thats not a bad idea... I think I may have some scrap laying around for that
this is my latest progress: http://images.myonlinesite.com/cnc/construct/20070527/
Nice. looks like it moves Smooooth
yeah... its not bad :)
Its about 1/32" high on the left side, but I can shim it to fix that
good enough for wood work ;p
be back in a bit
03jepler 07TRUNK * 10emc2/src/hal/classicladder/ (arrays.c manager.c): rtapi.h should be included before any rtapi_xxx header
What are the "trunk" messages?
code changes committed to CVS generate messages from CIA
TRUNK refers to the branch of development that the commit was for
Trunk is like cutting edge developement
no pun intended :)
that would be a great name for a CNC newsletter
"The Cutting Edge"
that is so bad it's awsome.
* Ziegler shakes head
what bin do I start emc2 with in sim mode...
(ubuntu 7 install from source)
03jepler 07TRUNK * 10emc2/src/Makefile:
* make sure PYTHONPATH is set to include lib/python when building, so that the sys.path hack isn't necessary in comp or jmk's new script when running from make
* install fixes for sim systems (bfload, bfmerge do not exist -- well bfmerge doesn't exist anymore anyway)
Good evening! Is anyone here that can help me sort out an M5I20 servo homing problem?
you're the guy with the proxes, and the "home on limit, but limit disables the amps" issue, right?
Yes. One prox now feeds both limits and the other prox is within the permitted range of motion.
ok, what is the problem?
1st, 1 Q: Is it permissible to home at either end of an axis (by negating velocities)?
the sign of search vel tells it which way to move (and thus which end it expects to find the switch at)
OK. Search vel=-0.5, latch vel=0.01. Axis travels to limit, reverses at slower speed, but doesn't see the index pulse (which I can see getting to the input on an O-scope).
I said limit but I meant home prox.
so it sits all day inching along at 0.01 inches/sec?
are you talking to me from your EMC PC?
Sadly, no. I'm doing a relay race back and forth across the shop. :)
no network on the machine?
(I guess that explains the long delays between replies too)
I am embarrassingly linux illiterate. Would it help if I got on an IRC client on the machine itself?
yes, very much so
gaim works ok
so that machine IS on the network?
I was afraid you were going to say that. Yes it is on my net.
is it an ubuntu machine (EMC live CD, or ubuntu install)
if so, gaim is already installed
Recent Ubuntu install.
ok, he didn't have to leave....
or did he.
just fire up gaim on the ubuntu machine
theres room for two of him in here
* Jymmm lol @ jmkasunich
How far do you think an buffered signal could go cbale length wise?
100' total loop length?
how long is a piece of wood?
what kind of buffer, what kind of cable, whats on the receiving end?
you ask very open ended questions
jmkasunich: I want to run the chargepump signal thru the ESTOP buttons and back into the control box. The idea being that even if a cbale opened/shorted, it would still work properly.
so you're gonna be running a single ended signal, no nice twisted pair or anything
I got shitloads of CAT5 I could use.
on the other hand, you also don't give a crap about the signal - its not like you're sending serial data down a link and need to worry about line noise - its either toggleing or it isn't
Well, I want to make sure that the noise does falsely trigger the charge pump circuit either
you do NOT want to send the outgoing signal in the same pair as the return signal, or you might get enough crosstalk to do exactly that
I'm considering using this http://www.cnc4pc.com/Store/osc/product_info.php?cPath=23&products_id=51
jmkasunich: It be a single wire and the return path.
Nope. Back on other PC.
btw, there should be no need to shut down one IRC client before starting another one
there's room for two of you in the channel
I can't get gaim to sign on. I do have net connectivity.
let me try
Do I need to configure a different screen name?
you need to set up an irc account with gaim
in the accounts section
yes, try EEEE_Chris =)
(I've never used gaim, using xchat right now - so whatever I have to do will be the same as you have to do)
I had a hard time figuring out gaim too
EE_Chris: There can't be duplicate nicks on IRC
i found bitchx... it's pretty slick
Oh - Me stoopid mebbe. I think I used the wrong name for freenode.
you did set the protocol to IRC, right? It defaults to AIM or some other teenager chat thing
hey, its me!
when it connected, gaim printed this: (23:57:52) freenode-connect: Received CTCP 'VERSION' (to jmk2) from freenode-connect
then I had to do "/join #emc" and here I am
[04:00:12] <chr0n1c> http://nofingclue.com
<- my new domain name.. i'm setting up drupal on it right now (and i need to turn on domain masking)
i also bought nofingway.com and nofingclue.org
about 6:oo pm today
i was bored!
oh gawd, gaim prints graphical smileys
registered.. not bought i s'pose
hey, its chris
jmk turn em off.
* jmkasunich turns gaim off
* chr0n1c loves smileys
Head temporarily extracted from posterior.
jmkasunich: Did you see my question about ESTOP (f2) from about 4-5 hours ago?
I vaguely recall something
but I don't know the answer
jmkasunich: So, is ESTOP both input and output from/to emc?
there are two separate pins
one is telling the world the EMC needs to stop the machine
hal or paraport pins?
and one is the world telling EMC that something stopped the machine
oh.. i was wrong.. again
its up to you whether they go to the parport, or a mesa card, or nowhere at all
jmkasunich: Ok, just wanted to make sure I understood the concept correctly.
EE_Chri1: ok, where are we
I never realize that EMC can "power up" the controller box.
I recall reading something about an aborted homing cycle hosing up the M5I20 fpga. Is that a potential problem while I'm debugging this?
