Does anyone know when 2.1.4 will flush the probe file to disk, its had several hundred hits, but the file is still 0 bytes long...
[00:02:04] <ohiopctechDOTcom> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Videos
<-just added my vid link to the wiki vid page
gene: refresh the file manager?
RT GPS dont work well with usb2serial either
damn USB polling
Long live UARTs!!!
what's RT GPS?
Real Time GPS data collection
your GPS is military grade?
Jymmmmmmm are you using that on your cnc table?
No, but it has both serial and USB interfaces... I use the serial iface to connect ot my radio
a lot of gps units are 1Hz, fast ones 5Hz
"RT GPS CNC XYZ DATA HAL MODULE"
My radio understands the NEMA protocol
my cat understands french
does it really understand, or is it just nodding it's head going "uh huh, uh huh" ?
works with the cat or the radio
that's the kind of capacitance you can really sink your teeth into
i have one
damn thing is rated for 25 wvdc
I have some 100,000 some uf 5v ones.
[00:29:57] <SWPadnos> http://www.digikey.com/scripts/DkSearch/dksus.dll?Detail?Ref=42115&Row=460751&Site=US
42V, 200F (yes, farad) caps
only $5245.68 each at digikey
and for 1/1000 the price, you can have one of these: http://www.digikey.com/scripts/DkSearch/dksus.dll?Detail?Ref=42283&Row=79020&Site=US
I have a 16V 12.5A PS, I wonder if I could get 50VDC out of it
but like I said, the cap is 70000uF @ 25 WVDC
that's not enough WVDC, I think ;)
what WVDC is I have nfc
Working Volts Direct Current, I believe
shit, well if I have to buy just a cap, and can use this xmfr, would be a good thing
got a rectifier?
Acme Electric sal tlake City
40A 600V bridge
I've had it for 1.5 years
Acme Electric sal tlake City, 200B16M
Hmmm it's at 115 nominal, but has +- 10% taps as well. plus the 230 too
there is NO focus adjustment on this webcam... it does have a stupid head tracking feature though.. wowo, neat
gringos is now known as Unit41
Linux newbie Q: For a dedicated EMC PC running Ubuntu (an old P3, if that makes a difference), is it a good idea to keep up with non-EMC software updates? There are 75 or so I haven't installed.
if it's net-connected, and you're interested in security updates, then it probably does
def. yes if it's hooked to the net*
you can also disable all but the security (and emc2) repositories, and you'll only be notified of security and emc updates
So updates generally haven't appeared to create new issues with EMC? Are the platforms that are being used by the developers for testing kept current in terms of available updates?
others.. maybe less often
i know debian is sometimes behind the times
Are any M5I20 gurus present?
depends on the level of guru-ishness you need :)
I'm figuring out wiring for my installation now, and I have an Opto22 24-module rack I was planning on using.
Is there some way without diving into the VHDL source that I can figure out which discrete I/Os can be reassigned to a different connector by massaging .hal files and which can't?
no I/Os can be moved from within HAL
you can't decide what's an input or what's an output, you can only decide how to hook up the inputs and outputs
with the current configurations
Yes, I get that. And of course the servo interface (7I33) I/Os can't move.
I may have misunderstood the question
But I think I can move inputs or outputs from the default locations to the same I/O type on a different connector, right?
can't you map any pin to any signal with hal?
As long as there isn't anything in the firmware that depends on it being in the original location.
*that's what i gathered as far as i dived into the config...
Chris_sub_1, yes, without changing types, you can decide what "HAL function" will be connected to any pin
ohiopctechDOTcom, no, you can't change any hardware configuration from within HAL, you can only decide (with complete flexibility) what any hardware connections mean to HAL/EMC2
... i've never used a m5I20 so i may be mistaken
ok.. thanks for explaining it SWPadnos
that's actually true for all hardware. you can decide what hardware to use (as long as it's installed), you can make very limited decisions about how it's configured (like parallel port IN vs OUT), and other than that, all you can do is use the I/O however you want ;)
[02:21:24] <ohiopctechDOTcom> http://www.youtube.com/watch?v=Q5nmspVOz_Y
As a for-instance, can I change the limit switch inputs to different physical inputs, or are they used by hostmot?
you can change the limit switch inputs to any input you like
nothing inside EMC2 deals directly with hardware except for HAL
there are so many ideas involving reprogramming a Prius to run machine shop tools....
that youtube link i just posted is "somewhat" cnc related... it has gears in it
Is there a block diagram available showing what is going on in hostmot?
err - I don't think so
the only things built into the FPGA in that configuration are PWM/DIR outputs and encoder inputs
I'm close to saturation with new information. VHDL source might make my head explode. ;)
do you want a pinout or a block diagram?
pinout is much easier to grok, and is really all you need :)
Mostly, I just want to know what just goes through a register (or some such) as an isolated input or output bit, and what is tied into other functions in the FPGA.
ok, then the pinout is what you need
I'm not sure where the pinout is, actually
I figured out (I think) all the I/O assignments, but I don't know anything about the FPGA code uses.
the first connector is used by PWM/DIR and encoders (like the 7i33 expects)
the second and third connectors are 16in + 8 out ( I think - could be 16 out + 8 in) each
I looked at Show HAL Configuration, and then verified against the HAL files to be sure.
I guess it would also be nice to know if anyone has any plans to dedicate I/Os that are currently unassigned in the distribution.
I plan on dedicating some to limit and home switches
there's quite a bit of work being done on a new 5i20 FPGA config and driver
remember - the only pins that are actually dedicated to a certain function are on the first connector, everything else is a user-assignable input or output
Jymmmno, not stable yet
it's for step/dir though correct?
that'll be the first output type available, yes
hey - here's a 5i20 guru now ;)
Lube On is in P4, and I'd like to move it to a spare output in P3 (which I wouldn't expect to be a problem). I don't need axis enables that are also in P4.
SWPadnos: Like PeterW would know anything about it
I don't need the amp fault inputs and would like to use those for other things.
Just wanted to say that the HOSTMOT configuration has 24 I/O bits - each bit can be input or output, but I'm not sure if the driver gives you access to the data direction register...
oh - interesting. I don't think you get to choose at the moment
(24 I/O bit per free connector)
I'm also not sure that's in the version with EMC2 (though it should be)
SHould be , I/O has always been that way
ok - I'm looking at the driver source now - it looks like it's hardcoded
but it shouldn't be too hard to make it user-configurable
The manual entry seems to use the I/O directions appropriate to the 7I37, but the configuration has no such limit
hmmm - I guess it'll take a little more work inthe driver - the inputs/outputs are defined with a shift amount which would have to be calculated, and maintained separately for the two connectors
but still not too hard
is there a (configurable) pull-up for inputs?
