hmm is that MEEP bad or MEEP good ?
difficult to tell
some good things ... some bad
on the good side, we opened our second factory today :)
on the bad side, well, I cant tell you about that .. not until after the court case anyway :)
robin_sz: hmm might turn out nr 2 is good, just look at SCO vs IBM :D
but can I wait 2 years ?
robin_sz: :( that's the bad thing with the US legal system, if they don't have a case they can always outspend/outwait you
anyone who tries to outspend IBM better have DEEEEEEEP pockets
robin_sz: Well SCO was backed by patent trolls, they don't have as much cash but they do have the manpower ,, but IBM was probably to bif of a bite from the start
and their stock is where today?
robin_sz: lol, not worth it's weight as toilet paper
heading for $1
its even possibkle Darl will be sued personally
I try to follow the story as best I can ... thought its difficult to find a balanced, neutral reporting
robin_sz: I clearly don't follow it as well as you, mostly get what seeps into slashdot
yeah, not very balanced
grokalw is just as bad
groklaw .. if not worse
[01:25:08] <SWPadnos> http://www.groklaw.net/
they're anti-SCO, but accurate as far as I can tell
yeah right... not.
take todays "Novell sells out" piece
ah - I was referring to the SCO coverage, not the site in general
I haven't read it much
its not bad, but the bias is getting a bit too strong recently
it used to be a lot better
well, again they;ve got a particular slant, but is the information inaccurate?
SWPadnos: I might actually get the parts tomorrow.
mmm ... difficult to say, the actual logging of facts is very accurate, the news and comment is dubious
skunkworks, cool, as long as the bad weather around Chicago is headed for us instead of you
the problem is the "PJ is right, anyone who disagrees is a troll" menatlity. it stifles proper discussion
sure - comments are from individuals, and I'm not sure how tightly moderated they are
well, that's not so great
well, if you agree with PJ its not moderated .,. disagree and it usually gets wiped within an hour or less
and then you get "sandboxed" :)
I haven't seen that, but again, I don't monitor that site
well, you wouldnt would you :)
see it I mean
no, I wouldn't
it is her site though, so I guess it's OK :)
once a user is tagged for sandboxing, anyting they post from that IP is invisible, unitl approved
they can see it so they THINK they have posted, but no one else can
man - Dolphin CAM is expensive
$6495 for Dolphin 3D CAM level 2
PartMaster is much less
mmm ... typical I guess
remember its aimed at a commercial enviroment
well, on CCED, a guy just posted that it's now available in the US for "a low price ..."
6K is ballpark
I think even CAMWorks is less than that
(or SolidCAM - whatever the SolidWorks plugin is)
Good evening folks, been awhile.
I'm getting a joint following error, how much more should your minferror be over your backlash?
steppers or servos?
steppers , Hi there Steven.
I'm not sure min ferror is related to backlash
the physical error will be, of course, but ferror is calculated from the output of stepgen, not the actual machine position
The problem just started after I changed the lash figure on Y axis to 15 minferror is 0.010
hmmm - well, try increasing it ;)
should .005 over do it?
if backlash is 0.015 (?), and ferror is related (which still surprises me), then I'd make it > 0.015
but I sure hope I'm wrong about that
Ok, I'll give that a try.
That did the trick, works fine now. Thanks a bunch. Vern...
bummer - you're welcome
See Ya later.
see you - have fun with it
you guys sure talk a lot ;-)
:D we do
(just got done reading back)
Anyone want to try out a prototype dxf to gcode converter ?
Been making one in python for the last week, still far from complete but already better than some of the things out there
jmkasunich: what are you views on a differntial opamp and a current sense resister in series with the motor?
the differential amp needs to handle the full motor supply voltage
whats that for you? 80V, 100V, more?
that ain't your average opamp
im off to bed, gn
no - I was actually looking at specific dif amps that could handle it.
what amps are those?
hold on - let me find the link
the switching edges have frequency components of hundreds of KHz, maybe even 10s of megahertz
look at the common mode rejection of the diffamp at those frequencies - I bet it ain't much (considering that you need to reject 100V or more of common mode and you only have a 1V or less signal from your resistor)
[02:34:41] <skunkworks> http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1154,C1009,C1126,P7521,D5087
look at the first graph on page 11 of the datasheet
at 100KHz you have 30 dB CMRR
that means a 100V 100KHz signal will appear as 3V on the output
god - people should not be so smart :)
if you are just trying to prevent motor overcurrent, you can probably filter the snot out of it, since the motor inductance limits the dI/dT - anything really fast on the diff amp output must be noise
but if you are trying to protect against a short at the motor or output terminals, you need fast response, so you can't filter
and the common mode noise would cause false trips
I was looking using the pluto board which is pwm at 20khz
* skunkworks is just trying to keep up.
its not the PWM frequency, its the rise and fall time
Ok - I thought I was missing it again.
