indicator is mag-mounted to the bed, indicator is touching the head
Directly above the screw?
close, but not directly.. within a couple inches
seems awfully consistent for it to be something like that
Try it a couple of inches the other side?
Well, try other stuff first. That is a bit of an out-there suggestion
i still advocate just changing the pitch and being done with it
Yeah, lie about the axis scale and see how it goes.
It returns to zero, so isn't missing steps.
presumably the help has the method the scale figure in the ini file is calculated?
e.g. SCALE = 20320.0
i set this up in stepconf
steps * microsteps * pulley ratio * tpi at a guess
What pitch leadscrews?
(but i work in inches)
200 step motor?
200 step motor, 10 microstep drive, 10:20, 5.08TPI
if you don't have gage blocks, use a 123 block for 1" and test the same zero on the dial indicator - I'm suspicious of those cheapy ones.
i have cheapy 123 blocks too.. :)
It isn't accidentally a 5tpi screw is it?
andy: i don't think so, all of the ball nuts and screws were the same in the kit
the crappiest 123 block will be 1" +- 0.001
it's hard to foul up grinding more than .001
i also set my calipers to 25mm and checked several points on that screw, it falls dead nuts on the little grooves on the tops of the threads
Put a flag on your leadscrew pulley and dial in a 50mm move. Is it _exactly_ 10 turns? If it is, then the issue is definitely leadscrew pitch and nothing electrical
it could be exactly 10 turns and still be wrong
MarkusBec is now known as MarkusBec_away
misreading like an idiot
as best as i can tell by eye that seems to work
10 revs, the flag seems to come back where it started after 10 revs
i had to do it in inches, moved to 1.9685.. should be close enough
Is it the same at both extremes of travel?
(ie if you re-set the DTI and try again?)
towards the column vs away from the column?
if it is, then you can either fiddle the axis scale and forget about the "why" or look further
i have tried this at a couple of points on the table anyway
rather, with the table in, and out
Yes, try all the way in and then all the way out. See if the problem is consistent on different parts of the screw
i'll try again, it's easy enough
I think you might need to contact the manufacturer of the screw. The fact it returns so very nicely to zero makes me feel that pitch error is most probable
it looks similar at two ends of the screw
Can you clamp a block to the table, wind it all the way in, measure block-to-coumn with the calipers, wind it out all the way and measure again?
If that also shows the problem then I would definitely be thinking a dodgy screw.
i'm not quite visualizing what you're suggesting
As the DTI has only a limited range, use calipers. Either column to table slot, or column to block clamped to table.
ah, i think i see what you're saying
let me see
I am just curious what the magnitude of the cumalative error is.
i was able to clamp my calipers to the column and to a block on the table
poor man's DRO
And the conclusion?
just running it now
-0.5: -0.507, -1.0: -1.015, -1.5: -1.522, -2.0: -2.029, -2.5: -2.535, -3.0: -3.041
out of travel
Very consistent 15 thou per inch then.
Almost exactly what you would get with a 5tpi rather than 5mm pitch screw
Hmm, perhaps not.
yeah, 5.08/5 = 1.016
(5.08 tpi for a 5mm)
But not exactly that, as it is then 2.032 and then 3.048
yeah.. close enough to be interesting, but unlikely to be the error
We have determined that the leadscrew is going round the right number of times. There is not a lot left.
Try changing that scale to 20,000,rebooting and seeing what you get.
i'm running off of the live CD, no stored configurations yet
It rather depends on whether you want to make it work, or want to understand the problem. (myself, the latter every time)
OK, so Machine->Hal CConfig
i just ran it in the other direction for giggles: .505, 1.011, 1.516, 2.024, 2.530, 3.038
6, 5, 8, 6, 8
goes back to zero within a thou though
(after backing through the backlash of course)
See how this works: Go to the Machine Menu and View HAL COnfiguratiob
ok, i have the HAL Configuration dialog up
Then navigate to Parameters-Stepgen-1-position-scale
ok (though i'm working off of axis 0 still, it's still hooked up to my X hardware)
shows 20320 as it's set in the ini file
What's the number?
