What is a XEON cpu equiv to these days?
allo from taipei :)
oops 9:03am gotta run
JymmEMC: Core Quad.
I can't recall... who was the cheap bastard buying *USED* DLT/8MM tapes?
could have been me - often people retire a bunch of good tapes when they get a new drive, and the tapes are fine
cradek: was it 8mm or dlt?
but I don't need anymore - I have a few dozen.
cradek: realisticly, how much data to you usually get on them?
and do you have more binary or text data?
my drives are all 35/70 and it depends on the data - but never 70
I *think* these are 40/120, but I could be wrong
40/80 is the most common tape
no such thing as 40/120 afaik
40/80 sounds better, I'm too lazy to walk over and look =)
actually, I use 40/80 tapes on my 35/70 drives
all (?) real tape drives are scsi
Heh, they have eSATA ones now
heh, knowing that does not make me want to retract my earlier statement
at least I didn't say USB =)
parallel port? floppy?
Watch it now... don't make me play the trump card on your ass!
Colorado fdd iface tape drive =)
I had some QIC-80 (floppy) stuff - it sucked.
I also had, but never even tried to use, a full-length isa card that would hook to a VCR (seriously)
Hey, back then, it's all we had
I think I rememebr something like that
I really dont believe in tape these days anyway.
Do any of you python guys (jepler?) have any idea about this:
mshaver@M2A-VM:~/gdepth-0.2$ ./gdepth.py -x -10 10 -y -10 10 -z -2 0 -r 72 /usr/share/emc/ncfiles/spiral.ngc
vee_common 45 1.0
vee_common 30 0.57735026919
Tool: -0.0625 -0.0135316469341
Traceback (most recent call last):
File "./gdepth.py", line 233, in <module>
a = numarray.zeros((sx+ts, sy+ts), "Float32")
File "/usr/lib/python2.5/site-packages/numarray/numarraycore.py", line 1463, in zeros
retarr = NumArray(shape=shape, type=type)
ValueError: invalid shape tuple
before I break down and actually have to learn something?
shoulda thought of pastebin
print sx+ts and sy+ts
maybe they are silly numbers
ok, hold on...
yep, 3/4 of those 4 numbers are silly
so, 1/4 is silly
what are those 4 numbers?
they come from gdepth.py
an array of size 0 or -1264 is almost as silly
I mean what did you print? I suggested you print two things, you pasted four
it's the output of: print sx+ts, sy+ts in python
two lines? really?
I know nothing of python....
>>> print a+b, c+d
yep, let me look again
oh I see now that the 0.0 inf was in your original paste
your new debug info is the last line you pasted
so you are asking for a matrix of dimension -1264 by 156, that's the cause of the error
yep, i just found that myself
i make things -1 inch long all the time! :)
hmm, well that gives me a leg up on understanding this code
I cut things 1" too short sometimes...
I could just find out how to take the absolute value of something in python...
...or figure out why the number is wrong in the first place...
that seems unlikely to be the right fix...
yeah that :-)
well, i guess.........
i'll dig back in and see what i find, THANKS!
I bet gdepth has bitrotted
(I suspect interpreter changes)
hope you find it - would be nice to have it working again.
yea, it was supposed to not work from 2.2 onward, but I really want to see it
yeah it's neat.
i downloaded this game development engine panda3d, and was fiddling with it
for 3 axis stuff anyway.
i was able to view a 3d dxf in only a few minutes of poking around
cradek: Thi is kinda cool, it has a built-in web server http://www.areca.com.tw/products/ide_to_sata_ll_sub.htm
i have visions of something like world of warcraft, but with machine tools instead
mshaver: What, sit around all day waiting for ev3eryone to show up for 90 minutes to kill some ugly sucker?
so i felt the need to get something going on a less grandiose scale ;)
I saw "vericut" at IMTS. it's a polished gdepth...
JymmmEMC: my son in law is an expert at this game and he's toured me around the gui - quite a piece of work!
mshaver: the graphics are awesome, the game is, well, boring....
it's basically a social network with something to do
yep, I think the difference between a lot of gpl stuff and stuff costing $$$ is that last polish and packaging
hang on mshaver...
i'll see what i can do with gdepth
mshaver: or the difference might just be salesmen and money
mshaver: it doens't have to cost a thing... http://www.wesnoth.org/
Not WOW mind you, but still...
cradek: could be, could be...
no, not WoW. I don't play computer games at all myself, but my kids and son-in-law do, extensively. It's why we still have Windows PCs :( around. I'm going to hunt me up a few level 27 bugs in Pythonic Gulch, I'll see you later!
fallout 3 is on the way!
