastebin.com/7fye7xzU
old-lathe.hal:4: Warning: File contains DOS-style line endings
my hal file that gen the eror
I saw that
but just the one line
or am I reading it wrong
how do I get rid of the dos line endings?
I am using the linux editor
[03:47:38] -!- kwallace [firstname.lastname@example.org] has joined #linuxcnc
I think I have ultra edit installed, that would convert line endings
The line ending message is just a warning. The "already linked" message is an error.
I used this as a template
[03:52:24] <Aero-Tec> http://pastebin.ca/2262383
but had no idea how to make it work
getting it to load with out error is one thing, getting it to work is another thing all together
remark out line 68
you mean I got the rest of it right and just messed up not remming out that one line?
69 and 71 need to be hooked up to net spindle-index-filtered instead
I am on my nook - so this is a pain
[03:57:53] -!- tronwizard [email@example.com] has joined #linuxcnc
If one unlinks a pin, does it unlink both the straight and "not" version of the pin? I don't do much unlinking.
Oops, gotta go.
anything I can do to help out?
skunkworks2: thank you for the help
well - remark out 68 and see what the next error is
pin spindle index does not exist
that is the new error
net spindle-index <= debounce.0.0.out
in stead of net spindle-index-filtered <= debounce.0.0.out => spindle-index encoder.0.phase-Z
ok, will try it and be right back with a report
[04:13:05] -!- skunkworks3 [firstname.lastname@example.org] has joined #linuxcnc
[04:13:05] -!- skunkworks2 has quit [Read error: Connection reset by peer]
skunkworks3: you still there
it loads but no spindle speed
could the debounce be to high?
I think it is 20
not sure what that means
[04:20:22] -!- Cylly [Cylly!~cylly@p54B11FF0.dip.t-dialin.net] has joined #linuxcnc
will try to gen a sig by hand and see if that works
[04:21:24] -!- Loetmichel has quit [Ping timeout: 260 seconds]
[04:21:35] -!- ve7it has quit [Remote host closed the connection]
do small CNC routers for wood and plastic typically have or even need servos as opposed to steppers?
that did not work
most use steppers
[04:23:27] -!- ve7it [ve7it!~LawrenceG@S0106001c10b7770f.pk.shawcable.net] has joined #linuxcnc
servos are faster if done right
and less likely to loose steps
they have a feed back loop so the drives know when the steppes have not been done
set debounce to 1 and see if it works normally
when would you use servos. I read a post on CNC zone. someone had two similar machines one with stepprs and one with servos and stated that in practice both are just as effective
You can also have closed loop stepper systems,..
they cost more so if you're building a system on the cheap, steppers are it
servos have feedback
if you loose steps with a stepper you don't get em back
you can use encoders on steppers but it's not that highly recomended
If you are converting something like a huge vertical knee mill, is that something where servos are highly recommended?
even so, you can get large steppers
Could you say that servos have a higher 'power to weight ratio'? which for a 3 ton mill , who cares
you would belt or gear drive a servo
and possibly even a stepper on that
[04:34:46] -!- sliptonic [email@example.com] has parted #linuxcnc
How do they even CNC the Z axis of those? via the quill?
[04:35:04] -!- sliptonic [firstname.lastname@example.org] has joined #linuxcnc
[04:37:19] -!- sliptonic [email@example.com] has parted #linuxcnc
[04:38:28] -!- ve7it has quit [Remote host closed the connection]
did not work
tried to undo it all and see if things worked, it did not, not sure what I did wrong to undo it all
[04:40:11] -!- sliptonic [firstname.lastname@example.org] has joined #linuxcnc
odd - in have to sleep. be back tomorrow.. if you don't have it by then
ok thanks for the help
[04:52:04] -!- Aero-Tec has quit [Ping timeout: 248 seconds]
[04:52:42] -!- ve7it [ve7it!~LawrenceG@S0106001c10b7770f.pk.shawcable.net] has joined #linuxcnc
[04:52:46] -!- ve7it has quit [Client Quit]
[04:52:52] -!- Valen has quit [Quit: Leaving.]
where do the lost steps go?
[05:03:03] -!- hdokes|werkin has quit [Ping timeout: 276 seconds]
[05:03:11] -!- skunkworks3 has quit [Ping timeout: 260 seconds]
[05:37:28] -!- Aero-Tec [Aero-Tecemail@example.com] has joined #linuxcnc
skunkworks: got it working, thanks for your help, you had it right the last try, and setting 1 and 20 both seem to work, not sure if 1 would have fixed the spindle speed problem but 20 seems to have it fixed
the problem of why it did not work was a hardware glitch I have fixed
thanks very much for your help
is there a way to setup jogging when paused?
[05:51:36] satyag is now known as zz_satyag
or do a Gcode pause that would allow jogging?
[05:59:23] -!- kwallace [firstname.lastname@example.org] has parted #linuxcnc
[06:03:29] -!- Fox_Muldr has quit [Ping timeout: 260 seconds]
[06:05:02] -!- Fox_Muldr [Fox_Muldremail@example.comWays.net] has joined #linuxcnc
[06:07:36] -!- lwizardl [firstname.lastname@example.org] has joined #linuxcnc
[06:08:26] -!- AR_ has quit [Ping timeout: 255 seconds]
[06:08:45] -!- davec [email@example.com] has joined #linuxcnc
what are some good programs for building the layouts of materials.
[06:11:36] -!- cevad has quit [Ping timeout: 244 seconds]
looking at these so far. Xara Xtreme, Skencil, sK1 Project, Inkscape
[06:49:42] -!- minotaur [firstname.lastname@example.org] has joined #linuxcnc
[07:04:00] -!- kwallace [email@example.com] has joined #linuxcnc
[07:08:27] -!- kwallace [firstname.lastname@example.org] has parted #linuxcnc
[07:09:14] -!- mhaberler [email@example.com] has joined #linuxcnc
[07:12:20] -!- jpk has quit [Ping timeout: 248 seconds]
pcw_home: How's your ether board coming along?
[07:17:38] -!- firephoto [firstname.lastname@example.org] has joined #linuxcnc
Aero-Tec, nah but i did ask for help in the channel and people are very helpfull =)
If using a hand-held router as a spindle is it not possible to set the RPM in EMC or Mach3?
or monitor for that matter
[07:55:34] -!- DJ9DJ [DJ9DJ!~Deejay@unaffiliated/dj9dj] has joined #linuxcnc
RyanS, if you slap on an encoder of some kind yes youc an monitor it .. or tachometer
controlling it ... hmm they are usaly controllable from a knop on them, just hack that in some way? :)
but generally you would have to control it manually if you want to keep it simple?
