yea, think of an upside down U that is flattened out some and has a screw in the middle... the ends contact the work and the table
Connor: actually people have had issues with PCB's not being paticularly flat
when it comes to cutting with a v shaped bit anyway
I'm new to this, so, I haven't really looked.. but, any of the ones I've seen are pretty good.. maybe that's not enough.. ?
depends on the feature size i guess
the bit could also 'suck up' the board
the idea of making a picket jig is.. I can mill it perfectly flat, because, my table is a little off.. fine for 2D work.. but, PCB requires a little more than that.
theorb is now known as theorbtwo
I got a V bit from PreciseBits.com
along with a PCB Drill Set.
What forum are you guys talking about ?
[00:44:47] <JT-Shop> http://i47.photobucket.com/albums/f163/johnplctech/hunting/HPIM1566.jpg
[00:45:05] <JT-Shop> http://i47.photobucket.com/albums/f163/johnplctech/hunting/P1010063.jpg
oh, time for dinner
SWPadnos_ is now known as SWPadnos
Jymmm, you awake?
LawrenceG: Define awake?
hey... sure has been quiet in here
LawrenceG: Especially at night
LawrenceG: whats gin on?
just watchin the tube and waiting to go to sleep
how are the birds doing?
LawrenceG: SSDD... snoopin and poopin
LawrenceG: Havng issues with airflow *sigh*, will try something new tomorrow hopefully
carbon filter box still under development
how about a layer of pellets between two furnace filters..... add more thickness of pellets until airflow is close to your min flow required
davec is now known as Guest56437
davec_ is now known as Guest86038
one HW question: I have a stepper motor on bridge - is it better do have longer put the stepper driver close to motor and have short power lines and long signal lines or would be better to place the driver on frame and have short signal lines and long power lines?
one HW question: I have a stepper motor on bridge - is it better to put the stepper driver close to motor and have short power lines and long signal lines or would be better to place the driver on the frame and have short signal lines and long power lines?
depends on quality of ground wires, screening etc
I've wondered about that too. Professional machines seem to put the drives near the computer, and the motor and encoder leads are long and carefully grounded/shielded.
but that may be to keep the drives safe more than anything else
this where a scope can be useful when you get it wrong
unless your motors are big and your power is high, rest assured that thousands of hobbyists seem to do an adequate job :-)
stepper systems are very forgiving
a long wire into the drive acts as an antenna picking up electrical noise, screening it stops that
encoder wiring is where noise problems can really show up
my spindle vfd output noise got into my stepper inputs :)
I moved the vfd to the machine in that case
I have my vfd output wires and spindle encoder wires in the same conduit! it was originally that way (originally a resolver though)
are the encoder signals on differencial pairs
all our 'real' machines have drives away from the equipment.
rizo: is it you -- Drmn4ea@forum?
no, this is me: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Videos#Wire_cutting
but you both are intersted in emcrsh as i recall :)
i see :) ...i am reading the forum
We have a (non-emc) CNC customer that used integrated drives (drive on back of motor) for a while
turned out to be disaster as the drives are much more delicate than the motors
that seems to be an obviously bad idea, but if people make them...
anyone use PMAC stuff (Delta-Tau)?
(speaking of bad ideas)
They've got some smart engineers (that used to hang out in sci.eng.control) and good stuff
heh... it's good when it works. When it doesn't, it always seems to take me >1 day to figure out wtf it is breaking.
the motion controller of EMC was originally implemented on a Delta Tau DSP-based motion controller board
that's one of the reasons that the task controller, IO controller, and motion controller are separate programs - they used to run on separate CPUs (2 CPU cards in a VME or S100 cage, plus the DSP)
I have some VME ones, they have daughterboards for DSP and dual ported ram
or maybe one is just for the other 4 drives
the latest problem was some magically missing dual port ram configuration... things on either side worked fine
I'm hoping you aren't talking about DEC RL05's
heh. I am :)
(I know, wrong machine, but still :) )
wow, you must be ancient :)
I dont' know that I've ever seen a 5, lots of rl01/rl02 though
I think we had the RL05 attached to a PDP-11 in one of the physics labs in college
but I'm not sure
our high school computer club took one apart at a bank - they upgraded their system and let us get rid of the old one for them
(not just the disk pack, an entire minicomputer)
you want some strong motors, those things are beasrs
I have 3 pdp-11's running in production
hmm. now which combo is better? http://www.newegg.com/Product/ComboBundleDetails.aspx?ItemList=Combo.572765
the specs are a little different so it's hard to choose :)
get one of each
good idea... send me the one you dont' like
oh. good idea :)
ries_ is now known as ries
if only the server one had more RAM
buy two of first combos and you have double RAM!