They are configurable for TTL or for RS-422 differential, but I don't see anything about a pullup (either way).
No, The card has 3.3K Pullups. If you wanted to, you could pull the resnets, and make a config with pullups/pulldowns on a per pin basis, but thats a different config
(Thats on the 7I33, the 5I20 is TTL/CMOS/LVTTL only)
You're right...thanks. Wrong doc.
is JMK? around, had a PWM gen idea for him
check in -devel
SWPadnos: Ever heard of a IC mfg called
ST Micor, yes
Micro, that was
I was in another channel, looks like a good 5i20 conversation going on
* jmk-solo reads back
SWPadnos: Bummer, I found a datasheet for this uC but it's all in chinese
I've heard of chinese
google translate it
it should be quite funny
petev: it's a pdf, and google dont do pdf's that well
I'm sure google has excellent chinese -> engrish translation software
[02:48:46] <Jymmmm> http://news.thomasnet.com/images/large/023/23600.jpg
I was planning on using the 24-I/O Opto22 rack on P3 for all the 'heavy lifting', and a 7I37 on P4 for lower-current loads and additional inputs.
the 24-I/O rack - is that inputs, outputs, or both?
Either. Pin-type plug-in modules.
and all sockets can take either kind of module?
Chris_sub_1: the new driver will let you use any pin thats not a special function (like encoder, pwm, step/dir, etc) as either input or output on a pin-by-pin basis
the existing driver limits you to 16 in 8 out per cable IIRC
It takes a cardedge connector on the rack and the cable pinout is compatible with the 50-pin connectors on the M5I20.
but either way, in HAL you have complete flexibility of what signal goes on which pin
jmk-solo, actually, with some relatively minor changes, the current driver can allow the same thing on the second and third connectors
jmk-solo -- We are doing a biggerdual HBridge (20A 200V) and had a (simple) idea to reduce DC bus ripple - put a 3 or so bit adder in reference counter MSBs so the PWM on sequential channels is phase shifted 90 or 45 degrees
SWPadnos: I assume so, but I'm not at all motivated to work on the existing driver
patches gratefully accepted ;-)
I'll take a look, and if it's easy enough, I'll do it this week
PeterW: clever - we did something like that on a large DC-DC converter at work
unless I can't get my machine going with RT
JMK-I figured out how to use the hal files to move I/Os around, but my question was about which I/Os have a location dependency in the FPGA.
what do you mean by which I/O?
in the FPGA, an input is an input, it cares not at all what you use it for
as long as its not dedicated to an encoder or something
likewise with outputs
That's what I meant.
you can put things like "X hi limit switch" on ANY input pin
and "mist on" on any output pin
That's what I thought, but I wasnt' sure which were off-limits for relocating.
none are off-limits
Gawd, I'd hope not
So all the non-servo (P3/P4) I/Os can be moved without breaking anything?
One thing you need to watch out for when driving OPTO 22 racks with the 5I20 is that the 5I20 outputs only swing up to 3.3V. So if you have 5V input modules, you need to drive the outputs in open drain mode - that is set the output bit low and manipulate the DDR bit to control the output
another thing the current driver doesn't do
somewhat more complex, especially if you want separate mode control lines for each pin (input, open drain output, output)
SWPadnos: dealing with those issues is the main thing thats made my progress so slow
it's too bad there's no way to change read/write mode on a pin after module init
a HAL pin, that is
if you wanted to be able to use a physical pin for both in and out, you'd export both in and out hal pins (and probably a "dir" or "output enable" hal pin as well)
yeah, though the OE parameter would have to be mode, since the OC mode is needed for >3.3V outputs
which brings us to another "too bad" - it would be nice if we had enums for HAL parameters (and maybe pins)
I'm thinking the config options would be "make input pin available (Y/N)", and "output mode: (normal, OC, tristate)"
correction "output mode: (normal, OC, tristate, none)"
any reason to not export the -in for all pins?
Gotta love the catagory.... http://news.thomasnet.com/news/explosives_armaments_weaponry
other than namespace pollution
namespace pollution is the main reason
there is also some shared memory usage associated with every HAL pin
holds the names, after all ;)
remember, every input has 2 hal pins, -in, and -in-not, so the lists get long fsat
What is the default I/O configuration? LVTTL?
5v tolerant inputs, 3.3V drive on outputs
another thing I plan to allow, is input HAL pins even for physical pins that are associated with some function like stepgen or encoder
so you can halmeter/halscope them
I'd make the mode parameter also control inversion, so you don't need the -in and -in-not construct
well, the canonical digital input has -in and -in-not
I don't think it should
likewise, the canonical output has -out, and a param -invert
I know the parport does, and there it makes sense because you'll either have the input and an -invert parameter, or -in and -in-not
the 5i20 inputs would be the same
but for something that already has a mode control param, you can lump all the possible modes, including inversion, into the mode param
the mode control is NOT a HAL param
parts of it could be, but parts can't
you're talking about the new configs, I'm talking about what could be done with the current config
the part that says "do or do not export a -in pin, or a -out pin" needs to be known a driver load time, therefore can't be a hal param
in the current driver, that is hardcoded, so not an issue
for the current hardware config, I was considering adding two insmod parameters (sets actually, one per card)
one for each connector, that has the number of pins to use as output
that would define for each connector a new INPUT_SHIFT variable, and lets the driver construct the input and output masks
since my goal is to deprecate the old driver asap, I'm not going to comment on proposed changes to it ;-)
use a negative number to use upper pins as output (otherwise they're on the lower numbered pins)
remember - the old driver is compatible with the various add-on cards, which is a plus for many people
you mean the 7i33 and such?
Oh oh, got a problem guys, with the probeing.
the new one will of course be compatible with those things - I'd be crazy not to support them
gene: what's up?
emc went away, long before it was done.
thats rather vague
It left the probe file, onnly about 19k in size, and it was supposed to go 4.3" in x, but died at 0.050"
lemme look at the logs.
slight mistake, it went .140" out of 4.3"
died as in "decided that was far enough", died as in "following error on joint NN", died as in "motor went the other way" ... ???
let out the magic smoke?
as in the screen where emc was running, now only has the shell that launched it.
any messages in the shell?
there should be something, either normal shutdown messages, or error messages or...
that is similar to what I have been seeing
emc exits with no apparent error
well, not quite, it left this behind in that shell:
/usr/bin/emc: line 565: 6037 Segmentation fault nice $EMC2_BIN_DIR/$EMCDISPLAY - ini $INIFILE $EMCDISPLAYARGS $EXTRA_ARGS
Shutting down and cleaning up EMC2...
EMC terminated with an error. You can find more information in the log files
as well as in the output of the shell comma nd 'dmesg' and in the terminal
segfault.... thats nasty
and also I think it's fixed
yeah, but the rest of the system seems to be ok yet...
