hi :D, i saw that I can get button input to emc2 using hal_input, but how can I get position output if I want to e.g. use an external DRO?
mrpinky: it depends on how the PC and the DRO communicate
mrpinky: you can look at the (not well-documented) emclcd program, which communicates with certain serial and usb lcd displays; I'm not familiar with it personally
or you can write a "hal userspace component" in C, C++ or Python, which repeatedly gets the position from halui and outputs it to your DRO in whatever way it requires
enabling halui, description of its pins: http://linuxcnc.org/docs/html/gui_halui.html
a python userspace component that uses (usb) serial to communicate with an avr: http://emergent.unpy.net/01198594294
hi jepler, thanks for all the advice!
I hope the links help you
i am trying to work out the most appropriate way to interface, without knowing any of the possible ways yet (because i am new to emc), so they will all be helful thanks
i am building the DRO myself, i am considering to make it emulate USB HID class. so input is easy using hal_input, but i think output is impossible with this module. i have to write a hal userspace component to do it this way, may be OK
mrpinky: as far as I've been able to tell (I'm the author of hal_input), the linux "input" layer doesn't have any support for outputs except keyboard LEDs, even though HID *does*
I know that you can go below the input layer to the HID layer if you want to / need to, but hal_input doesn't know how
if you have the expertise to do that, we'd love to have you contribute it
(does HID have a data type that is appropriate for a position display? The "analog inputs" I've seen in joysticks are fairly low precision and range..)
it looks like the value is a signed int
(in struct hiddev_event)
then again, that's for a read. it may be different for written values
jepler, you are right, it looks like hiddev isn't low-level enough for control transfers :(
mrpinky: for the next level below that, there's libusb
i have lots of usb experience on windows, but i have only been using linux for 6 months (and i haven't touched the code)
there's a lot of information ni the kernel Documentation directory
I'd get a copy of the kernel source, then look in Documentation/input and Documentation/usb
what about speaker?
oh, you may want to have a look at linux/hidraw.h too
can't you output position as a stream on a port and just read the port?
you can probably do something like that, but it requires new software everywhere between libusb and HAL
because that's what he's building
or at least something that emulates USB HID
well ther'es hte usb to serial chop 3.3v levels
and even fbsd have a driver for it
sure, an FTDI or similar chip could be used
or a USB-enabled AVR/PIC/other micro
then could use th hitachi LED backpacks
no micro eeded
hitachi LED backpacks?
sure, there are several USB or serial (or SPI) LCD "controllers"
my floppy copy of the internet is being slow
which of course have their own microcontroller on board ...
and others that are similar: SED1536 (I think)
want me to find the usb to serial chip?
oh, that one is for graphic displays actually, not usually used in character modules
FTDI FT245 or similar
then wth you harrassing me for then?
and there are others, like from Cypress and others
[01:14:49] <skunkworks> http://www.youtube.com/watch?v=lHJOvZK2GNA&feature=channel_page
I just thougth it was funny. cannot really tell but she kept getting her tongue stuck
please don't post the hairball video
SWPadnos: double perjury trousers thermal!!!
man is that boring *wait*
how do you guys watch your mill while its running hour long gcodes?
i get on my laptop and just sit there.
until I hear the long bang ;)
haha, yep =)
but you stay in some reach nearby.. some feet
wait till the screeching stops?
just within earshot
if it's a proven program, I have been on the other side of the building
depends on what it's doing, as wbell
most tool failures don't cause too many problems, and if i'm suspicious of tool wear i'll throw the op stop on
we now got a videostream.. its just too noisy
that way if the tool roasts 30 minutes into a 60 minute roughing pass and I'm gone the whole time, the finishing tool doesn't crash
but really, laptop
laptop + discretion when to leave and when it requires babysitting
sitting by mine... having the stream on and a emergency-stop in this room
when a circuit board takes an hour a side on our slow machine - I just leave and come back.
