what is that, an inverting buffer?
it looks like what you want is "up/down" PWM, not "pwm+direction". that is pwmgen output_type=2
type 2: up/down
Two output pins, up and down. For positive inputs, the PWM/PDM
waveform appears on up, while down is low. For negative inputs,
the waveform appears on down, while up is low. Suitable for
driving the two sides of an H-bridge to generate a bipolar out-
ill test tomorrow
UUUHHH NO... today ;(
$44.82 ea (qty 10 pricing) for E5s encoders with index... plus connectors, add about $15 for 6ft shielded.
jepler have you written the manual?
insmod: error inserting /home/jepler/emc2/rtlib/hal_parport.ko:
floppy12: I've written parts of the manual
just like the rest of emc, many developers and users give their time to improve the manual
jepler where stand this? [02:05] <jepler> type 2: up/down.................................
floppy12: in the terminal, type "man pwmgen"
floppy12: or online, http://linuxcnc.org/docs/2.1/html/man/man9/pwmgen.9.html
thank you a lot
i hope it work
those weird russians. http://www.dailymail.co.uk/pages/live/articles/news/news.html?in_article_id=471324&in_page_id=1770
how does one go about finding out the actual amount of backlash a machine has?
use a dial indicator
preload it a bit, then set the dial to zero
command a .1: move and see if you get that
then a -.1" (inc) move and see if you get back to zero
the readings will tell you how much backlash is under no load
(at that spot)
(will be more or less if lead screw is worn in different places)
not from X.1 to X-.1 which would go .2"
noticed circles I was cutting
are a tiny bit elliptical
yep 3-6-9-12 - where the axis drives have to reverse direction
is shorter in the x than the y
old CNC's had a 1/4 circle limit - had to continue on next line to go past a reversal point
I was doing small line segments anyway
having problems with my CAM program generating emc compatible G3's
* Skullworks-PGAB likes new CNC's that cut a full circle with a G3 J-1.
which post are you using
version 21 or 2007?
I have 221
have not tried the post yet since the hardware is still ing process
program gets G3's right alot of the time
but it doesnt like it sometimes
using inch to the 4th or 5th decimal
well EMC2 has a bad problem (if the manual is correct) that requires a X or Y value on any arc command
yeah I have noticed that
most commercial CNC's don't.
you can force BCam to place them... at least I think I can
well there is an "Output ALL Coordinates" option
but that buts an X/Y/Z on very line
file size balloons
yup... I got tired of messing with it, so I incremented the arc to line segments
I am going down to pick up a powersupply tonite to build another linux box.
I will be setting it up to run sims
The arc issue is one that I think should be addressed
It is a fossil remenant of Allen Bradley's NIST involvement in the early EMC developement
and it remains one of the biggest compatability issues that keep people from adopting EMC2 in industry
how dug in is it? Ie.. how much effort to change?
I have a great library of macros that I would love to port to EMC but they will error if this is so
not sure - I think its coded in the interp
but I doubt it would take much
since most other values are accepted as modal until changed
who is the interpreter guru?
JMK or Alex I guess
Though ray might have a handle on it to
* Skullworks-PGAB can't write C - only debug it
I can read it...
writing it... not usually with good results
John K what are your areas of expertise?
I know you've been doing alot of the Mesa drivers
and you avoid stick GUI issues
I work on HAL and realtime stuff like the motion controller
drivers are part of hal, so I work on those too
homing code, jogwheels, that kind of stuff is what I do
in interp_convert.cc start by removing the lines near NCE_X_AND_Y_WORDS_MISSING_FOR_ARC_IN_XY_PLANE);
function convert_arc; lines 170 through 180 in my version
are full circles such an important case? They are of course a pathologically bad case for R-format arcs
so maybe the check should be made only for R-format
* jepler doesn't know what's best
that would work
what's the exact incompatibility?
well lets say you wanted to mill a full circle
the command could be g3 i-1
that's the whole problem?