in the Test HAL command: box type setp stepgen.0.position-scale 20000
looks like it took
Now repeat your test
.498, .995, 1.493, 1.992, 2.491, 2.990
that's in the original direction i was testing
Closer. So that would be an option
-0.500, -0.999, -1.499, -1.997, -2.496, -2.992 in the other direction
it's very different in the two directions
almost dead nuts at first, then it drifts off quite a bit
You could keep fiddling the number till it works. See it as tuning out the machines imperfections in software
are you doing this from a center point
i'm not sure how i'd tune out the fact that it's so grossly different in each direction, otherwise i'd be with you
are you moving the carriage back to some central point before switching the indicator
and then going in the other direction?
toasty: not sure i know what you mean.. i'm using about 3" of travel on the table, i run it one way going 0.5" at a time, then re-zero (and fix backlash) at the other end and repeat going the other way
don't do that
go one direction, then stop. flip the indicator, and go back to zero.
This is with a lashed-up DRO (caliper clamped to table and column)
what andy said
The other axes are significantly better are they>
yeah, much better
afk a few mins, gotta fold a load of laundry
Looking at the numbers I think you could tune to 1 thou per inch, which is probably good enough for practical purposes. However you will probably always wonder what is wrong.
Too late for me on a workday. I think that it looks like a duff screw. I can think of exotic ways for a nut to fail, but they seem unlikely (loose ball track rotating round geared by the balls so it is consistent, but I am struggling).
Talk to the manufacturer tomorrow, is my advice.
thanks for all the help tonight, i think i'm about done for now.. i'm sure i'll be back.
in case anyone is still alive and following my saga... I posted a video of the poor man DRO test, as well as a table of results
[02:44:50] <gweepprefect> http://www.tangentaudio.com/?p=185
gweepprefect: i just got back, i saw the link to the new test.
i think ...
it does not look nonlinear, it does look like a linear error PLUS some lack of precision.
if i fiddle the values barely, you have 506 1012 1518 2024 2532 3038,
which is a very linear error. that can be scaling, and can be needed because the actual pitch is not the advertised pitch.
the remaining error may just be that you dont have Grade 3 to 5 screws, but a rolled balled screw
( what i saw looked to be a rolled screw )
i bet you can use scaling to reduce the numbers to 500 1000 1500 2000 2502 3002 not perfect but thats the screw
just for a test, multiply your present scale by 0.9881422925 and see what happens, you can always go back to the original value
and your vids and your work look beautiful
tomp is now known as tom3p
which work do you mean? i can't take credit for the conversion mechanicals, those are from a kit.. they are really nice though, i am impressed myself.
like the amp wiring, very clean
ah, that is mine.. i'm an electrical and software engineer and probably slightly OCD, i can't help it. :)
does EMC2 have screw mapping?
pitch-error comp? looking now. there's several kinds in cnc, and what emc2 has?, well gotta goto the books...
[03:43:11] <tom3p> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Screw_Compensation
better here, looks pretty arbitrary (thats good ) http://www.linuxcnc.org/docview/html//config_ini_config.html#sub:%5BAXIS%5D-Section
looks like it's mutually exclusive to a backlash compensation, though i assume you get some of that by making a compensation table
its not meant to correct for a screw advertised as 1"pitch that is really 0.998xxx"pitch, if used for that, it would jerk at each point. i'm not sure how the error value is added to the motion (not sure if theres any smoothing in the error addition at the summer or it just suddenly goes to the suggested position, also with 256 entries and a short screw it can be smoother due to smaller deviations)
i would hope they did some kind of interpolation between the points in the table, but i don't know much about what's going on in the motion engine
well you can try it, and you could just try the scaling. just record the initial values so you can get back.
i will try the scaling, probably tomorrow evening
we only had 64 points on the heidenhain 4xx series, and unidirectional.
tom3p: I think emc2 uses linear interpolation for screw comp so there is no position jump.
i am off for the night.. thanks again for the help. i'll probably stop in tomorrow and let you know how it's looking
EbiDK is now known as EbiDK|AWAY
MarkusBec_away is now known as MarkusBec
MarkusBec is now known as MarkusBec_away
MarkusBec_away is now known as MarkusBec
alex_joni: You interested in any rack space by chance?
Jymmm: hmm.. thanks, but no
got two unused racks here :D
alex_joni: Well shit, where do I ship the 1U box to?
unused as in not connected to anything, cause I didn't need them :P
but if you want I can easily store a 1U inside :D
alex_joni: with power and connectivity =)
that wouldn't be much of an issue.. but like I said, no need here
I only have one public ip though
could probably get more, but.. didn't need them :D
alex_joni: Just thought it be nice to have a box in EU
I need a server in each continent for my MMO :)
alex_joni: I'm working on starting a colo co-op and looking for ppl that are interested.