(which is not ip based, but a real wire)
how many times have you used e-stop
most crashes during a run on proven code are things you could walk back to
tool fails, machine just plods along not knowing any different
mostly when one doesnt know how to get something fixed finally yet
what do you mean?
trying to mill your fixiating gear instead of the workpiece, too low clearances etc
and mounting workpiece.
what are you doing that is taking an hour of spindle time
milling aluminium without too much cooling
no fluid yet, adding oil every layer
what's your chip load and sfm
[01:27:42] <roh> http://m21.hyte.de:8000
eeeh.. its a 3-spoked 6mm diameter head. torroidal
it's trying to connect but going slowly
about and i
3 spoked 6mm diameter head?
eh and im'm in the middle between 300 and 3000 rpm in the upper gear on the spindle
you can top your machine out
1mm layers, 6 mm depth pocket with 2mm sidecut
if you're not taking deep passes
usually you don't need e-stop; feed hold is sufficient if you are careful
roh, how fast are you feeding the tool
if you run the machine at 3k rpm, you can easily double that
I hope that is not IPM ;)
toastydeath eh.. jap
nah.. metric land here
i'm sure there is a faster way, but i think i should have working cooling for that
you aren't even beginning to push the tool temperature
the only thing coolant does in aluminum is remove chips (and control part expansion)
it does not cool the tool
oh.. thats new to me
the energy required to shear aluminum is not very high, so there isn't much heat generated to begin with
how fast do you think one could do what i do?
i think you can run your machine flat out
if you had more, you could run more
bump it up incrementally, obviously
but the numbers I'm getting are 540 mm/min and for a 3000 rpm max spindle
assuming your 3 flute, 6mm endmill.
if you have a bigger machine, you can just do that and be fine, but if you have a desktop mill i'd step it up
also, on the initial slot go slowly.
its about 750W spindle power
spindle power isn't the issue, it's the rigidity of the machine
syil bf20 vario
got a picture?
the one pictured here: http://hackerspaces.org/wiki/Raumfahrtagentur
i would step up in a couple increments
and see how it does
it SHOULD go fine, but better safe than sorry.
'makes more noise' for sure now
if the machine's quiet, it had better be a finishing pass
i like roughing at school with a big facemill, you can hear it clear down the hallway
but back to your thing, with a 2mm stepover you should be pretty good for higher feedrates
slotting is really the only killer you'll encounter - if your tool is going to break, it will break during a slot
currently its pocketing.. can see it on the video
the video won't connect for me, i tried =(
what i mean by "slotting" is any time the endmill is cutting full width
so like, anytime it steps down, that is slotting
ah.. yes... 3 times the force _after_ it stepped down
what kills the endmill in slotting is that the chips don't get cleared without air blast or flood coolant
so a chip comes around, and wedges between the tooth and the wall and doesn't shear
itty bitty 6mm endmill can't crush the chip, tool flexes and breaks
yeah.. halved the feedrate for most of the 'middle slots' which initiate a center startin pocket
you can tell a surface that has chip recutting because the finish is different - it'll have almost white vertical lines that look like noise or snow on a tv set
and it comes and goes
as the endmill loads and throws chips then clogs again
yeah, good policy to back it down
i just wish there would be a way to tell heekscam to do that automagically ;)
will think about hacking it in or suggest it to dan
does your cam thing have an option for trochoidal looping
or any orbital/circular pocketing routine
eeeh.. whats that?
it has pocketing
instead of like, milling in lines
it mills arcs until it hits a hard wall
then it mills out that corner again in loops
and profiles. also 'adaptive rouging' and zigzag.. drilling is in the works still
but what trochoidal looping/orbital pocketing does is eliminate slotting
because as soon as you plunge, you've created a place for the chips to go
instead of going off in a straight line
basically everything which works well for us here is open and free sw
also if you have the option, a bigger endmill would help.
for roughing, anyway
Jymmmm is now known as Jymmm
Are there instructions anywhere on how to build configs for HM2?
start with the sample
I'm not sure where the sample is.
geo01005: FPGA configs?
PCW: Yeah, more than anything I'm just curious. I would like to learn to use the FPGA tool chain.
I believe there isn't a sample config that e.g. duplicates the old m5i20 setup
oh, if you're tealking about making new FPGAs then the best place to look might be at the work jmkasunich did on that stuff
I guess I didn't understand the question
it's in a branch on the CVS server, there's a makefile that can generate a bitfile
If you want to build a config, you really should use the latest source. its now much easier to chose options
hmmm. that could be useful
it's not what I'm thinking of
PCW, By the way the 3X20 looks awesome.
Its a custom card we will try and sell as a general product
Is there any documentation on how to use the latest source?