I don't know what you read or where, but I doubt it
what kind of machine is this? sherline sized, mill-drill sized, bport, or machining center?
So, I come back off the home prox and then the index pulses are never seen...
Jymmm, what would you host at the domain, nofingway.com
Hurco KM3 (factory CNC knee mill)
ok, so pretty beefy, you can't easily spin screws by hand to check encoders, and you need to be carefull
(just getting myself up to speed)
Scary machine, yes.
you are checking the presence of index pulses with a real oscope, not halscope, right?
checking them where? at the 5i20 breakout board?
chr0n1c: chinese driving directions
I'm crawling slowly enough that they could not conceivable be missed, even by this PC. :)
the 5i20 is supposed to spot them in hardware, so the pc doesn't even matter
Have I possibly missed hooking something up in the ini or hal files?
but if they're not making it to the FPGA, or the FPGA hasn't been told to look for them....
thats why I'm asking questions
where are you scopeing the index (physically)? at the 5i20 breakout board?
That 2nd part is interesting...there is an enable generated by EMC during the homing cycle?
where are you scopeing the index (physically)? at the 5i20 breakout board?
Yes, pin 8 IDX0
Jymmm: driving directions in chinese or directions to drive to china
Jymmm: please clarify
* chr0n1c rofl
low idle, high pulse
EE_Chri1: I'm not sure what polarity the FPGA expects, but at worse you'd be off by one count because of triggering on the wrong edge, so lets ignore that for now
have you ever used halscope?
can you start it up now (I assume you have EMC running?)
oh... what config did you start with?
Its running. Trigger on m5i20.0.enc-00-index?
did you uncomment the index related lines near the end of m5i20_motion.hal?
The ones that said they were commented for pending driver changes?
I bet your indexes aren't hooked up
I bet you're right. Try it now?
however, I might be turning red soon, because I'm not sure the existing 5i20 driver properly handles indexing
if scope is still running its easy to check
try to put a "probe" on the signal "Xindex"
if there isn't such a signal, that would be a problem ;-)
FWIW, I did try using Show HAL Status to look at all things index-0 related and didn't see any activity.
the index-enable signal is bidirectional
its supposed to run from the axis.0.index-enable pin to the driver's index-enable pin (not sure what its called on the current 5i20 driver, but hal show pin will tell you)
when EMC wants the driver to detect an index (during homing) it will set the signal high
when the driver finds the index, it will set the signal low
one wire handshake kind of thing
if the signal isn't connected, then emc will tell nobody in particular to detect an index, and nobody ever will...
Found the section in m5i20_motion.hal - Uncomment the whole index section?
yes, I think
you really need to know the proper name for the driver pin, it might not be right in that section
if emc is still running, open a shell, and do "halcmd show pin m5i20.0.enc"
that will show you all the encoder related HAL pins
one of them is probably named index-enable or something similar
it might just be named index, but I hope thats been changed, since its confusing
03 bit OUT TRUE m5i20.0.enc-00-idx-latch
03 bit OUT TRUE m5i20.0.enc-00-index
03 bit I/O FALSE m5i20.0.enc-00-latch-index
03 float OUT -0 m5i20.0.enc-00-pos-latch
latch-index is the one you want
(its the bidirectional one)
you do have a hardware estop, right?
when you try the home sequence, be next to it
# Connect index pulses to motion controller.
# This has been commented out pending changes to the
# driver to work with the canonical encoder interface
newsig Xindex bit
newsig Yindex bit
newsig Zindex bit
linksp Xindex <= m5i20.0.enc-00-index
linksp Xindex => axis.0.index-enable
linksp Yindex <= m5i20.0.enc-01-index
linksp Yindex => axis.1.index-enable
linksp Zindex <= m5i20.0.enc-02-index
linksp Zindex => axis.2.index-enable
That's the existing (uncommented)
change m5i20.0.enc-00-index to m5i20.0.enc-00-latch-index
(and likewise for 01 and 02)
Is there a way to re-init EMC or do you just close and re-launch to re-read the config files?
close and re-launch
Just curious has anyone HERE used the chargepump ?
we used A chargepump on the Mazak at the workshop
the steve stallings board
jmkasunich: which one is that =)
PMDX-122 I think, but not sure
it was about 2 years ago
hard to keep names and products associated any more, lots of snake opil out there now
steve doesn't sell snake oil
if he did, we'd hit him with something
jmkasunich: Well, what do you think of my idea of routing the signal thru all the eston buttons?
it could work
I'm trying to prevent SPOF primarily
My only thought is if the chargepump ciruit failed.
to be honest, I think you are overthinking this
your machine isn't big enough to chase you across the shop, or destroy itself
jmkasunich: Oh I know I am. Just got this fire paranoia thing.
Heh, you should see it start fires though =)
how does the power get to the spindle? extension cord, right?
It's the ultimate rubbing of two sticks together.
* jmkasunich hands Jymmm a hatchet to keep next to the cord
will be via SSR when I get done.
hatchet is more reliable
jmkasunich: Yeah right.... tool cutting into live power lines.... been there, done that.
hatchet with ametal handle?
I'm joking of course, but my point is, you can easily just unplug the thing
no, dicks with rubber insulated handles =)
I think you mean dikes
keep a loop of it tied to your leg.. then when you dive away it will unlpug
jmkasunich: This is so I can "walk away" from it.
you mean let it run unsupervised?
jmkasunich: Eh, dikes be chics with dicks.
in that case, your problem isn't SPOF in the estop chain
Hey, I just homed!
its the fact that red buttons can't smell smoke and press themselves
could you monitor the current to the spindle motor and cut estop in hal if the amps get to high / low of a load?