I have not had it crash from a shell yet, maybe it's related
gene, what HW are you running on?
cradek, what was fixed?
amd xp1400, running at 1600, half a gig of ram IIRC, Mach Speed mobo.
I have been running from a shell with no spontaneous exits, but I also have updated to latest head the other day
totaly different that my machine, VIA MB 256MB ram, embedded graphics
[03:23:23] <cradek> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/src/emc/usr_intf/axis/extensions/emcmodule.cc.diff?r1=188.8.131.52;r2=184.108.40.206
this is a seeparate nvidia card, a gforce2 5200 IIRC, nv driver
gene: update emc
is it in adept now?
unfortunately it didn't get in the changelog, but a segmentation violation like that was fixed
yes the update is available
you know, this will break my program?
heh - so it will ;)
I'm not sure I understand just what you're probing
do you know of any photos online that I could look at to see the problem?
the new code throws an abort if the probe doesn't touch, and it will miss about 10% of the area, its a rifle floor plate, and not by any means rectangular
I'm not sure exactly what shape it would have, but can you cut out a piece of cardbard or something for the probe to hit around the edges? (then you'd have to filter out any numbers below $something as being misses)
So the only thing I can do is grab a sheet of brass, drill a hole in it to clear the latch on one end and the lock lip on the other, so the probe will hit a solid surface on every stroke.
wouldn't it normally hit the table anyway (assuming you have the depth for it)?
Cardboard isn't going to be strong enough.
the plate is locked in a vise on the table so there aren't any holddowns to worry about clearing.
DUCT TAPE FTW!
chuckle, light sidem or dark side?
light, so the probe stays put ;)
guarnteed to trigger
you guys never quite do you :)
you guys never quit do you :)
never quiet either
until i need a question answered...
never was a quiet quitter
I guess I'll have to rewrite the routine so it stays on the plate, but then I lose my edge references
there's another sneaky way to do it
put another switch in parallel with the probe switch, but attach it to the column so that it triggers when the quill gets to a certain depth (preferably adjustable)
an artificial "brass plate"
or "probe depth stop" :)
...attached with DUCT TAPE!
just add a thicker layer... that's adjustable!
or use electrical tape, it stretches.
SOB it does exist.... Optical Radar
Could you use hal to do the same thing? Use a limit block tied to the Z position to trip the probe input in software? Please note that I know very little about what I am suggesting.
yes, acutally you could
good idea :)
a comparator and an or block would do it
knows more than me... HAl the computer on space odesy
now this is making sense for a change
I'll let this run for now Steve, but keep that thought in mind in case this run can't be extrapolated and I have to go back to square one.
Is any of this in the user manual, if so what page?
there wouldn't be anything in the user manual about tricking the probing code, but there should be docs on the various HAL modules
ie, there is info on how to use the gun, but no directions on how to shoot yourself ;)
Thanks for the help, all. Good night.
SWPadnos: Has "shoting yourself"EVER needed instructions?
dunno, never checked
[04:23:49] <Jymmmm> http://www.ncrr.nih.gov/ncrrprog/btdir/OpticalSpect.asp
i'm noticeing my y motor is hiccuppinh backwards occasionally, but always goes to where it belongs, about half a turn backwards, no rythm to it.
on 20 TPI screws?
shitty unipolar steppers ?
or bad driver ?
has anyone else seen this in a step and repeat routine? yeah 20 tpi screws, xylotex kit
ok - running at 8x microstepping?
hmmm. that's 800 steps out of 32k/inch, or 0.025
do you have backlash compensation set?
looksabout right. it self correcting though
hmm. nothing jumps to (my) mind
me either, and if it wasn't for the odd noise, i'd plumb miss it
well, its running, and I suspect it can run without me, lights out as they say, good night all
about time for me as well
03jmelson 07TRUNK * 10emc2/src/hal/drivers/hal_ppmc.c: correct polarity of index signal
lewin1 is now known as lewing
03jmelson 07v2_1_branch * 10emc2/src/hal/drivers/hal_ppmc.c: correct polarity of index signal
Anyonw know of a manual step/dir pulse generator?
a 555 timer?
I can reverse the direction with a 555/556 too?
use a switch..
oh, duh! lol
I was thining some digital setting, but would need a uC for that *sigh* My mind is a terrible thing to waster.
I was thinking I needed a pulse train for the direction.
Jymmmmmmmmmmmm: I thought you were low on cash.
as that's what you always talk about.. when looking at stuff..
whats that, lack of funds?
In respect to *MY* machine, I REALLY REALLY hate the iea of just throwing money to hope that it'll resolve the issue, without knowing what's is causing the problem. If I got through al this toruble in making a PS then getting the geckos, etc and it turns out (as exmaple) to be a bad cable or crapping motors, I woul dhave wasted the resources in the wrong way.
looking at $700 to do this.
did you just pull that figure out of your ass?
you forget you get nickeld and dimed to death too
Geckos alone = $489.39
it's not like you need to order 3 just to see if that's the problem
ah well, I dont give a shit.. it's not my machine that's not moving.
anonimasu I have to build a $100 PS
Jymmmmmmmmmmmm: and it's absolutely impossible for you to borrow one?
I don't have a 50VDC PS
Call a local college and ask them nicely..
I still hae to order the gecko
Heh, "borrow" if I was in a class, sure.
Just some guy off the street wanting to "borrow" a PS to take home.
I guess your social skills suck.
ultimately it's your own problem and your machine..
if it's a bad cable/crapping motor why arent you making parts already?
Making parts on my machine you mean... I have been, but when I get 90 minutes into a job and it gets ruined 5 times, it's stating to cost in material.
03petev 07TRUNK * 10emc2/src/hal/components/at_pid.c:
-Initial attemp at an auto-tuning PID using a limit cycle to characterize
the process and determine the ultimate gain and period. The Ziegler-Nichols
method is then used to calculate PID from the ultimate gain and period.
Needs more evalutation and testing.