No, but its easier ;-)
So where do I start?
Ill try and write up some startup instructions eventually...
Basically you chose a top level file depending on card type (all top level files start with TOP_)
in the top level file there are 2 options (card type and pinout)
if you want a custom pinout you need to create a new pinout file
for 7I43 there are only 2 card types (200K and 400K)
The pinout file drives the build (choses which modules and how many are installed, and the actual pinout)
(this is all for the new source with pinouts in individual files, not all lumped into a deathstar)
So where is this source? Is it inside of trunk?
its waiting for Setbastion to commit it...
Ahh, I see.
(I can send you a zipfile, not very big may 1/5 m)
also has stepgen fix and SSI interface module added
I was trying to build the bitfile from the hostmot2 source from your website. I really don't know much about the whole thing those.
Do you still have my address?
Its not up to date either (who can I blame for that?)
Is it best to build from the command line?
Yep its in the sent mail spool
I use the GUI, (windows) I tried from the command line
but it bailed with a license violation but thats 9.2 maybe the latest fixes this
(the GUI will make a command line for you)
I'll give it a try in the GUI. See if I can figure it out.
Thanks PCW :)
jepler: I plan on learning some of the syntax just not sure how a post processor works. Have used manual mills in past just purchased bridgeport CNC to learn and use.
a postprocessor, like, a mastercam postprocessor?
anyone know if a standard shop vac can run continously for hours at a time?
i don't know if it's a good idea but i've done it
~4-6 hours runtime
jepler: The post process is not so bad It looks like a header with a command to do this and then a trailer. OK
jepler: It is just an alias.....
Hey all its Valen here
This is my fathers PC and its his mill we are converting
We are thinking about taking the current 80+kg head off and putting a spindle type setup on and were wondering about how they go doing low speed operations, tapping or face milling kind of thing
and does anybody have any reccomendations on automatic tool change systems?
pjm__ is now known as pjm
Russell01: that's a pretty hard question to answer over IRC
otoh it's night in the US, where most guys are, so you might want to ask again later
or even better on the emc-users mailing list
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-06-11.txt
alex_joni: you into video at all?
hey archivist, what do you think of spindle motors for low speed stuff?
tapping, face milling etc
(its Valen here atm btw)
Russell01, depends in feedback and torque etc
I was thinking a generic 1000 count encoder
i was mainly wondering how they were for torque as a rule
they seem to be high speed things in general
face mill with a big cut needs a lot of grunt
tapping needs good feedback
Jymmm: what do you mean?
alex_joni: HD mpeg2, streaming, etc
watching some from time to time though :)
[Global Notice] Hi all, In a couple of minutes we will be performing a clear-out of the Services database, removing deadwood from the NickServ and ChanServ databases. This will slow down services and could temporarily flood them off. It shouldn't take particularly long though! Thank you for your patience and for using freenode!
[Global Notice] Hi all, The services prune is complete! If your project is noticing issues with channels in your namespace, please get in touch with freenode staff and we'll sort you out swiftly! Thank you for using freenode and have a great day!
pjm_ is now known as pjm
whee, a debate about the metric system, just what we need
and only 30 years too late
782 21/32ths fortnights
* archivist uses both
maybe 4 wythnos
I think everything should be based on planks distance.
how many barn is that
the "Man-Eating Seals of Antiquity" are out
Hardware purchase for EMC2 on with feedback from linear scales. 1-Mesa1 5i22 PCi card. 1-7i33T 4 axis servo amp interface. 2-7i37T 16 in 9 out isolated interface. Does this make sense or am i missing something?
jimbo: is this you? http://www.cnczone.com/forums/showthread.php?t=82659
that's 8 outputs, not 9
otherwise, it's about right
if you expect to use scales and encoders, you'll want to look at the I/O timing of the 7i37T, since one of those would be used for the secondary encoder inputs
or you could get a second 7i33T, and use the spare analog outputs for spindle and whatnot
(using the encoder inputs as secondaries for the furst 4 axes)
skunkworks: Yes this is me. The encoders on the servo motors look like they are tied into the NC400 servo drivers. would I then be able to tie my linear encoders into the 7i33T?