I am sorta back to - see if it will work the way it is ;)
thats a good choice
put a fuse in series with the main power supply (_not_ between the big cap and the FETs)
that is the plan.
what is your target current again?
this is the caddilac approach: http://www.lem.com/docs/products/LA%2025-NP%20E.pdf
put that in series with the motor
the current sense signal is completely isolated from the motor circuit
no common mode problems
only problem is it costs $22
it will measure up to 36A
and is fast enough to use for both normal feedback (current limit) and short circuit protection
if you can tolerate a +/- 18A range, you can even get ground fault protection with it
that is nice.. I have been doing lots of readin on h-bridge current sensing.. And I still know squat.
if the $22 isn't out of line, I can explain how you would use it
It isn't for a solution.. Could I ask you about it in the future? I have about 8 projects before I get to this problem.. if it is a problem.
Thanks - very much.
jmkasunich: did you find ball screws?
turns out mcmaster has 5/8 diameter 13/64 (0.203) lead screws for less than $2 per inch, and nuts are less than $30
those are (i assume) rolled?
0.004" per foot accuracy
but at that price, I can buy two nuts and preload them with 100 lbs or so
you are going to try to preload them with 2 nuts?
and we have screw error comp in emc now ;-)
I think the ballscrew addition is down the road a bit
I'm gonna stick with acme for the moment
don't want to bite off more than I can chew
oh - I never have that problem.
when I do the CNC conversion with acme, I'll still have the handwheels
so if I need to do some lathe work to get the lathe working, I can, even if the CNC isn't running yet
I just found out our z axis spins the nut instead of the lead screw. going to require a bit more "engineering"
once I go to ballscrews, I won't have the handle anymore
that is a good Idea
so the CNC needs to be working first
do you have a scale you can use to calibrate the screw error?
I actually have a set of gage blocks
lot of work to calibrate every 1/2 inch or so, but I can do it if I'm motivated enough
to add myself to a group, I just add my name to the end of the relevant line in /etc/group ?
* jmkasunich reads man group
jepler: nice write-up on your blog
ve7it is now known as LawrenceG
need more help
skunkworks: cradek or I did something to kill the pluto-p last night :(
skunkworks: luckily he'd ordered two more, one for him and one for the inevitable screwup
do you know what happened?
so later this week I'll get back to working on it
it's possible that it was my fault -- I was trying to probe at one of the FPGA's tiny pins, and the probe went ZIP! across the whole row .. cradek tried to fix up any bridges I'd created, but it never worked right again
or it might have been something else
before we realized it was ruined, but maybe after it was already ruined, we saw that VCC went up to 4V (it should be 3.3V) when the parallel port is plugged in---whether that is another symptom of whatever happened, or whether it's the cause, I don't know
yeck. Those pins are so small.
I read your blog last night. Nice write up. You had already posted that cradek had bought 2 just in case.
oh had I?
well as usual he was right
So you think the avr's that you play with are a little more robust?
oh i've destroyed some of those too
nothing is immune to my ability to f**k stuff up
:) the other thing is - you guys really have no isolation between the the
"why is this pdip package melting"
high power circuit
skunkworks: I'm not sure why cradek doesn't feel more uneasy without some kind of isolation
because he has never had problems. :)
but you guys are starting to push currents that will probably start to be an issue.
I only have optos on the h-bridge. I probably won't isolate the encoder.
how did your new code work on the encoder pulse?
we didn't get a chance to test it before I broke it
it's time for me to drive to work .. chew on this for a bit and tell me what you think: http://emergent.unpy.net/files/sandbox/pluto-servo-pinout.pdf
jepler: Our apologizes for this error and we have arranged to send the correct CD to you. The files are quite numerous and difficult to send as an email attachment
there may be one or two PDFs on there I want
oh I should try that tracking number again
yes - make sure I didn
didn't transpose some numbers
Scheduled Delivery: 12/06/2006
anyway -- talk to you later
cool - your not that far away.
skunkworks: any thoughts on the pinout? My basic goal was to put a good selection of pins on the extended main header (the 26+4 pins that can be connected with a single 34-pin IDC connector), and then to put the rest of the inputs together and the rest of the outputs together
It looks good to me. But I am easy :) makes sense
I wonder how many pluto boards get sent back because the 'stop' working.
jepler: floppy drive cable has 1 more pair than the bottom row.
I would just cut off the flipped end (drive a:)
It would be cool if you guys could isolate the h-bridge from the rest of the circuit.. Then see if you still have noise issues.
and I suppose spindle drive.
my setup right now - powers the breakout board from the computer power supply. This is what also powers the encoders. - the h-bridge is on a totally different power supply - it is controleed with the breakout board through opto-isolators on the h-bridge board.
a.out in free(): warning: junk pointer, too high to make sense
does anyone have a clue why this happens?
skunkworks: actually the spindle drive is isolated -- it has an entire separate controller from the board that has the X and Z servo h-bridges
anonimasu: it's some kind of bug in the software you're using. A pointer passed to the "free()" function was invalid.
jepler: it's my software and it shouldnt be wrong..
char *token = malloc(10*sizeof(char));
and then I'm doing free(&token);
anonimasu: it's unlikely that you've discovered a bug in your platform's free() and malloc(). It's much more likely that the bug is somewhere in your code.
yeah, strtok ate my pointer.