I had an exciting moment due to my home offset having the wrong sign on the first try, but I had the stops well inboard for just such an occasion.
jmkasunich: No, I want to make sure that if needed to push the red button, it'll do it's job.
jmkasunich: I'm installing one outside the room.
if you are around to do the pushing, there is nothing more reliable than pulling out the plug
the "s" key works fine for me
even through vnc
if you push the estop and it doesn't work, (which is EXTREMELY unlikely even with a conventional estop circuit unless you totally fsck it up), you can pull the plug
its not like you have a 50A three phase feed
jmkasunich: If I can GET to the plug you're assuming.
sometimes I wonder why I even try to answer your questions
you know what your situation is
you know what you want
so do it
jmkasunich: I nevre said I wasn't. I just wanted to know if anyone has run the chargepump and how it worked for them. Did the long length cause any issues, noise, etc.
I've never heard of running a chargepump signal thru the estop buttons
Jymmm: can you send me half the money you are throwing at this machine and i will pray for it
I would be somewhat worried about crosstalk between the outgoing signal and the return signal, such that even with a button pressed, you might get enough signal to keep the pump happy
It just made sense, dind't know if it was doable though. or if I had misunderstood something.
a conventional relay circuit won't do that - it takes non-trivial mA of current to keep the relay pulled in
jmkasunich: Yeah, it's things like that why I asked. I know a relay works, but it the cable every shorted, it would fail .
not if you route it right
power to first button, to next button, to next button, to last button, to relay, in a loop
don't put the outgoing power in the same cable as the return wire
You're talking about a NC circuit, right?
so they can't get shorted together
and if anything gets shorted to ground, that will kill the relay too
yes, NC driving a relay that energises to allow power to the machine
and you have never had a switch wfail with loose contacts and short out?
not a decent switch
you get your red mushroom estop switches from harbor freight or something?
so you have. Well that's why I was thinking of running the CP signal thru th eestop buttons, it wouldn't matter if it was a $100 or $2 switch.
besides, even with crap switches, the vast majority of failures are fail open
Hmmm. Jogging works fine. Then after I home, I get a following error traversing to the home position (because nothing is tuned and my speed is set too high), After that when I re-start, the position doesn't count and a discrete jog doesn't stop.
if the switch fails shorted, it won't matter what signal you are running thru it
Is it still stuck in a homing mode because of the FE?
if it fails shorted, the CP ciruict would trip the ESTOP
Jymmm: shorted to what?
if the two terminals of the switch are shorted to EACH OTHER, it doesn't make a damn bit of difference what signal you are running thru it
shorted or opened, the CP cir would trip estop
EE_Chri1: It shouldn't be still in homing mode
what exactly happens - it finds the index, position display jumps to the proper value, then it begins the rapid to the home position... then you get an ferror, right?
I changed some speed settings (not knowing what I was doing) and I may have broken something. I set them all to smaller values though.
jmkasunich: I thought I was on to something there.
and after you reset the ferror, you try to jog and what? it moves off at jog speed even after you let go of the button? or it runs away at high speed? or?
Not sure about 'proper value' but it does begin to rapid, then FE.
jmkasunich: a watchdog of sorts.
Moves at jog speed but pos count doesn't change and a discrete jog doesn't end.
Jymmm: there is a small set of possible failure modes that your scheme could protect you against, that a regular circuit might not detect. BUT!! there are other ways in which your scheme could be riskier than the conventional approach
jmkasunich: Ok, o how do you prevent a cable short pre-switch?
and if you have to ask what those cases are, you don't know enough to design the safety circuit
EE_Chri1: something is truly confusing - a servo system can't move in a controlled fashion without feedback
and if there is feedback, your numbers ought to be changing
Well, the pos display isn't updating.
start a halmeter, and look at the actual HAL position feedback signal, and/or the position command signal
when you say "discrete jog doesn't end", do you mean incremental jog? ("jog 0.01 inches each time I press the button")
Yes, incremental was the word I couldn't retrieve. :)
or do you mean regular jog? "move when I press the button and stop when I let go"
so do regular jogs work?
jymmm: any safety circuit relies on having some "source", that goes thru all the buttons and gets to some "load"
if the source ever stops getting to the load, then the load turns off
it doesn't matter if the source is 120VAC, 24VDC, or as 12KHz charge pump signal - what matters is the source, the load, and the path between
Whoa! I set for slow continuous, and it just took off (rapidly) and gave an FE. Maybe I touched a speed ini/hal setting in a bad way.
jmkasunich: Right, but have you ever had a tool (or whatever) that had the power cord twisted and short out?
so you are talking about shorting two conductors in a single cable
jmkasunich: You have never had a cbale pari rub againest a sharp edge and short out ?
doesn't matter at all what signal is on those conductors - if you run the source lead going to a button and the return lead coming from a button in the same bundle, and they short to each other, that button doesn't work anymore
because its bypassed
if you short one of the wires to ground, that is different
jmkasunich: Right, I understand that now.
