[#emc] "This is the #emc channel - talk related to the Enhanced Machine Controller and general machining. Website: http://www.linuxcnc.org/,
wiki at http://wiki.linuxcnc.org/"
genehacker, you at UT?
why are you at UT?
you wouldn't happen to be the guy with the EMC2 based CNC machine on campus would you?
I went to UT when Janis Joplin hung around the student union :)
ries_ is now known as ries
hey all, what's the difference between STEPGEN_MAXACCEL and MAX_ACCELERATION in a stepper configuration? Or , where can i read about what they do...
MAX_ACCELERATION is a limit that the trajectory planner uses
STEPGEN_MAXACCEL is a limit that's generally applied to the stepgens themselves
since the stepgen always lags a little bit (it has to be told to move before it starts moving), you need a little bit of "headroom" in the stepgen limits, to allow it to catch up to what the motion controller wants it to do
so... STEPGEN_MAXACCEL is close to the physical limits of joint (dangerous edge), and MAX_ACCELRATION is further from the physical limits of axes? (sort of, DontFrequentlyChangeQuickerThanThis ?)
bah due to my experimentation Ive forgotten which values that were good for Alu :P
or well only for 1 tool though
so cant be hard to find :P
frallzor, for that teeny tiny endmill you were using?
for my 6mm
im getting to figure its my vfd that the cunt in this problem
under heavy load
its not made for milling
so the curve isnt made for it
frallzor, you had a fixed toolpath but a variable depth of cut, right?
yeah same toolpath and in depth increments
frallzor, do they make roughing endmills that size?
I just use regular 2 flut mills =)
something like that
We've got some 0.02" endmlls, just got them in
got one of those
I think they go down to 0.005 or 0.002 or so
didnt work too well
frallzor, they leave a shit finish (hence roughing endmills) and tend to vibrate a bit, but they really chew through material and for the same feed and speed you should see improved chip evacuation and a smaller HP requirement from your spindle
well I rather get a proper vfd that can handle the load better
since i dont use nearly my 2kW according to it
are there any torque boosting options on your vfd?>
nothing like that
as I said, it only got 1 curve setting and its made for pumps and fans =)
not really the same
but I was told it wouldnt matter since a motor is a motor
i like my http://www.driveswarehouse.com/Drives/AC+Drives/Sensorless+Vector+VFD/SJ200-007NFU.html
but Now I see it make a shitload of difference :P
Hey all, I'm having some trouble setting up a HAL file for a Mesa 5i23 board for a gantry router
tried selling it but not much response :P
so Ill just have to keep it as a spare or just kick the shit out of it for fun
We're trying to use the m5i20 reference files as a guide, but can't figure out where the variables corresponding to the Mesa board signals come from
I probably will kick the shit out of it in pure anger
for example, where do variables like m5i20.0.in-00-not come from?
it'll tell you exactly what pin goes where
m5i20 is deprecated, 5I23 is hostmot2 only so like alex says, use dmesg (or the appropriate .PIN file)
I've got a .bit file and .pin file that Peter from Mesa sent me, specific to the daughter boards I'm trying to run
When I do dmesg it doesn't seem to say anything about the Mesa board, but I do have it in the PCI slot?
correction: dmesg reports a "HM2 failed to load firmware" or something like that
I'm rebooting the machine
lepton: no need to reboot, probably
lepton: like pcw said, don't use the m5i20 driver, use the hostmot2 driver and the hm2_pci driver
look at either hm2-servo or hm2-stepper as a starting place for your hal file
Sounds good, I'll stop looking at the m5i20 hal example I printed out :)
Am I correct in thinking I should now be trying to use the Mesa utility to put over the bit file they sent me?
aww you're not in the other channel
When, you and genehacker are already over here!