03petev 07TRUNK * 10emc2/src/ (Makefile Makefile.inc.in): -Added at_pid module.
emc freeze every time on this step: /usr/bin/halcmd -i /etc/emc2/sample-configs/sim/axis.ini -f /etc/emc2/sample-configs/sim/core_sim.hal
any idea ?
curiosity: using Comp, how do I read the time to use it in the hal component?
i dont know :)
for example the component siggen, I know that the sin output should be sin(wt) where w=f/2pi...or something like that, but how I get t?
zelva don't worry, we will wait someone who knows it :)
mybi you know solution for my problem ?
emc freeze every time on this step: /usr/bin/halcmd -i /etc/emc2/sample-configs/sim/axis.ini -f /etc/emc2/sample-configs/sim/core_sim.hal
xemet: siggen keeps track of the time internally, storing the needeed information in 'index' between invocations of the function
zelva: the most common reason for whole-system freezes at emc startup is that you have set BASE_PERIOD (or another thread time) too low
where can i see that ?
it is in the file "axis.ini", but if you haven't modified that file then that is not the problem.
no i didn't modify anything
what version of emc are you using? does one of the other example configs work properly (for instance tkemc.ini)?
no in every config same story
did you use any earlier version of emc?
you're using ubuntu 6.06 or something else?
no debian etch
have you run the latency test, or any other program using the realtime kernel?
in emc realtime scripts works too
when you say "emc freeze" do you mean that the whole system freezes, or just that emc doesn't start?
whole system freeze and only hard reboot helps
zelva: I don't have any more ideas right now. You could add some lines to the core_sim.hal file to find out where it dies
zelva: on two separate lines you would write: "loadusr -w echo Reached step X" and then "loadusr -w sleep 3" to print a message and then wait 3 seconds
Dallur: Hello, I made some progress, I got Axis up with your config, so thats a start, but there is some problems
if you see the message you know it's at a later point that it locks up.
ok ill try
You could download and try the Ubuntu 6.06 + EMC2 live CD -- if that doesn't work either, it points at the hardware
jepler: so, I could increment a counter every time a realtime function is executed
jepler: but I can't know the period of the threads before
you could use the 'halrun' script to interactively set up a small HAL environment -- there should be an example in one of the manuals
halrun work too
and ubuntu livecd works too
only problem is emc script
xemet: oh -- comp provides "fperiod" (period in seconds as floating point) and "period" (period in nanoseconds as long integer)
Martin_Lundstrom: great news :)
time for me to go to work .. bbl
Martin_Lundstrom: what problems are you having ?
jepler: so if I use period of fperiod in a comp component I can make calculations without worrying about the period of the thread, right?
xemet: consider the 'ddt' example at the very top of the documentation. ddt needs to know the time of the prior sample compared to this one, to compute the approximation of the derivative..
xemet: so, basically, yes
right, I'm seeing it now
Dallur: I changed to trivkins in kinematics.hal ...
thank you jepler
Dallur: comp.1.equal in line net AtPosition comp.1.equal => and2.7.in0 in thc is not valid
Martin_Lundstrom: that should be fine, if you don't need gantrykins it should not affect anything else
is any problems with emc2 2.1.4 and debian... becuse it's freeze when start loading core_sim.hal...
Martin_Lundstrom: equal is a new pin in comp, I don't remember if the patch is in head already
Dallur: a intresting
ah, jepler, a question. Do you remember my modifications in order to add the nurbs to gcode? it is not complete yet, but is it possible to include it in the development version of EMC2? I remember that your spline feature was added. If yes, what should I do?
Dallur: oneshot.0.time-left in line net PierceDelayTimer oneshot.0.time-left => pyvcp.Num-PierceDelayTimer in thc_vpc.hal makes problems and I removed it for now
xemet: prepare a patch
Martin_Lundstrom: I converted time-left from parameter to pin
Martin_Lundstrom: I need to send for a new THC/Breakout so I will probably have time to commit some of the changes during the week
xemet: cvs diff -u > file.patch
this will put your changes (differences from the CVS version) into a file
ah ok, I will try
Martin_Lundstrom: I will try to do it over the weekend, i'm pretty sure everything works, I'm just having hardware issues atm
Dallur: I also had problems with the initial tags in the pyvcp xml file so I had them deleted for now
Martin_Lundstrom: yup, I made some changes to pyvcp components to enable initial values to be set
alex: should I run this command from...? the emc2 head directory?
ok, will try
How would I found and get the equal pin for comp? I have the latest trunk from cvs
you can also run it from a subdir, if your changes are local to that dir
Martin_Lundstrom: I'm not sure if I have committed the change,
Dallur: It would be nice to be able to test
Martin_Lundstrom: Ill do it tonight :)
Dallur: are you at work?
Martin_Lundstrom: yup, but compiling :)
think I heard that word before
alex_joni: it means someone else pays for the coffee
well.. then I geuss that's teh reason I don't like it
I don't drink coffee
i hate myself and my stupid computer
Dallur: anyway, Im keen on helping out to test the config further.
Dallur: I stand by
Martin_Lundstrom: great, i appreciate the help
zelva: push it all onto the computer
no reason to hate yourself ;)
if this solve my problem, then i do it
Martin_Lundstrom: I have tested everything on my table, the only issues I am facing now is that the damn high frequency always gives me false estop/limit signals, going to try to replace the breakout board/THC but if that does not work I might just have to do some more config to ignore any signals while the arc is being started
Dallur: and this during no operation!?
Martin_Lundstrom: ? It just means that after initiating the arcON, and untill I receive an ArcOK I would ignore all limit/home/external estop signals
Dallur: HF problems... More shielding :)
Dallur: thats a nice feature!
Martin_Lundstrom: it's hard to shield HF
* alex_joni just had big issues with the same thing
Martin_Lundstrom: at this point it's impossible to do more shielding, all cabling is wire shielded, all is terminated at only one end, I have a brand new grounding system measuring <1ohm, I just built a custom aluminum housing for my computer and I run all my cabling inside the table(3mm steel)
on a TIG filler wire drive unit
I tried using my tig with AC and constant HF, no problems with that, but that plasma has a bad a$$ hf module
I also have ferrite beads on every single cable, at this point there isn't a single trick left in the book, so I'll try changing the THC/breakout and see if that helps
Dallur: there's still room for voodoo
alex_joni: capacitors on the other end of wires and ground em to ?
When there is HF involved, I pray ;)
alex_joni: any advice is greatly appreciated :)
Dallur: sorry.. I'm fresh out
woodo, I love it
I feel like i'm chasing ghosts :P
Dallur: is your torch wire shielded?
Martin_Lundstrom: it happens even with the hand torch, just cutting any material with a parallel cable connected to the breakout causes the problem
even when the parallel cable isn't hooked up to anything, and keep in mind that the parallel cable does not run the limit/estop signals, it's only for step/dir/enable so the breakout is a likely candidate
Dallur: but the torchwire works as a bad antenna, it helps to shield it
Martin_Lundstrom: I might try putting a sleeve over it to test it out
Dallur: this it what I did
Dallur: and it got better for me, I think it was the woodoo effect ;)
But I still think ignoring all signals while starting arcs and perhaps disabling motors might be a good idea :)
Dallur: its a great feature!
Dallur: So you think I should wait for a cvs check in from you before I continue?
probably, it would make things a whole lot easier for you
Dallur: Your work with the user interface is great!