How about a smart TDR module looking at a NC estop chain? ;)
and the result depends on the signal - if the signal is +24V, and the negative side is grounded, you just shorted out your 24V supply, and the relay will drop out
if that signal is chargepump, and you short it to ground, the charge pump will drop out - same difference either way
jmkasunich: I have seen some real shitty wiring jobs that I've had to cleanup. I'd just like to add that lil extra if reasonable, to "just in case"
EE_Chri1: I would never even try homing or jogging until I had the motors tuned and all velocity and accel numbers set correctly
correctly means you personally calculated the answers and know that they are appropriate for your design
Jymmm: one of the risks with the charge pump approach.... if you are using a relay, it needs a fair amount of current to pull in - you know that current can't flow thru the air
but a charge pump circuit has fairly high input impedance - stray capacitance could possibly couple the "source" signal (a high frequency square wave - those things just love stray capacitance) into the "load" circuit, without going thru the switch
that would be bad
jmkasunich: What I was thinking is pc to control box to multiple mushrooms to the chargepump circuit.
Jymmm: why not run that exact same path with 24Vdc?
now if you KNOW that the charge pump circuit has a relatively low input impedance and isn't vulnerable to stray coupling, it would be fine
but you don't know that
jmkasunich: I could, I just thought that if the cable ever rubbed againest a sharp edge, it would be "detected"
and I sure as hell don't know that
like I've said before - there are DIFFERNET failure modes
some can be detected by both circuits
some can only be detected by one
you need to understand all of them
"rubbing against a sharp edge" doesnt' cause the problem
jmkasunich: Ok, so I would need a smarts at each mushroom to detect open/short/ failure. etc.
something getting shorted does
I've bene thinkinf og the cp signal as a "heartbeat" of sorts.
I can't seem to get thru to you that there is nothing magic about a cp signal
jmkasunich: I already said I understood
jmkasunich: (2007-05-30 22:01:47) Jymmm: jmkasunich: Right, I understand that now.
the only thing it tells you that a steady 24V won't tell you, is that the thing GENERATING it hasn't gone out to lunch, it doesn't tell you squat about the switches is passes thru
jmkasunich: Yes and I said (2007-05-30 22:08:30) Jymmm: jmkasunich: Ok, so I would need some smarts at each mushroom to detect open/short/ failure. etc.
any open in the loop is going to drop out the relay, you don't need any "smarts" for that
and if the cbale got shorted out befor ethe mushroom, it will never work as intended either.
a switch that fails shorted (input to output short) is impossible to detect except by pushing it and seeing if the machine stops
I've got to finish some wiring for the other axes and then begin tuning (steep learning curve expected). For the moment I'm going to assume the servo amps are reasonably tuned since I didn't mess with them other than dialing in the null settings. Any advice beyond reading the references linked in the EMC docs/wiki?
if the cable gets a short between the outgoing "source" and the returning signal, then it won't work
if the cable gets shorted to something else, it will work
If you only knew how many time I see big corps have all this data disaster recovery setup for backups, offsite storae to neer check them to see if they can restore and then loose millions. for something so stupid.
you prevent shorts between the outgoing source and the returning signal by not running them in the same cable
can still happen and you wouldn't know .\
test the estop before each day's run
this is pointless
I'll never let you design my estop circuits, and you can do whatever you think is right for yours
jmkasunich: No it's not. you are trying to say KISS, and I'm saying thhat safety is not that simple as you think it is.
Jymmm: I design 1000HP motor drives for a living. I know a little about safety and estops for machines that can kill people
if cab;e A shorts againest metal frame, then cable B shorts againest he safe metal frame a week later there goes your ESTOP
not if the frame is grounded
and grounds have come loose before.
if cable A shorts against a grounded frame, it shorts out the poiwer supply and the machine stops
if you lose your frame ground, you got worse issues
you are describing a three point failure
bad ground, and two independent shorts to the frame
A loose ground would be a SPOF, where is the other two?
in what you were describing that is.
cable A to frame, cable B to frame
also known as "everything gone wrong that could go wrong"
aren't signals from PC opto isolated typically?
24VDC estop circuits are not isolated
(not isolated from ground)
that way a short to ground will kill the relay
if you isolate it, then you can have the 2 point failure that you described, A and B both to frame, path completed thru frame, buttons don't work
but if you ground the negative side of the 24V, and use the positive for all the wiring, then any short to frame will kill the estop
ok, that makes snes.
It's difficult for me to think of DC ground and AC ground and chassis grounds being all "connected"
thats cause you have a machine thats at least partly wood
not the machine, just the enclosure
all industrial stuff is made of metal, the control electronics and electricals are in metal boxes, there are metal conduits connecting everything, and its grounded to the building power distribution system thru at least 2 redundant paths - the green ground wire, and the conduit or mounting hardware
jmkasunich: I've always has issues understanding the difference between chassis earth signal floating gorunds.
not just electrical but electronics as well.
electrical just uses them more
I hate the word ground sometimes, because it can be vague
especially for signal stuff, one ground isn't always electrically the same as another
jmkasunich: Take a switching PS as examle.... I can't connect two of them toghether becasue of the grounds. I get it, but I don't if you know what I mean.
thats another reason why safety circuits don't use tiny high frequency signals, they use nice strong 24V stuff
I don't know what you mean with the power supplies
connect two 12V switching Ps to get 24V as example.
or to double the amperage
you mean two PC supplies
PC or popen-framw switching PS
Just think of everything in terms of return currents, and then it either all makes sense or your head explodes thinking about it. ;)
switching power supply is a very general term - there are plenty of switchers out there where you CAN connect outputs in series
because the outputs are floating
but PC power supplies do not have floating outputs, the common is tied to chassis
I was just using that as example of "grounds"
I need to go to sleed
Me too (snore). Any comments on tuning before I go?
its hard to describe
Start with links in wiki?
OK. Thanks much for all the help.
hmmm if isolation is the only reason why you can't series supplies....
why not put isolation transformers on the input side of the supplies?
its not the input that is the problem
the negative side of each supply is tied to its case
so mount each one in a plastic case
that would work
but there are other penalties - not grounding the cases leads to more RF noise being emitted, etc
always trade offs
Good night, all.