it depends on where you want the PID to be
if the velocity control in the servo drives is good, you might be able to get by without connecting the encoders to the PC
those aren't encoders
if they go to the amps, they are tachs
true, I'd be surprised if a drive from that era took encoder input
you don't hook up tachs to a 7i33
they can also be resolvers
oh, I read the question again -- linear encoders yes - read them with the 7i33
those do not hook to the amps
SWPadnos: without hooking the encoders to the pc? what are you talking about?
oh right - I was unclear in that I meant you could possibly get by with only the linear encoders and not the motor-mounted ones (which are more likely tachs)
ah I see
since I had recommended another servo card to read the secondary set of encoders
each axis has (7i33 dac AND analog tachometer -> nc400 -> motor), (encoder -> 7i33)
jimbo: do you have a good picture of the motor and all its markings?
hey how often do you guys restart/shut down your emc2 computers?
I expect it to run for days or weeks without trouble, though I typically shut it down when I'm not planning to use it again that day.
same here, shut down when finished using it for the day
12:47:44 up 38 days, 1:37, 1 user, load average: 0.00, 0.00, 0.00
^^ this machine runs emc
I shut mine down unless it has buildbot duties at the end of the day
my wife gets tired of pedaling the generator after a while
* archivist_attic hands BJT-Work a larger whip
BJT-Work- hahaha my dog does too
she can shoot better than me so no whips are allowed
i scrapped a part yesterday, it lost position
it's been acting a little weird
freezes here and there (the computer)
I have mechanical problems to sort on mine
archivist, like what?
not stiff enough
that's what she said
archivist what kind of machine?
Danimal, if your PC freezes from time to time, you should run a long-term memtest, and possibly something that exercises the CPU a lot
SWPadnos, think my mem might be bad?
it could be
there's definitely something wrong with the PC if it "freezes here and there"
unless it always freezes when doing some particular thing
I know my memory is bad or at least I think so
for 24 hours
ram for that thing is probably $$$, it's an old pentium 3 box
Danimal, homebrew 5 axis mill
archivist, nice! any pics?
Working on the picture. had to go buy batteries for the camera.
remember the internet before cheap digital cameras? we had to use scanners for everything. sometimes polaroids + scanners.
dang your old cradek :)
cradek: What do you before? I still use my scanner
still got my scanner somewhere
I remember 5 1/4 floppies
and still have a bunch
I not only have 5 1/4 inch floppys but I have a couple of 8" hanging along with a ZRdos machine...
8" floppy rulz
Actually, I have the coolest scanner around... 1" thick (including cover) and USB powered.
should i bother considering a computer upgrade for my machine? it's an old 1.4ghz pentium 3 compaq server.... the jitter was like 17000ns
if it has enough ram, it'll be fine
Though, I am looking for a color AIO that's network ready, ADF,and duplex printing
somewhere around 520mb or so i think
compaq server should be able to take more ram
it can take 2 processors but only has 1
512 is probably good enough
yeah, more would be better
might need registered PC133 though - it can be kind of hard to find
thats what i was thinking.... i think i'd replace the computer before i delt with ram in it
Danimal: what model
it has 640mb of ram
Danimal: what model?
Jymmm proliant ml330
I'm gonna guess a proliagn
thats all it says on it
lemme look at the back
Danimal, stacked A on B http://www.collection.archivist.info/archive/DJCPD/PD/2009/2009_05_27_cnc/IMG_0246.JPG
oh it says g2 on it near the top
how many slots are in it?
and how much would you liek to go up to max?
max ram 4GB
oh 4 slots.
hold on, lemme pop the cover
it's 4 slots
4 256s shouldn't be too hard to find or expensive
i only use it for emc2, so what do you think would be best?
$29.85/ea for 256 pc133 RE
[18:13:05] <Danimal> http://stores.channeladvisor.com/4-All-Deals/items/item.aspx?itemid=5847344
21 bucks for 1 gig, not bad at all
up to 4 GB too
what do you think would be sufficient?
640MB is sufficient
768MB would be plenty, 1GB would be luxurious
512M or 1G would be fine, if you do find that your existing memory is bad
640KB is all anyone woudl ever need
oh, I've never even needed that much
... on my watch
do you think i should try to find another p3 processor and toss second one in there
I don't have a Dick Tracy GPS watch
No GPS, just two-way video communicator
Danimal: for emc2 it doesn't pay off, unless you want to change the kernel to take advantage of SMP
a second CPU? no. without rebuilding or using experimental packages, emc only uses one CPU or core
archivist, very cool machine, i like the 5th axis unit. where do you think your rigidity issues are coming from?