oops, of second:)
buy ten of those, i guess that's enough for most things
and some extra CPUs to put on eBay
heh, my friend was buying things in combos, taking parts he need and selling not needed parts... sometimes with prices higher then he bought :)
ah. here's one with enough memory for a few months: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=270674861900
nobody will ever need more than 256GB RAM :)
do you think I can make it power a clock?
I have 30 people sharing a much smaller machine (64 GB)
that one might speed up compiles a little though :)
i guess so
we were working on a supercomputer mainboard that had 12 sockets for the 12 core Magny-Cours
I was wondering where the 8-socket motherboards (or quad+quad add-on) were
you could feed all the source files of most projects simultaneously to it
there were several for the older Opterons
I remember our server had 12 users on 8 meg Altos 1000 386 with unix on it
it sure could be wast
the new AMD server chipsets only support up to 4-P
oh yeah. I remember installing an NEC 386 machine with either Xenix or SCO or something, along with a 32-port serial card for all the users to connect their dumb terminals
a 386-33 or thereabouts
I don't think it had as much as 8M
I think git is an excellent example of how the computers we have today should seem to us
i guess you could serve quite a many serial terminals with a 4 cpu 256G machine
I'm always amazed at how quickly it does what it does
now that I have a almost always on connection is there any need for additional buildbots slaves?
IIRC the entire cluster had >1K Magny-Cours and 2K of the dual GPU hd5970's
and 1000A 480V power cord
I wonder if the Fermi architecture would be better for GPU computing
it sure seems fast as hell on my machine
(and glxgears is pretty fast too, 143000 Frames in 5s or ~28000 FPS once I disabled desktop effects)
Sandia and Los Alamos consider the gpu based clusters to be failures since nobody wants to have to rewrite all their software
there are C compilers for GPU
and fortran I think, maybe from PGI?
SWPadnos: according to HPL-GPU devs radeon cards are a lot faster
even if they are cheaper for an order of magnitude...
but nvidia has far better support for cuda/opencl now
yeah.. my very quick lookup of HPL says that the code was written for the AMD chips, so it's not surprising that it's slower on NVidia
[15:59:20] <L84Supper> http://www.top500.org/system/10587
[16:00:13] <L84Supper> http://www.top500.org/system/10484
same for #3
oh, that was #1 :)
they probably don't let them go home until their code is optimized for the GPU :p
then again with 4PFlops peak, they might just be settling for ~25% efficiency
any suggestions for spreading a layer of powder 50um deep across a plane (~25cm x 25cm) as quickly and cheaply as possible and with the least amount of dust?
toner cartridges with the built in drums seem to be a proven approach
I've been looking at the reprap, makerbot etc projects and the 3D print speeds can increase by > 10x by using inkjet + powder vs extruding melted plastics
srpski is now known as Dannyboy
what about premade sheets of the filler (like paper). lay a new sheet on, paint, repeat
srpski is now known as Dannyboy
psha_ is now known as psha
pcw_home: that system also works, but it fills in the hollow area with paper, that you can get out, think about printing a hollow sphere with the paper method
Water soluble binder (wash when done)
That was rather random, Pete
good night all
for the 3-D printer?