so it has nothing to do with R format (you can't cut a full circle with R format no matter what)
threadmilling macros will output an I | J and a Z
even commerical machines sometimes error out on a valid arc command when an R value is used
if nobody wants to fix it right now (I don't) the best thing you could do is file a bug report or feature request
if you can test the change I suggested and mention in the feature request that it works, that's even better
well - strictly speaking it follows the NGC definition
but not RS 274
and it isn't a killer issue - it just makes headaches because of how many post processors are defined
its an all values or only changed value type choice in the post
I can see how it's probably typical to write (if the XY target is the same as the current position, don't output XY words), and that means this case is broken
crap - keyboard still acting up.
great, will you file it? I think it is a feature request - I checked the docs and they do say at least one X/Y word is required
sure - I guess I can do that - I'm about to head out the door now though.
ok, whenever you want, I just don't want it to be forgotten
trust me - I won't forget - it makes my $2200 engraving software useless, and my macro library too.
I'm surprised we haven't heard about this before
I hope to port that library and once proven to be fully functional/compatable offer it up as an option
normally - all the cheap post-processors allow you to persist or not the coordinates.
the NPT thread milling macro would be a huge help to some people
its got my vote.... I cant code C, but I would be willing to take some active effort if possible
bodies to hide???
bbl - must fetch that pws
omg, dmessier is a live
Ziegler: i think the simplest ball screw backlash test is to mount an indicator to the nut and put tip onto flank of a screw groove. then move the nut fwd & bwd ( not rotation) to show any backlash. This only shows lash between the nut and screw.
actually dudes usually measure from the spindle
putting an indicator on the spindle, and pointing it, in line, at a parallel
take the backlash out, set zero, then move an inch
move back an inch
deviation from zero is the backlash
if you have an interferometer you can do it for all three axes at once.
I've done something similar to what tomp describes, only I used a lathe and fixed everything so that nothing could rotate. Then I mounted a dial indicator with the plunger touching the nut. At that point, I pushed and pulled on the nut
this was when I was reloading balls into ballscrews
trying different value balls in .0001" increments
the problem with that is it doesn't show you the actual effect on the machine, and also doesn't reproduce the forces the machine will generate
I was looking at the screws themselves
yeah, but you don't use the screws by themselves
then we looked at the screws in the slides later
checking the thrust bearings
this was for a bunch of screws and slides
I do the indicator against the spindle trick on my machines here at home though
that's the ASPE standardized test for machine backlash
i need to get a copy of that document
what is the prefered way of controlling spindle speed with EMC on a lathe? (PWM? a command through the serial port, a DAC?)
no, just "fff"
I'll take your word for it
anyone have any ideas on squaring up an aluminum blick and shaving off about .015" or so?
uh, like, you are saying you have .015
and that's it
is that per side or what
to true something up, you pick a side
put that side face up in a vise against the hard jaw
er fixed jaw
put a good sized bar next to it and clamp the part through the bar
this is side one
mill side one
flip it 90 degrees so side 1 is on the fixed jaw, repeat clamping with the bar
now side 1 should be perpendicular to side 2
flip it 90 again, still with the bar
so side 1 rests on the bottom of the vise and side 2 is resting on the fixed jaw
machine this side, then flip it 90 again, remove the bar and clamp it
that is a much more frustrating process then you'd think
it's not at all, and it work
i do it for production crap at work
forget to just hammer down one side and you have a funny shaped block
don't hammer it down!
that will put warp into it
if I don't get get a parallogram shaped
the vise is lifting the part then, carry on
not hammer hammer...just tap it down so the parallels no longer slide around
if you use a toolmaker's vise that won't happen (or shouldn't)
toolmaker as in the screwless ones?
but hammering does put extra stress into the part
a good one, not a cheapass HF vise
I was doing it on a kurt
yar, any screw vise will lift
kurt anglock vises do it less than other brands
but they still do it
I dont have a mill
like I said, it is frustrating =)
Jymmm: same proccedure on a 4 jaw chuck
just keep facing it
and rotating it
I dont have a lathe
was thinking of a face plate
face plate works too
I dont have a granite plate
two jaw chuck, four jaw chuck, faceplate
4J is another one of those super annoying things
any of the above
I have duct tape, bailing wire, and an angle grinder!
not going to happen
you could file it
got a machinist square as reference?
i have serious doubts as to whether or not this will end well
*as to whether
toast: me too
file some, hold vise to light, note high spot, repeat
not square + peice to light i mean
Jymmm: you have a lack of equipment necessary
to make what you want
toast: tell me aboout it
go buy some
how big of an Al block are we talking about?