Y'all know that there are 8 core systems these days, right?
ok, cool, that saves me 20-30 bucks
one coe per axis!!!!!
Jymmm: that isn't as great an idea as it sounds, since each axis has to be moved in lockstep with the others
i keep getting joint errors, even when i turn down the velocity
every once in a while the pc will lock up when i'm jogging
jepler: Well, though logic in there will a, eeessh! ;)
then i have to restart the computer
jepler: logic as in your arument, not as a solution
Danimal: that sounds bad
Jymmm, there are 32-core PC systems these days. They've beed around for a couple of years
random lockups are certainly a nono, and usually a sign to run memtest over night
or check temperature, or other things that can cause lockups
seems to be ok when i run a program, it's only when i jog that it acts up
SWPadnos: But nothing to take advantage of them, and as jepler pointed out too.
that's surely something you are imagining ;)
Jymmm: RTAI can take advantage of SMP
I can take advantage of them quite nicely, thank you very much :)
from what it looks like to me a dual core could have a considerable advantage...
it's just that building packages for every SMP out there is not possible
over single core
expecially in the z axis... for some reason it hates jogging in z
some SMP require combinations of ACPI, APIC, etc that on other machines won't boot at all
SWPadnos: What, one pr0n site per core?
mozmck, it depends on the type of multi-core system, and oh how well RTAI can be configured for it
i gotta keep the velocity below 45ipm
Jymmm, no, but I can think of some image processing tasks that could be sped up dramatically :)
I know debian had a number of kernels at one time, but I'm not sure if they still do.
SWPadnos: But only like 5%
SWPadnos: those might still be used for pr0n lateron
depends on the images
mozmck: they still do, and ubuntu does too
Jymmm, no, the speedup is nearly linear with the number of cores
you can run multiple copies of the processing software
depends on the app though
SWPadnos: what app?
if it's threadable
almost any of them, but Imagemagick comes to mind
Jymmm: say batch-processing of a folder
in my recent experience, a single instance only used one core, and possibly sometimes two
do you think we might could do something like that at some time?
mozmck: some people already did
but as I said, it's very computer specific
what runs for them, might not run for you
in experimental right?
mozmck, it's possible now, but I wouldn't expect a fully-configured SMP liveCD (with all the isolcpus/cpusets defined) any time soon
so you'll have to try it (no support ;), or do it yourself
Guys we've had this discussion before and unless the app is multi core, not gonna happen.
Jymmm: or the app can be run more than one time
Jymmm, you can always run more than one copy at a time if your app is single-threaded
as in logical separated operations
you can even do a pipeline where multiple copies do different things to the same image, but they do it at the same time on consecutive images
does each mb chipset cause different problems, or does it have mostly to do with the cpu?
Yeah, doen't help when transcoding a mpeg2 stream to have multiple iterations
mozmck: it's a combination of chipset, processor and BIOS
you then have a latency "problem", but while your pipe is full, you get linear speedup
and also configuration
a dual-core machine should have isolcpus=1 (probably)
SWPadnos: at least for 5-6 more kernel versions
a quad-core (or triple or 6-core) might want to use cpusets and IRQ steering
after that isolcpus will be deprecated
archivist, is your z axis a ball screw, or just a thread?
I left on the basic acpi with the thermal stuff, and still got very low jitter btw
alex_joni, ok, the CPUSet equivalent of isolcpus, but with or without IRQ steering is another issue
turned off all other acpi stuff
with 3+ cores, you can even have RT tasks (and the timer interrupt) on one core, EMC-related userspace apps on another, and the rest of the OS on the other(s)
mozmck: in your case that might be ok
on the next machine it might not
if you have only 2 cores, does it hurt to put more than that in the rtai config?
that's why our kernel has the options that might make sense on the most machines
mozmck: depends on the rtai version :)
but I think it's safe to put 8
alex_joni: I figured that re the acpi. I just expected worse on my machine.