I think that doing 'free(&token);' is incorrect, it should be: free(token)
it gives the same result
but I found what causes it..
but I dont know why
skunkworks: yeah -- isn't 34 pins the size of the floppy connector?
that's what I had in mind too
jepler: yes - 34 pins.
then all I have to come up with is an 8 pin cable :)
* skunkworks goes and looks in his cable box
if you want to use the main and right connectors, I think you'll want to use a 26-pin and a 6-pin -- if you span part of the right connector with a 34-pin there won't be room for the other connector
er, 26- and 12-
you are right
[15:04:17] <cradek> http://sourceforge.net/mailarchive/forum.php?thread_id=6091841&forum_id=33140
[15:04:20] <cradek> http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2004/2004-12-05.txt
AXIS is two years old, happy birthday
I was thinking about that just recently but had no idea which day it was
02:09:40 <alex_joni> jepler: will AXIS go on the emc cvs?
I didn't remember that people started suggesting this so early after the announcement
wow - paul seemed to be nice
yes, he was
btw - great work. I think axis is one of the main reasons I am using emc2
I wonder how easy it is to reuse these connectors - I have a few 26 pin plug but it breaks out to a joystick port and audio. (old motherboard)
hmm - I think pretty easy. Maybe only a vise to smuck it back togather.
other than I need to get a plug out
well that sucks - this plug they glued in. I just jammed a whole row of sharp spikes into my finger ;)
cool - that might work.
goodness knows I have enough ribbon cable.
one option is just to solder the other end into a generic circuit board. (protoboard)
looks like 12 pins is not very common
almost everything about that board was done to make it as small as possible
been that long already?
jepler: I have found a 16 pin connector I have
but - you are right - I don't have any 12 pin ones
lots of 10 and lots of 16
might just have to cobble something- looks kinda like a normal plug will not fit on the right side. well maybe it will the header will raise the plug. got get over the capasitor near the power jack
what the hell did I just write
jepler: the pci8255 is also (3) 34 pin headers - Cool
I know there was a reason I kept all of my floppy drive cables.
ask away. Someone should be able to help
please i need a 2 parallel port in my sistem
i don´'t run if both 2
what version of emc are you using - and what exactly happens when they are both installes?
Guest101: for emc2, determine the address of each port, and use it in the 'loadrt hal_parport' line. For example, if the ports are at the standard LPT1 and LPT2 addresses, you might write: loadrt hal_parport cfg="0x378 0x278"
then HAL will create parport.0 and parport.1
for PCI parallel ports you can often find the proper address in the output of "lspci -v -v"
jepler: AXIS surely has come a long way. congrats
alex_joni: thank you
that goes for cradek too :)
* alex_joni remembers quite a while ago trying it out on emc2
and having to hack parts of it to make it compile :)
before we switched to emc2, you were the one who "ported" it
I know chris switched afterwards to emc2, and continued the "porting"
2 years ago it sure didn't seem like emc2 was ready .. but that has sure changed
I'm glad you feel that way
I was stuck on emc1 because I didn't have realtime on any of my machines
heh.. I just remembered rs274py
that was before we came up with the snappy name
I think I looked at it even when it interpreted the g-code by itself
or were those some screenshots? I can't remember
there might have been a tarball or two from that era
in fact I think there was even an axis 0.9.x before the switch to the emc interpreter
remembering those fun times is nice
* alex_joni remembers the first time beeing helped out by cradek :D
if dont talk
in many minutes
in from brazil
no more good
Guest101: hello from romania ..(opposite corner of the world)
so, you said you have problems with the second paralel port
but very tanhs for you
Guest101: emc or emc2 ?
you have 1 parport now, and need another one?
i need 2
how are configure?
a need outputs
my ports is:
port1 address c800
port2 address b000
you are running stepper configuration?
how write in standard_pinout.hal?
there is a line in the standard_pinout.hal which says: 'loadrt hal_parport cfg="0x378"'
Guest101: you need to make a copy of the configuration to your home directory
hang on a bit
there is a line in the standard_pinout.hal which says: 'loadrt hal_parport cfg="0x378"'
you need to change that to 'loadrt hal_parport cfg="0xc800 0xb000"'
addess of pins
to be able to edit the configuration you need to copy the config from /etc/emc2/sample-configs/stepper/... to /home/user/emc2/configs/stepper
it is described in detail here: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?CustomizingConfigsOnUbuntu
ok, the pins are named like this:
one time for test...