Dallur: regarding your HF problems, some motherboards are more sensitive to HF problems
Dallur: you might wanna try with another computer
Martin_Lundstrom: great, but my problem is not the MB, I have addon pci parallel cards, and the issue is happing in the breakout for sure
Martin_Lundstrom: I have an opto-isolated breakout so the computer is rock stable, i'm just getting false signals from the breakout
bbl, going to a vendor meeting
Dallur: have fun
anyone know what this error mean?
rtapi/rtai_ulapi.c: In function 'rtapi_exit':
rtapi/rtai_ulapi.c:202: warning: implicit declaration of function 'test_bit'
rtapi/rtai_ulapi.c: In function 'rtapi_shmem_new':
rtapi/rtai_ulapi.c:456: warning: implicit declaration of function 'set_bit'
rtapi/rtai_ulapi.c: In function 'shmem_delete':
rtapi/rtai_ulapi.c:536: warning: implicit declaration of function 'clear_bit'
../lib/libemchal.so.0: undefined reference to `test_bit'
../lib/libemchal.so.0: undefined reference to `set_bit'
../lib/libemchal.so.0: undefined reference to `clear_bit'
collect2: ld returned 1 exit status
make: *** [../bin/halstreamer] Error 1
zelva, what type of computer are you compiling on?
SWPadnos: i386 I guess
but it's not ubuntu afaik
hmmm. the error tells me (maybe incorrectly) that bitops.h isn't right
zelva: you said debian etch?
zelva: can you pastebin the Makefile.inc ?
yes just a sec
i have to get it from cvs again...
because i am stupid and i deleted it :S
# Makefile.inc. Generated from Makefile.inc.in by configure.
# on Mon May 7 15:33:05 CEST 2007
ok i have makefile but it's too long
# Makefile.inc. Generated from Makefile.inc.in by configure.
# on Mon May 7 15:33:05 CEST 2007
doctor doctor, it hurts when I do this
where i can put Makefile ?
[13:43:37] <jepler> http://pastebin.ca
is a good place to show the contents of large files
[13:44:33] <zelva> http://pastebin.ca/475479
here is my Makefile.inc
and here is error log: http://pastebin.ca/475482
iam running out of ideas :(
can you post the output of configure too
just a sec
[13:50:18] <zelva> http://pastebin.ca/475485
checking for usability of asm/bitops.h... no
and what is this ?
that should be fine
what linux kernel are you running and what platform is it built for?
220.127.116.11 pentium classic
what emc version are you trying to build?
trunk or release branch?
huh... i dont know that... i used cvs -z5 -d:ext:email@example.com:/cvs co -rHEAD -dhead emc2
not sure what -rHEAD gives you but I assume it's the trunk
hm -- looks like the definition of test_bit were removed between 2.1 and TRUNK
.. by me .. revision 1.6
date: 2007/03/13 00:54:33; author: jepler; state: Exp; lines: +0 -350
remove stuff that is not used by rtapi
i don't know... this command found in http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Debian_Etch_Compile_EMC2
jepler: when I did the ppc, I'm sure I only had to do test_and_set, test_and_clear
cradek: yes but you didn't compile rtai_ulapi.c
and on ubuntu realtime we don't exercise that code, because <asm/bitops.h> is usable
zelva: you should probably try getting an older version of the file rtapi_bitops.h -- just a second and I will tell you the command to do that.
from the src/ directory inside emc, type this: cvs up -p -r1.5 rtapi/rtapi_bitops.h > rtapi/rtapi_bitops.h
then try to "make" again
no error yet
now try it
freeze again :(
03jepler 07TRUNK * 10emc2/src/rtapi/rtapi_bitops.h: restore definitions of clear_bit, set_bit, test_bit because these are used in rtai_ulapi.c
06:47:58 <zelva> in emc realtime scripts works too
06:48:34 <jepler> when you say "emc freeze" do you mean that the whole system freezes, or just that emc doesn't start?
when i run scripts/emc and pick anyone config system freeze
06:48:54 <zelva> whole system freeze and only hard reboot helps
I also had that problem with kernel 2.6.19. kernel 2.6.17 works correctly.
I was using rtai3.5
but i need 2.6.19 kernel :(
zelva: any reason?
and my professor wants it
[14:14:36] <cradek> https://mail.rtai.org/pipermail/rtai/2007-March/016900.html
yes i saw this
it sounds like multiple people have had the same problem, whatever exactly it was
I doubt Alexander Kern's application was emc2, either
zelva: I don't think we can help you other than advise 2.6.17 until the rtai group figures this out
do you have config file for 2.6.17 kernel ?
this is what I am using on my two processor machines: http://timeguy.com/cradek-files/emc/config-2.6.17-rtai3.5
as i said... once emc2 works with 2.6.19 + rtai 3.5
now no more
compiling new kernel... ill report later
cradek 2 processors so thats like 64 bit then ?
no, I have dual P2 and P3 machines
the price on all mobo's just dropped
those new dual 64's just tookover it seems
and with quadcore comming its only going to get worse
worse for all of us better for some and most wont even notice type thing
jepler are you there? I've download the last trunk version and I'm trying to add the nurbs feature I had added to a old version, do you know if the function CHKF is still present?
ok, solved, it compiled now so the problem doesn't exist
alex are you therE?
so, I've the patch file for the nurbs now, could you explain me how to use it if I want to patch the original version I download from CVS?
the pathc file is here: http://www.pastebin.ca/475781
usually you go to the new checkout
and do "patch -p1 </path/to/path"
maybe the p1 is not the right number (then it will complain about missing files)
you can try p0 or p2, etc
ok, I will try
you also need to run in the right directory
it looks like you produced that diff from the src/emc subdirectory, so you'll need to run patch in the same directory
so...the passage are: 1) I download the trunk version from CVS
-pN means "strip off N levels of directories from each file named in the patch"
2) I go in the /src/emc/ directory of the trunk version
and run the command patch -p1 </path/to/m patch
and you'll probably need to hope you don't get merge errors
you'll probably want -p0
you can also use "patch --dry-run ..." to have it report whether it would succeed or fail
* JymmmEMC hands xemet a dead chicken and magic powder to sacrafice to the CVS Gawds.... May they have mercy on your code!
... [Something being hummed in latin goes here]
that's chicken BLOOD!!!
quid quid latine dictum sit, altum sigitur
so, if I get errors I've to change everything manually...right?
ds2: Hey dead chicken is just a recyclable container for the blood, mkay?
is there any possibility to include these changes in the CVS version one day, so I can have it right when I download the last developmente version?
xemet: patch will do everything it can, and save the rest to ".rej" (reject) files .. those you have to do manually
the gods want a virgin container!
ds2: Awww, I like alex_joni, dont want to sacrafice him.
JymmmEMC: thanks :D
alex_joni: your welcome =)
anyways.. 2 more weeks of beeing a bachelor
bride stressing over the arrangements?