You are not going to change RPM during a job... Surely.
yeah i guess you will only change rpm with change of cutter
and cutting speed
[08:00:17] -!- Nick001 has quit 
monitoring rpm could be a good thing, if it drops to 0 you can stop the machine from linuxcnc :P
[08:02:06] -!- c60 has quit [Ping timeout: 256 seconds]
is structural steel angle any good as a frame material in a DIY CNC router?.... I can't imagine it being dimensionally accurate to any great degree
[08:08:58] -!- lwizardl has quit [Quit: Leaving]
[08:14:15] -!- FinboySlick has quit [Quit: Leaving.]
im building my router with 40x80mm box tubing
tho i do not expect 0.0001mm accuracy either :P
RyanS, what do you mean by dimensionally accurate
if you mean cold rolled steel, no, it's not accurate
Does that matter if you design design the machine properly?
what expectations do you have on accuracy in the machine ?
you can use structural steel as the frame material and it will be far, far more rigid than, say, MDF
mrsun: nanometers to 12 digits!
but you will have to test and correct whatever axis guide you're using
if you use something like linear rails, you can shim the rails up and down
into rough alignment
mrsun: I want to THINK about having to fart and toss it out of alignment!
Guppy, heh... http://www.youtube.com/watch?v=arytH0E5xCY
[08:18:52] -!- motioncontrol [email@example.com] has joined #linuxcnc
Accuracy I don't know.. You will probably laugh 0.1mm perhaps?
Too much to ask?
[08:22:04] * Jymmm gives toastydeath a noogy!
Is this sort of design good, average, crap? http://www.kronosrobotics.com/krmx02/index.shtml
I like the idea of reinforcing the 80/20 with steel
why even use 80/20 at that point
the whole point of extrusions is not having to cobble together all kinds of angle and plate and drill holes all over the place
idk is it more accurately sized than cold rolled steel?
Perhaps alu. profile gives accuracy and steel rigidity ?
[08:26:57] * icee shrugs
you can always scrape steel flat
With what some industrial processing machine.....
aluminum extrusions aren't exactly super precise dimensionally either
no, a set of hand scraping tools and a good datum and some layout blue
it goes way faster than one would expect
That is the most clusterfuck of a concept I've seen yet
RyanS, making machines accurate is a hand-worked process
no machines involved
ok so crap design?
design doesn't make much sense, no.
i do kinda like the bed-of-extrusions
you'll be sure to have enough t slots
why? I'm just wondering because I have no idea
They tried getting cheap by using steel plate and skate bearings instead of linear rails
or v0groove bearings, whatever.
too many bolts, too complicated ?
RyanS: I honestly don't understand their design decisions at all
bad design.... they dont have a mill to fab, they only have a drillpress and a saw
It said something about lower centre of gravity....
Is that BS?
a low cetner of gravity is nice to the point.. that the machine doesn't tip over
any lower doesn't matter (much)
horseshit... the 2x4 extrusion is top heavy alone
or is that 2x6?
I like gantry machines where the crossrail can be raised and lowered
much, much more rigid
and cheap hardware too
not even ss
adding a 1/4 of the current distance halves the static rigidity of the machine
As in something like because the sides are raised the gantry pillars are not as tall? Perhaps that is why?
RyanS: DONT BUY IT
I wasn't planning on
does anyone make a kind of.. general purpose gantry machine?
i'd really like it if
RyanS: They only good thign I see on it is the base.
i could hot wire, plasma cut, and route full sheets of ply from the same thing
hot wire no
and can build it myself if i have to, but it'd be nice to have a shopbot++ good-enough design
i realize hot wire makes it a lot harder
but i've seen a machine that had plasma, torch, and router
which makes it optional
I know how to use solidworks pretty well but I'm no techie
also a drag knife
solidworks doesn't design the machine
solidworks is how you go from a design to a set of prints
...so why bring it up
solidworks can be fun to play with concepts in
calculate all your moments, verify things actually clear, etc
Because I have the ability to document a design
At least I don't have to learn that bit
I'd rather see that machine as two large pillars riding on a rail that's below the table
and the crossrail bolted to the pillars
toastydeath: do you know offhand who makes the multipurpose gantry machine you saw?
also the router is way overhanging the bottom of the Z axis for no apparent reason
icee, negative, it was like 6 months ago
iirc all i was doing is searching for "cnc router"
So rather than dropping the bed it's better to just have a taller Z?
and looking for 4x8, 3 axis machines
no, it's better to have as little kinematic path between the part and the tool
[08:37:05] -!- mhaberler has quit [Read error: Connection reset by peer]
"taller z" doesn't totally summarize that
[08:37:12] <icee> http://www.plasmaroutecnc.com/index.php
ah, i found this
[08:37:18] -!- motioncontrol has quit [Ping timeout: 244 seconds]
longer z stroke
[08:37:35] -!- motioncontrol [firstname.lastname@example.org] has joined #linuxcnc
Long z stroke, and the position of Z should be adjustable if possible
didn't think of that..
[08:38:04] -!- mhaberler [email@example.com] has joined #linuxcnc
if you have a water table, you're not doing vacuum bleed holddown
put some posts in the water table
[08:38:26] -!- automata [firstname.lastname@example.org] has joined #linuxcnc
build a vac table
If you make it too long do you loose rigidity ?
RyanS, let me draw something
toastydeath: is it pr0n?
toastydeath: is it ascii pr0n?
toastydeath: is it colorized ascii pr0n?
If I was in the US, I'd buy this and have no design headaches. http://romaxxcncrouters.com/cnc-router-store#ecwid:category=2549172&mode=product&product=10142335
it seems like a good price for turnkey
[08:40:11] <toastydeath> http://www.twiddla.com/1019356
it can be ascii porn i you draw it
[08:41:04] -!- automata has quit [Client Quit]
toastydeath: I think I'd ave prefered the ascii version =)
toastydeath: not bad on the text drawing =)
i have a wacom
yah, i don't see any examples of water table + vacuum table
[08:42:56] -!- Cylly has quit 
toastydeath: how is it for vector drawing?
[08:43:20] -!- Loetmichel [Loetmichel!~cylly@p54B11FF0.dip.t-dialin.net] has joined #linuxcnc
toastydeath: and what size do you have?
i guess it's not a super common use case
it's tiny as balls
and it's kinda pointless for vector drawing
toastydeath: how so?
dunno, that's just how i feel about it
i always use the mouse
it's really hard to adjust points and splines and things
toastydeath: pita for node editing you mean?
also bad for single-line drawing
it's best for digital painting
where you don't have to be accurate
on the other hand
[08:46:42] -!- Belerafon_L [Belerafon_L!~Belerafon@220.127.116.11] has joined #linuxcnc
the nouveau-spaceball-ish / 3dconnexion stuff
is the best thing ever
toastydeath: I thought it might be nice to program certain things into it to relieve the mouse PLUS kybd comannds
it might work for you
it doesn't for me
Not that big a deal, just cumbersome sometimes
I pretty much use it for math tutoring/online whiteboarding only
flipping back and forth between tools is getting old.
I've never used a drafters tablet, but that seems more of what I need.
or whatever they are called.