03jmkasunich 07TRUNK * 10emc2/src/hal/drivers/mesa_5i2x/firmware/Makefile: update FPGA build process to use new tools
03jmkasunich 07TRUNK * 10emc2/src/hal/drivers/mesa_5i2x/firmware/ (stepgen.mspec toplevel.vhd): template and module spec files for FPGA build process
ds2: People want to use PC Ps's becasue xfmr's can be expensive, and PC PS's are oh so abundant.
ds2: so you use insulated standoffs inside a metal case to keep the EMI/RFI minimal.
you can take the case apart and remove the case ground wire/traces
and do that...
i know you can do paralell that way
or CAN you.
some people can...
do you have the tools and balls to try it?
I have two extinguishers in the garage, so yes.
All rated ABC =)
C as in ELECTRICAL fire =)
and the huge ass blower to exhaust the stinking fumes for the next 4 days afterwards too =)
* chr0n1c bites the head off of a live bat... (i thought it was a rubber bat)
Jymmm: except transformers on the input are relatively low current
but, yes PC powersupplies can be had for free :/
provided you steal computers
I just need to say I am looking for old computers and they just appear
i'm sure you're violating some DMCA nonsense by doing that
prepare to be sued.
Ha, like you have to steal an old computer.... just place a free ad for recycling old computers, and people will give them to you.
<- has a wall covered with old towers ripe for plucking parts from in the basement
Jymmm: need more? =)
i emailed some machinery sales place to get a price
Hey, that's an idea... I need to find an old AT case
ds2: PS or cases?
you want an old AT case?
how many do you want?
on a 20"x120" american pacemaker
ds2: I might be able to use a mid tower case, depending on it's config/layout
which was loaded - coolant, lights, a ton of chucks, toolposts, a horde of toolholders, taper attachment, 15-2000 rpm
AT cases I just remembered have the whole shell attached.
Jymmm: I have an old AT (desktop style) case + PS I would like to dispose of
plus it had a copy tracer
but the dude wanted $199,000.00 for it
ds2: 6" tall?
a manual lathe
Jymmm: around there... it has 3 5.25" bays
one hundred nintey-nine thousand dollars
i have a 6" tall pc case.. it was an external hotswap HD case
the shell slides out
ds2: Yeah, that was the icky part of those iirc
Jymmm: i'd let it go cheap ;)...
chr0n1c: free and free shipping
yes, but it is solid metal unlike the frys $15 specials
ples a bottm hidden bay for the ps that was in it
Jymmm: do you really need/want a mid tower?
Jymmm: ehh not that cheap
ds2: screw frys, I'm talking asking for them from weird stuff or the place next door to them
maybye like 5 bux+ shipping
ds2: I'm not sure yet, I have a sweet case I'm using for the new control box, but I'm not sure if eerything will fit in it yet.
chr0n1c: Deal, you pay me $5 to take it off your hands.
it needs a plate made for where the two HD bays were
Jymmm: let me know... got stuff to get rid off
ds2: where you near?
Jymmm: i'll take that as a "no thanks"
chr0n1c: I'll never pay for PC cases
ds2: behind old IBM?
Jymmm: that general area
Gawd, I always forget the simple shit... like how many I/O's are on paraport. *sigh* My mind is a terrible thign to waste.
much rather give stuff away to people who will use it then deal with those so call "recyclers" who turn around and re-sell stuff
ds2: I understnad that one. Got any relays or SSr hidden over there too?
Jymmm: relays, yes. no SSR, AFAIK... the relays I think are in octal base form
ds2: got the bases too????
don't think so
they are standard tube bases so....
excess sol'n might have them in their piles
ds2: but have you ever tried mounting a tube base =)
not yet.. I am looking for suitable tubes (got some?) to try to build a tube receiver just for the experience
ds2: I have no tubes of any kind
CRT being the closest
shitloads of PC HW
one day I'll find those
modern PC HW?
ATX or newer era
lol, yes =)
so you do have modern stuff... I have too much AT stuff
ds2: are you needing anything specific?
a PS with P4 tail connector
normally, I'd goto surplus computers but since they closed their store... :(
ds2: compusa's dumpster
you mean the 4c molex connector from the PS ?
you mean the square 4c molex connector from the PS ?
chr0n1c: all compusas within 40miles are closed
I might have one grab a pic of google and let me see what you are thinking of
[06:09:23] <ds2> http://support.intel.com/support/processors/pentium4/sb/cs-007994.htm
the +12V Power connector
ds2: Yeah I think I have one of those, not sure if it's in use or not.
ds2: what wattage?
250-300 should be fine
iirc P4's need 350+
*shrug* I just have a board and need a PS for it
you might want to check the requirements for the mobo
easier to start with what I can find
oh if any machine builders are still awake, what do you guys use to couple to a lead screw?
we had a mobo and a 300W PS, would power off
just the mobo?
yep, needed to be at least 350W
no the whole thing
if that is the case, I shouldn't come across many/any sub 350W P4 connector power supplies
it depends on the mobo specs mostly, just fyi =)
Hi Folks, anyone know of a bug in pyvcp whereby the LED indicator fails to honour the on_color and off_color tags?
hcseb: not yet
alex_joni: Not yet meaning that the feature isn't implemented yet?
not yet, meaning I didn't know of the bug
Oh, ok. Well, in this 2.1.5 version, that appears to be the case.
I'll have a quick look at the code..