I have nfc :D
LOL, good for you =)
NFC? is that like KFC but from Nebraska?
NFC = No Fscking CLue
getting close to lunch so...
alex_joni: still think you should just go to the judge ;)
* JymmmEMC hates the idea that a marriage has to be sanctioned by the state
* SWPadnos hates the idea that a marriage has to be sanctioned by a church
* skunkworks hates the idea that marriage has to be sanctioned by church
there is always common law ;)
shoot it, bubba
SWPadnos: Well, maybe not church, but at least "spiratual in nature"
comm law marriages are sanctioned by the state ;)
Not if Calif =)
I guess it depends on the state
there is the "church" popular with the BM folks
Ah, too popular now, more of a fad than anything else.
xemet: oh, you will also have to copy the new file (nurbs_additional_functions.cc) from the old directory to the new one..
I've been reading Marriss's white paper and he mentions amps per phase... how many phases's are then in a stepper?
4 IIRC is common
jepler: do you know if the CHKF macro is still in the interp_internal.hh?
when I downloaded the last trunk and I run diff from the version I had modified (maybe about two month ago) it found that I had to add the CHkF macro to that file...however, I added it and it worked
there are two phases in a bipolar stepper
which is the only kind the geckos will drive
how are phases properly counted?
(you can also have 5-phase bipolar, but geckos do't drive those, and they're pretty uncommon)
Can they both be driven at the same time, requiring double the amperage?
ds2, hmmm - I'm not sure how to phrase that :)
thought it was the number of coils?
JymmmEMC, yes, it's possible for both to be driver, but I think Mariss says that the gecko will never drive more than 0.6* the total (I don't remember where that info is)
ds2, well, I'd say it's the number of independently excitable coils
SWPadnos: I do, jsut a sec...
or number of independantly driven coils (to seperate it from a pole)
you can have an 8-wire motor, but it makes no sense to power one half winding separately from the other half of that winding
[18:32:30] <JymmmEMC> http://www.geckodrive.com/photos/Step_motor_basics.pdf
ah that makes sense for a bipolar stepper
JymmmEMC, that 0.6x figure is for geckos, not steppers in general
ds2, even fora unipolar: if you escite A and A' simultaneously, they cancel out (and that would be a 4-phase stepper anyway)
SWPadnos: I still dont quite understand how that could be. Looking for the passage now
arrgg power electronics
Ok, here it is...
The easiest factor in choosing a power supply is its current rating. The current rating of the supply
is based on your motor choice. The drive will always d raw less than 2/3 of the motor’s rated
current when it is parallel (or half-winding) connected and 1/3 of the motor’s rated current when it
is series (or full-winding) connected. That is to say, a 6 Amp / phase motor will require a 4 Amp
rated supply when parallel connected and a 2 Amp rated supply when series connected. If
multiple motors and drives are used, add the current requirements of each to arrive at the total
power supply current rating.
yeah - I'm not sure why his drives are that way, but it's a good thing :)
He says 2/3's in parallel, and 1/3 in series.
I wonder if that has anything todo with BEMF
I don' think so
SWPadnos: Tis okey, I speak typo
some one should include a hal component that calculates sin and cos
there is one example in the hal manual made with comp
I made it, but would be nice to have it available in the official release
also would be nice to have a component that calculates acos asin
What are folks using for a heat sink on geckos?
WooHoo I found a PS http://cgi.ebay.com/NEW-48-VDC-800-amp-Rectifier-Emersen-Network-Power_W0QQitemZ7544257610QQihZ017QQcategoryZ58288QQrdZ1QQssPageNameZWD1VQQcmdZViewItem
and an enclosure for it too... http://cgi.ebay.com/48-VDC-10-000-amp-Power-Distribution-Bay_W0QQitemZ290079300535QQihZ019QQcategoryZ58288QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
get a real power supply: http://cgi.ebay.com/DYNAPOWER-309-DC-Power-Supply-400-V-1200-Amp_W0QQitemZ190102204224QQihZ009QQcategoryZ71391QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
SWPadnos: Ah, the "portable" PS
Did you see the bank of caps in that thing?!
yeah, only 480KW, a baby
I've seen the insides of much larger ones at Dynapower :)
21x12 lathe... http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=3845991642&ssPageName=MERC_VI_RSCC_Pr4_PcY_BIN_Stores_IT&refitem=190102204224&itemcount=4&refwidgetloc=active_view_item&usedrule1=StoreCatToStoreCat&usedrule2=CrossSell_LogicX&refwidgettype=cross_promot_widget
powering a small city?
or a large plating vat
already CNC too
480KW to plate? I can see refining aluminum but...
Out generator at work can power a small city for a week.
is that one of those semi trailer generators that all the other colos use?
nope, it has it's own bldg though.
actually, that suply is a bit small for mining operations
see "HOW TO SPECIFY THE TRANSFORMER"
JymmmEMC: were you effected by the fire this weekend?
ds2: I'll say no, what fire?
there was a 3 alarm grass fire in south SJ on Sat
could smell it all the way in campbell
ds2: No, didn't even smell it
Under the "HOW TO SPECIFY THE TRANSFORMER" section at http://www.mcitransformer.com/i_notes.html
Is the last item on that formula: the recipical fo the SQRT of 2 ?
or 0.707 ?
which iirc is the RMS voltage
yes, the factor 1/sqrt(2) ~= 0.707 converts from sinewave peak voltage to RMS voltage
Ok, just wasn't sure about the grpahic is all =) TY
note that for DC bulk supplies, the derating factor is 1.8 (ie, you take the VA rating of the transformer and divide by 1.8 to get the wattage for the resulting DC supply)
So, by that formula, I'd need a 51.4VAC xmfr to produce 48VDC (unreg)
no, you need less AC voltage to get more DC voltage
48 * 0.707, or 48 / 1.414
err isnt that the other way around?
if you have 48 VAC, then you will get 48 * sqrt(2) VDC out of the supply
which is 48 * 1.414
yeah not / ;)
or 48 / 0.707
no, if you want to get 48 VDC, then the needed AC voltage is 48 * 0.707
that's disregarding the 1-1.5V loss from the rectifier, so 34-35VAC is what's needed
36VAC *1.414 = 50.9 VDC
33.94 VAC will give you 48V peak, but since you lose a little in the rectifier, you add one VAC to get 49.4 peak, then subtract the 1 to 1.5V from the rectifier, giving you 47.9 to 48.4 VDC
the 36 VAC xformer (much more common than 35,I bet) would probably be fine, unless there's a hard limit ("absolute maximum rating") that would be exceeded
Those motors will turn as smoothly as servos. Use about a 48VDC power
supply, a 1/4W 33K current set resistor and adjust the trimpot +/-
1/8 turn for min vibration (per datasheet).
that's from Marriss.