[08:48:34] -!- archivist has quit [Ping timeout: 260 seconds]
like, the actual drafting machines?
i have one of those
toasty so the crossrail is the horizontal beam on the gantry or Y axis?
the crossrail is the Y+Z axis assembly
And to making adjustable height on that means you can tram the machine?
you HAVE to tram the machine
every time you move it, you have to realign it
[08:54:54] <toastydeath> http://www.abltechnology.com/equipment-images/large/7469.jpg
here's a good example of a machine that has a movable crossrail
notice how the Z and X axes (since it's a lathe) are on the crossrail
but the crossrail itself moves
Having a movable crossrail also means you can cut on a slant
if you design it cleverly
But you would not CNC that bit?
generally no, the crossrail never moves
and so isn't cnc
personally, I'd just make it bolt-on
the Bullard in that picture weighs 50,000+ lbs, so the crossrail is probably 8000-10k lbs
so it has to have screws to move it
alu extrusions, by contrast, are easy to move and tram
So how do you tram the X axis if the gantry is 'nodding' forward or backward?
the x axis on the bullard or on a 3 axis mill
3 axis mill
Moving gantry style
that's dependent on how you do the gantry
Like the typical DIY router
since X is the entire gantry moving, the alignment is relatively fixed
let me draw again so we have the same frame of refg
[08:59:37] <toastydeath> http://www.twiddla.com/1019361
talk in here though
what kind of nod are you talking about
I guess you can shim the linear rails for X
and that's what I recommend
it kills rigidity but scraping is probably beyond your skill level for some time
I need side view
RyanS turn your monitor sideways
[09:02:48] -!- mhaberler has quit [Read error: Connection reset by peer]
that doesn't happen
if it does, it means your X is out of alignment vertically and you shimmed it wrong
But not difficult to fix?
umm and ..
if it nods forward that means the rails are bowed
[09:05:05] -!- archivist [email@example.com] has joined #linuxcnc
the Z axis nod in the Y direction is fixed by unbolting the rails a little and hitting it with a hammer
until it's in alignment
(use a deadblow with a soft face)
Unbolt the spindle moount?
And of course the Y by shifting the crossrail
I think I'd make the base out of hulking great pieces of structural steel
if you can find iron, that's best
it makes a big difference for chatter
I know why the KRM has the be spilling over the front
because it's not a very good design
it doesn't matter WHY it's like that
Because crossbeam pillars are vertical and not slanting backwards
no, that's fine
i draw the bigass slanting pillars because it's rigid
the closer to the base you are, the more important it is to have rigidity there
so very often in machine castings you'll see a tapered profile
because it makes no sense to have a really thick top
because there's no stress there
[09:10:09] -!- rob_h [firstname.lastname@example.org] has joined #linuxcnc
I wonder if you could just buy triangular shaped aluminium flat to use as the pillars.. I don't have access to a milling machine
use whatever you got
just make them really thick in the X axis
in both, if you can
because they're the major element absorbing the milling force in the X and Y direction
the crossbeam absorbs most of Z
and by absorb, i mean contributes the most to deflection
We have lots of plate but nothing to cut with besides an angle grinder or a hacksaw
which is why the Y axis on a gantry should ALWAYS be the smallest one
but an pretty beefy drill press which managed to drill through 6mm stainless plate without dying
Is that not an achievement?
depends on the size of the drill
oftentimes small drill presses stuggle more with themselves, in that they don't deal well at slow speeds
rather than with actual drill or material considerations
It ate 1/4" cobolt drill bits
Ran pretty low RPM too
Cut really hard
SS also work hardens
which may be why the tool died
yeah we tried to cut aggressively
by hand I like to cut as slow as possible and basically hang off the feed
with 1/2 of my bodyweight or so
[09:15:26] -!- mhaberler [email@example.com] has joined #linuxcnc
nice thick chips mean no work hardening
You basically have to 'gouge' metal rather than scraping
the counterintuitive thing is that the bigger chip you cut, the less horsepower per volume per minute it takes
so if you cut the speed in half, you will be able to more than double the chip load
which means you'll be cutting even faster per minute
the tool will go 1" at speed 1, and 1.1-1.2" at speed .5
I have a SS still made with sanity tube fittings.... I just need to get around to finishing the electronics wiring and and assemble the thing
as long as the machine can handle that larger load
We had to drill through the end caps to make a shell and tube heat exchanger
Those were 6 mm
It was a lesson in budget blowout and getting carried away in CAD
[09:28:19] -!- nextstopsun has quit [Quit: Colloquy for iPhone - http://colloquy.mobi]
[09:53:02] -!- mhaberler has quit [Quit: mhaberler]
[09:58:46] -!- mhaberler [firstname.lastname@example.org] has joined #linuxcnc
[10:11:39] -!- neverho0d has quit [Ping timeout: 256 seconds]
[10:18:00] -!- vladimirek [email@example.com] has joined #linuxcnc
[10:18:27] -!- cncbasher has quit [Ping timeout: 256 seconds]
[10:20:59] -!- Belerafon [Belerafon!~Belerafon@18.104.22.168] has joined #linuxcnc
[10:20:59] -!- Belerafon_L has quit [Read error: Connection reset by peer]
[10:21:33] -!- mhaberler has quit [Quit: mhaberler]
[10:29:10] -!- yuvipanda has quit [Quit: yuvipanda]
[10:55:25] -!- racycle has quit [Quit: racycle]
[11:09:33] -!- motioncontrol has quit [Quit: Sto andando via]
[11:13:39] -!- mhaberler [firstname.lastname@example.org] has joined #linuxcnc
mhaberler: hi. any insanely-great developments lately? (as usual! :)
hm, visitors present so pipeline stalled ;)
ok.. I am trying this atom-board, which unfortunately has "cedarview" graphics.. well hopefully it will work
with RTAI I assume?
awallin: I know I need to write this up, but that branch is still shaky and wip, so leave a bit of time until it approaches merge candidate..
we dont have any tested PCI drivers yet, for instance
[11:33:54] -!- ybon has quit [Quit: WeeChat 0.3.8]
I'm trying just a normal ubuntu install now, no real-time. it seems either the cedarview package is broken, or it takes a LONG time to compile/install
I bet there are no cedarview drivers for 10.04... if I wanted to try RTAI (or can I use an old RTAI kernel on a 12.04 distro?)
[11:45:36] -!- asdfasd [email@example.com] has joined #linuxcnc
[11:46:07] -!- r00t4rd3d [r00t4rd3d!~r00t4rd3d@unaffiliated/r00t4rd3d] has joined #linuxcnc
I would think that works
what changes is: you get the set of drivers supported by the RTAI kernel, not the set assumed by precise, so one or the other hw-dependent peripheral might not work as expected
ok... what I would want is either xenomai or rt-preempt on 12.04lts with reliable 1ms servo thread, then work on gtk+HAL standalone things..