* spasticteapot is downloading Fallout 1 / 2.
hey that is really cool... USGS released high-res topos of all the US available for free download
kind of like google earth, just zoom in and click the quad you want, then download it
[01:58:57] <dareposte1> http://store.usgs.gov/b2c_usgs/usgs/maplocator/(layout=6_1_61_50_2&uiarea=2&ctype=areaDetails&carea=0000000009)/.do
went there to order one and behold i just downloaded it instead
guess nobody else likes topos
if only they were 3-D
i think you can get the GIS data free from them too, but i never tried
3d doesn't do any good on paper though, humping it through the back woods
it's the bumpy maps I like :)
it would be pretty sweet if you could pipe the GIS data into a 3d printer of some sort....
i guess not much use though
topos are 3-d if you know how to read them
Or you put them into Excel.
and then what??
good luck inputing co-ord's
You can put the coordinates into Excel.
Rows and columns are X and Y, and the value in the columns is Z.
I know this because I used it to make maps for Dungeons and Dragons games when I was a kid.
* spasticteapot is a bit of a nerd.
except that the world is a lot bigger than a spreadsheet can handle
I don't know the limits of Excel, but I bet it's 1k-8k columns
Sorry, I got in here a bit late.
I assumed you were mapping a hill or something.
nope, topo maps of the US, from the USGS
65326 rows max
SWPadnos: wha ya doin with the topos?
nothing. dmess mentioned that you can zoom around and download maps from the USGS
I have digital topos, and then the full res for all of calif
that takes like4 dvd's
or a day download (at home anyway)
Oh no, took a week
I had to contact them too
they're like 120mb each, a few like 200mb
sounds a bit like the high res blue marble images :)
I only recently got enough storage to actually save them all
I got those too,
all 8 at full res was over a TB
I only grabed the ones I liked
I still have access to the DC if needed =)
SWPadnos: Oh, looky what I found... http://www.areca.com.tw/products/ide_to_sata_ll_sub.htm
I just picked up 5 1TB drives though, so like I said, I could at least save all those images now
SWPadnos: Ha, I have 300 DVD-R's =)
how much is that box?
I think I looked at it, and it was much more than the 3Ware cage/controller
cheapest I found is $684
(a couple of years ago, so it may have been a different generation)
that's not bad
But the thing is you can connect it to ide or sata
so you can move it anywhere
or you can get the scsi version
It has a web server buikt in too
5 drives in 3 5.25" bay
it says host connectivity is IDE or SATA 1.0
so that's toward the disks, or toward the computer?
disks is sata II
that seems pretty silly
Disk Bus Interface SATA ll Compatible -3.0 Gbps(300MB/Sec)
aggregate 5 SATA2 disks to a single SATA1 interface?
SWPadnos: Maybe, but I wnat redundancy, not so much speed
good, cause you won't get it ;)
if I want speed, I can get the SCSI iface instead
the ARC-6020 == SCSI iface to PC
But I like the versatility of IDE and SATA
since most of my systems are IDE
all of them I'd bet (capability-wise anyway)
There's this too... http://www.areca.com.tw/products/scsi_to_sata_ll.htm
But 4+1 hot spare seems reasonable
Unlike the all-in-one boxes, I can have any OS/services I want and still have redundancy if case of hw failure
the danger is that you're dependent on their RAID software and controller, unless they publish the algorithms they use to distribute data across the drives
similar to the Drobo
I'll find out if I can yank a drive out and recover data using linux
the controller does it. the question is, can you stick the drives in a PC and keep using them if the box dies?
that's what I mean
ah, ok then :)
[03:48:23] <JymmmEMC> http://www.kaleidonet.com/storage.html
might as well go with an Aberdeen box or something at that point :)
not familure with it
[03:50:55] <SWPadnos> http://www.aberdeeninc.com/
I be too poor
Anyone here know how to put a 5-micron (or thinner) layer of aluminum on a 7.5-micron layer of polymide film (Kapton)?
get a spin-coater
hmmm. no, not with Alu
sputtering may work
duct tape and bailing wire!