I'm thinking 60+12+1 cpu config on each one tho
MattyMatt: 60+12+1 ?
yeah, whole planet sim, with each of the 60 managing one patch like a soccer ball
12 aggregators, 1 manager
that'll do for testing, until the number of users rise :)
Jymmm, two of us here serve on adsl with our own dns setup
archivist: ok. and?
its an even smaller coop
archivist: Heh. And no ups + generator, multiple peers, nor across 10GigE pipes
I have a ups and a generator
archivist: and cooling too?
no need for one pc
archivist: and multiple peers?
archivist: I have 18Mbps/1.5Mbps here
But still working on colo co-op
my exchange has not got adsl2 yet :((
I already have a few ppl, just trying to find a few more to get the pricing down.
archivist: Actually, I have SDSL, not ADSL.
18Mbps/1.5Mbps is no sdsl
I host a bunch of clients off my dads 20/1.5mbit connection
the "server" is also his TV computer
Valen: serving what? web?
web + imap mail
i make all the web sites etc, it all runs in a few VM's inside his tv
all the email is replicated to my place which is a secondary mail server
I'm not convinced VM are worth the effort, except for testing
they are pretty handy, beat the pants of configuring new stuff if you change hardware
small, one-task boxes are best
if i did that I would have 6 computers under my tv
idiots running mysql cluster on VMs take the biscuits for dumbness
I can see why you *might* want to do that but its not really a good idea
archivist: a cluster.... eh, it really depends.
I'd rather have a stack of tiny machines than one big one that needs 6 fans
* alex_joni imagines a cluster of NSLU's
maybe 2 mil. of them
if your running cluster your going to be pretty intense, the VM overhead on disk-io is probably going to hurt
dbs and vms dont mix
seems (as usual) others have tried that too: http://www.beowulf.org/archive/2007-March/017568.html
currently my tv (quad core 2.6ghz) is running an accounts vm, a pbx, a firewall, a web server and 2 mail servers (one of them is unused, its just got old archival stuff in it)
oh yeah and its also a mythtv front and back end
tilera is beowulf on a chip . latest has 100 cores. it doesn't help disk io tho
unless you use some of the 12 GbE to attach a NAS
upside to all that is my firewall has 3 cores and 512mb of ram with 12gb of disk
its way faster than the bare metal 500mhz P3 it used to reside in
yeah the real benefit of VM is the shared RAM
physically shared, I mean, so you only have to buy one big lump
ahh, I find that the biggest down side to it
instead of a stick on each little mobo
that machine has 6gb of ram in it currently
and its starting to hit the limits of what it can handle just due to running out of ram
to me the biggest benifits are cpu sharing, so that i can get loads of performance on stuff that i only use occasionally
power saving, that machine pulls ~60W
yeah but cpu is not the bottleneck in most server situations
and hardware independance, I can pick the VM's up and drop them on any other hardware without worrying about it
The biggest savings is being able to recover and backup within 30 minutes
soon i'll be getting some new cpu's for my other machines, so i can live migrate all that stuff to my desktop when i want to reboot the tv
if you are actually performance constrained then a VM is a bad idea, most servers spend most of their time doing nothing
jymmm, that was a friend's idea, but backing up VM images is causing him to escalate his RAIDs
hdd's are cheap.
if hdd are cheap, put 2 in each tiny box
he should be able to backup a diff from the "master" backup
tiny box wont be able to do the "big job" when you want it too though
and will use more power
I looked at running my mail server on an atom based board
it was rather slower than it is inside the VM, the mail server I use uses mysql as a backend so it gets hurt somewhat from running inside the vm too
it was slow to the point it harmed my "end user experience" on it
sql is the first thing I'd put on an external box
if you dont need to why bother?
on dads machine which runs the same mail server (replicated) I moved the database to the host to give it some more speed
whats unreliable about a vm?