10.04lts + RTAI will just continue to generate more and more problems with new motherboards I feel
[11:59:54] zz_satyag is now known as satyag
[12:00:41] -!- Belerafon has quit [Quit: KVIrc 4.0.4 Insomnia http://www.kvirc.net/]
[12:06:02] -!- mattions has quit [Quit: Leaving]
[12:07:30] -!- mhaberler has quit [Quit: mhaberler]
[12:10:40] -!- hdokes|werkin [hdokes|werkin!~IceChat77@173-165-40-177-Illinois.hfc.comcastbusiness.net] has joined #linuxcnc
[12:11:33] -!- hdokes|werkin has quit [Client Quit]
[12:11:55] -!- hdokes [hdokes!~IceChat77@173-165-40-177-Illinois.hfc.comcastbusiness.net] has joined #linuxcnc
[12:15:02] -!- rizo [rizo!~klemen@BSN-61-60-156.dial-up.dsl.siol.net] has joined #linuxcnc
grrr cedarview driver install seems to have failed now for the 3rd time. just reboots the machine in the middle of the install and when it boots again the screen is just garbled up and unusable
Hello, one short question: The output of PID position loop is velocity. Ok, but how can i connect the velocity to torque input of sine servo amplifier?
the output is just a number.. you can connect it anywhere
What on earth is this used for ? http://weldingweb.com/showthread.php?t=44756
I know it can be connected anywhere :) but i suppose it should be connected in a way that make sense
rizo: torque mode will work, the interpretation of what the gains mean/do just changes
i am still not sure: the output of pid is scalar, but i need a vector to control the servo amplifier. I need to specify the currents through 3 windings.
rizo: oh, there should be some bldc components for that, I think by andypugh
do you have a mesa fpga-card for pwm-generation?
no...some coustom made DA converter.
The voltage is proportional to phase current.
do you have a linuxcnc driver for this then?
no, but i will write a code for simple hal component so that i will control the desired voltage.
rizo: try http://www.linuxcnc.org/docs/2.5/html/man/man9/bldc_sine.9.html
I thing figuring out where exactly the rotor is, so that you get the sine-commutation right is the tricky part
jeez hand scraping is crazy stuff
[12:33:44] -!- mhaberler [firstname.lastname@example.org] has joined #linuxcnc
yes, I fear so
that combo should work fine (pci is in the works, Charles is on the pci shim case and has a 5i25 to test)
mhaberler: so I fear I'll have to abandon the dn2800mt atom-board because of the cedarview graphics :(
awallin, i see, i think this bldc_sine is exactly what i need :) Will try that, thank you!
never heard of it, need to look that up
[12:37:45] <awallin> https://www-ssl.intel.com/content/www/us/en/motherboards/desktop-motherboards/desktop-board-dn2800mt.html
apparently the graphics chip on those is well known to be really really badly supported on linux. I guess I should have checked before getting the board..
is there a driver issue, or latency?
I can't get the graphics driver installed
you mean on rtai?
no just normal 12.04 install
and that: ? http://daily.siebler.eu/2012/06/ubuntu-12-04-driver-for-intel-cedarview-atom-n2000-und-d2000-serie/
you likely need to rebuild from source because of the old kernel
yeah those are now supposed to be in the repos. and ubuntu automagically suggests them - but the installer fails
just reboots the machine in the middle of the driver-install
after that when booting the screen is garbled
did anyone try rtai on amd fusion cpu?
fusion.. whatzat, 64bit?
comparable to intel atom
low power low performance etc
that looks indeed a bit desparate: http://askubuntu.com/questions/168986/cedar-view-drivers-arent-working
the first answer looks like he got it going
not for the faint
is "intel HD graphics" a well defined chipset? :) with good linux support?
[12:47:52] <awallin> http://www.intel.com/content/www/us/en/motherboards/desktop-motherboards/desktop-board-dh61ag.html
this is terry incognita for m
[12:53:36] -!- cncbasher [email@example.com] has joined #linuxcnc
[12:54:07] hdokes is now known as hdokes|werkin
[12:54:24] -!- Simooon [Simooonfirstname.lastname@example.org] has joined #linuxcnc
[13:01:19] satyag is now known as zz_satyag
[13:03:12] -!- morfic has quit [Ping timeout: 252 seconds]
mhaberler: any idea if it will be difficult to get PCI(E) drivers working on xenomai/rt-preempt?
not really; it's going to be all memory mapped
no interrupts used, so straighforward; Charles is working on replacing the pci kernel support with a shim include file/maybe a bit of support code to make drivers run unchanged except for the include/define/link mods
ok, sounds good
it's going to be the same for all userland threads, so you'll be able to test a hm2 driver with simulator provided that makes any sense;)
still requires the fpga-card to be present though?
well yes, we cant simulate that just yet
sim really was a misnomer - it means 'no drivers'
(and reduced kernel requirements, i.e. none)
I am fairly sure that the beaglebone $85 board with stepgens in the onboard RT CPU will blow away any RT threading we have up to now
(sw stepping I wanted to say)
it comes with two free 200Mhz 32bit 'arduinos' on board which can talk to pins
but then non-x86 is always going to be trickier than plain x86..
well its really the intel/pc monoculture vs the ARM board djungle
it is really per-SoC
not per board
[13:15:53] -!- newleaf has quit [Ping timeout: 245 seconds]
I'm trying to import a python file into my gui but for some reason I get an error, I copied the file with sudo cp jog.py /usr/share/pyshared/gui7
now the strange part I did this a while back (gui6) and I can say from gui6 import jog and it works
but when I say from gui7 import jog I get this error ImportError: cannot import name jog
I'm using touchy as a guide
do you have a duplicate name (a proper module with .so extension and a python file with same basename but .py) ?
in that case the python interp goes for the .py file and fails to import from the .so
I don't think I have a .so file
or an old .pyc file with same name
(precompiled python, appears after first run)
let me look
is /usr/share/pyshared/gui7 on the PYTHON_PATH?
try sys.path.add(..) that directory to make sure python looks there
ImportError: cannot import name jog
['/usr/bin', '/usr/bin', '/usr/lib/python', '/usr/bin', '/usr/lib/python', '/usr/lib/python2.6', '/usr/lib/python2.6/plat-linux2', '/usr/lib/python2.6/lib-tk', '/usr/lib/python2.6/lib-old', '/usr/lib/python2.6/lib-dynload', '/usr/lib/python2.6/dist-packages', '/usr/lib/python2.6/dist-packages/PIL', '/usr/lib/python2.6/dist-packages/gst-0.10', '/usr/lib/pymodules/python2.6', '/usr/lib/python2.6/dist-packages/gtk-2.0', '/usr/lib/pymodules
/python2.6/gtk-2.0', '/usr/local/lib/python2.6/dist-packages', '/usr/share/pyshared/gui7']
seems to be on the path
[13:33:50] -!- sumpfralle [email@example.com] has joined #linuxcnc
permissions on /usr/share/pyshared/gui7/jog.py ?
or the directory /usr/share/pyshared/gui7 ?
it's useful to include a git commit id or some other revision string in the modules so you can check what you have right after "import"
I use `git describe --dirty --always` for a git scent mark
[13:39:06] -!- The_Ball has quit [Ping timeout: 264 seconds]
the file permissions for jog.py are the same in both /usr/share/pyshared/gui6 and gui7 -rw-r--r--
can't figure out the command to see what the directory permissions are yet
ls -ld <dir>
both directories are drwxr-xr-x
cd to that dir, start python, and try 'import jog', would that work?
yes that works
so its either a path issue, or it picks up some jog.py on a a dir before yours
this will prepend that dir before all others
so if some other dir has a jog.so or jog.py or jog.pyc, it wont suck that in
oh, I see /usr/bin in your path.. why?