Can you sputter that much on?
And don't you need massively expensive equipment?
oh, it's expensive. you didn't say cheap
I saw a guy who had a guide to gold sputtering.
there's probably a chemical plating-like process too - I can ask my mother if I think of it
Chemical vapor deposition, I think.
Triethyl aluminum is used, according to Wikipedia.
well, that works on metals, but I don't know about plastic
Goodness knows what triethyl aluminum is, or where I can get it.
there's a technique used for CDs/DVDs though, so it may be possible
that thin CVD is the only choice
then again, they can also make those with a thin foil that gets bonded to the plastic, so they don't necessarily need deposition for that
Plan B is to just fine some 2.5 micron aluminum foil.
Although I don't know if I can realistically work with that or not.
static deposition might work
I also need to buy a rubber ear.
that sure sounds thin to me, but who knows
* spasticteapot is building headphones.
Or I could stop being thick and just build electrostatics instead.
charge the plastic + and use cermetal
its a finely ground aluminum in an epoxy binder
but al powder would work too then you'd have to seal it somehow
yes. good idea. good night all
dmess: Wait...I need more info!
I can't find this stuff anywhere.
It sounds like nifty stuff.
I just can't find any.
I might just cut open a capacitor and pray there's nothing too nasty in it.
do you need optic
I just need it to be conductive.
the rework stuff is al specs and epoxy - urethabond 100 if i recall after 16 beer
as light as you can apply..NO epoxy is light compared to CVD or PVD
Chemical vapor deposition doesn't sound very practical.
I can get the 12.5 micron kapton with 12.5 microns of copper.
dmess: Because I have access to a high school chemistry lab.
And that's it.
And I don't even know what the hell tri-ethyl aluminum is, let alone where I might buy it.
Or how I'd use it.
wheres the budget for the science fair??? i wanna blow IT
find the local college or university
they handed me heaps of FREE stuff over the Sci-Fair years
Worth a shot.
you NEVER been told no yet??
Mostly just ignored.
there's a classic silver reduction reaction for mirroring glass
sell the project 1st.. and find a friend... whos interested in it.. take your time..
similar to electroless nickel
[04:14:39] <fenn> http://www.make-stuff.com/formulas/mirrors.html
fenn.. what can MASK for electroless Ni
he needs to mask?
I don't need nickel.
I need aluminum.
silver is more conductive than aluminum
Yes, but aluminum is lighter.
I do.. V22-Osprey piston.. nobody on the planet wants to quote it
How hard is silver-plating?
^^ read that link
There's the other reason I'm using aluminum.
Unlike thin layers of copper or silver, it won't corrode when exposed to air.
Gold would work, except that it costs a lot and most of the things that will etch it will also destroy the polymide.
spray some sealant on it?
or gold plate..
many steel components are flash silvered prior to final coating
fenn: That might work.
and gold is softer than cad..
I'm trying to create very thin traces on the polymide.
over $$$ let em die
Ideally 1mm or so.
as in the past
spasticteapot: is this supposed to be a circuit of some sort? like an inductor
fenn: It's a transducer.
i dont know what that means
it's a headphone speaker, but how does it work?
[04:18:54] <spasticteapot> http://en.wikipedia.org/wiki/Ribbon_microphone
Like this, but backwards.
You put it between two magnets.
cermetal is THE Cadmium replacement... 787 has NO cad or chrome... cermetal and HVOF
If current goes through the traces one way, it moves forward; if current goes through the other way, it moves backwards.
why dont you just get some of the ribbon mic stuff?
[04:21:24] <spasticteapot> http://i7.photobucket.com/albums/y285/heildriver/Maiordisassembled.jpg
fenn: Ribbon microphones have an impedance of a tiny fraction of an ohm.
They also have a transformer.
Building a transformer that would work from 20hz to 20khz is very very tricky.
As it turns out, Hobby Lobby carries a material known as Silver Leaf, which is pure silver sheets that are about 3-5 microns thick. They come in a package of 25, 5 inch by 5 inch sheets for $6.99.
aw what the hell, now you go and make it all complicated
just get some of that kapton circuit film and etch the traces with standard circuit board techniques
it doesnt matter if the wires are thick or thin
fenn: It does.
The traces are wired in series.
Every time you cut the width of the traces in half, you double the number of traces, quadrupling impedance.