2" x 2" x .5"
and what kind of squareness call tolerance needed
it's the ballnut mounting block
ending so badly
Jymmm: do you have any reference angle plate
that you can use
toast: No, but I was thinking of ordering a 1-2-3 block if need be
I'd never use it again.
1-2-3 blocks can be useful man
i use them a lot
for measurement crap
but i digress
I dont have a mill though
they're not really for mills
so it just needs to be square, no need for a hole to be bored?
you aren't really supposed to set up using them, but you can if necessary
also what i was thinking is if you have an angle plate and some glass
if you have time, sign up for MCNC71 at deanza and use their bridgeports
set the reference 90 deg thing on the glass, put some sandpaper on the glass
and you can sand it
but i like ds2's idea
Are HF mini mills "square" ?
not at all
no mill is intrinsically square
you have to tram it
heheheh, your block right now is probally more square then they are as purchased
even if you tram it, the HF mills are... iffy
if you change your Z height
on an HF mill
you will no longer be in tram
Jymmm, where did the block come from?
most extrusions are pretty decent from the mill
oh so it has been worked on already
Yeah, they fucked up the machining when they made it the 1st and 2nd time, I just never realized it was "THAT" bad
might easier if you buy a peice of 2" wide, 0.5" thick bar stock and file the ends square
like I said, most extrusions are pretty square from the mill.. which means you will just need to clean up the ends you hack/bandsawed
so just 2 surfaces instead of 6
It's not an extrusion...it was machined
so there are other features on it besides just 6 flat surfaces?
Ohhhh that complicates things a LOT
you don't just need a square'd block, you need it squared relative to certain features?
I moght have to remove about .030"
No, just sqaure it up againes the one edge
so is square now and just the wrong size?
I jsut dont have anything to remove THAT much material and sqaure it up.
ds2: (00:36:57) Jymmm: Yeah, they fucked up the machining when they made it the 1st and 2nd time, I just never realized it was "THAT" bad
cuz if that is the case, maybe super glue it to an angle block
and use that to how against a belt sander or disc sander
you asked for square.. it could have meant they made it a parallogram prism
it sounds like you just need to shave off 0.030
but how do you square the sander to the material?
you square it to the rest or table
and the angle block would give you the 90deg
and use the angle block to run it into the abrasive surface
touch a bit, mic it, and repeat
no sander, jmkasunich made me return the one I got from HF
damn thing is, I could probably do it on my machien, if it wasn't the part I needed to drive my machine.
if you had a mill, it'd be a easy fly cut job
know anyone at san jose city college?
sign up at the techshop?
what ya think? http://www.use-enco.com/CGI/INSRIT?PARTPG=INLMKD&PMPXNO=947676&PMAKA=630-4010
pew pew pew
they're probably not going to last very long
i wouldn't trust them as much more than blocks of metal
and i can find cheaper blocks of metal than 8 dollars
[08:35:50] <Jymmm> http://www.use-enco.com/CGI/INPDF?PMAKA=NV625-8511&PMCTLG=01
27 bucks is still kind of cheap for 1-2-3s
still up eh
sandpaper+glass will only get it square in one direction
should just make it on his mill.. it does work after all
wow I hate glade more every second I use it
click the wrong thing -- hit paste -- all your work from a session is gone, because there's no "undo"
My tts was reading Cory Doctorow's "Someone Comes to Town, Someone Leaves Town" and in it they talk about mesh nets. I see that gumstik has a new wifistik that can do meshnets. The idea of a free open community network is great. http://www.gutenberg.org/etext/18224
i am a litte more happy now
i can control with hal 0 to 5 V output ;)
on a prallel port?
i used pwmgen in mode 2
so i can give a number between 0.0 and 1.0 for 0 to 5V on pin A and a number between -1.0 and 0.0 for 0 to 5V on Pin B
ok, and you are low-pass filtering the pwm? or driving an amp directly?
no, a lowpas
but my amp gonna wrong :(
sometimes it let flow 120mA, sometimes 0mA....