2 OS to break instead of 1
hell putting stuff into VM's gives you more reliability because you can move them around the network in case of a hardware failure/planned upgrade lol
well its 2 os + network cards + switches + cables if you push it to another machine
I have uptimes of 8 months or more on some client machines of mine, and thats with windows 2000 as the guest
yeah but when one function is stable, the others can't affect it as long as they use the right net protocols
Jymmm, its running some aincent version of sybase sql anywhere for their "one true app"
adding components decreases reliability
unless they are redundant
sybase?! Might as well have said dBase III
jymmm it came out in 1998 i think
it's cheaper to pair just the SQL box, instead of the whole host
i'm just glad they got 2000 not win98 as the server
your missing the point MattyMatt, if you need to use some big iron to run some heavily loaded database, then a vm is the wrong answer
if however you have a number of things that have "bursty" demands, a vm provides a way to minimise the amount of hardware you need to run it, and in my personal experience, that means you can get better/faster hardware to run things on
I'm not talking big iron, although the same applies there I guess, but tiny little quiet machines with one program in rom. appliances
I have found VM's to provide a good mix of continuity over hardware changes, easy backup/restore and improved performance Vs running lots of machines
they dont have the power to do the middle of the road stuff though
e.g. a home wifi router is more reliable than a smoothwall box, if less flexible
I wouldn't say that
it's certainly quieter
I have had several linksys modems shit themselves on me
now I run pfsense in a VM and have greatly reduced dificulties
(and when i piss the linksys off they go away completley ;->)
yes, but if you are going to have one computer running anyway, (IE to record TV etc) you might as well use it
oh, those linksyses were pulling 10-15 watts too
well it's all speculation, as I don't have enough experience of either
last time I checked
I'm just relating my personal experience
careful matching of hardware to the task can reduce the the TDP if unused bits are turned off, but commodity parts reduces that opportunity
the housebrick sized transformer running at 80C wouldnt have been helping
somewhat. you can get 9W x86 and 140W x86
I'd rather one decent box running a number of lightly loaded VM's with the ability to do heavy duty work and get it done faster than a bunch of slower machines doing the same job
I'm talking in the normal small/medium office or the average geeks house ;->
when it says "unable to reach SQL server" you know exactly which box to power cycle
I don't have that problem ;->
I power cycle my modems and switches far more often than i need to do anything to my hosts or guest operating systems
normal offices have no tech at all, generally, except on the phone
all the ones I deal with have a computer on each desk and everybody has email
I then give them IP phones ;->
no technician, I meant
yeah, I'm talking corporate office not manufacturing
actually thats one other thing your not sposed to VM, IP phones
they are sposed to be bad for jitter ETC
I haven't actually had a problem with them so i'm happy
I use hardphones
mm jitter. this is why I want emc on its own cpu
but running the pbx in a VM is not a good idea
MattyMatt, called SMP with a reserved CPU
but not a reserved frontside
doesn't seem to matter
Valen: Eh, you can run asterisk on a ddwrt box, so...
I have a jitter of 2400
Jymmm, yeah but for how many calls ?
how many ddwrt boxes? :)
lol, you can power cycle your 15 ddwrts
stack 'em high, sell 'em cheap
I'll have an off the shelf core-2 for the same price doing a better job ;->
and saving all their voicemails
we used to have nosy switchboard operators with a good memory and ear for gossip, for that
I think the DoHS would like to install computerised equivalent :)
[11:35:06] <Valen> http://www2.vapourforge.com/emc/VM.png
should be a screen shot of what my server is up to at the moment
vapourforge :) love it
pfsense was just sucking 15% cpu from a file I just downloaded
yeah I like it too
need to re-jigger the website itself
oh yeah, thats coming out of a VM ;->
why did google chrome call it incognito mode
they should have called it porn mode
my collection of old xboxes won't get turned into appliances after all. gentoox is dying rapidly and pata drives are overpriced
collection = 2 now bro has taken one for dvd player
I stuck a blueray in the tv computer
blueray sucks for copy protection crap meaning I cant watch stuff on it
if I had cash and a yen for TV, I'd get a PS3
wow they are spensive
yeah, not the best value appliance, but you could hopefully use the gpu for stuff
the mbo I got has a 9400 on board so it can run CUDA stuff
they were the cheapest way to get BD by far for a long while
means that when i come across a bluray i can actually get off the disk I can watch it at ~10% cpu load
I got a 9500 w 1GB of DDR2. great value and I was bragging about my vram for ages
I haven't done any CUDA tho :)
I'll get straight into CL instead I think
hopefully it'll all be fit for market at the same time
havent heard of it?