And driving a load of less than two ohms with ordinary stereo equipment is very tricky.
then add an inductor
it's foolish to try to start out with making the best
get some experience with easier materials and processes first
then at least you will have something to compare to
maddash is now known as Guest25174
maddash_ is now known as maddash
* archivist reminds the brits that the midlands model engineer exhibition in on today till tuesday
I bet I will overspend :((
I sure would
it would cost thousands just to get there ;)
annual trek for us
quick, someone tell me that single-point knurling won't work, before I'm dumb enough to try
it will, but it wont make crosshatching
anonimasu: I could cut both ways
making diamond patterns, like on handles
[13:07:10] <anonimasu> http://en.wikipedia.org/wiki/Knurling
cradek: that might work, however I think the rolls need to be synchronized for it to look nice
anonimasu: there wouldn't be any rolls
I can't find any real documentation on halscope in the manual. Am I just not looking in the right spot?
oh, cutting with a cnc lathe?
yes 'single-point' knurling
I'm sure it'll work, but your ^ wont be raised :)
like with a threading tool
looks like a couple of G33 moves
true it probably won't look like a real knurl
some CW some CCW
try it ^_^
alex_joni: some left, some right
as long as you have enough side releave on the side of the tool..
you can cut knurs apparently too
cradek my helix code done both sides
skunkworks_: yeah that's where I'll have to be careful. I'll have to figure out what angle I can get.
mill or lathe?
ah I would mill
yeah a 4-axis mill could do it, but only with a pointy tool
think multi star worm an do a left on top of a right
hmm.. a 3 axis too and some neat interpolation
is it possible to make a pdf open to a particular secion thru the url?
(found halscope in the hal manual)
time for live tooling on the lathe as well :)
I passed my chem test -_-
just barely, but I passed.
I'm in electronics lab at school
finished the assignment that should take ~4 hours in about 7 minutes.
And since we /HAVE/ to be here,.. I'm just wasting my time a bit... Should 've brought my controller and finished the soldering work on it :(
* anonimasu nods
here - watch a exciting short video http://www.youtube.com/watch?v=eUmFKOVepYY
"short"? how is that going to help me kill 2 hours :p
who with internet access can't kill two hours!?
JymmmEMC: Looking through some logs I see that I can get as little as 52GB and as much as 85GB on a tape in 35/70 mode
some of my data is more compressible; fortunately my backup software handles that all smartly
are you using Amanda?
is there anything else? :-)
hoot hoot! *owl*
amanda is great
if I remember correctly, it's a pain to set up but very powerful
I've used it for 15? years
I wonder if there's some relatively easy setup program with Ubuntu server 8.10
yes it can be a challenge to set up.
I don't know... it's probably getting easier over time.
heh - that could be experience talking ;)
for one machine you probably don't need it, although it's still nice. but for a network of machines, I can't imagine going without it.
Im tending to backup on other machines these days
it's possible that we're actually at the point where hard disks are a better backup medium than tape
Description: sophisticated graphical program which corrects your miss typing
you may be - I'm sure not
wots a tyop
(I know a lot of people have been saying this for years, but I didn't agree - I may have to change my mind soon)
Lol. the way it "corrects" that is by passing 2 20 wagon trains at a slow rate in ASCII form.... love it
fragalot: oh I thought you were making fun of "miss typing"
cradek: no. :) "sl" look it up :p
tyops are part of IRC
yesterday I saw a plumber's van nicely painted with their slogan "Professionalism @ it's Best" (including the quotes)
it just wouldn't be the same without it
archivist_emc: but "sl" is a console ap
yes, @ sign; yes, "it's"
yes, unnecessary quotes
prophetism at it's best.
I hate it when it's faster to file a piece than cnc it
how accurate is your file though?
to a chalk mark
as the drawing Im working to is lacking in dimensions, it's a guess anyway
it's a bit like our old motto (in the College of Engineering and Mathematics): "Measure with micrometer. Mark with chalk. Cut with chainsaw" :)
gas axe engineering
I did something nice at work the other day.. recycling...
found an old bucket of paint that solidified completely... cut it to pieces, and now the paint serves as dampeners...
small compressor :p
I don't expect it to last long, but it was funny
just add/spill paint thinner near it some time
accidently of course
I was kindof surprised at how flexible paint remains after it's completely dried out
has to be as most painted items stretch/bend/expand with heat etc
and for wood it has to breath
so is porous
noticed that too,.. it's like chocolate cream if you tear it :p
yet baby smooth if cut.