If anyone wants to try out my dxf to g-code converter you can download it from http://dallur.com/fileadmin/user_upload/dxf2nc_beta1.zip
It's far from complete but comments, pointers and feature request are welcome
Rugludallur: I'll look at it
jepler: Great :D it requires vpython though
unfortunately, I don't have that
it's surprisingly small :)
maybe when i'm at home and can install additional software
jepler: you on win or linux ?
jepler: hmm let me check if I can package it somehow
that's OK, I'll just wait and look tonight
Rugludallur: I see that you're using python2.5
Rugludallur: I think it's just python2.4 on my home linux machine (ubuntu 6)
jepler: hmm I upgraded because of vpython, let me check if it is required
Rugludallur: the 'partition' method of string objects doesn't exist before 2.5.
i come back with my question
Guest101: I'll come back with my answer
how to mount rtai_serial.ko properly
jepler: I can probably get around that though
which its email? therefore the test must delay some minutes, thus with its email can enterar in contact. - alex_joni
jepler: vpython depends on 2.5 , just checked so no matter :P
Guest101: you can use alex_joni AT users.sourceforge.net
use @ instead of AT
Rugludallur: I tried some dxf file I have and I get this error:
Unsupported Key Type for Geometric Object:
jepler: ok, ill get right on finding out what object type 70 is :D
jepler: and fixing it :D
Rugludallur: that's the spirit
look for 71-80 while you're at it
Rugludallur: I have no idea if this is a good DXF file -- I got it from someone else, no idea about the version of autocad or anything
File "dxf2nc.py", line 319, in parse_dxf_with_pythoncad
fsock = open(filename + '.tmp', 'w')
IOError: [Errno 13] Permission denied: '/usr/share/qcad/patterns/honeycomb.dxf.tmp'
hm, this is not good -- it should not be required to write a file in the same directory as the input file
jepler: it sounds like it's after acad2000.. pretty common these days
jepler: check, will change that to system tmp directory
somebody knows what is the equicvalent for hyperterminal of windows
Rugludallur: you should use the 'tempfile' module to create temporary file names
probably tempfile.TemporaryFile() or tempfile.NamedTemporaryFile()
Guest338: the easiest way is cat /dev/ttyS0 > file and echo "foo" > /dev/ttyS0
Guest338: but there are others.. minicom is sometimes appreciated
jepler: ok, in this case I wanted to be able to view the file ater running the conversion but for release tempfile is better yes
Rugludallur: I think the dxf spec requires that a program reading a dxf ignores what it doesn't understand
for alex_joni: ok! the test must delay some minutes, thus that we can we will enter in contact. Debtor for the time being.
as in "we'll crap whatever we want into the file and it's your job to politely skip over it"
Guest101: no problem
cradek: probably best to warn & ignore then
that's the best way to make a file format, IMO
"NOTE Accommodating DXF files from future releases of AutoCAD will be easier if you write your DXF processing program in a table-driven way, ignore undefined group codes, and make no assumptions about the order of group codes in an entity. With each new AutoCAD release, new group codes will be added to entities to accommodate additional features."
you are busy, i will come back later, bye, and thanks :)
NOTE Accommodating DXF files from future releases of AutoCAD will be easier if you write your DXF processing program in a table-driven way, ignore undefined group codes, and make no assumptions about the order of group codes in an entity. With each new AutoCAD release, new group codes will be added to entities to accommodate additional features.
SWPadnos: the only problem is when they replace the old entities that used to work - like replacing POLYLINE with LWPOLYLINE
well, that can be a pain in the ass
jepler: thanks for finding that
I like the "principle of least change" - use as old a storage format as possible for the given data
* alex_joni heads home finally
see you Alex
or later alex :D
at least you didn't: DROP TABLE alex_joni;
you can alter alex later
wth is alter anyway?
ALTER TABLE is an SQL command
alter alex_joni add column 'SQL92';
it can add or remove rows from the named table, among other things
jepler: it's also an emc leftover related to comp
Question: What does the error "joint 2 following error" mean - after increasing the "INPUT_SCALE" to a larger number, Or what else should be changed if the larger value is needed for my stepper motor system?
and that was a quote from a source file ;)
alex_joni: it is?
EldonB46: you probably requested a higher step rate than you can get at your BASE_PERIOD
jepler: apparently they imagined alter as an offset which you can set, to compensate for various things like thermal expansion of the screws
Going from 4000 to 16000 - should I increase the BASE_PERIOD by 4?
EldonB46: For instance, if your BASE_PERIOD is 50000 (the default) and you do not have additional setup or hold times defined, then you can get at most 1 step each 2*50000ns, or 10000 steps per second
but it's 1-dimensioned
EldonB46: you would make your BASE_PERIOD a lower number, such as 20000, to increase the maximum number of steps per second.
ideally you would make it 4 times lower, but that probably won't work
Thanks, I love the fast answer that are available here!! I'll do some more testing, THANKS!
come back if you have more questions
EldonB46: if you lower it too much, it will lock your computer
be careful though - if you set BASE_PERIOD too low, the computer will lock up
like alex said ;)
50000/4 = 12500.. kinda extreme
I've ran that for a while though
7500 was my max I think
you should calculate the period you need based on your desired rapid speed
no need to set it lower than necessary
you had it down to ~6 uS - I remember jmk mentioning that it was only 7 ticks of the timer
right.. it gets jitter
It's currently on a very old PC, 360MHz, would upgrading to a +2GHz help?