PCW: good idea, that would work unless its trapped inside a hollow like a printed pingpong ball
powder + inkjet printed activator has the same problem with trapped uncured powder inside a closed object
PCW: if you used a water soluble binder then the actual filler "paper" would have to be insoluble in water or your printed part would also dissolve
L84Supper, powder inside a ball would also be a problem - you still have to poke a hole or two in it and flush it out somehow
like you said :)
I mean weak water soluble binder for filler paper. non water soluble binder for printing
( or laser fuse)
the catch is using thermal printheads that use water as a vehicle
piezo heads are too problematic and closed source for DIY
if a $50 thermal printhead clogs you toss it out and replace, piezo heads can run into the several $Ks each
thermal heads are also closed source but easier to obtain , reverse engineer and far lower cost than piezo
Well forget inkjet and use a laser (then you can do metal as well)
the problems with laser for DIY are safety and cost
do not look directly into laser with remaining eye.
i guess somebody could eat the powder or get their fingers caught in the inkjet carriage
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2010-12-16.txt
Can you get enough volume out of the inkjet that you dont need a filler? (2 heads with 2 part fast hardening glop)
I'm just looking at ways to speed up the DIY 3D printers, the reprap types all extrude at glacial rates
or single thermal or UV/blue catalyzed hardening material
yeah, that also works but then there's the UV lamp safety issue or you can project a UV image from a DLP projector and cure an area of pixels vs one at a time
DLP costs + complex optics + UV safety issue
Bulk light (blue even) is OK since ink is only where needed
photpolymers are printable with piezo heads, not cheap thermal heads yet
Well thats just a paint problem...
and you can't print an overhang since you'll need to also print a support material
with microvalve yes, inkjet head no.... but now you also trade 2 $100 valves for 640 inkjet nozzles for $40
I think the powder system I looked at had a table that lowered for each slice and had a dust dump/scrape cycle
water + ethanol printed on gypsum and acrylic powders are pretty safe
We have a nice machine at work that bonds sand, then pours molten iron in....
yeah, how to spread a new layer of powder as quickly and without leaving a dust cloud
Cad model to cast iron part in one machine.
andypugh: yes, that what a DIY 3D printer based on thermal heads can do
Zcorp uses HP heads and are really slow, yet several times faster than the molten extruded plastic types
[21:32:32] <L84Supper> http://www.zcorp.com/en/Products/3D-Printers/spage.aspx
As long as you can seal up the working chamber, let the dustt fly
wow - that was painless. we had a sentry safe that we lost its combo. fax a notorized letter to sentry - pay $20 and recieve the combo through email.
if the dust comes into contact with the binder in the inkjet nozzles then you have problems :p
The obvious solution is to omit the support powder and simply print in zero gravity
the same problem with UV lamps and UV injet heads
yeah, no good DIY for that yet
the gravity insulator!
[21:36:31] <L84Supper> http://open3dp.me.washington.edu/
looking to improve on what they started
if the head shoots through an aperture with slight positive clean air pressure (not so much you blow the ink around) from the nozzle side you could keep the dust way from the head
tricky since there is only a ~40um layer of powder
and not moving the heads too fast and close, so no dust gets kicked up
skunkworks, that seems fairly creepy to me
not least because email is highly snoopable
shit yeah. got my 30 amp breaker wired up in the garage
PCW: I like the aperture idea! I can see a way to hold the head in a way to keep the air turbulence to a minimum as it scans
SWPadnos: a bit yes ;)
and also wipe the nozzle surface every pass or so
a single printead can scan a 10" x 10" area in 10 seconds, it would be nice to place a dust free powder layer in the same time or less
moving a 10" x 10" stage 40um that might be 10Kg max load is not a problem in a second
why are all airport announcers trained at jack-in-the-box? ( mumble mumble fiitzz pop mumble )
they are related to train station announcers
on some stations workers dispatchers talk to each other using loudspeakers :)
it's funny - once i was waken by loud voices talking about son (doughter) who goes to regional city into local university...
at first i thought that i've either sleeping or crazy :)
anyone have any good schematics for a speed controller for standard AC based router? I found out how to disable the onboard speed controller for my Bosch Colt, and want to hookup a computer controlled RPM setup.
by 'disable', you mean 'leave it turned all the way up'?
No. as in, Short all the leads on the Triac inside of it.