I'd say that 49-ish is "about 48" ...
if you're talking about geckos, then the higher voltage isn't a problem
In marriss' whitepaper he says vDC * .8 = xmfr secodary VAC
my bad... wrong citer
well anyways.. .how's everyone's day going?
i had a mechanical question.. i was looking in mcmaster-carr's catalog and they list several types of acme thread.. i need to span about 5 feet with the threaded rod on my 24x48" machine i am building... would i want to use 1/2" or maybe bigger around the 1" range like a bridgeport would use?
* was going to go ballscrews, but i am going to hold off on that untill the machine starts making me some money...
5 feet is a 1" screw
also if this is going into a CNC you're probably not going to make money using a leadscrew
i'm not talkin about millions of dollars here...
do you mean because of the precision?
er uh backlash?
and.. screwing up parts?
precision doesn't factor in
that 5ft/1" figure is a result of how much the screw bends (related to modulus of elasticity for steel), right?
unless you have cradles for the screw
so in theory, he could go with a thinner carbide screw if one existed and was cheap enough?
and even then you want a 1"
well, no not really
carbide may shatter
carbide is expensive too
really damn expensive
but from the bending standpoint, that would be true?
it will only span about 4'4" maybe...
and a thinner screw is going to deflect more
ignoring all the other "issues" w/carbide
the rest is for a gear or whatever on the outside of the frame
if we pretend we are in an alternate universe where no other factor at all existed
including any sort of load
carbide is really expensive
toastydeath: lol, "because it's cool"
they do have titanium leadscrews....
i'd still want a cast iron leadscrew
*nod* just verifying my understanding of things as carbide is the only other thing I know that is stiffer
carbide also ""snaps" in half easy
or a tool steel
I think the screws are out of steel.. of some sort
thought all steels have the same stiffness, just tool steel will bounce back?
type in leadscrew
no, steels vary wildly
I'm too lazy..
some steel is spongy some is brittle
there are over a hundred different kinds of tool steels
ds2: it depends on what kind.. what kind of mixture
also how it's hardened..
[21:01:07] <ohiopctechDOTcom> http://www.mcmaster.com/
just type in threaded rod
i'm not sure i would buy acme threaded rod if i was going to build a machine
i would buy a leadscrew.
anonimasu: so you are saying if I take a HSS boring bar and anneal it; it will deflect more? (let's assume I am boring a bar of wax)
hss doesn't anneal like other steel alloys
it depends on your alloy..
toastydeath: there are steels you can harden yourself..
via heating and dipping them in oil/coal powder/water
uh, i can anneal most steels by myself?
I know you can soften HSS... done it before
* ohiopctechDOTcom is checking out the june ed. playboy mag... fresh in from the airlmail stork
hss requires a much higher temperature than other steels to get it to begin annealing
it's not really representative of the process
and it is a pain to reharden
really wish that JC materials class was a multiquarter thing... it was fun
we were annealing all sorts of stuff; then doing tensile and compression tests
all i know is i am going to build a preloaded but and some acme threaded rod... and the machine is going to move. then from there i am going to tweak the design. but that's what i'm going to use to begin the project and get it functional
50ton hydraulic press + coin = nice toy
ohiopctechdotcom: what are you doing
well, that'll work..
i was just curious on the span/size combo i should use
I guess designing a anti backslash nut isnt impossible..
my boss told me at least 1"
automatic backlash eliminators use geared nuts
i thought i could use 1/2" still after he said that
toastydeath: geared nuts?
a pair of geared nuts and a differential
toastydeath: building a 24x48" gantry table
how long span did you say?
sitting in a pillow block
oh... forget it..
that sounds kinda high-tech toastydeath
that's more hightech then I've ever seen
heard there are plastic nuts that's good for backlash, turcite(?)
that's what they used back in the 30's through the like, 80's
it will span about 4.5 feet
underneath the table
turcite isn't really a plastic
you need a pretty thick screw to do that..
it's more for restoring ways
there is something injectable to cast a nut
turcite or moglice
moglice is it
I think a preloaded set of nuts would do pretty well
anyone have a 48" of travel ballscrew they wanna sell cheap?
contouring on a leadscrew is always going to suck, even with backlash eliminators
they're designed for climb milling, not curves
i realize the impulse to get it up and running is strong
toastydeath: everyone dosent have a shotload of cash..
well, i could buy them
and with a limited budget you need to define what, exactly, you can accomplish
ohiopctechDOTcom, I have a bridgeport ballscrew (and power feed) I can sell you :)
cheap machines will work just fine..
i want to make the design right before i dump the money into it
would it be cheaper to go with a double acting hydraulic cylinder setup to avoid the whole lead screw issue?
so you do not get upset when problems come out
actaully, it's not a ballscrew, it's an acme
ds2: huydralic control is a mess.
but it is 1-3/8 diameter, I believe
and i am having other people do the frame work so i want to get it working while they are still motivated
and then i can do the fine details after they are done
though it works better with really big cylinders and heavy workpices..
getting effective hydraulic control costs more than ballscrews
I'll sell you a valve package..
3k eur ;)
in fact, if anyone would like to buy a set of Bridgeport X+Y acme screws, with nuts, yoke, and X power feed, let me know :)
the most expensive part of hydraulic would be the valves and the controller for them
if you wanted to do hydro stuff, your best bet would be getting a retardedly large used copy mill
*to interface with a pc
SWPadnos: do they meet you under some bridge, afterdark and have to pay in small unmarked bills? =)
then pull all crap off it
and hook it up to your control
ds2, hmmm. no, I'll accept PayPal ;)
but only paypal FROM UNDER A BRIDGE
oh, of course
i plan on at least polishing the acme thread a little bit
even if you account for the servo motor and stuff, hydraulic is that much more?
do not TOUCH the thread
*if it's not to awful bad
do not TOUCH the thread
I was going to say that, but a little nicer ;)
i was just looking at a small chunk at work... and it was a tad bit rough
you could lap it
with the nut you intend to use
yes.. thats what i meant
ds2: it is
ds2: a valve for pwm control is a few thousand dollars..
and they arent really fast either..
i could make one with a knee mill and a servo motor and a small pile of stock... maybe a pivot pin and a cotter pin...
ohiopctechdotcom: every time you touch one spot of the screw, it will throw it off a little bit in that one spot
like, if it has crap on it
that one spot will always have an innacuracy
(the knee mill is only for machining) * not used for actual pwm hydro valve
something like 0.10sec delay..
if I remember my datasheets right
so i'm definately not going to power the new machine with pwm hydraulic cylinders
i guess you could try to grind a ballscrew
you'd have to make some fixturing though.