so i must search the mistake
but if you close the loop with a PID some small imperfections or nonlinearities in the amp should not matter
yes i know
but i want a perfect system :)
the olpc project uses mesh net for kids in computer-challenged countries http://www.youtube.com/v/7kHIZXYJbWY&autoplay=1&rel=0
Makes it much easier for kids to IM 24/7
I was in Fry's yesterday, there was a girl (maybe 14-16yo) on her cellphone at the Mac display just talking up a storm with her friend. Was giving me a headache she was talking so fast, and had nothing to say!
weird i just read "down and out in the magic kingdom" this morning
fenn: good stuff
horribly depressing if you get to thinking about it too much tho
i was bouncing around the idea of making a free mesh network using usb 802.11 thumb-thingies and dumpster-grade pc's
wow, you're kurt from the story ;) and you're the guys EE from the olpc project ( shoes for the people! free huey, free wifi )
well, i'm on page 0, i'll tell you if you were right i guess
OLPC is great; i just wish they'd give them to american kids
fenn: that book is a lot of stories, the mesh net is 1 story. its not separate, it's weirdly threaded/knotted with the other stories, so you wont get done with one without the others
and olpc, i wish we could pay our taxes in donations of units to kids. btw you can get the os
any one have machine set up with rigid tapping ??
I only know of the mazak that we set up during fest
hmm.. and lathes that can thread???
only if the spindle can reverse
so its servo/spindle control thats the biggest issue??
you can't rigid tap without that
fenn: were you being serious?
fenn: Sorry, I meant about the free mesh network
i have an old mazak with a fanuc 3 - available for retro... but wasnt sure i wanted to play with it... any thoughts??
i used to volunteer at a place that fixed up abandoned bikes and gave them away, we were talking about a similar idea for computers
and then the problem was how to get internet access for all these people with free computers
but if we install the software we can run mesh networking on them as well
fenn: I've been considerign the idea using linksys routers and modified OpenWRT
usb thingies are $5
fenn: The problem is no ext antennas
run a usb cable to the roof with your thingie sealed in heatshrink tube
USB has a 12' cable length
only if you're a wuss
its 16m anyway
er, 5m bah
hm, looks like the emc script can't handle having your configuration directory named "Jeff's Mill" and your inifile named "Jeff's Mill.ini"
so much for user friendliness
either way, too short
should work hanging out a window too
i was planning on running usb over cat5.. would that work better?
definetly add range..
fenn: Well, the linksys has two antenns, I'm not sure if that's in a dual antenna configuration or dual radios.
fenn: The other thing is the A/B/G routers, using the A (5GHz) for control/uplink
It has to act as BOTH: an AP, and a repeater.
jymmm this may be interesting to you http://www.cuwireless.net/
i talked to these guys four or five years ago when it was just getting started
fenn But that requires pole mounted radios, no?
they put them on rooftops
i also liked the ronja
[18:15:12] <fenn> http://ronja.twibright.com/
fenn: Yeah, but commercial rooftops, not residential, right?
fenn: Meaning that city/county permits are required
its residential, i dont see why that would be a problem
What, gettign permits?
why would you need a permit to put something on your roof?
fenn: What you talked about was on commercial roof tops
jmkasunich: are you near the kybd this fine afternoon?
did you ever get a complete eagle project for your fet bridge?
technically its skunkwork's project
I am playing with a pwm chip to do current control...
I just did some layout, to optimise stray inductance, etc
I have one layout, but it seems from the very early stages
so you are looking for the power board layout?
yes... with a couple of added current sense resistors, I think it could be driven with the uc3638 as a torque (current mode) amp
maybe... to a certain extent, an H bridge is an H bridge, you can do whatever you want with it
yes... but keeping the smoke in a brdge is one of the hard parts when dealing with 100v/20a servos.... your layout seems to be surviving
I'm having trouble finding it
rather, I've found something that I think is it, but I'm having trouble invoking eagle to take a look
(I haven't touched eagle since I did that layout)
and its on an older computer
the files I have are called amp.sch and amp.brd
I have those too
mine are dated....
oct 7, 2006
after I sent those to skunkworks, I believe he added the LED side of the optos - connectors, LED series resistors, etc
this is the latest I have - http://www.electronicsam.com/images/KandT/servostart/schem/
The one that actually cut the board I brought to the fest
nothing is really labeled correct though ;)
I think LawrenceG should start with yours, not mine
thankyou skunkworks .... I see you have the ngc files as well
skunkworks: have you been running your h bridge much?
some testing. about 50v 10a so far
have not blown it up
it would be interesting to observe the FET drain-source and gate-source voltages during switching
drain-source for turn-off overvoltages, and gate-source for ringing etc
I have been meaning to do that.. It is at the shop now where the scopes are.