[22:22:13] <Connor> http://www.cnczone.com/forums/diy-cnc_router_table_machines/112658-super-pid_new_low-cost_router_speed_controller-26.html
I found info on the Super-PID, but, I think it's too expensive..
I can build my own no problem..
Let EMC2 handle the PID
that's not the issue..it's the hardware side to control the wave form of the AC.
In fact, EMC2 could handle the AC control.
You have a 1kHz servo thread, plenty fast enough to gate a 50Hz mains on and off
I do 60hz in US.
Didn't think on/off was best method for AC Control.
Does that not deform the AC Wave and cause strange things to happen?
I think that if you can think of a way to send a timing pulse to EMC2 at zero-crossing, then EMC2 can turn on a triac, then off again after a certain number of nS.
That's all your original Triac control did, I bet.
Yea. That's how it works.. it's the Zero-crossing part that's tricky. I would rather take a PWM signal out from EMC into a control board that handled all of that.
I DO have some 25Amp Solid State Relays coming..
that could be rapidly switched on/off I think.
[22:30:24] <andypugh> http://www.tmworld.com/article/321998-Not_your_father_s_power_tools.php
andypugh: most interesting on that link is "A battery charger built into a radio provides charged batteries and entertainment."
yea. Didn't find too much I understood in that article.
Yeah, I have seen quite a few of them.
Connor: It's talking about battery tools anyway, I think. Not very relevant.
Google will find any number of triac AC power control circuits. You need to get clever to go below 50% duty cycle.
I have a feeling that it ought to be easy to spot the point when current switches from flowing from neutral to live to the reverse. (diode, resistor, capacitor?) and then software might as well turn on and off the power electronics.
I may be over-zealous in my enthusiasm for moving as much analogue electronics as possible into softwae.
the SSR units I have coming ar FOTEK-SSR-25 DA units.. not sure they'll be useful for speed control.. I plan on using one for a PID based Heater control for a reflow oven.
tom3p: Attention Miazziz Hari please come to the courtesy counter. Those announcers? :p
If they are zero-crossing type then they are no use at all for speed control.
But they are great for preventing VFDs from tripping RCDs
RCD ? That like a Ground fault?
They key, is the Triac, and a Zero Crossing detector circuit.
[22:52:16] <Connor> http://multyremotes.com/ac-motor-speed-control.htm
I can build the Zero Crossing Detector circuit..I have the parts with exception to the Transformer.. but, I bet I have a power brick that's AC output that would do the same.
I have a basic config question. I am setting up a new machine, just have the controller together and trying to get a single motor to run on the bench. If I run StepConf Wizard and get to X Axis config and select test this axis should I be able to see step pulses on the parallel port output at that point?
I would have to build the rectifier out of individual parts.. as I don't have any bridges in stock.. no biggie.
andypugh: Look at that link I posted.. I think that's what I'm looking for.
hmm, I am not seeing any...
Connor: Aye, that looks good. Replace the microcontroller with EMC2 and a very simple HAL module, and the job is done.
or a ATmega 168/328
I am thinking that you want to control the speed via EMC2 and a PID, the more you can bring "inside" the computer the better.
I'm a little concerned they say to short all 3 pins on a Triac to disable it.. would it not be best to just short across it .. I.E. leave the base alone ?
triacs dont have a base
maybe base isn't the right term.
grommit: How are you testing?
I have an oscilloscope on the wires that would normally be plugged into my Geckos (320x).
looking at the schem, of D7, would it not be better to short the two that are in series with N, and leave the one going to ping 5 on the Opto alone ?
I also have a multimeter, but currently, no steps and 0v on the par port pin 2 (X step)....
I would expect a scope to work, but you could try adding a pull-up or pull-down just to be sure
between the scope probe and the signal gnd?
I am clutching at colour-banded straws here
I assume you are confident you have the base address right?
Nothing concerning in dmesg?
dmesg gives me: parport0: PC-style at 0x378, irq 7 [PCSPP]
I used 378 in the config screen
Sounds good to me, but I know very little
yeah, it isn't supposed to be hard at this point :-)