it's the universal replacement for CUDA/Stream and Larabee
the 1.0 spec is available, and AMD have a cpu-only sdk available
but it's not ready for mainstream yet
it's the future of multicore and gpgpu tho
[11:49:18] <MattyMatt> http://www.khronos.org/opencl/
sweet. NV have released a wrapper over CUDA
only M$ forging their own path now
I wonder if you could push general purpose computing into it
virus scanners would be great to run on a video card
you sure can, if it's a parallel task
yeah but many things don't work over the trip to and from video memory
I'm thinking database queries and such like
1GB vram :) one time journey hopefully
its more the query > response time
I'm thinking a mail server with a state machine based layout with all the protocol stuff being handled by the GPU ;->
I see all this massivley parallel processing working great for stuff like CFD or similar things, but I can't see it making my desktop faster yaknow
we'll see video cards with sata, eventually, or gpu turning up in other appliances. NV already sell one with no vid port
tilera is the fun chip of the moment. 100 cores each running a linux kernel, and interlinked via internal fast IP network
that'd be cool
what sort of performance per chip?
per core rather
the 64 core version has been on sale for 2 years. about $1000 for the card with 12 GbE ports
If only I had a use for it lol
concentrating video streams is their selling point
it would make a nifty web server tho :)
they seem to have no ram
yeah only the cache on each core
if you do it right a P3 300 can flood a 100mbit link ;->
If it had ~512mb of ram per core then it'd be really usefull
this has 27Gbit internally, IIRC
thats only 32gb of ram, its not too insane
oh hang on it has access to DDR
that might do the job
yeah on the TileExpress board
ooh actually they would be really kick ass for VoIP servers
faster than contending with the cpu for the mainboard ram, hopefully
yep, that's another of their selling points
hmm single 10GbE port now. that's new
it's a shame PC are so cheap these days. it makes a $1000 card look expensive :)
you could have a pair of SSD for that price, and see more instant improvements
actually for its applications $1000 is pretty cheap
stuff like VoIP encoding/transcoding is quite cpu intensive, one of those in a backplane would support quite a number of E1
probably ~10000 phone calls or so siumultaniously
in a box ~4Ru high ;->
ok, 20RU then lol
(best E1 cards support 8 E1s
and it'd be cheap
these people pay that much for 1000 calls, and thats without routing them
can you get pci-e backplanes?
I think tilera sell other formats too, but I'm only interested in the PC board personally
[12:26:38] <MattyMatt> http://www.tilera.com/products/platforms.php
yeah was looking through that
I'd love to release a game that requires one as minimum spec :)
I never really got my head around how you talk to stuff on the PCI bus when it is itself a CPU
do you implement some kind of device driver
use it for a-star
access shared memory over DMA
its all scary lol
i guess you load a program on it or something?
if it's like PCI, there are lines for the cpu to seize the bus
bus-mastering. some vid cards do it which is why I suspect they break emc
although as Matrox can do it, and they don't break emc, it's more likely a X/driver issue
I have my 2400 latency with nvidia driver
SMP is the win
carefully selected cpu for each task (in my case selected from the junkpile)
if the the k6-2 500 runs my router, I'll be happy
that machine has no AGP tho, so I guess it's down to the onboard vid
i have used EMC on a k6-2 laptop
allthread & crappy motors & no feedback should keep it conservative too :)
axis runs too, it has a trident graphics chipset
which i guess doesn't have too much acceleration support from X
amiga managed OK with shared RAM. there are advantages that could outweigh the contention
lol sis suck
for video support under linux
I had X running fine on it in Mdk7.1 days
but mostly for games I used fbdev
screw X, gimme direct access :)
yeah, you do that for a new nvidia card
the last thing I did directly was make the freebe/af driver work on mga400
that gives 2d accel in DOS and in theory on Linux but that rarely worked
It used the traditional CRTC so there was a hard limit of 1024 pixels wide, IIRC, but nice within its limits
I just wonder if anyone out there is using EMC2 to control a RepRap (RepStrap to be exact) machine...I hit an issue that am looking for solution..
just ask the real question
It's a machine with 3-Axis motor, controlled by paraller port. And a extruder which push hot plastic out to "print" a 3D model out. The input is GCode, with some M100~ range custom code to control the extruder.