I'm not sure the tags exist though..
hmm.. the code says it should work
Yes, I agree - I was lokoing in pyvcp_widgets.py
And that code appears to be there.
Assuming that TRUE in the hal evaluates to 1 in the python code.
Which it must, because, apart from the colour, the leds operate as expected for me
what colour did you try?
The Tk interface doesn't expect an rgb value for the colourd oes it? I tried"green", "red" and b"blue"
no, I think literals are expected
red / green etc
Ah. I can place either <on_color>"blue"</on_color> or <on_color>blue</on_color> and the interface will run, but neither gives the expected colours.
On another subject, if you were to want to run a cnc drililng machine which has X, Y motion and then will drill a hole or make a cut across the whole piece, how would you go about it? I am assuming I need to modify the tk based user interface and emcmot..
and I somehow need to link the relevant G commands to "start drilling" and "start torch" and "start cutting" etc. (it's a flame cutter for steel plate)
(off for lunch now)
you don't need to modify the interface
you only need to set up your configuration for only 2 axes (X,Y)
and then you'd probably use some custom M-codes for the drilling
I already edited the configuration to remove the Z axis. So, just the custom M codes for the drilling, cutting, etc.
What machines do you run with emc alex?
hcseb: I mostly do sw development for fun
I have a small mill (with steppers)
I'm having an interesting time with some largish machines in a structural steel fabricator. I'm doing this as contract work for the owner, who is a Linux enthusiast
linux enthusiasts are always welcome
for work I usually work with welding robots
not emc controlled though
We have a 9 axis driller/saw cutter which is the next machine on the list.
I understand I will have to hack emc a bit to get all the axes in... But they are simple axes, it's not complex motion like a robot
then you can probably get it easily done
the motion controller as it is now supports up to 8 joints
but that's just an arbitrary limit.. changing it should only mean changing a constant and recompiling
(and probably fix a couple of typos/thinkos) :D
Yes, I looked in the code and it seems like there are some defines to change, then think about.
The pyvnc system is pretty good. Shame the toolkit is ugly :(
What was the rationale behind moving from gtk to tk?
hcseb: development of the old vcp (written in C and gtk) had stalled .. I talked the author of the new vcp into using python+tk so that it could integrate within the AXIS gui window
I thought it might have been for AXIS integration
Damn damn damn... Just found 25' XLR to 1/4" phone cables for $4
It's only 3 conductor =(
ds2: I plan on hitting weird stuff today
is it a good idea to have the "enable line" break the motor circuit to the servo motors?
that depends on the machine
but it's probably as good as anything for a small servo machine
it's not a machine that will move if no power is applied
what if it's already moving? ...
then it'll come to a quick halt
or are you talking about enable, not e-stop?
yes, enable. that bit gets enabled with the "power" symbol in the gui right?
I'd like to stop the servo from singing when not in use, but have the powersupply still active
I think so (I'm not sure exactly which bits go on when)
ah. unless you have brakes on each axis, you probably want the servo amps enabled - they need to correct if the machine gets bumped
"machine on" turns on the various axis.N.amp-enable-out
the gecko will keep track of up to 40 encoder steps before it goes into error mode though
I like to use that to enable pid, choppers, etc
then you can turn "machine off" and move the machine by hand
geckos are actually +/- 128 counts
oh is it
but if you disable it, it won't do anything
i was thinking of putting a relay on the +- circuit for the motor...
oh I don't know anything about the 'gecko way'
the enable line is also the error line, so if you drive it low (I think) externally, there will never be a fault
don't ever disconnect the motor leads from an active drive
you're just about guaranteed to fry the drive if you do so
oh, won't do that then ;0
The_Ball likes letting the magic smoke out
heh - good plan ;)
SWPadnos: I was thinking of you... http://www.weirdstuff.com/cgi-bin/item/20008
read up in the geckodrive archives about that. I know Mariss says it's a Bad Thing, but I don't know the specifics of why
yes, I know
err - thanks
I wonder if I can still win a Sony home entertainment system
SWPadnos: Yeah, he told me that on the phone... The sure way to blow a drive is for the motors to become disconnected from it while in operation.
I know that's true of the stepper drives (except the vampire), but I'm not absolutely sure about the servo drives
SWPadnos: He said even vampire
but things might have changed since then too
hmmm. I thought the broken motor wire was one ofthe things he had solved
vampire was "in the works" at the time
you'd think that if you can dunk it in water while operating, you'd be able to disconnect a motor lead
SWPadnos, "the enable line is also the error line, so if you drive it low (I think) externally, there will never be a fault" im not sure of the behaviour on this "err/res" input. When not connected the gecko will not clear a error condition, when connected to +5V the error will clear after 3 seconds. don't know what happens if i drive it low, might disable the drive? I will go downstairs and try
well, I'd look at the manual before believing me :)
I have a set of geckos, but I don't work with them every day
* JymmmEMC ponders WTF am I looking at T1/T3 equipment for
PP-JTAG Controller with Printer Port Interface
SWPadnos, shorting to ground will enable error mode, shorting to +5v will clear it
ah - ok
the drive outputs a low, which will stop all the other drives if you just parallel the err/res pins
and if you leave it floating you can read the error state, nice design
well, it actually causes some problems when you design an estop circuit, but it does work
it's a dual-purpose pin, and you can't tell what it's being used for at any given time
you need to separate the err/res lines from multiple drives, if you want to be able to tell which axis faulted, so the fact that they can be paralleled isn't as useful as it could be
and you need to connect an input and an output pin to it (can be the same pin, depending), so you can both detect faults and reset the drives
SWPadnos: Damnit... http://www.halted.com/ccp21319-toroidal-transformer--250va-21058.htm
250VA - wimpy ;)
SWPadnos: 120/2=60V*1.414=84V =(
if you're thinking of running geckos, that's probably fine
cradek: what kind of torcher did you put your lathe thru?