I'm using 3.7 from cvs
archivist, throw an indicator on the spindle and try lifting the head, and try moving it side to side. linear rails tend to not be very rigid... you might have some issues there. seems like they almost amplify chatter
mozmck: 3.7 is released, you can grab the latest
but either one should be ok
I think I got cvs because it had a patch for linux 188.8.131.52 and the release did not.
man i didnt have my retract hight set high enough, and i cut right through a big clamp yesterday
I was just curious if rtai would need to be compiled different for different smp configurations as well.
couldnt believe it... carbide half inch endmill, machine was in rapid, probably a .700" cut, and the machine didnt even flinch
blew right through it like butter, no damage to the cutter
heh, that's probably how you're supposed to cut with it
mozmck: in the olden days (rtai 3.2 or so) we found that an smp rtai kernel wouldn't boot on some single processor machines
yea, i've been severly underworking the machine lol
smp has apic, so rtai compiled for smp assumes apic, but not all single processor machines have apic, particular in older systems
(or assumed, in the 3.2 era)
apic, lapic and there was another one still
jepler: I was think along the lines of maybe having 4 or 5 or so kernel versions like debian does, k8, k7, i686-dual-core,
mozmck: it takes a lot of time to build and test only one
that the end user would choose from depending on his system.
mozmck: that's fine, as soon as you supply me with a dozen stout men to build all the packages
I can see that
mozmck: a different build of emc is required for each different kernel
maybe 3-4 weeks of work till it's stable enough
doing that 6-7 times is crazy imo :D
jepler: because of rtai?
because we supply kernel modules along with emc2
mozmck: because emc uses kernel modules, and kernel modules are not portable from one kernel to another; certainly not from a kernel for cpu type A to cpu type B
those need to be the same version as the running kernel
we "could" one day provide uncompiled modules, and have them compiled on demand
I should remember that, I just had problems with that as I was getting my setup working.
like nvidia does, but that's a lot of work for limited results
or we could switch to lxrt
any of these options require someone who is dedicated to the idea
what is that? I saw it mentioned in rtai
it is a way for linux userspace programs to run as realtime
the way we use rt, only kernel modules can run as realtime
I had never thought of doing something like the nvidia thing, but that's an interesting idea.
since there's already infrastructure for it
but yeah, I agree (needs someone who is interested to figure it out)
there is an app in debian (and i think ubuntu) called module-assistant
m-a for short. it will download and compile kernel modules
dkms is the modern "auto-compile modules when necessary" system
(and there are a few non-module things that depend strongly on the kernel version -- emc_module_helper has paths that are kernel-version-dependent; halcmd and the userspace libemchal are conditionally compiled depending on rtai vs rtl vs sim; and maybe others that don't instantly come to mind)
hadn't heard of that one.
it's included in 9.04, and it works quite nicely in my experience
"dynamic kernel module system" or some such
I had the pleasant experience of updating my kernel, and having the NVidia driver get comiled on the next boot, with my graphical login showing up just like nothing had happened (though it tok a little while longer)
interesting. I tried to recompile the nvidia driver after putting linux184.108.40.206 with rtai and emc on 9.04 and it just complained.
you may need to register the RTAI kernel with dkms or something
I don't recall exactly how to do that
could be. I may look it up.
oh, did you make / make install the kernel, or install it as a debian package?
I made an RT SMP kernel on 9.04, but did it as a package, and dkms picked it up
make install was for emc
i figured the ubuntu kernel had some patches it needed, and I used vanilla
it's best not to have the nvidia driver and emc2 together
that's why I wasn't too worried about it. but I had wanted to try it on this dual-core and see what it did.
it seems to work, though I do get some higher latencies when I'm flipping the desktop cube around
I see. did you set isolcpus?
gotta go. bbl
no, I don't think so
cradek: The pictures you asked for are up. http://cnczone.com/forums/showthread.php?t=82659&page=2
aha, tachometer 19 volts/1000 rpm
so that answers that
but, the tach may be in the motor itself. if so, and if there is something stuck on the end, it may be an encoder or resolver too
cradek: 1.9V/1000 rpm
hmm.. no, 19V is right
I bet it's 19
yeah, the 1 that's a bit lower (serial number) is the same
MAX VOLTS 11 is curious
I bet they mean 110
I think the NC400 has a DC bus of 100 volts
the max volts is for the motor, not the tach
and it does say .311 V-s/rad
I know - but 11??