I seem to be having a copy problem, I changed the from gui7 import jog to from gui7 import status and I still get the jog error... let me check my copy script
[13:50:43] <jthornton> http://pastebin.com/VXW9Ci3V
monkey see monkey do, I copied some from touchy
hm, you're using subpackages: from gui7 import status
have an __init__.py file in that gui7 dir?
I think I see the problem I had gui7 and gui7.py in /usr/bin
so no matter what I did I still got the old file me thinks
no __init__.py in the gui7 directory, I'll add one
oh it works now, thanks
[14:04:15] -!- mhaberler has quit [Quit: mhaberler]
i am looking for bldc_sine component (http://www.linuxcnc.org/docs/2.5/html/man/man9/bldc_sine.9.html).
There is man.9 for the component, but not compiled in latest release 2.5 and also there is no source code at latest [linuxcnc.git] / src / hal / components /
is this hal component still in development phase?
[15:09:19] -!- mhaberler [firstname.lastname@example.org] has joined #linuxcnc
[15:23:36] -!- IchGuckLive [IchGuckLiveemail@example.com] has joined #linuxcnc
Hi all B)
it appears the man page has preceeded the comp
rizo: ask on the forum and Andy will be sure to see your question
BLDC_sine is obsolete having been replaced by BLDC
then the man page needs to be vaporized :)
hmm it is not a hand written man page
and both BLDC and its manual are in 2.5
[15:36:42] * JT-Shop is at a loss how the man page is there for bldc_sine
[15:38:20] zz_satyag is now known as satyag
BLCD are getting cheeper and cheeper
the BLDC comp does everything the BLDC_sine comp did and more (its kind of a kitchen-sink comp)
pcw_home: Is the 7I80DB the ONLY ether I/O card?
Jymmm: did someone already write a linuxcnc driver for that board?
awallin: I'm not sure.
[15:56:32] -!- r00t4rd3d has quit [Quit: Leaving]
[15:57:44] -!- Nick001 [Nick001!~Nick001@clsm-74-212-21-50-pppoe.dsl.clsm.epix.net] has joined #linuxcnc
[16:00:35] -!- the_wench [firstname.lastname@example.org] has joined #linuxcnc
Well there's the 50 pin I/O version (7I80HD)
driver is being worked on
pcw_home: 2x 50pin?
pcw_home: how do you get 4 extra pins by chaning connectors?
7I80DB uses Parallel port compatible pinout so 17 I/O x4 =68
"port compatible "
[16:06:26] -!- AR_ [AR_!~AR_@22.214.171.124] has joined #linuxcnc
pcw_home: Ok, I dont understand the differnce then
both 7I80s have 5V tolerant I/O
only difference is connectors
Oh, the 17 pins being the ieee-1284 compatability
Yes (well and all PC parallel ports from the Dawn of time compatible)
so just matching the pinouts, not different signal voltages
got it, are the other 4 pins "available" ?
parallel ports are just TTL compatible no more no less
Bummer, might hae been nice for extra LED's or whatever.
I'd rather the cards were identical except things that have to be different so the support firmware source and drivers are identical
[16:11:53] -!- motioncontrol [email@example.com] has joined #linuxcnc
I understand. What is "LX16" ?
there are 2 different FPGA sizes available (well possibly 3 but the smallest is pretty cramped)
pcw_home: It's not a DSP of course, but can you think of any reason it couldn't be used as such?
FPGAs are often used as DSPs
pcw_home: That's what I mean, and I was asking if the 7180 could be used as such
Probably one of the more common uses (Cellular base stations etc)
pcw_home: what kind of bandwidth would you guestimate?
The host bandwidth is not very high over Ethernet
[16:22:22] -!- rizo has quit [Quit: Leaving]
pcw_home: Not as a pseudo DSP sampling bandwidth
pcw_home: No, as a pseudo DSP sampling bandwidth
but there lots of processing power available (I think the are either 32 or 48 MACs available)
say 100 to 100s of MHz
pcw_home: 100MHz sampling bandwidth, really?
is that pushing it to it's limits?
Thats why you will see most high end A-D connect to FPGAs
pcw_home: Is there an equiv clock speed *I* could use to compare to an (lets say) intel cpu?
and the Spartan 6 supports LVDS serdes on all pins to close to a GHz
pcw_home: (BTW, the idea I have is a SDR Transceiver)
pcw_home: using the ether as UI I/O
medium sized FPGA to Intel CPU Raw floating point much lower pipelined fixed point equal or higher for 1/50 the power
pcw_home: I meant more like i5, P4, 286 as *I* have no comparison
It is rather apples and oranges
pcw_home: I know it's apples and walnuts, but just as an idea
get a lemon
archivist: I got LOTS of those =)
Its also really dependent on the skill of the FPGA firmware designer
If you have a Ray Andraka to design it it will fly. If I do it it will crawl...
pcw_home: and I do it, it will collect dust =)
pcw_home: Xilinx® Spartan® 3A-DSP 3400 FPGA
as i viewd around today of bad weather otside iv seen so many minimills are now build in facturies around the globe
also lots of lathe
i got also the new Deskproto installt today and it holds now 250 postprozessor of all kind of mashienes
including EMC i will tell him the nameconversion
it has now also a ROBOT post
lots of folks still use EMC
[16:39:57] satyag is now known as zz_satyag
so the demo is nice to be installed at a low cost pc every 30days :D ;-)
JT yes yesterday if been on 2.3.5
Question if i have steplen,stepspace,dir,... set do i then need the parport reset time
[16:46:35] -!- ktchk [firstname.lastname@example.org] has joined #linuxcnc
No sure, where is the reset mode documented? is there a man page?
i did also not found infos about this particular function
I have an entry in my pport machine. I assume stepconf put it in for me.
[16:55:58] <pcw_home> http://www.linuxcnc.org/docs/2.4/html/hal_parallel_port.html
seems to suggest that 1 for steplen on 0 for stepspace is OK if using reset mode
jdh: what is the number
setp parport.0.reset-time 2000
i added 5000
(U32) parport.<portnum>.reset-time -- The time (in nanoseconds) between a pin is set by write and reset by the reset function if it is enabled.
maybe i shoudt set up the hole mashine new with a higher linuxcnc version
[16:59:36] -!- ktchk [email@example.com] has parted #linuxcnc
IchGuckLive: if you dont need 5 usec I would reduce the reset time to the minimum as the PC just spins during that time
(so you will slow the PC for other tasks if you waste time spinning in the reset function)
i just edit the hal and in i to get some of my friend to a better speed
i woudt have loved to get I-net access at that tiime but the garage where ofline
k by its X-mas festivity in tow n
Christ child is hedding to the church tower to set of the tree
[17:22:59] -!- IchGuckLive has quit [Quit: ChatZilla 0.9.87 [Firefox 16.0.2/20121025205401]]
Hi at all. I command in position the exsternal axis with pid.x component, but for have ramp acc. and dec how can reailize it ?