I remember you mentioning that before
what fets are you using at the moment?
and the drivers? ir2111?
yes - looking for the list
I had the list on cnczone - but I think it is down right now. give me a second.
oh neat -- http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Koppi's_Toy
looks like "g0" doesn't work :-P
looks like the pen doesn't lift
right, which he says (Z not working)
skunkworks: right, that's what I meant
... lloking, but just get an emptywiki page
may need to write %27 instead of '? http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Koppi%27s_Toy
LawrenceG: fdh44n50 - fairchild 44a 500v http://www.newark.com/jsp/search/productdetail.jsp?SKU=82C2497&CMP=AFC-TL10000001
the Rds is a bit higher than i would like..
for me - 200-250v mosfets would be better.
Skullworks-PGAB: how about the diodes across the fets?
hold on :)
and ic4 the opto isolator
thks... updateing the schematic...
That would be cool :)
* skunkworks hates the cleanup work
jepler: koppi's toy is hektor http://www.hektor.ch/
stepper cog belt driven polar graffitti painting
was that emc controlled? then the kins is like polar (left motor and linkage) & complimentary polar (right motor and linkage )
the macintosh path generator has complicate dlooping designed to 'steady' the swinging mass
[20:59:19] <bjorkBSD> http://www.instructables.com/id/EBZ3QQVF3HY3SL2/
have anyone of you tried that? ^
no, but it looks nice, and a big one could make the parts for little ones ;)
it also sounds interesting to me
Hektor's software is based on Scriptographer, a scripting plugin for Adobe Illustrator"
regards, the wide table there's always the possibility of a wide table twisting during motion. i just imagined a band type motion where the band was as wide as the table, and in the band were diagonal slits that went over a driven worm... no twist, force along entire edge...
jepler: yeah, the scriptographer demo movies show the weird looping
a script running in an interpreted language in a proprietary application on windows is about as far away from emc as you could get
i saw chinese guys painting with water on concrete, the painting only lasts a while... maybe hektor/koppi could do same for local townships, to advert stuff.
no ones done anything but gcode in emc so far, but the innards allow for line/arc primitives. only got a gcode interp so far ( i know ... lazy me ;)
tho hal could stream ..
how fast is the parport pwm?
floppy1234: afaik it can change state once per base period
for the wide band positioner, the worm is unneccesary, the band keeps the wide table from crabbing. it can just be driven by the axle
lerneaen_hydra i want to calculate a lowpass... so i must know the maximum frequency
floppy1234: that depends on your base period
the max stepping rate will be 1/base_period
(again, IIRC ;) )
should be somewhere in the 10-59khz range
BASE_PERIOD = 20000
also 50 kHz
But I don't remember - I think you need to divide that by 2. You need 1 period for off and one for on..
yeah that's true
and the value?
i mean this is much slower
25khz then I would think..
between 3 and 10 khz
spreadsheet at the Bottom of page: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TweakingSoftwareStepGeneration
i gonna read
JymmmEMC: that is for steppers - floppy1234 is working on pwm.
skunkworks: no idea, jmkasunich write it up
says it's for step gen
floppy1234: I would man pwmgen.. I think it tells you everything you need.
[22:01:20] <floppy1234> http://linuxcnc.org/docs/2.1/html/man/man9/pwmgen.9.html
i gonna read
i cant find something about
floppy1234: here's a vid of smooth motion with similar rodless cylinder from Bimba
[22:44:24] <tomp> http://www.bimba.com/bandsvideo/
tomp: yeah i can identify with kurt
WOOHOO! Got the mobo mounted in the case... only needed to use an angle grinder and the drill press for the cpu fans!
Will take pics =)