.311 V-s/rad = 32.6 V / 1000 rpm
* 3600 (max rpm)
yeah it must be 110
might be max current
and 2.75 lb-in/amp
-> 11 Amps max
they swapped the RMS amps and voltage ratings on the nameplate
ohhhh it says 117 amps
they reversed them
heh, our math is right :D
now with labels for the eager engineer!
jimbo: so is there anything on the end of the motor?
cradeck there is a bump that the wiring is fed int. I will take another pic.
alex_joni: got time for a weld question?
cradek: looks like it: http://i559.photobucket.com/albums/ss35/jimbo-colleen/95a7_11.jpg
So then the next question is - what are the scales and what do they output?
we'll see if also for an answer :D
though my big problem requires a drawing so I can show it properly
anwyway, we have some problems with some stuff breaking out of cracks forming in the welds..
ie, weld acts like a stress riser, and stuff cracks after enough time
we are pondering some design changes to make things hold(I'll throw you a pic when I get it drawn)
how well, does tig treating welds work to combat thoose kind of things?
anonimasu: depends when the cracks appear
did you inspect the weld itself?
yes I did
did you use penetrants ?
I'd suggest you try that
hm, the crack was visible enough without
oh, so you can see the crack just after welding?
that's really bad :D
the welds are good after welding
I mean right after weld
you can show cracks even if tiny, and you miss them by eye
* anonimasu nods
get some sprays
usually a set has 3
got them already
next is material type
* alex_joni wonders if we should switch to private or some other channel..
cradek: Images are up
skunkworks: The scales are heidenhain LS903
ok, that's just wiring - no encoder there
the motor has extra non-powered brushes that act like a tachometer. those are the two small wires.
along with the big motor power wires, those go to the servo amp too.
can't find specs for those scales. I hope they have index somehow...
hmm - looks like the scales can hook into a QUADRA CHEK II READOUT - kinda makes you think quadrature
surely they are quadrature if they were used for cnc?
if used for dro, who knows
in other news, who knew that DROs now look like consumer electronics or something from startrek, with random useless swoops and curves: http://www.metronics.com/digital_metrology_readouts.php
QC-100 comes with a tricorder!
heh - I cannot believe there is still a market for dro's ;)
new ones even
whats better than a dro??
i need something for my bridgeport
sick of using the dials
cnc is better than a dro
interesting way to add 'screened' looke to a pcb (print out a paper and apply it to the board) http://www.cnczone.com/forums/showpost.php?p=624948&postcount=134
yea, but i'm not putting in ball screws and all that crap
why would I want to look at a dro and manually move the machine into place when I can just command it ;)
i use my manual machine over my cnc probably 90% of the time
quicker for onsies and twosies
then you don't have a jog wheel on your cnc...
no, but even if i did...
i've been running manual machines all of my life, i'm faster on a manual mill than a cnc
i use the cnc for big hole patterns or stuff that you cant realy do on a manual easily, like radi and angles
and also production work
but i'm a toolmaker, so i dont see much of that
roughing alot of material is good cnc work too
* Jymmm hands Danimal a drawing on a cocktail napkin... make me one of these
sure, napkin drawings are my specialty lol
Danimal: it's a tap btw, not a sex toy
one mans tap....
ribbed for her pleasure
Danimal: I never siad it couldn't be used as such, just that that is not what it is
napkin sketches are open to interpretation
good night all
man it would be pretty cool to emc2-tize my manual lathe
i think i need to finish my mill first though
I run emc2.deb in a simulator mode?
i.e. without compiling it from nthe source?\
alex_joni: went to sleep already
so, can I?
there are simulator-only packages
* alex_joni came back from sleep for that :)
alex_joni: can i train myself on a laptop with that "debian" way of installation through packets
[19:57:52] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?EMC2_Pure_Simulator
* alex_joni is really off to sleep now
ok, thanks a lot.
compiling takes too much time
cradek: The LS 903 scales tiee directly into the TNC145 controller.
jimbo: you can probably forget about heidenhain scales, most likely
they are often ~
[20:30:02] <anonimasu> http://www.auto-met.com/heidenhain/sealed_le/LS_187C.htm
so there output is a sine wave
And what I neet is digital
What you are saying is that most of the time they are defective. And I need to scope them to see if they are good?
anonimasu think your scales might output sinwave. You need square wave. If you cannot find any info on them - checking with a scope is about the only way.
they may plug in to a convertor/conditioner circuit that converts them to quadrature. if so, maybe you can keep that part.