normal axes/joints will have limited speed and acceleration (from INI). spindle might need a limit2/3 block.
i thing no, because i can connect it only on pid.x.command. example stay at position 0 pass new postiton 100 at limit 3 , it ramp in acc , but not in decc.
limit3 does both
yes buf if you change the target position , but you set only one position 0 -> 43
limit 3 is ok if i have 0->100 and 100 ->0
but when command a axis i command only one targhet position
limit component is good for spindle acc and dec
I'm not sure I follow... A-axis should have limits on acc and speed in the INI. In g-code the feedrate may be given in 'inverse time' format?
but not for position
for axis declare in ini is kinematic at create acc and dec i thing
Not sure what the problem is you should be able to dynamically change the .in value and limit3 should do its best to follow with violating its limits
trajectory generator create ramp
if it does not do this either your hal file or limit3 have a bug
limit 3 creates a trapezoidal profile just as the trajectory generator does
a axis declare in ini, when commanded in position first accel , after decel
(if you are using the trajectory generator (normal axis) you dont need limit3
with limit 3 you can only acc , for dec you change again targhet position , it not is correct
i want use pid component for exsternal axis not declare in ini file
If that true, ita a bug in limit3
are you saying that if you change the endpoint, limit 3 violates the acceleration limit?
the problem is end point, not have dec.
how are you setting the target position (not with the limits I hope)
limit is ok for spindle . example : spindle is stop, now m3 s100 have ramp acc, now m3 s10 have ramp dec
I don't know how the limit could know if you change the end point on the fly.. Say if you go from 10mm to 5 and it is already almost to 5..
it could be in the cruise phase - then if you change the endpoint to something that would require a faster de-acceleration than what is set.
it should always do the best it can under its contraints and current endpoint
otherwise it is broken
oh - I see - it should shoot past the end point...
motioncontrol: limit3 should do what you want either you are using it wrong or it has a bug that should be fixed
I am using limit 3 for the z axis temp offset.. but I am only moving .005" ;)
It really does not seem likely that such a big bug has been missed
motioncontrol: do you only see the problem if you change the endpoint when in motion?
thing at exsternal linear axis . now stay at 0 position.want targhet 100. ok pass 100 at limit 3 and it ramp on pid.x.command . pid rad ramp and executed it, this is ok. now you thing the targhet is very long, the axis after some second have velocity costant and arrived at targhet 100 position. The pid output fast is 0, not have ramp on limit some second first the targhet position. i thing limit approch not is corre
yes the problem is motion in end position
that is limit,N.in?
sounds like the acc is set incorrectly...
do yo have enough headroom ion the position limits? (they should be larger than the actual travel)
if use pid component only is good for control temperature on m19 spindle , but no for move a axis
the command the pid is problem, because only a traject generator can calculate the ramp and velocity
false limit 3 does
exist in emc another approch , thing loadrt another kinematict or another trajactory generator ?
thats is thats one of its purposes, to do a acceleration limited trajectory
limit3 should work
motioncontrol: are you saying that even if you do not change the endpoint (limit3.N,in) in motion, that limit3 does not generate a trapezoidal motion profile?
no not ganerate it
then there is a bug but this is surprising since that is exactly what limit3 is supposed to do
Are you sure you have .min and .max set OK and not changed (you should only change .in)
limit fuction thinking not is generate trapezoidal motion, but only acc and dec, but for this you change two value, 0-100 acc , 100-0 dec
motion condition is only 0-100
what? limit has no such pins
limit3s pins are in.out,load,min.max.maxv,maxa
limit have limit value pin and pin input variable and output pin variable control
yes ok for pin
please use the pin names from the limit3 component otherwise its hard to know if we are talking about the same thing
min.max.maxv and maxa are pinf for parameter the limit for select limit plus and minus and response the acc and dec
the pin input is input value the variable targhet
the pin output is variable calculate with acc and dec.
and .in the only one you should be changing for a trapezoidal profile
and the min and max are untouched and larger than you travel?
yes is limit min and max the output
right so they should be larger than you full travel (and never changed)
in limit when input increase have acc in output when decrease have dec in output
for have a complete acc and dec i change first increase value and after decrease, change two value, for motion you have only one targhet command and if use limit you can have only acc, because in limit when you increase value after some secont output is egual at input and not decrease if you dont' decrease the input
[18:11:19] -!- AR_ has quit [Ping timeout: 244 seconds]
if i put a step input into limit3 (.in pin) it will generate a complete trapezoidal profile
(otherwise it violates then acceleration constraint)
so it output simply follows the input while obeying the acceleration and velocity limits
what this step input ?
0 one sample and 100 next (and 100 till limit3 is stable)
I'm not saying limit3 may not have a bug but looking at its code it does what the manual page says: that is its output follows its input and the velocity and acceleration constraints = trapezoidal profile
motioncontrol, write up a test case to show any fault you think there is
I guess I could do this in my sim install at home
yes limit 3 fuction ok, but with 0-100 you have only acc, when stay arrived at endposition not have dec this is the problem.
limit3 is _not_ functioning OK if it does not decelerate with the same accel values as it accelerated
[18:51:21] -!- r00t4rd3d [r00t4rd3d!~r00t4rd3d@unaffiliated/r00t4rd3d] has joined #linuxcnc
[19:07:02] -!- AR_ [AR_!~AR_@Dyn148117.cc.lehigh.edu] has joined #linuxcnc
[19:07:54] -!- L33TG33KG34R has quit [Ping timeout: 264 seconds]
[19:14:43] -!- morfic [morfic!~morfic@unaffiliated/morfic] has joined #linuxcnc
[19:18:45] -!- L33TG33KG34R [L33TG33KG34R!~L33TG33KG@S010674ea3aa162f7.cg.shawcable.net] has joined #linuxcnc
[19:24:10] -!- tjb1 [firstname.lastname@example.org] has joined #linuxcnc
[19:27:16] -!- AR_ has quit [Ping timeout: 244 seconds]
[19:30:02] -!- AR_ [AR_!~AR_@Dyn148117.cc.lehigh.edu] has joined #linuxcnc
[19:31:20] -!- motioncontrol has quit [Remote host closed the connection]
[19:32:51] -!- yuvipanda has quit [Quit: yuvipanda]
[19:34:47] -!- c60 [email@example.com] has joined #linuxcnc
[19:37:24] -!- grummund has quit [Ping timeout: 264 seconds]
[19:38:49] -!- tjb1 has quit [Quit: tjb1]
[19:40:13] -!- grummund [grummund!~user@unaffiliated/grummund] has joined #linuxcnc
[19:45:07] -!- AR_ has quit [Ping timeout: 256 seconds]
[19:48:04] -!- maximilian_h [firstname.lastname@example.org] has joined #linuxcnc
[19:49:16] -!- maximilian_h has quit [Client Quit]
i need to control the speed of a single phase induction motor load.