Just bring fenn over and it will be a full hplusroadmap party
What version EMC? I think 2.4.1 and maybe 2.4 have a bug with loading 5I23 firmware
I'm on 2.4RUnning version 2.4.1
pcw_home: any more info on the bug?
lepton: fenn *is* in here
haha, and here I was trying to make a joke
small world... ah... freenode, eh?
the 5i23 loading bug is fixed, and will be in the next release of 2.4
the 7i43 bug has a "we don't know why, but this works" fix
if you install the 2.4 package from the buildbot, you'll get the bug fix before 2.4.2 comes out
and apparently only shows up when parport_pc is also loaded
hmmm, is there a precompiled buildbot package link?
fixed 2.4 packages here: http://emc2-buildbot.colorado.edu/~buildmaster/
the 2.4.1-47 package has the 5i23 firmware loading fix
Sweet, from a server just up the road for me :)
I'm guessing I want the master branch, and stable is what I'm already running?
you probably want the 2.4 branch
are you near CU?
Yeap, graduated from there a few years back, now work in town
cool! i'm in boulder too!
well i'm in ann arbor right now for the cnc workshop, but i live & work in boulder
We should have an EMC2 party
just the two of you ? :P
We're a start-up / semi-hacker space sort of thing
CNC router, incoming 3d printer, we do a lot of embedded electronics development
nifty, i've been looking for a hackspace in boulder
When do you get back in town?
saturday eve, but i'm swamped all next week
maybe we can get together for a coffee the week of july 12th or later?
Yeah, that sounds good
my email is email@example.com
do you know about this: https://groups.google.com/group/boulder-hackerspace-public
just trying to start up, nothing's happening yet
I see meeting are at the cup, I take it there's no space in the space so far?
greetings (from the Norfolk VA hackerspace) to the Boulder CO hackerspace
lepton: right, it's just hackers at this point ;-)
greeting telmnstr :-)
All the same, very cool. I've been wondering when Boulderies would self assembled into something like there
these people I bought my lead screw from, I told them it arrived bent
now they won't talk to me anymore. Heh
There's a burning man group not to far from my shop space that has a moderatly open workspace (Pheonix assylem?) but it's more metal and fire art focused then hacker/maker focused
i've heard of phoenix asylum but haven't gone and met them yet
I haven't either, I've just run into some of the guys from there in my Boulder travels
I've recently over committed myself to CNC work for a Burning Man camp this year, which has introduced me to a fair bit more people in the area
cool. have you met any of the apogea folks? mostly burners, with some of the usual maker/hacker overlap
Yeap. I was gonna go this year, but got caught up in start up company madness
cant seem to get the same cut quality as before now, Im sure its dull now due to all failed tests :P
I *might* go to Vayab (sp?) in Crestone in July, similar groups of people
but definetly going to Burning Man this year
what cnc stuff are you doing for burning man?
inkjet plot acid on paper lol
Oh gosh, too much. Furniture like structures from wood, LED lighting (part of what we're doing as a company) installations, some scuptural sort of cut outs for the inside of a dome
and hopefully we'll be 3d printing some PLA based seed containers that are fancy geometric shapes that can be cracked and burried to plant things
that's a nifty use of pla :-)
I hope it works out well, I'm a bit excited to try it
my wife's really into gardening, I wonder if she has any input on useful seed container shapes
We're wondering if the seeds are going to stick to the PLA, I think it might me neat for them to be floating in internal cavities
so you can shake it and hear things inside
the PLA can be clear, too, so you can see them, or even potentially see them starting growing
I don't know if this scheme will work out at all, but I'll be trying it next month
are you planning on closed internal cavities with pores to get water in to start the germination, then the sprouts crack the shell?
Something like that has been discussed
Also something as simple as seeds in cavities (full closed) inside a yo-yo like object, where you twist the crack it open and expose the seeds
but before you do so it has a logo/emblem/location & date or something else on it, so you can pass it out like a trinket
Though I tend to hesitate with randomly handing things out, seems to generate unnecessary waste. At least you can some plants out of it?
Alright! Forward progress. the 5i23 is coming up properly with the EMC2 update
thanks seb_kuzminsky and pcw_home!
hello from the Toronto hack lab http://hacklab.to/
Great, now HAL should have some actual pins
What step am I missing to persuade emc2 to run on a kernel other than the 188.8.131.52-RTAI that it insists it needs?