That is part of the main controller but i am not against gutting the controller. It sounds like this is the only sticky wicket...
I have to go make someones computer work and someone elses AV system sound good........
skunkworks_: most heidenhain scales do tath
there's converters for them though
Converters is good... ? where do I find or who sells them..
if they have a resolver type output, you can use the pico systems resolver-to-quadrature boards
I used those on my lathe retrofit
heidenhain sells the converters too
[20:54:17] <anonimasu> http://www.heidenhain.com/wcmsmimefiles/208_961_24_16952.PDF
[20:57:28] <anonimasu> http://bbs.homeshopmachinist.net/archive/index.php/t-21140.html
looks like they make ttl output scales also - I wonder how lucky jimbo is. ;)
wow - the windows viruses are getting nastier and nastier.
I think he's unlucky.. :)
I have not powerd up yet I will work on this tonight.
For my sake I want to get lucky tonight.....
your control dosent look so awful ...
(though I might be looking at tnc-150) docs
My control is only a 2 1/2 D. All is good except for the monitor which is toast.
they are sinusoidal scales..
1Vpp or 1uApp
I wonder who figured out this clever quadrature decoding algorithm: http://emergent.unpy.net/files/sandbox/weird_quadrature_decoder.py
-- I still don't follow it, but it does work.
jepler, yes... that is a mind bender.. one table lookup is so much easier on the head
hm, I wonder if mesa has considered shipping 7i43s with a step-to-pwm-servo fpga firmware. 4 axis servo plus 24 I/O points with some weird multiplexing to the parport. might be a way to capture mach customers.
maybe they don't want mach customers :-P
a question for the circuit board milling guys: how do you make vias (side-to-otherside connections)?
Spida: I make only single-sided boards, so I don't deal with that problem
Spida: but basically, just put a tiny length of wire in the via and solder on both sides
thaht wouldnt work below my smds...
[22:08:59] <Spida> http://www.spida.net/tmp/platinen/whirlygig-bottom.jpg http://www.spida.net/tmp/platinen/whirlygig-brd.png
I think you can get this pretty flat on one side but not on both
place the board on a flat surface. Insert a wire prepared to be only as thick as the board. solder that side. after soldering all wires on that side, flip and solder the second side
the tip of the wire will be flush with the copper on the second side, so the only added height is the solder
guest3673456 is now known as FlyingElectron
Hi All! I have a machine where I installed 2.3.1 from the live CD. I downloaded and built the latest source (2.4.0 pre) but when I try to run it I get a "HAL: ERROR: version code mismatch" and a "HAL: ERROR: could not init shared memory" errors. Anyone have any ideas what I can try to get it to work?
FlyingElectron: when building emc yourself, use the --enable-run-in-place flag when you ./configure, and always "source scripts/emc-environment" in the terminal where you will run emc commands.
i think i built with the --enable-run-in-place flag, but let me try it again. From the src directory, i do a "./configure --enable-run-in-place" followed by a "make && sudo make setuid"
right so far
then what commands did you enter to run emc? You should have done something like: cd .. (to go out of 'src' to the top directory), source scripts/emc-environment (to set up the terminal to run this version of emc instead of the installed version), emc (to start it up)
in the scripts directory, i did a ". ./emc-environment" followed by a "./emc"
i deleted the source code and am doing a fresh checkout
i had an old 2.3.x source code that i did an update on, and i'm thinking maybe something didn't update right
maybe i had some changes that I made that didn't merge correctly
what you did sounds correct
anyway, I have to go. bbl.
thanks for your help!
"The quality is the most excellence!"
haha, that sounds like some of the foreign documentation i get sometimes
FlyingElectron: have you tried running through sudo?
mozmck, though that's probably good advice for determining what's wrong, if you need sudo then by definition something is wrong
i figured it out
i had a bad update from cvs
from an old 2.3 to the new 2.4
SWPadnos: true, but it sounded somewhat similar to the problem I had with limits.conf
i deleted the source and did a fresh checkout from cvs and it made my erros go away
mozmck, ok - just checking ;)
a make clean should fix just about anything for RIP
i think i may have had some modified code in the old source so maybe the merge during cvs update didn't go right