(not software control... just an appropriate drive to do so).
i know router speed controls can be abused as slip-controls for induction loads but that's not really good.. and i know of /have used/ have found lots of 3 phase VFDs
better to change the motor
I you woudl be a lot better off to use a VFD and a 3 phase motor
but.. any 1 phase VFD or similar that could fulfill the task out there? seems hard to find. load is a 120V 1HP 2pole induction motor
yah, i'm not going to re-engineer the shitty blower i'm trying to slow down
how big a motor?
induction motors have a false phase to create the rotating field, just change the motor to 3ph and use a vfd
it may over heat if you try to slow it down
You _might_ be able to slow a shaded pole motor but less likely to be able with a motor with a starting winding
skunkworks: you were 100% with your edit of my hal file
pcw_home: before, i've cheated and .. started similar motors at full speed and slowed them down
which i know is naughty
Yeah but its all pretty iffy
[20:00:42] -!- c60 has quit [Ping timeout: 265 seconds]
[20:01:28] -!- munktell [email@example.com] has joined #linuxcnc
the encoder wire was moved from a lower position to a higher one and had oil on it, the oil ran down the wire and onto the opto encoder so it stopped working
drip loop FTW
Is it possible to get the current dynamic tool offset set with g43.1 from any parameter? Want to use it in a sub as a variabler.
have you read this http://linuxcnc.org/docs/html/gcode/overview.html
skunkworks: thanks a million for your help, the spindle speed indicator is now rock solid, the debounce worked like a charm
Yes, this is not lister as a parameter. You can get tool offset from tool table from parameter, but not the g43.1 ofset
g43.1 just adds to the current offset
Yes, but I need to know the offset before i do a g49. Its for a M6 remap thingy
the current tool offset?
I was just thinking, there should be a place where code can be posted for INI and HAL and any other things that is EMC related.
Yes, men not the offset from tool table, The dynamic offset
a resource area that one can have there code posted
if you programmed the dynamic offset you should know what it is
have a email address where one can send the code and then have someone look at it and put in into the library of code and snippets
just a thought, or does this already exist?
would be nice to have a location one can go to to see how others dealt with some problem or how they did some amazing thing with EMC
Hmmm, yes. But it's in a M6 remap sub, and I can not be sure no-one have set a new g43.1 between the sub is executed
wiki seems to be the collection spot and the forum
after awhile most things would be covered and be less work for the guru's here on the forum
the forum would be ton of searching, not even sure how one could or would search the forum for info
one thing that helps is browsing other hal files
the forum search works OK (used to be broken but seems OK after the forum software upgrade)
and the wiki is cool and very help full, but I did not see a good code example for debounce, little on debounce code for a spindle, little on debounce code for a spindle with just one index input
[20:19:41] -!- sumpfralle has quit [Ping timeout: 256 seconds]
I am sure someone else will need to do just what I did and there is no real place to look for that code
well you have to generalize component usage to see the power of hal
I could save the z position before g49 and then restore the z pos to that after the g49. Would work
so one can search the forum now?
still leaves things scattered and spread out
[20:21:54] -!- FinboySlick [FinboySlickfirstname.lastname@example.org] has joined #linuxcnc
also I am looking for a way to shut down EMC and and have it start up using the exact zeroing it had when shut down
[20:22:46] -!- munktell has quit [Remote host closed the connection]
[20:22:58] -!- munktell [email@example.com] has joined #linuxcnc
like t1, I zero t1 and all tools are zeroed off of t1
so now I have to put t1 to z0x0
then shut down, start up
move the jog to dead dead dead slow
[20:25:09] -!- AR_ [AR_!~AR_@Dyn148117.cc.lehigh.edu] has joined #linuxcnc
jog the z towards the chuck just like I did to go to zero, and x towards -x just like when it went to zero, home it and zero it
is there a quicker and better way?
the jogging is to over come back lash, I have the backlash turned on and set
but have to set jog to slow so it will not move the axis but will set the back lash in the right direction
you dont have home switches?
it is a lathe
not sure where I would put them
I have like 5 chucks I use
that would be the standard way, though they work better with servo than step motors since you have th eindex
the collet chuck is in very tight and if I have a big 3 or 6 jaw in there I can not get to the home for the collet chuck
I have servos on the lathe
not the spindle
a home switch can be anywhwre
It could be in the middle of travel
so throw a switch anywhere?
cool, could not do that with mach
had to be a the end
theres a home offset
so that would be the + and - limit that is in the INI
I may just have to put some in
what are the best ones to use?
optos can get junk in them
hal effects can get metal suck to them
micro switches look to be the best but not sure if they are accurate enough
not sure its a compromise (inductive are tough but inaccurate)
thought the hal effect switches were dead accurate
[20:35:36] -!- sumpfralle [firstname.lastname@example.org] has joined #linuxcnc
I think people are getting close to 1 mill repeatability with microswitches
they have such a good response time
I dont think you can get 1 mill without a pretty exotic Hall device
only thing is if metal chips get near them it can mess things up
and there is no limit or end to metal chips in a cnc
Normally with servos, 2 step homing is done (first to switch and then to index)
so only +- .1" switch accuracy is needed
(assuming 5TPI ballscrew)
will have to think on a switch, something that will not wear fast, and no leavers for chips to get into and bind up
[20:40:07] -!- sumpfralle has quit [Ping timeout: 244 seconds]
that would be if the servos are driven by EMC
mine are servos but controlled by gecko, so step and dir
[20:41:08] -!- vladimirek has quit [Remote host closed the connection]
[20:41:10] -!- sumpfralle [email@example.com] has joined #linuxcnc
there is no index ether
yeah no homing on index is a disadvantage
I can install new optos with index
it _might_ be doable with the gecko but theres no direct support for this (using index with a step.dir drive)
I guess if I had a index from each axis, one could work from that even with step and dir
so, back to the problem so, what home switches would be best for my system?
currently index feeds into the encoder component so its a little awkward
right now my axis have no index
just the spindle, and it only has index
probably better asked on the LinuxCNC users mailing list to reach a larger number of machine builders that have experience with homing switches
I started with a shoe string, and no cnc experience at all
there is a mailing list?
how do you get on it?
thought this was the place to be for EMC
theres the forum and the mailing list as well
if you go to linuxcnc.org then select 'community '
the first link is a form to sign up to the mailing list
[20:51:35] -!- munktell has quit [Remote host closed the connection]
[20:51:48] -!- munktell [firstname.lastname@example.org] has joined #linuxcnc
Aero-Tec: so it works?
like a charm
your a EMC wiz
it was some oil on the opto that made it not work, your code was perfect
thanks again for the help
[21:01:07] -!- kmiyashiro has quit [Quit: kmiyashiro]
well I am now on the mailing list as well
both user and developer
is there a way to better understand the HAL programming?
I have done programming so know some things about coding
it is more of a virtual wiring scheme
but I am not understanding HAL, yet
I hope to be well versed in HAL programming at some point, soon I hope
the more you do it - the better you get :)
for sure, but right now I feel more like a deer in the headlights of a train the there are black things on ether side of me....LOL
not sure where to start with HAL programming, is there a youtube vid or something
HAL programming 101
the manual has a hal 101
I have read some of the wiki, but it does not really fill in the blanks for me
[21:08:09] <skunkworks> http://linuxcnc.org/docs/2.5/html/hal/intro.html
will look for it and reread it if I have not read it all ready
is there a hal programming tool of sorts?