SWPadnos: my motors need 50VDC
ah, so they'd get a bit warm with that xfmr
SWPadnos: I see multiple leads... I wonder if it has +/-10V
skunkworks: I cut an inch part on my metric lathe using multiple tools with tool offsets
it works great now
(the tool table stuff was pretty busted before)
however, I found that I have a fairly weak power supply that's causing problems
no, it's a toroid transformer that's unexpectedly wimpy
cap it ?
cradek: very nice.
Can't wait to see it in action.
would a capicator help ?
Unit41: nope, I think more caps wouldn't really help much
when I deal with electronics stuff I just try to throw capacitors in random spots
somtimes it works... most of the time not
250VA is what about 210W ?
do kernel modules get realtime speeds ?
found this great article http://linuxgazette.net/122/sreejith.html
Unit41: in emc2, only special kernel modules that are "hal realtime modules" get realtime performance.
emc2 is a bit more complex than the source code shown on that page :-P
I knows it :P
JymmmEMC, you need to take the VA rating and divide by either 1.414 (for direct DC watts), or divide by 1.8 (the recommended derating for transformers used in DC bulk supplies)
jepler, i would like to ask you opinion on something, since you are the pluto expert. I have two servos arriving any day now: http://cgi.ebay.com.au/ws/eBayISAPI.dll?ViewItem&ih=008&sspagename=STRK%3AMEWN%3AIT&viewitem=&item=180119980489&rd=1&rd=1
which can accept step/dir or +-10V, I also have a pluto coming, would you use step/dir with the ac drives internal pid control or would you use pluto to control them? Im not ever sure if pl
uto can be used because i believe the motors have resolvers, not encoders on them
The_Ball_: pluto doesn't do either step+dir or resolvers
take the 264W you want, and multiply by 1.8 ...
i know, i would use a parallel port for step/dir, and im not sure if the listing is correct regarding resolver or encoder
that gives ~475VA, so 450 would probably be OK (since you won't be using full power most of the time), and 500 would be slight overkill
SWPadnos: That torroid is 250VA though
The_Ball_: the pluto also uses PWM output. So you would have to build up a bit of a circuit to turn the pwm into a +/-10 signal.
so emc runs from command line then ?
JymmmEMC, yes, which means it's undersized for your application ...
SWPadnos: damn PF! LOL
skunkworks, oh, ok, nothing a voltage divider and a op-amp/voltage follower couldn't handle
Sounds good ;)
[15:14:25] <Unit41> http://www.callowayengines.com/cal3.jpg
live long and prosper!
purpetual motion Yo
its a gravity engine
I emailed the dude and he said a video goes up in 2 weeks
it looks to me like they have the rotation backwards
there is a trick there though
see how the ball goes through the slit
if you look at the first position, the torques from 5+4 cancel, 8+1 cancel, and 3+6 cancel
you need to have an opening in the center of the devide
that leaves 2+7 for net torque
remember 7's gotta roll yet
7 is at a larger radius than 2, so it has more torque, but it's counterclockwise
and you dont cancel them out like that
use ones on the same devider
I'm looking at it from a static->dynamic perspective
ie, if you start it in position 1, what will happen
you need to spin it up first
it rotates at about 60 rpm
ok, that's a different story then :)
a video goes up in 2 weeks... :)
[15:37:47] <skunkworks> http://www.youtube.com/watch?v=xWIN_b95GoI
reminds me of the huygens? gravity escapement
according to the text, the "free energy" part is not shown here (and compressed air is used instead), but it's in the plans
I saw that ;)
the gravity escapement seems a bit magical at first too, it's very clever
Bill doesn't get sucked into these projects?
you mean magic machines? um, no
cradek how does big ben work ?
[15:51:36] <Unit41> http://us.geocities.com/mvhw/gravity.html
this looks plane retarded
Unit41: are you saying that big ben is a perpetual motion machine?
that animation doesn't work for me, but that is a gravity escapement
well, I got it to animate, but it's bogus, it doesn't show how it works
I think the animator may not have understood the escapement actually
[15:58:20] <Unit41> http://www.geocities.com/mvhw/bpinwhl.html
building my own clock just went on my future todo list
maybe a water drop clock
that makes volts to power an led and uC
[16:06:47] <Unit41> http://amasci.com/emotor/kelvin.html
you know it is a big oil consperocy.. same with the 900mpg carberator.
that they bought the patent and hide it
you know it..
I heard it was 150MPG using an MG body and various parts
but I don't remember which university
all I know is the education system sucks and were all forced into conformity
Unit41: GET BACK IN LINE BIOTCH!