longer pwm cycles mean higher resolution but slower response times. shorter cycles mean lower resolution but faster response. is this correct?
yes pretty much. the longer "response time" is because you have to use a longer time constant on the external filter to smooth it
i thought it was because one would have to wait until the end of the current pwm cycle for changes to take effect
if you use PDM instead of traditional PWM, the filter components don't need to be as latge, and the adjustment is significantly quicker
the filter needs a time constant of roughly 10 PWM cycles for traditional PWM - I think that's the main cause of slow response
pdm == pulse-density modulation?
by 'filter', you mean that RC circuit that converts the digital pulse to the appropriate analog output?
ok, so when the pwm duty cycle is held at a constant rate, does the filter output a constant analog voltage? or would it be sinusoidal?
it would be variable, but not a sinusoid
it's a repetitive capacitive charge-discharge cycle
so it's a logarithmic approach to the high value and then a logarithmic approach to the low value
oh geez. so is there any way to get a steady analog signal from digital besides a true DAC?
it depends on the time constants involved
depends on the definition of "steady"
for pwm/pdm your tradeoff is "steadyness" vs responsiveness
steady, meaning a constant digital input ==> constant, non-varying analog output?
non-varying over what time scale?
for a spindle speed control, a time constant of a full second might be fine, which lets you use a lot of filtering
if you're driving a spindle motor, you can have small variations on a millisecond (or longer) time scale, and it won't matter
for a servo amp, a time constant of a second would be a disaster
for a laser galvo, microseconds could spell disaster
else sample and hold per cycle
when you say "time constant," you're referring to the period of the analog variations?
it's more the response time of the thing being controlled
as long as the controlling signal is much faster than the thing being controlled, it should be fine
I almost pasted the wikipedia link, but it wouldn't make this any clearer
for a "constant" analog output, you probably do need a dac
(unless the thing in question samples and holds rather than filtering - then you get problems since the sampled waveform is varying and you don't know where in the vyvle you'll sample)
but you may be mistaken that you need a "constant" analog output
[15:09:59] <cradek> http://en.wikipedia.org/wiki/Time_constant
even DACs aren't constant
many of them output a small spike whenever a new value is written, even if it's the same as the old value
maddash: what's your particular application?
and then there's noise on the lines
it's basically impossible to have a constant analog value anywhere, unless you're in a well-shielded room at absolute zero
but DACs don't have inherent periodic variation in their output
there is always tolerance and drift
they do if they're not high quality and there's a periodic refresh
tolerance and drift are unrelated to noise
(in this context)
what is this context?
I wish we knew
talking about getting a constant value output
heh - me too :)
cradek: sending an analog signal to an amplifier to control voltage? see, I have a 20V range to choose from, and I either use PWM or a resistor ladder (troublesome, and noticeably discrete steps)
I think maddash is probably asking the wrong questions, and he's getting some pedantic answers to them - nothing that will really help him solve a problem he's working on.
PWM also has discrete steps, you may nust not notice them
maddash: what are the timing requirements and acceptable output quantization?
erm, I guess a 200-400 us response time is enough?
true, pwm does, but pdm practically doesn't, though it gets bad at extreme values
right - the filter bandwidth comes into play at the fringes
but it's still good advice ;)
* renesis 24db/oct active filters ur moms
guys im hungry
SWPadnos: 16-bit looks fancy.
it is :)
and massively adjustable (both in full scale range and offset value)
DAC, the AD5764
SWPadnos: I'm trying to be as cheapo as possible and use whatever is available on the ucontroller already.
noise+error is prob always gonna be greater than a discrete step on that
on lower signal voltage stuff
omg, wouldn't a dac be prone to the same sort of emi as an adc?