I was reading somewhere and it looked to me as if there was some sort of thing one can load into the gui that would allow one to do some hal programming
people have worked on that... but I don't thing there is any that works 100%
one very cool thing about EMC is it can do so much and is so versatile, the over whelming thing about EMC is it can do so much and is so versatil
maybe I can code some tools
first I need to figure it out
Its the differece between a configuration lanquage and selecting items on a menu
a language is much more flexible and powerful but has a steeper learning curve
maybe a open source tool set for EMC, complete, setting up of a machine, something more complete then the wiz
mach was selecting, quick and easy to learn, but was VERY limited
I like the configuration of EMC, but your right the learning curve is steep, more so if your running a company and your the only guy in the company, LOL no others to hand work off to, if I do not do it it does not get done
Even with a helpful GUI, understanding the configuration language frees you from the limits of what the GUI writer imagined were rasonable choices
hard to find the time to learn with out interruptions
I do love EMC
but yes for one time use LinuxCNC setup can be daunting
it is very cool
so back to, I have no home switches and would love to have EMC shut down and start up and be exactly where it was when shut down
[21:21:37] -!- vladimirek [email@example.com] has joined #linuxcnc
for zero of axis
for a given tool
[21:22:31] -!- tjb1 [firstname.lastname@example.org] has joined #linuxcnc
JT-Shop: You here?
I have a system and it seams to work,but takes some time
lots of steps and easy to mess up
also can one set up joging in mid program run
I have a case where I sometime get some tool chatter and would like to halt and recut that cut manualy
[21:25:21] -!- tjb1 [email@example.com] has parted #linuxcnc
jogging is locked out during program run, and I like that, but would be nice to have a gcode that would unlock it for one small bit of code and then relock it
any help with ether of the problems?
Probably need JT or some other gcode maven
[21:33:10] -!- ve7it [ve7it!~LawrenceG@S0106001c10b7770f.pk.shawcable.net] has joined #linuxcnc
[21:33:20] -!- AR_ has quit [Ping timeout: 244 seconds]
Well I can't duplicate motioncontrol's problem with limit3, it works as advertsed
[21:46:32] -!- DJ9DJ has quit [Quit: bye]
[21:49:13] -!- munktell has quit [Remote host closed the connection]
[21:49:26] -!- munktell [firstname.lastname@example.org] has joined #linuxcnc
[22:07:07] -!- munktell has quit [Read error: Connection reset by peer]
[22:07:23] -!- munktell [email@example.com] has joined #linuxcnc
Aero-Tec: look at http://linuxcnc.org/docs/2.5/html/config/ini_config.html#_traj_section_a_id_sub_traj_section_a
specifically POSITION_FILE = position.txt
[22:08:27] -!- factor has quit [Read error: Connection reset by peer]
pcw_home: I think he was changing the accelleration on the fly...
That would be my guess
Aero-Tec: you cannot jog during a program run
I can see why not
but would be cool to be able to make a small window where you could
some gcode or m code or something
just for the extra tweak from time to time
the tubing I have to use to make a part is not consistent, so sometime it chatters and most of the time it does not
would be nice to knock the tops off the chatter for the times it does chatter
Aero-Tec: did you see the 'POSITION_FILE = position.txt'
was just going to it
[22:16:44] -!- phantoneD [phantoneDfirstname.lastname@example.org] has joined #linuxcnc
just what I needed
[22:17:52] -!- vladimirek has quit [Remote host closed the connection]
I will set it up, would that file path be the config dir for the running config?
I think so..
I don't think you need a path - and it puts it where it should
will find out soon enough
I feel like I should be sending you money or something
you have been so helpful
[22:19:54] -!- phantoxeD has quit [Ping timeout: 264 seconds]
heh - no
pay it forward....
spread the word... ;)
just expressing my gratitude, and would gladly have sent some money if you had asked for some
will do, I can do that
any ideas on dealing with chatter?
[22:25:47] -!- AR_ [AR_!~AR_@126.96.36.199] has joined #linuxcnc
I have a 0.005 per rev infeed, 3 in .25 wall pipe, inside cut for O ring
have tried different hold times for end of cut, ut to long and you have short quick chatter
not long enough and you can have long deep chatter
boring bar is as short as possible
skunkworks: not sure but motonworks seemed to say that it stopped instantly at the end point
and thats not what I get (it does a trapezodal velocity, parabolic position profile)
6061 tubing, with aluminum cutter fluid
it does not always chatter, most of the time it is good, or only very slight
this is where a variable speed spindle would come in very handy
[22:35:53] -!- munktell has quit [Remote host closed the connection]
[22:36:07] -!- munktell [email@example.com] has joined #linuxcnc
[22:54:41] -!- tjb1 [firstname.lastname@example.org] has joined #linuxcnc
[22:57:27] -!- joeg has quit [Ping timeout: 260 seconds]
[22:58:23] -!- joeg [email@example.com] has joined #linuxcnc
[23:01:08] -!- andypugh [firstname.lastname@example.org] has joined #linuxcnc
[23:05:07] -!- munktell has quit [Remote host closed the connection]
[23:05:22] -!- munktell [email@example.com] has joined #linuxcnc
Anyone else having trouble posting on cnczone?
[23:08:40] -!- tjb1 has quit [Remote host closed the connection]
[23:08:53] -!- tjb1 [firstname.lastname@example.org] has joined #linuxcnc
[23:09:18] -!- RyanS has quit [Ping timeout: 252 seconds]
[23:14:20] -!- RyanS [RyanSemail@example.com] has joined #linuxcnc
[23:16:35] -!- factor [firstname.lastname@example.org] has joined #linuxcnc
[23:25:18] -!- Simooon has quit [Quit: Leaving]
Anyone at a computer with mastercam?
[23:39:00] -!- ve7it has quit [Remote host closed the connection]
I have it but not set up yet
I could get it up and running if needed
I need a mcx file converted, guess that wont do any good :)
what are you needing done?
what are you wanting it converted to?
that's the opposite direction than you usually go
true, but it may work
Actually I got it now, friend was having trouble but he got it to me :)
glad you got it
so did master cam do the conversion?
or did he use something else to convert it for you
probably just saved it as dwg
I should install it, had it running on my old computer, I may still have it there, but have not used it much so have not gotten around to installing it here
I was going to cut the school mascot thing for everyone in my senior project class…not going to work
I have art programs for converting pix and art to a 3D cnc gcode
if that would help
all I would need is a pix or graphic and I can convert it for you
No, its just the way it is, wont be able to cut it with plasma
scaled up to 8.5"x8.5" there are still sections with only .023 between them
master cam should do the same thing I would think, have not used it much
ok for a plasma, I was thinking mill
one could edit out or move the fine detail around some
I know…its not really that mascot if I edit all that out
do a more impressionist version of it
do you have a pix of it?
would like to see it
[23:57:34] <tjb1> https://dl.dropbox.com/u/3090951/Screen%20Shot%202012-12-02%20at%206.56.57%20PM.png