do you have a lathe
it could just be 2 magnetic blocks mounted on the surfac grinder at an angle.. with the acme thread on one side and the new ground thread on the other side
*at the same centerline
we have a 12x31 mag. chuck i think on a thompson
we do have a tool post grinder for a lathe though
now we're getting somewhere
why? maybe they used to have a lathe
you could assemble a machine to grind threads
then, grind a ballscrew
and tear aprt 25 ball bearings to make the nut
and then sell ground ballscrews on ebay for cheap
and never have to work again!
or if you just had a lathe.
you could just order the ball bearings
you'd also have to make a crush dresser
just grind some bb's ;)
the problem with trying to make your own bearings is that it's very hard
everything about it screams "maybe this is something we should just buy, dudes"
lol how do you figure nah
grinding round stuff down to 0.001+/- is easy
grinding round is easy
is not so much
heh, I mean spherical..
and .001 isn't good enough for a bearing
what apparatus do you have that produces ground spheres
on all sides
I just roll them around by hand..
then I use my caliper to measure..
and it goes to half a tenth.
no. it goes down to 0.01mm..
stick it in a E field with a reference electrode that encompasses it and measure the capacitance
then how are you grinding to .001
toastydeath: I just do ok?
anyway to make an actual bearing
seem recall the wood working guys put it on a vibrating abrasive
bearings are rolled
in ball mills
or maybe 3 abrasive balls working in a manner similar to a centerless grinder?
they're a set of plates with grooves cut in the face
[21:43:29] <anonimasu> http://en.wikipedia.org/wiki/Ball_bearing
rough balls get dropped in the top
AND IT GOES
working like a rolling mill.
they show it in "how it's made"
acme threads are rolled also sometimes
and screw machines roll threads
i don't see the link for how it's made
screw machines don't usually make screws
and they're certainly not making a screw longer than 8"
I couldnt find it
[21:46:19] <toastydeath> http://science.howstuffworks.com/question513.htm
03flo-h 07TRUNK * 10emc2/src/emc/task/emctaskmain.cc: include i18n error messages, needs review
toastydeath: sqeeze between your hands..
03petev 07TRUNK * 10emc2/src/hal/components/at_pid.c: -Make error output update every cycle during tuning, regardless of state.
AS HARD AS I CAN
bearings +/- .001 mm will result!
okay time to do other things now bye guys!!!!
I was just thinking, have any of you ever heard of an airline giving away empty seats for charity and using the donation as a tax writeoff ?
Been thinking since I have done a lot of work for airlines, perhaps I could convince some of them that it would be in their interest to donate empty seats to people in the OSS community so they could attend meetings and such
(thinking up creative ways to go to the fest :)
that sounds interesting :)
skunkworks: It would be really good marketing for the airline/alliance, plus the tax writeoff plus doing something good for the community
Ok, I might be able to get my hands on some switching PS's, but the thing is they're rated at 5.5A and I need 6A. Is there some way to parallel two switching PS's "safely"/"cleanly"?
they're open frame, not PC PS's
what did we tell you about switching PSUs?
actually 5.5A will be fine, you are most unlikely to ever pull over 3A or so anyway
By Marriss thingy, it's 6.2A (that's at 2/3rd's)
thats the theoritcal maximum
no room for margin/loss
I challenge you to pull more then 3A
Marriss white paper just says to use a cap on switching PS
IFF you can get all three motors to hit their knee point simultaeneously, you'll pull 6.2A ... thats each motor hitting peak current, at the knee point, ie just about to stall
even then, it will only be for a fleeting moment, and in practice, it never happens. I'll lay money that you can;t get the ammeter to go above 3A
a 5.5A will cruise it
you ordered your first Gecko yet?
gecko should draw enough current at idle to keep any switcher regulating shouldn't it?
with a 3A motor on the far side, they pull typically about 100 to 200 mA
most modern switchers are happy even on no load
I dont really like regulated or switched psus for stepper drives a toroid and a cap is much better
a lot of surplus switchers are 20 years old
like the ones in my basement
mmm ... possibly, they were just about starting to be common 20 years ago
it's 2007, right?
toroids are great becaus they have HUGE overload capability
I keep forgetting
so you can get away with really small ones
I agree with your choice
but considering surplus electronics in the U.S., it's usually a lot easier to get a switcher
the trouble with switchers is they have a habit of folding back on the slightest peak
thus the capacitor
assuming its big enough ...
but .. if you are buying the capacitor .. why not just buy a rectifier and a toroid and be done
simple, cheap, effective
toroid is going to cost the same as a gecko
he only needs 300VA at most
that's because you are over there in fancy-pants euroland
well it would cost even less to me
thats because I have toroid winding machines :)
I have a box full of toroids in my basement, I think they are 2x20v
perfect, send one to jymm
I expect it will still take over a year for him to wire it up though. the current "shall I buy a transformer" thread is at least 3 weeks old ... I fear its still in its infancy too
I keep thinking about putting them on ebay
I can't talk about that, I have projects from 1985
well yeah,. me too
I've been "working" on my mill since 2001
hmm If I had projects from '85 they would probably be made out of lego :P
I have electronics projects that will never happen as technology has passed them by
lego is too expensive
hell I have projects with valves in them!
that doesn't count
valves are fun
dont start me ...
I have a 2C39 on 1.3ghz
or if you call them by their proper name, tubes :)
I have another project that has a 4CX1000 in it ....
eric_u: they are still good for an amp though ;)
know what that is?
never heard of it
1000W VHF tetrode ...
I have a project using core memory..
and thats just the driver stage :)
core memory would be cool
the main amp is a 3CX15000 ... 15kw anode dissipation :) ...
good night guys
my oscilloscope uses nuvistors in its input stage http://en.wikipedia.org/wiki/Nuvistor
Ive used those .. in the head amps of a video recorder
two of those are the only tube/valve in the whole thing
Ampex 'Quad' VR1000
wait .. yo need to see this ... I'll find one
[23:47:14] <robin_sz> http://www.lionlmb.org/quad/vr2000_2.jpg
that brings back memories
4 track audio?
that was my first guess
with 4 heads
didn't know why they would need 4 tracks
you got 15 lines of each head
so 20 rotations per field roughly
the head ran in an air bearing
I had no idea when they invented air bearings
it's been a while apparently
we had two compressors on the roof, plus each machine had its own backup compressor in the back
no pressure is bad, the guys in the lab next door have a backup system for theirs and lock them down at night
the trick was to align all 4 heads to match .. or you got 'banding' on the picture
and edits .. oh edits were fun :) ...
want to make an edit, easy. find a spot on the tape ... cover it with a little magentic stainless steel dust and alcohol, now you can see the referenc pulse adnthe video tracks ...
slice it in *just* the right place with a razor blade ...
get your other bit of tape and a bit of special aluminium sticky tape ...
now thats REAL editing :)