SWPadnos: remember that pic i was talking about?
use an internal PWM if you have one left
if you're running with a 20+ MHz clock, you can do 20 kHz PWM with 10 bit resolution
softpwm is pretty easy if you have any timer ints running already
: "Settling time: 10us maximum" <-- is this the time constant you were talking about?
thats the switching transistion time
are you doing an audio amplifier, a motor amplifier, an escalator controller - what?
if you're doing something like a thermostat setting, you have a time constant in the minutes (ie, it takes minutes for a new setting to take effect, so there's no sense in changing the setting faster than the house can warm up)
you should think about the mechanical response time you can expect from a motor
they're pretty good second (?) order filters
if you set the motor to 100% current (from 0), how long does it take for current to actually build up to 100% in the windings?
hold on, I'm trying to figure out how to went from 20mhz to 20khz/1024 steps
20 mhz/1024 ~= 20kHz
the 10-bit PWM needs 1024 clocks per PWM period
and 20khz means a 50 us wait, minimally, after I switch to a new duty cycle
the minimum is one clock
though you're right - the average won't be "correct" until the next full PWM cycle has finished
so it's actually 1 PWM cyccle +1 clock to 2 PWM cycles
(up to 1 cycle for the new setting to be used, then another one for it to be output)
FREQ_OSC = (resolution in # of steps)*(length of pwm cycle)
er,, that's not right
so FREQ_OSC = (resolution in # of steps) / (length of pwm cycle)
(length of PWM cycle) = FREQ_OSC / (resolution in # of steps)
that would be freq of pwm cycle
err - whatever. tiem for more coffee
heheh k good luck
shit i have to go
sorry for creating such a pointless stumbling point
man. marketing materials are so stupid
I just got an offer to switch to FairPoint for voice and data
they are generally selling to the stupid :)
they tell me I'll get a $200 gas card and there are no setup fees
they tell about some of the features of the service, but nowhere do they say the price of the service
not even a "starting at $39.99/month" statement - nothing
they want you to ring up so they can employ a rabid high pressure salesman on you
the whole website is the same way
I actually called recently because their advertising implied that fiber should be available in my area (it's not, nor will it ever be)
can anyone tell me what G91 G10 L2 P1 X1 does on a fanuc control?
syntax error (I can't parse it)
?REDO FROM START
too verbose ;)
"ERROR 0001" (compliments of BOSS)
maddash is now known as Guest28095
no idea if that link will work for you...
yes, it does
see the bottom of page 51
(or do you care about this?)
sure - I'm interested (at least from a functionality conversation standpoint)
so it looks like the G91 version is supposed to "move" the offset by that amount
EMC doesn't do that
I think that would fall out of the current calculations anyway, no?
does it ignore G91?
(in that case)
right, G10 does not consider G90/G91
well, that seems like an error to me
well, if we were fanuc, it would be
unless it's specifically in the docs that abs/rel is ignored
(the EMC docs)
[19:26:51] <cradek> http://www.linuxcnc.org/docs/devel/html/gcode_main.html#sub:G10:-Set-Coordinate
well, that's clear as mud
skunworks\ is now known as skunkworks
skunkworks, its alive!
Yes - so far - Only running about 2.5a and 20v. But the current limit seems to be working so far.
cool... are the fets getting warm?
The voltage across the sense resistor goes right to the sense voltage and clips.
I am sure they will at 10a and above. they have a pretty high Rds
they are around 44a but as jmk schooled me - they can't get the heat away from the die fast enough so 20A is about all I will probably be able to get.
good old i^2*r always wins
but it gives a good peak headroom :)
but I can always find some better fets. this is just what I have on hand.
do you have a larger servo motor to play with?
[19:37:37] <skunkworks> http://www.electronicsam.com/images/KandT/DSC_0242.JPG
[19:38:16] <skunkworks> http://www.electronicsam.com/images/KandT/servotag.JPG
cool... I have been playing with ltspice from linear technology, trying to design a 2kv 2kw supply... I decided their switchmode controllers arent what I need
150v gives me the max rpm of these monstors - 1200rpm
yea... those should cause some distress to a controller
cradek, I think the EMC docs description is compatible with the Fanuc implementation
can you point to a specific thing to reassure me?
wherever it says "absolute", it seems to me that the purpose is to distinguish which coordinate system the offset will be relative to, not the coordinate mode in which the offset is specified
I also have a puma arm now with no control. (one of the driving forces for this current limited design) (need 6 of them)
yes I also don't believe that "absolute" means anything wrt g90/g91
I think a pic with pwm, some software will be much easier to get softstart working better.... charging a 150uF to 2500volts is a little bit of an issue
so I don't think our docs specifically cover the Fanuc version, but they also don't preclude it
the question I'd have would be - does the G91 stick, or should it be treated similar to G53
g91 always sticks
well, that answers that question :)
so you don't think the ngc docs say what G91 G10 does, so we may as well do the compatible thing?