EldonB46: what is the speed in inches per second?
the best answer is " maybe"
EldonB46: I recommend you do the calculations first, then decide if you need an upgrade
but upgrading to a 7-900 MHz would
from there to +2GHz .. it's pure luck
EldonB46: you really need to see what max velocity you need
THANKS, love the help!!
say you need to move 1 inch per second at 16000 steps per inch. That's 16000 steps per second, so a BASE_PERIOD of 25000 would be small enough
say 100 IPM, at 16000 pulses/inch -> 26.6kHz
jepler: you forget that you need two base_periods for a pulse
alex_joni: did I? 25000*2 ns is 20kHz
ohl.. sorry ;)
but my parel port not in
you can't load the driver if the hardware is not in the computer
you want inputs?
because 2 ports!
ok: loadrt hal_paport cfg="0xb800 in 0xc000"
oops - c800, b000 :)
a need 1 exemple
`halcmd show pin parport` will show all parport pins
a want address of pins
not need imputs
i need outputs
[17:35:51] <SWPadnos> http://www.linuxcnc.org/docs/HAL_Documentation.pdf
diagram on page 54
need a 2 ports paralel works in one time
I don't understand
you want both parallel ports as input?
i have 2 paralel ports in my pc
i need more outputs
more than 2 parallel ports?
no i have 2
how many input pins do you need?
and how many output pons do you need?
Thanks, guys, BASE_PERIOD of 25000 works great for my system! (12500 was NOT).
how configure standart_pinout for this?
how many pins do you need for input, and how many for output?
i not speak more inglish
sorry - I speak no portuguese :)
how configure standart_pinout.hal?
for works 2 port
and addres of pins
change the parport load line to: loadrt hal_parport cfg="0xc800 0xb000"
for input, put "in" after the address:
loadrt hal_parport cfg="0xc000 in 0xb000"
the names of all the pins are on page 54 of http://www.linuxcnc.org/docs/HAL_Documentation.pdf
the first port address is parport.0 (0xc000), the second port address id parport.1 (0xb800)
sorry - I meant this:
the first port address is parport.0 (0xc800), the second port address is parport.1 (0xb000)
in or out?
two parallel ports ...
parallel ports are bir-directional, they have input and output pins
Guest101: the example on page 53 shows how to request the "in" mode, where pins 2 through 9 are inputs to the PC rather than outputs
the loadrt version of that command would be: loadrt hal_parport cfg="278 378 in 2a20 out"
[18:39:26] <jepler> http://blog.modernmechanix.com/2006/12/05/robots-stage-realistic-prize-fight/
Guest101: I would like to but I don't understand the help you need.
I want to work with 2 ports parallel
at the same time
have you tried the modified 'loadrt hal_parport' command shown earlier?
plesase one time
the 2 must make the engines to function
set us up the bomb
0xc800 goes to command engine X
0xb000 goes to command the engine y
Guest101: did you 'addf parport.1.write base_thread' and 'addf parport.1.write base_thread' so that the values created in the emc2 software are actually sent to the parallel port?
yes it is possible
I am trying to help you do it
but I don't understand at what step of this process you've encountered a problem
please one teime
does that mean you've had success?
i believe its yes
eholmgren: main screen turn on!
you are on the way to destruction!
you have no chance to survive make your time
ha ha ha
... for great justice!
can't someone program CIA-8 to translate between languages?
(well, maybe, but I didn't want to tell you that ;) )
eholmgren: you surely got where that was from
base base base base
People very tanks
who was it in here that have _never_ seen that before?
Good bless Ypu
good luck Guest101 :)
would it be fairly trivial to control 110V power relays with a C program though the parport?
it would not be fairly trivial?
the control is easy, but electrically, relays tend to need more power than the port can output, and they also have nasty spikes from the coils
trigger smaller transistors with external power from a wart?
there have to be circuits for 110V power control designed to work with logic
that should be fine, but add a catch diode or you'll probably fry port pins
I use those X10 modules to control a few things, but the cost of each module adds up
well, you could always use an SSR as cradek mentioned. opto-22 compatible ones are $10 or so at digikey
I'd rather just make a bank of computer controlled outlets
I don't think you'll get below $12 or so per outlet
the cost also goes up quite a bit as the current capacity increases
$12 doesn't beat X10 modules in price?
it does, but you also need to add lots of time, and also the outlets themselves ...
or at least not by much
I also don't really need them all over the house either though, just in one location
[19:40:46] <jepler> http://www.electronics123.com/s.nl/it.A/id.446/.f?sc=8&category=35
but this isn't standard outlets
sure - relays are cheap. I was thinking of SSRs for the $12 number
yep, you're right
then again, relays that can switch 15A at 120V (the standard outlet spec) aren't all that cheap either ;)
this says 5A?
oh - and you need DPDT probably
err - DPST
maybe there's a good idea here? http://www.hackaday.com/2006/11/19/screen-saver-power-switching/
looks like maybe the whole bank of outlets is switched on a single signal, I guess that's not what you want
actually, there are 8 outltes, and 8 data bits - coincidence?
yep it's 8 SSRs
he says he got a deal on ebay and had the 8 left over from another project
bummer - I only have 7 AC modules
I think I am going to solder the other ends into a protoboard.