The extruder is controlled by an microcontroller, which is connected with serial port. Currently, I wrote some Python script to accept those M100~ code and to send out whatever serial command needed to the microcontorller
in this scenario it's not really fitted for emc2
you can run one with emc2, but directly
(control the steppers and solenoids and whatnot from one or more than one parports)
Right, but then, how could I? i mean...
I know how to hook them up electronically...even so, how could a GCODE result in a change of a HAL pin?
[14:06:57] <skunkworks_> http://objects.reprap.org/wiki/Builders/EMCRepStrap
The left hand side picture on that wiki is my machine.
I think I forgot to mention the problem but you might have guessed it...the problem is--there is too much lag time in executing the python script. As you can imagine, the motion paused and jerk when it reaches the M100~ command. So the end result is bad print output because the there are too much or too few plastic at some point.
does the extruder have to be controlled with a microcontroller?
EMC is capable of controlling temperature with a PID loop (if you have the right hardware for feedback)
EMC can also control "pump speed", possibly with a G-code axis (like C) or spindle speed command
I am looking for a GCODE command, which could affect a HAL pin. Then I could either hook that pin to some physical HAL pin, or write whatever code/script needed to monitor pin. I wish there is a magic GCODE Parameter (those #1000) which maps to some HAL pin.
there are motion-synced I/O pins in the motion controller, used with M65 or something
that reminds me of a factor we had in a weighing machine, that is knowing the in-flight time so you cut off early enough
hmhm- SWPadnos almost have that right - yes the EMC is capable to do all the calculation that the microcontroller is doing
to control the extruder - we need at least two output
1. the rate (motor speed), 2. the temperature
may be somehow I could map the Spindle output to the motor control, but then I still need another output "command" to set the temperature.
does the temperature change often?
oh yeah you got that right. hm...may be the spindle output would do the job..after all, it's the only thing that's latency sensitive..
thanks folks. that's very inspiring...let me go back and see if I could put them together
ok, have fun :)
sam0737, think about using an extra axis output as the temperature or motor control as well - that would be an output that can be ramped over the course of a move
spindle speed won't do that
oh. never thought of that too
let's see if I could put them all together. Thanks a lot.
the trip to springfield was uneventful other than the flat tire. First one I have ever had. spooky
I found out I can change a tire in less then 5 minutes when the traffic is flying by at 75mph
note to everyone.. check your valve stem seals for failures.
if you had air in your spare, a lug wrench, and a jack, you're in the top 2 percent of the population
heh - new enough car that that stuff was still in there ;)
I had a flat on a motorway slip road in London on the trailer and had a good spare, yes you work fast with cars at that speed flying by
shirley was wondering how long I thought it would take - said around 10 minutes. We got back into the car and she was supprised to see it way less. I manliness has increased ;)
JT-Work: how is the lathe?
it's still on the trailer :(
so it is safe for the moment
MarkusBec is now known as MarkusBec_away
MarkusBec_away is now known as MarkusBec
[Global Notice] Hi all, as you may be aware we're in the process of changing irc daemon -- as we are nearing completetion, we're now looking for some help in testing the new software in preparation for the move. If you would be happy to help test our new ircd, please /mode yournick +w for further information on how you can help. Thanks in advance!
i have a die grinder to use as a spindle on a micro mill. the motor is brushed and is labeled 110V 210W 60Hz 25000rpm. can the speed be controlled?
hook it up to dc and see if it turns.
most likely though
im looking at live tooling
to see how they make the bearing arrangement
skunkworks: i suspect its a universal, if it is(runs on dc), can it be pwm'd? is this practical for a die grinder->spindle?
looking for help with subroutines on emc2
Guest844: ask the question
it wont recognize the o for some reason
www.pastebin.ca <-- can you paste an example?
bad charactor error
what verion of emc are you using?
I dont have an example handy but its a simple drill subroutine from a fanuc post.
even the examples from the wiki source say the same thing
[18:56:47] <Guest844> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?SubroutineSample
I tried that but still has an issue with any line of code with an "o"
I usually work around subroutines but I've just added an A axis and subs would be ideal
can you paste exactly error message you get?
just "bad Character o near line...."
wiki examples loads clearly for me (2.4)
gcode file was created in windows or linux?
windows using mastercam x2 with a generic fanuc post
try load gcode to gedit and see if ot looks properly (line breaks and such)
everything looks great and all the other programs post perfectly it just has issues with any "o" I put in the program
I found version 2.3.4-1 is that correct? I cant seem to locate 2.4
hmm without seeing your error file I can help much
Guest844: 2.3.4-1 is latest release version
my is development version
when I get over to the shop I'll copy the file and post it.... OK I'll install that version it's 1 release newer than what I have
EbiDK is now known as EbiDK|AWAY
Guest844: you have already installed the latest emc version
exmape works for me also. 2.3.4, i suggest copy and paste an run to insure one reproducable error with known src.
tom3p: I would think it would work ok.. PWM'ing the universal motor..