correct on both counts
LawrenceG: for a car supply?
a generic ham radio amplifier supply using microwave oven transformers.... I would like softstart, regulation, adjustable voltage, and current limiting
ah - cool
the switch mode controllers are designed more for 50khz+ ferrite cores..... 60hz with big iron doesnt work very well... soft starts dont run the push pull outputs symmetrically which give large dc current in primary
current limits seem to work the same way..... iron core will saturate
Its possible the MW oven transformers will work better at slightly higher than 60hz.... they are design with as little iron as possible... they run awful close to saturation
where can i check what pins are doing what??
ask your local voodoo witch doctor
or use halscope/halmeter
well thats YOU to me
oh. I guess I should stock up on chicken beaks or something then
i'm THIS close to hooking up 1 motor
in hal.. can i just swap the connect physical pins to the signals section
was that supposed to be yoda-speak?
i have a working motor.. probably NOT optimally tuned.. but it spins... in manual mode
manual = jog?
turning by hand
so i managed to swap the correct pins
good night all
skunkworks, any suggestions on cheap 400v fets? avg current >10A.... peaks 50+
I bought my last ones on ebay.
LawrenceG, browse SGS
thanks guys.... wonder how sgs is on samples?
them and http://www.irf.com
do sexy fets
Lawrence... my motor runs
i have no torque.. but i just grabbed the default nist-lathe config and re-arranged the pins to matche the board
dmess, a stepper or a VFD motor?
the stepper for today..
we'll see about wiring the bigger stuff tomorrow
one step forward is always good
do i have to tell EMC im running 1/4 step.. or is tha all on the board controlled??
dmess: if hand-editing the inifile, you include that in your calculation for SCALE aka INPUT_SCALE. if using stepconf, it's broken out as a separate item "driver microstepping".
i'm on a 6.xx handing it... full sten on the board and the motor just makes noise
seems it i tap the down key it moves 1/4 turn
for example, my system has a 200-step-per-rev motor, 1/8 microstepping on driver board, and 1-start, 16TPI leadscrew (advances 1/16 inch per revolution). This gives me a SCALE of 200*8*16 = 25600
(that's 25600 steps per inch). a similar example in metric would be a screw that advances 2mm per revolution, which would give a sacle of 200*8*2 = 3200 (steps per mm)
1/8 step gives me the torque i need but is quite slow... unhooke from the macine of course.. it may be plenty fast on my little machine
er, whoops, metric example is wrong. a screw that advances 2mm per revolution gives a scale of 200*8/2 = 800 (steps per mm)
yeah, microstepping quickly brings you to the limits of software step generation
i havent dug out the screw specs yet.. just happy to have a turning stepper
and in your case, you'd use 4 instead of 8 if you have 1/4 stepping instead of 1/8 stepping
(and a gecko with 1/10 stepping would use *10)
im 1/8 now and a lot better
torque wise... speed is low
im .1" per rev
all of a sudden i have unexpected realtime delays... other than the 5 other windows ... whats the problem
EMC really doest seem to like bouncing in and outta chat
wifi, USB, graphics drivers, ACPI hardware, other power management, chipset, SCSI, SATA, journaling ...
any or all of the above, and other things I didn't list, can cause RT problems
hey i have ALL of them... go figure
i'll have to check i when all alon... but the test program seemed to react the way i thought a drive would make it
im dreaming up the spindle set-up as we speak
who can check out a HAL mod "apperantly" for the C6 board for EMC
email or post somewhere
what is step gen for the spindle... and does it require another drive??
[21:34:20] <skunkworks> http://www.cnczone.com/forums/showthread.php?t=66530
a happy punter
2 motors and golden
get to 4 motors and play daisy.ngc
runs thr axis and i feel the dir& speed of the motors and get a HARD on
i have only 3 drives
* archivist thinks daisy should be updated for orchestras of 3,4,5 steppers
my drives seem cold as cucs... made 4-5 pawns
is that GOOD
steppers warm to hot ish if unmounted is ok, drive electronics cold is ok
all seems cold.. currently.... any idea how this will affect my x10 stuff on the SAME box???
its rs232.. serial
fenn: Anyone here know anything about electronic fuel injection?