[18:01:28] <skunkworks> http://cnczone.com/forums/showthread.php?t=103639
andypugh: compile it?
./conficure --enable-run-in-place ; make clean ; make ; sudo make setuid
(I am trying what I think was the kernel from the livecd to see if that helps with my "only not-wierd in sim" issue
Linux CNC 2.6.24-16-rtai #1 Tue Sep 30 22:54:33 EEST 2008 i686 GNU/Linux
I guess I can try compiling the 2.5~pre on this kernel instead of the 2.3~pre I just did...
(2.4~pre that is)
I probably should get an actual released version on here somewhere. :-)
the key is coolant all the time
I'm currently writing a hal file for a gantry router with stepper motors and a Mesa boards (steppers DO have quadrature input). Am I correct in thinking that I don't need to make PID threads, even though I do want to read the quadratures?
SWPadnos: BTW, the cords have a lifetime warranty on them.
lepton: yes you're right. pid is only if you want closed-loop servo control
Any advice on what I need to be doing within the Hal file to make the stepper quadrature encoders useful?
I'm referencing the hm2-stepper files, which don't seem to use quadrature input?
what do you want to use the input for?
lepton: right, the hm2-stepper configs don't read any feedback
from the machine
stepper machines usually dont have any feedback, encoder or otherwise
whats is the backlash command in the ini? =)
Valen, actually mbed won't be running emc2 any time soon. 20kB ram = forget Linux
i2c ram chip would fix it, if they are available
haven't seen an i2c RAM (EEPROM yes, RAM no)
I just googled, found a 32K one for $4.56 aleady
why not use non-volitile SRAM?
strange. I guess I had never looked for ant
since it's so very very slow compared to actual RAM, and the cost of an external chip (and board/routing space) is usually more than the cost of the next bigger micro
bootnecklad: they all seem to be non volatile
[20:05:38] <MattyMatt> http://www.ramtron.com/products/nonvolatile-memory/serial-product.aspx?id=109
I need a 20bit addr one!
so it's meant to replace an EEPROM :)
I think I'd rather use a plain mpu + d/sram with a cpld/fpga as the glue chip
yes, it's an ic :)
32MB dram are common and cheap
how much ram would you need for a minimal RT kernel and emc2?
that's left as an exercise for the asker :)
as I thought experiment, I'm guessing even 20kB + 128kB is not gonna cut it
but 32MB probably could, without too much stripping down
no X obviously, or maybe not obviously if it's a midget LCD
ah well. one day I'll have a breadboard like pfred1's and time & money to play with it :)
* MattyMatt googles beagleboard
eek 3.4Mhz max speed on the i2c ram chips. forget that
could you guys point me to resources where i can learn about diy cnc mills?
on the wiki i see a page on stepper devices, but i didn't find a list of diy projects and their complexity/price/precision
ah there you go, case studies
so which one is suggested for diy?
[20:24:46] <kanzure> http://diylilcnc.org/
DO NOT make this :P
i'd suggest maybe one of the gingery machines with a cnc conversion
unless you want to convert a mill that you find somewhere?
'conversions' are where you pick up some manual mill and thorw in the electronics on your own
corecode, do you have lots of plywood lying around?
[20:30:42] <MattyMatt> http://imagebin.ca/view/VYrny5J.html
I think if I was starting again, with a slightly higher budget, I'd pay the $35 for solsylva's plans
when you say cnc mill, are you thinking heavy metal or desktop wooden?
my dremel pusher is a mill, and it's cnc :)
i have a dremel pusher
can you control hal parameters with a pyvcp slider?