I could then use screw in terminals for testing.
it is a lot of I/O for the size.
yeah it is
there are even more I/O on the FPGA, just not brought out to the headers
I asked you yesterday - going to solder directly to the pins for that I/O ;)
sounds like, though, there isn't much more you can do with it being at 95%
I won't be surprised if I end up devoting the rest to index pulse filtering
still don't understand what makes the index pulse more of an issue. Just too short?
(than the quaduture output)
quadrature is tolerant to noise, index isn't
as far as the fpga reading it (logic wise)?
a glithc on a quadrature input causes a quick count up then back down
ok - that starts to make sense
consider the first image here: http://emergent.unpy.net/projects/01149094674
and a glitch on index gets latched and reported...
consider those edges on "B" as noise
ok that makes sense
in regards to the index pulse - if we actually get the 'accupins' working in the future.. It will use a resolver to quadature converter - that has an index pulse. also
also ... ?
sorry I ment it has an index pulse also
how goes it?
So far so good. To many projects
heh.. I know how that is :(
mdynac: any luck?
well i changed the motenc encoder pulse thingees, and emc booted up, however when i home any axis it does not find an insex pulse, it just moves across the entire axis until it hits the limit....
that's what I was afraid of :(
it goes past the home switch?
cradek: it probably finds the switch then hunts for the index
it doesn't care then if it goes off the switch or not
mdynac: does the display change at all?
I mean does it look like the axis gets reset at some point?
no sir....you mean to green? nope
not to green.. the numbers
do they keep in/decreasing steadily?
or is there a jump in position at some point?
the numbers are moving right along......
ok.. so it's not working then
* alex_joni heads for bed :/
it seemsto not even find the index
cradek: the motenc worked with index homing while the motion controller didn't do the new canon stuff
now that it does it doesn't work anymore
the motenc driver probably needs fixing
homing with index pulse works fine in 2.0.4
who has one?
mdynac: you should ask on the user or devel list
cradek: I think peteg ?
or petev can't recall
the guy that wrote the driver
I'm sorry I can't stay longer
the vital system guy?
mdynac: for now use only limit switch homing if it's acceptable
mdynac: and ask on the mailing list for people with such a board
Pete V. wrote the driver - he's an independent developer
* alex_joni is gone
mdynac: do you have a spare you could loan someone?
maintaining all these drivers is tough because we don't all have the hardware
wanna take a trip to Sugar Grove, IL ???
in june maybe :-)
there is one in the mazak.....
the mazak doesn't home on index either (not sure if it even has index)
if it does we'll probably fix it in june
:P maybe some one could borrow that for a while...
so i must choose a limit to home on correct?
you have home switches too right?
then just home on the home switches
just turn USE_INDEX or whatever it's called off
and thats it?
I think so
'lo cradek, alex
you missed alex though
i hope it is dead nuts accurate without index.....
mdynac: depends on your switch :-)
so are the limits...
mdynac: it's got to be better than no homing until index is fixed
okay, i can work around the problem then.....no big deal
won't that have the effect of moving the home position a little bit?
you can't home to the index of a servo?
so you may have to adjust the soft limits a tad
Lerneaen_Hydra: with some drivers. others aren't updated to work right
oh, I see
Lerneaen_Hydra: each piece of hardware does index differently, and it seems like most of them are ... a pain
as long as i can get the axis to go green, i should be in good shape....
I don't see what the problem can be if EMC gets acces to quadrature and index
Lerneaen_Hydra: when emc reads that directly, it works fine
thats the problem - it seems to not have access to the index ;)
hydra, the pre2.2 cvs head version has index problems....
oh, so it's only if the drive bastardises the ouput that you get issues
Lerneaen_Hydra: the problem is hardware that counts for you - imagine between servo cycles you have 100 counts, and the index was during count 38
Lerneaen_Hydra: the hardware has to get that information all into emc, and they all do it differently
Lerneaen_Hydra: so it's just a driver issue - a hard one to work on without access to the hardware
mdynac, did this work in emc1?
if you guys could whip me up a 2.0.4 version that has M52 in it......
cradek: I can certainly imagine
never tried emc1
ok - just wondering if something got broken in the driver
works perfect in 2.0.4
ok - home with index works in 2.0.x, but not in head
... for motenc
that was the question: ... for mdynac ;)
he had said that already...
sorry - I missed that part
* cradek pokes SWPadnos
however 2.0.x has no M52.....
(works for mdynac, not the question was for mdynac :) )
a catch 22 situation....