[19:32:35] <skunkworks_> http://www.cnc4pc.com/Store/osc/product_info.php?cPath=25&products_id=236
tom3p, you can buy router speed control boxes that would probably work very well http://www.harborfreight.com/cpi/ctaf/displayitem.taf?Itemnumber=43060
thx i was looking at the ST control system, but 120$ is too much for a dev pcb http://www.stmicroelectronics.com/stonline/products/families/evaluation_boards/steval-ihm007v1.htm
maybe not too much, ST is closed loop if you add a hall effect or tach to the spindle. i spose hal could close a speed loop given the same, and output the new pwm. then again, open loop & 20$ is nice from HF :) thx again
did anyone broach internal splines on a gear=
with a homemade broach?
I have made a broach and used bought broaches
but not splines , it was a keyway slot, forces are very high
* anonimasu nods
so broach manufacture has to be good
im thinking of turning it out of drillrod
if I can get ahold of that
always one machine too little..
I need a edm -_-
my homme made one was carbon steel, got it hardened at a uni and then ground to sharped
I failed on the clearance needed
you cannot pull back so need space for all swarf for the cut
I have to drill a .120" diameter hole 2.75" in CRS. Would I be better off to rotate the part in my lathe to keep the hole straight?
cold roll steel
shouldnt be a problem at all
drill it and bore it(if you need a good finish)
drilling it in my mill or on my lathe?
might have hard spots to cause it to wander
sometimes it's better to spin the tool in the lathe and make the part sit in the toolpost
how would that be different from using my mill?
(but usually it's for big off center parts that wont get into the chuck or cause vibrations
if it were me, I'd drill it undersize then bore it
cant bore that size and depth
I plan on reaming it after drilling
part of one
oh .120" :D
the reamer is .126"
its in the realm of deep hole drilling
if tolerance is critical drill it in steps
the main thing is that the hole exits in the proper place :)
swarf is main cause of wandering, pecking is the order of the day (and night)
It would be easier to peck drill in the mill then
peck drill and ream in the lathe - very easy
cradek: sounds like a plan
also, get extra drills :-)
who me break a small drill...
is this the first or last operation on the part?
The second op
I'm not doing much to it until that hole is done :)
if exit location is the most critical, you could drill from both ends and then ream
reamer will help straighten the hole, but won't change the entrance or exit locations much at all
heh the meating of out of line holes is entertaining
yeah you'd have to get it pretty darn close...
dunno why I designed it like that :O
I want to have a working cnc :(
fix it and hurry it is almost Xmas time again
Currently 3000 miles away from it.
that's going to be kinda tough then
I know... :(
At least my project is only down the road a bit
hopefully my machine will move with me to NC
what part of NC?
I have a dumb question... Hot water heaters.... Most of them have an exhaust pipe that gets hot. Why don't they use the heat from the exhaust pipes to pre-heat cold water supply?
those are called - high efficency ones. ;)
I'd call it common sense. A coil of copper pipe has got to be cheaper than the fuel
Especially for you ice/snow bound folks
Is there a way to measure the exhaust BTU's?
* JT-Work heads home
mass airflow sensor from a car?
I arrived mid thread. But at a guess measure the temperature, time how long it takes to inflate a bin-liner, and multiply temperature x volume flow rate
and a thermocouple
Q = mc dT
acemi_ is now known as acemi
[21:54:56] <roh> http://yamato.hyte.de/packages/heeks/
has fresh heekscad and cam packages now for 9.10 and 8.04
I didn't realize it was that straight forward. What about heat up and cool down times?
Or would I just have to wait till the temp leveled off and go from there?
Wait, is Q (kcal) the same as BTU ?
so - does anyone know anybody in amsterdam?
skunkworks_: Joy Jugs
heh - male stippers? (wife has a lay-over (heh) there for a few hours)
I'm sure they probably do
[22:30:42] <Jymmm> http://www.executivemaleescorts.com/amsterdam_male_escorts.htm
boy howdy its been so long nickserve dropped me!