MattyMatt: did you convert it to cnc?
are any of these good to mill small pcb features?
mikegg: Yes, that is their main use
I'm continuing to work on my Hal file (it's finally starting to make sense!) and I'm trying to figure out how to tie together two stepper outputs for a gantry
as in, I have two seperate step generators / encoders on my Mesa board that need to work in unison to move my X axis gantry
corecode, mine is pure CNC. you'd want sth stiffer than mine for fine pcb work tho
plywood gantry is a bit wobbly etc
there's supposed to be some steel in the sides, but there's none yet
if you have $500 + shipping from taiwan, search for Sable 2015 on ebay
tom3p, have you noticed any downsides to the Sable yet? I read a forum where somebody was complaining about bad alignment or sth
lepton Have a look at the stepper gantry sample file. There is a special kinematics file for gantry systems.
andypugh: how? net spindle-trim pyvcp.scale01 hm2_5i20.0.pwmgen.03.scale is not doing the trick...
mikegg: It needs to be in the postgui hal file, the pyvcp pins don't exist to be linked until axis has launched
Sweet! M8 is hooked up and coolant flows blue!
JT-Hardinge: But is it flowing all over the floor is the question =)
JT-Hardinge: and are you sure thats coolant, and not one of those discs you toss in the toilet tank
question: How (or where) are the absolute coordinates for G53 defined?
could be toilet water :)
from the machine origin
davidf: when you home that sets the G53 zero for each axis you home
OK thanks, I figured that must be it, but I couldn't seem to find any documentation of that. Prob in section on homing where I didn't look. I was looking for a G-code.
Thanks JT-Hardinge !
np, where did you look to find the answer in the g code section?
how do you get more tune tabs to show up in the calibration window?
har har har...
mikegg: That's All Folks!
add more axis?
hmm G53 is still in "programmer-speak" I can barely understand it and I know what it means :)
JT-Hardinge: (story of my life)
I got it. set AXES=4 in the ini file
andypugh: I had it in a postgui hal file
Did you get it working?
no, not yet... went a different route.
my spindle controller is 0-10 VDC analog control
it's not really linear, was going to put a slider to adjust the pwmgen scale
so I could trim the spindle speed in
couldn't get it to work, so I did PID with the velocity feedback
it ought to work, the most likely issue is that the pins aren't called what you think they are called. comment out that net line, start emc, and look in machine->config
Hmm, I'm getting an invalid module format error when I try to call up gantrykins.ko in a hal file (following the stepper-gantry example)
Any ideas on how to move forward with that?
what does your loadrt line actually say?
it keeps telling me: pin 'pyvcp.my-hscale' does not exist
mikegg: Does it?
The pyvcp panel is loading OK?
Just like the example file (assuming your question was directed at me)
Yes, it was
ahh yes. No, it doesn't. But when I change it to pyvcp.my-hscale-f it then tells me that hm2_5i20.0.pwmgen.03.scale does not exist....which it doesn't
I think I am just too much of a rookie when it comes to hal. I've read through the tutorials on linuxcnc.org many many times. still confuses me
I just want to drive hm2_5i20...scale with pyvcp.my-hscale-f
Yeah, me too.
In the HAL file, just after all the loadrt lines, add a line saying (I think) "show all"
That should dump a list of all pins and parameters (possibly to dmesg)
Actually, it goes to the terminal, which means you need to start emc from the terminal.
mikegg: What's your "num_pwmgens" number?
i mean, it makes sense that the command is not working. hm2_5i20.0.pwmgen.03.scale is not a pin
Ah, it's a parameter
Take the pwm demand input through a scaling function instead?\
yeah, I could make a pin for the slider and one for the command value and then multiply the two before I send it to pwmgen value
(loadrt scale count=1 ; net pid.0.output => scale.0.in ; net pwmgen.03.command scale.0.out )
[23:22:18] <andypugh> http://www.linuxcnc.org/docview/html//man/man9/scale.9.html
well not using PID, but the idea holds
Then you can net your slider to the scale.0.gain to get what you want.
Closed-loop spindle control might be better though.
oddnamezor is now known as frallzor
that's what I thought, but I think the spindle motor controller is running it's own PID at a faster rate
which can be problematic
hmm if I want to calculate reduction for belts is it as simple as like 1 gear = 42 teeth and and is like 12teeth then you just divide them and get the reduction you get?
like in this case, 1:3.5
Any ideas on the issue I'm having with gantrykins.ko - dmesg reports "gantrykins: exports duplicate symbol kinematicsInverse owned by trivkins"