*(pEncoder->pIndex) = (status >> (MOTENC_STATUS_INDEX_SHFT + j)) & 1;
*(pEncoder->pIndexLatch) = (status >> (MOTENC_STATUS_INDEX_LATCH_SHFT + j)) & 1;
* Jymmm pokes cradek
wonder if this is one where you have to slow down to one count per servo cycle and watch for the index latch
I thought the motenc was one of the cards that resets the count when an index occurs (if it's in index checking mode)
I could be way wrong about that though
I think you're right
pCard->fpga[i].encoderCount[j] = *(pEncoder->pLatchIndex);
surely this is a bug
alex and I noticed that also
ok, this is not rocket science, but one of us needs a card :-)
I don't see any differences in the driver between 2.0.5 and HEAD, except for the pin naming changes (and removal of the automatic thread creation code)
or a r00ted machine with a card ;)
SWPadnos: but the homing code changed didn't it? I thought that was the point of the canonical encoder interface
the driver itself hasn't changed, except for the pin type ...
(I didn't know any index homing worked in 2.0...)
maybe it should be changed to clear the index RW bit ...
(which used to be output only ...)
that would explain the "runaway"
index gets set, motion controller tells driver to reset on next index, driver doesn't bother because it has no idea what to do ...
I think you know much more about this than I do...
no - it just looks that way ;)
I'll take a look at the changes to other drivers and see if I can come up with something for mdynac to test
I'd have to first, find the canonical encoder document, then second, read it
yeah - that would be helpful for me as well :)
mdynac, which board do you have?
anyone know where to get the register definition for the boards?
check the motec manual on vital system website
ah - found it there - thanks
you had a problem with the DACs being backwards also, right?
i don't think so...
i plugged it in and it worked...
ok. someone was saying that they needed to use a negative scale to get positive output
sounds like someone crossed some wires.....
random question, what type of education have most people here done?
no - some board (motenc included) use inverting buffers - the motenc outputs +10V if you write 0 to the DAC, 0V if you write 0x0FFF, and -10V if you write 0x1FFF
6 years US Navy f/a 18 apg-65 radar tech, component level repair 6 years, rf tech, 30 years electronics technician....all component level repair,
SWP is that a problem?
Lerneaen_Hydra: some computer science, but I quit before I got my batchelor's degree
not a problem if the driver is correctly written ;)
yeah - who needs a degree anyway?
ah yes college, the best 10 years of my life!!!
Lerneaen_Hydra: because I found a good job instead
heh - 4 years of school, but I decided to not bother working on classes because I had 2 part-time jobs, was starting a business, and was in the national guard (adding in the 18 credits of engineering was a little bit too much)
I have a BS in computer engineering (CS + EE)
a "master of none" degree
the reason people don't respect school: First you get a BS, and we all know what that means. Then you go on to a MS - More of the Same. After years of study, you can get a PhD - Piled higher and Deeper :)
hows that for motivation ;)
and still no real world experience....
I wouldn't say that
my mother is a perfect example of "no real world experience" ;)
then again, if you don't do stuff outside of the required stuff you won't have any idea of real world experience, at my EE course you can get an MS without having seen a soldering iron
a PhD and several masters (including CS), but still needs help to find the power switch ...
I however like to make lots of real world stuff
get messy in the shop and stuff
lots of EE has nothing to do with solder
text book circuits work in only one place....the book....
we learned plenty about silicon doping, signal analysis, etc.
yeah, but all I've done so far is simple analog stuff (don't forget I'm still damp behind the ears)
jw and all
i wanted an AM/FM/SW radio when i was 13, my dad bought me a kit, and i was hooked.....
my son wanted a pc so i gave him a PII and three slackware disks......that was his start....
and gee, he figured it out....
kids are too lazy nowadays....
that's what they all say ... ;)
our pc "expert" at work knows how to point and click in xp........
he is a moron.....
some computer science, and then an art "degree" :p
4 year BA
which one makes you more money?
bachelor of arts
as opposed to bachelor of science
still thinking about going back and getting an EE
or an ECE
my dopleganger must be idle on the mac at home
* eholmgren kicks ejholmgren
i tried a mac.....once.
OS 9 made me vomit uncontrollably whenever I used it
i know, thats why i walked away...
i hated when the mac would tell me in so many words "it's either my way or the highway"
did OS 9 even have a command line?
i didn't hang around long enough to check....
a great machine for a non computer user.....a kiosk thing....
do when you handed your son the PII
i said good luck.....
did you also give him the full set of slackware 3.5 disks
except the X
i was going to give him Linux from scratch, but i am not that brutal....
he also got linuz in a nutshell and running linux.....
he figured it out and he actually likes it.....he's one up on the mass of doze people....
okay emc guys, i will go ahead and home without index so i can test the M52 function.....
one more question....what CAD/CAM software posts out to emc?
I use edgecam/solidwords
mdynac: are you trying M52 for EDM?
tomp well not quite yet, had some homing issues with 2.2 head.....but i will be testing it soon...
gotta go fix a japax tomorrow
can I watch? ( not the japax :-)
wanna stop by the shop?
and see the machine?
when you try M52
and cutting power
gee, i wasn't expecting an audience....
here's my stuff down in elburn http://tag-edm.com
tomp are you well versed in emc2?
no ( is anyone?)
esp w M52
dont worry, no problem
hey do you live in Elbow?