* pfred1 just punched a hole in a block wall to get an ethernet cable to this system :)
been meaning to do that for oh 4 years now ...
but enough about that I jumped on to find out if there are any schematics out there for port buffers that EMC users use
hmm thats better
I have not used it - but have heard good reviews
skunkworks_ woo nice ta
I got a pile of the ICs in this thing
I use a slightly more brutal driver in my latest conversion a ULN2003 it meant I could drive 12v inputs
well, not ACs all of them mostly HC
archivist_emc I just want something to protect the port
I was thinking going optically isolated
this is just a buffer - not opto-isolated.. (but most of my projects have just been buffered)
printer ports are cheap. ;)
the uln is well rated for high current optos
skunkworks_ think I paid about $20 last time I bought one
besides thats what was in the scrap box :)
skunkworks_ not blowing them out is infinitely cheaper
I have not taken one out - yet
archivist_emc I have piles of 2065s it seems
yeah I'm pretty set on the motor drivers I've made so I need to get them hooked up
stepper motor drivers are touchy!
[22:49:00] <jt-plasma> http://imagebin.ca/view/lbkFvF.html
jt-plasma I met a guy who has a water jet he routinely cuts stacks of sheets a foot thick :)
jt-plasma and it looks like you hand rubbed the edges your whole life!
but hey not everyone has 100 grand to toss around
i thoguht i had that ... damn have to check my account now :/
bills going to hell
MrSunshine he ain't cutting foot thick stacks because he wants to
MrSunshine to pay for his machine he has to!
* pfred1 wonders why freenode-connect is versioning him?
next thing they'll be a/s/ling me
tests for spammers
archivist_emc and this used to be such a nice quiet network :(
archivist_emc: that screen shot was the second test of the THC showing the actual tip voltage
during a cut
they are getting pretty good at weeding script kiddies out, saves us ops some work
archivist_emc least back when it was OPN
pfred1: Freenode has been versioning for years
pfred1: Even back in the OPN days
Jymmm yeah I haven't been here in over 4 now
Jymmm no, never did for me
pfred1: They are checking for open proxys
jt-plasma, looks nice
pfred1: always =)
yea, I'm really stoked to be at this point
Jymmm I wouldn't know an open proxy if it bounced on my lap
pfred1: They have always checked for open proxies. lol
pfred1: blame gnaa for that
pfred1dont ask, you really dont want to know
you dont want to know
gnaa sound like something Monty Python did a skit about
pfred1: it's a hate group
iirc... gnaa == Gay Niggers blah blah
Jymmm and they hate freeload why?
I mean whats to hate?
pfred1: do you really have to ask that?
since poor Rob got run over ...
pfred1: do you really have to ask why one group hates something?
pfred1: far before lilo left.
Jymmm he didn't leave he died
pfred1: Yes, I know.
anyway.... how bout them lakers?
ot: anyone know how to do the inverse kinematics of a cartesian robot in real time?
for lilo /wallops "I lift this clavicle to worlds without number
wholly crap they closed down the #politics channel!
I think I might get one... http://www.makershed.com/ProductDetails.asp?ProductCode=MSGSA
this network really has hit hard times :(
pfred1, has it moved to ##politics
about channels are supposed to be ## and official #
unter I think you may need this: http://larrybettencourt.files.wordpress.com/2008/10/twister1.jpg?w=300&h=292
archivist_emc oh thanks
further OT: i meant inverse kinematics of a planar robot
unter spin it faster it'll work there too!
i can make it move no problem, I just want to make it move in a straight line
unter: what kind of geometry do you have?
two links in a plane, and i control 2 rotary joints
for example: http://hughjack.com/book_integrated/images/spatial18.gif
ok. isn't there kins in emc2 for scara robots and similar which could be used?
I was thinking there was, I went and looked, maybe I should look again
MarkusBec is now known as MarkusBec_away
* jt-plasma listens to My Ding A Ling
I'm wondering if the scara kinematics are correct
unless I'm reading the code wrong, the forward kinematics are wrong
why, oh why did those guys re-use variable names?????
so forward kinematics are correct
I'm tempted to say bad words about that
* jt-plasma listens to Destroyer... the Kinks
me waits for jt-plasma to say Lawrence welk