If your going to waste time on free energy - your time would be well spent working on a low temp differental stirling engine.
ebay wont ship my books till the 4th
I bought a bunch of tesla books and some heat engine books
you can spin a parabolic dish from plaster
but your still only going to get (in a perfect world) 50% of the engergy you put in back out.
solar conversion can be 90% efficient
of what to what?
but thats only if you wana heat a swimming pool
exactly - heat to heat
light to heat
ok - yes
anodised aluminum works best
I still have to take pictures of my unfinished monstrosity
I seen a neat article on solar refridgeration too
yes - same principle as gas powered refridgerators in rv's
you heat a tube with ammonia and ice remover for sidewalks
and it releases the ammonia
it then condenses in the fridge part and duri8ng the night it evaporates
goes back into the salt
I think an 8-pin "attiny" would make a nice current limiter / chopper for L298 servos. Pull L298 "enable" low when the current limit is reached, and set it high again after a fixed off time
the only external parts look to be the sense resistor and the reference voltage divider, trimpot, or whatever
isn't a comparator easier and just as small?
ds2: no, that's a bit tricky to do -- when doing "enable" chopping, you don't get current through the sense resistor, so it will look like the current quickly drops below the limit
I suspect a comparator is harder to adjust, and also harder to make it resistant to temperature changes and the like
harder to adjust the delay time, that is
the l297 chopper incorporates a comparator, flip-flop, and clock
the allegro part I looked at was similar, except that it uses fixed off time
03cradek 07TRUNK * 10emc2/src/emc/rs274ngc/interp_convert.cc: I actually tried/debugged this now
03jmkasunich 07TRUNK * 10emc2/src/hal/drivers/mesa_5i2x/firmware/configram.vhd: add VHDL for configuration data RAM
03jepler 07TRUNK * 10emc2/src/emc/kinematics/gantrykins.c: rtapi.h should be included before any rtapi_xxx.h header
03jepler 07TRUNK * 10emc2/src/rtapi/rtapi.h: fixes conflicts between kernel and userspace header files that caused problems on 64-bit systems
03jepler 07TRUNK * 10emc2/src/rtapi/rtapi_string.h: our definition of strcmp causes problems on 2.6/x86_64 systems; hopefully the comment about 'very early kernels' means this test for kernel 2.4 or greater is OK
03jepler 07TRUNK * 10emc2/src/rtapi/rtai_rtapi.c: using this old module_param syntax causes problems, and the use of the RTAPI macros should be preferred anyway
03jepler 07TRUNK * 10emc2/src/rtapi/ (rtapi_math.h rtapi_math_i386.h): the defintion of huge_val should be portable to all systems with IEEE maths, so move it out of the CPU-specific header file
03jepler 07TRUNK * 10emc2/src/module_helper/module_helper.c: this module is needed on some systems
03jepler 07TRUNK * 10emc2/src/hal/utils/comp.g: this header is not needed by most comps, so don't include it
03jepler 07TRUNK * 10emc2/src/Makefile: It is necessary to have the true kernel headers on my x86-64 machine, and it does not appear harmful on the 32-bit systems I tested (dapper/rtai, dapper/sim)
03jepler 07TRUNK * 10emc2/src/hal/Submakefile: if libemchal.so.0 is not an explicit target, it might not exist after the build completes
has any one using the DigiSpeed GX from homanndesigns for controlling PWM output to analog spindle speed?
I've generated PWM output directly from EMC2 in various ways to control spindle speed
(I don't know about that particular product though)
im trying to setup EMC with PWM output to feed into the digispeed to convert it into a 0-10v analog for the inverter
you can check out digispeed here http://homanndesigns.com/store/index.php?main_page=product_info&cPath=1&products_id=21&zenid=7aba984713efa5993762f6e9793b0a61
rob-h: can you tell us what the specific problem is (even if it's "it doesn't do anything") and show us the HAL configuration you're using?
oh ok, it's simply an isolator board
hmmm - cool little board. Should work nice with pwmgen
you should be able to run it with a parport output doing pwm, or various other things
simply yea, iv set it up as per manual, got it to give 0-10v output
im using the nist-lathe.hal as a starting block i have the PWM output fine, but the digispeed does nto seem to be giving me a output on analog
so you have a pwm signal showing up on your parport, verified with a scope or something?
yea put my scope on it (not EMC scope, real world one ;) )
that eliminates a lot of variables!
just in manual states active low pulse width mod.. with PWM periode range 2hz to 50Hz
about what period are you getting on the parport?
try setting the pwm frequency of pwmgen. the default setting, 0, means a PDM-like waveform which may switch on or off every base_period -- a lot faster than 50Hz
setting it to 50 will give 50Hz
"setp pwmgen.0.frequency 50"
where would be best to define that?
in a .hal file -- I'd put it in the same hal file as the other lines about the spindle
maybe right by the "setp" lines for the scale and gain
wow - says 2-100hz
2Hz seems awfully slow
it must have smarts - not just RC filtering etc
I suppose it is designed for mach ;)
03jepler 07TRUNK * 10emc2/src/ (Makefile Makefile.inc.in configure configure.in): "-msse -msse2" must be passed when building modules that use floating point for 64-bit systems
I will give that above ago in a bit, thank you
it seems like you can troubleshoot this with EMC and digispeed separately
if your scope says the digispeed input is something meaningful (within their specs) you can concentrate exclusively on whether the digispeed is working right
how many levels would that be at 50hz (period set to say 50us)
% units "1 / 50 Hz * 50 microsecond"
the device even has LEDs to help you troubleshoot
for instance, it has a way to show it's working but has no PWM signal
skunkworks: aboug 400
just tryed the "setp pwmgen.0.frequency 50" but EMC errors on it
looks like it's pwmgen.0.pwm-freq
you may also want to set dither-pwm off
see all the options with "man pwmgen"
Thanks for your quick helps guys, getting late here so ill take aplay with it again tomorow night
ok, good luck
yuck, I got a spurious 'joint on limit switch' while cutting
I wonder why
I hope it's that, rather than a pluto timing bug like the last one
well I had chips flying, and the switch terminals aren't covered
so I'd be loathe to blame the pluto at this point