Elgin, m/c in Elburn ( till delivered)
all jst up & down the river
well thursday i hope to test the andrew....
i fixed a novatech hole popper the other day.....
nova who? well hopops are pretty simple... just all different, I think EMC (hal) can control 'em
:=) so add a hopop to that Andrews line tracer :-)
i also fixed an andrew jig grinder control.....now that is old....
that is all my andrew can do right now is draw lines.....its an EF 330
till i get M52 going
oh, 'new' stuff, not the one that actuially traced a drawing on a drafting table
that is arelic......
we do have a linemaster 123 as a showpiece in the showroom....
with the ceramic wire tubes and all....
well, i have thoughts on M52 and would like to see what happens sometime.
they had another incarnation of M52 working at the fest......varying the axis speed with a joystick....
jiggle the stick and it'l jiggle the speed, not the position :-(
and you know it gotta jiggle fast
true, but it will vary the speed in proportion to gap voltage
i just gotta set my adc up properly....
you gotta vary the sign of the velocity ( fwds bwds) very very quickly
look at any edm cut with a scope
i am using all of the original Andrew gap sensing circuit....
just applying the output to the motenc adc
is it bipolar output? that indicates how well 'dynamic slow down' will work.
[Global Notice] Hi all. You're currently on a freenode server that's going to be restarted soon to upgrade the ircd code. This will happen at midnight GMT (a little over an hour from now). If you wish to reconnect to another server, feel free to reconnect to chat.freenode.net--none of the servers in that rotation will be affected. If you do not reconnect now, you can expect to be shunted from the server, but most clients will r
bipolar means it wanted to go back and go forth, not just slow down
ultimately it is a 0 to 5vdc signal from an opamp
i spent a couple years working on that route, so i hope you have better luck.
i am sure i can get the machine to behave like an old japax......
to start with,we haven't even considered "backup" yet.....
great stuff, maybe you can demo something for the thing out at cardinal eng.
[Global Notice] ... most clients will reconnect cleanly and the server should come back online immeadiately. Thanks for your cooperation and use of freenode!
not backup as in clear the gap, but jiggle as in... any edm
well, er gotta drag the andy down there....
anyone got a flatbed handy?
take movies :-)
no rigger needed
when an andrew backs up it goes exactly 50 thou back and forth until the signal goes high.....
thats not what i speak of
then she moves on forward until backup goes low again...
kinda like a subroutine of the O/S
while maintaining the correct gap it must go in 1um and soemtimes out 1um then in again, else it's ignoring the process.
the 50 thou is not what i mean
that is very coarse long term edm
not immediate dynamic control of position based on a live process
the pulley on a edm jitters while slowly moving forward
the pulley on a feed system just moves forward
one respects the process, the other crawls up behiond it or ignores it
slowing down to below the cut is not edm, it is feed
sorry for the rant...
i must figure out the absolute lowest gap voltage without wire break, then set up my adc to stop the movement at that point, and slow down as it approaches that threshold....
surfing on the edge is scary... but as the gap changes... the voltage changes, and you must change, and sometimes thats a teeny step bwds
cant do that with M52, the nice decided velocity will break the wire
correct, but we are taking the "baby step" approach, get this to burn without wire breakage, no matter how slow, then incorporate the backup stuff...
ok, the backup stuff i speak of is the jiggle, not the 'run away' you mention, it is the boxing of bantamweights
anyways, best of luck, you have some nice boneyard to play in :-)
it's the best......
i will keep you informed....
the Chmers will cut pcd and cbn now, so my customers that need this conversion are getting less and less....
do me a favor, look at the website, watch the pulley's, lemme know whatcha think
gee, its a sinker....what kind?
you built it from scratch?
rebuilt Eloc HRP and used generator from KC Tsai
will your generator work on a D10?
notice the hi speed hiccup on the pulley? thats dynamic... not to brag but it is easier to see edm on a good system
yeh i can put it on a D10 ( sorta like the JATO rocket car tho :-)
ya, hydraulic servo
(a d10 is a benchtop from Charmilles, 70's to 80;s vintage)
i use 3kw ac servo motor
a bit big for the D10
yep i have 2 with I80 generators(i hate those things....) way too complicated.....
and a D120 with a p25, much nicer....
yeh I know them... aguy up in... (brain freeze) north of us ... woodstock precision... bad one
they are a nightmare
80 amps is overkill for the spindle size... to much heat & too small a cross section
the D10 however is a nice tool.....
D10 is sweet, cut my teeth on them at ARC
the oil gets real hot....
not enough volume on a D10
hah, an infrared gun said mine hit 240 F !
you can heat your luch with it....
i had no di at all just a couple truck filter inside the tank
french fries :-)
mac. I'mm off to talk to someone about the same... see you later, best o luck