nicksydney has quit [Remote host closed the connection]
<wpwrak>
there you have the whole range of gcode - from the trivial set of a handful commands things like kicad use to writing entire programs
<wpwrak>
DocScrutinizer05: a very different question: how is the pictures quality coming along ? :)
panda|x201 has joined #qi-hardware
arielenter1 has quit [Quit: Leaving.]
nicksydney has joined #qi-hardware
<DocScrutinizer05>
I think there must be a nazillion of tests available in the intarwebs written by people more savvy about picture qualtity than I am.
<DocScrutinizer05>
bazillion* :-S
<DocScrutinizer05>
whitequark: you never coded in COBOL, eh?
<DocScrutinizer05>
you know that COBOL compiler creates code that stores reurn addr at and of subroutine code as an instruction?
<DocScrutinizer05>
of course COBOL also has calculated GOSUBs
<whitequark>
DocScrutinizer05: COBOL no, but I've did a little bit of PDP-7, it's the same there
<whitequark>
don't ask why.
<DocScrutinizer05>
anyway that weird method of storing return addr un the subroutine instead of a stack makes recursive or reentrant programming quite interesting ;-)
<DocScrutinizer05>
as far as sniffing goes without any wincrap to run fujicrap on
<roh>
well. not much to get around that.
<roh>
throw-away vm... as usual.
<DocScrutinizer05>
no such vm available
<roh>
then i wish you fun decompiling that .net pile
<DocScrutinizer05>
my last 2 PCs explicitly came without any windoze
<roh>
i wasnt suggesting buying one.
<DocScrutinizer05>
could try wine
<DocScrutinizer05>
oooh wait, I *might* have a working VM with installed XP somewhere
<DocScrutinizer05>
or maybe I simply should check TPB
<DocScrutinizer05>
and make sure my LAN is ready to cope with attacks from inside
<DocScrutinizer05>
hmm, indeed
<DocScrutinizer05>
-rw-r--r-- 2 jr users 8684 29. Sep 2008 Windows XP Professional.nvram
<DocScrutinizer05>
-rw-r--r-- 2 jr users 2182938624 3. Jun 2012 Windows XP Professional.vmdk
<DocScrutinizer05>
-rw-r--r-- 2 jr users 0 4. Sep 2008 Windows XP Professional.vmsd
<DocScrutinizer05>
-rwxr-xr-x 2 jr users 1491 29. Sep 2008 Windows XP Professional.vmx
<DocScrutinizer05>
-rw------- 2 jr users 278 29. Sep 2008 Windows XP Professional.vmxf
<DocScrutinizer05>
alas all it creates is a 0.05s bluescreen and the known friggin startup menu "protected mode", "last known good config", "normal" etc
<wpwrak>
(finalkey) hmm, seems unclear how it works. does it snoop the keyboard ? or are you expected to tell it via its serial console what password to send next ?
<wpwrak>
(mooltipass, design #1) ah, so they stole anelok's look and feel :)
derRichard has left #qi-hardware [#qi-hardware]
dos1 has joined #qi-hardware
rz2k has quit []
xiangfu has quit [Ping timeout: 245 seconds]
qi-bot has joined #qi-hardware
rz2k has joined #qi-hardware
<wpwrak>
hmm. and now it also killed an endmill. apparently because of a transmission error that sent the mill on an incorrect course or some screw-up in the controller. that was also the second to last such endmill :-(
pcercuei has joined #qi-hardware
lekernel has quit [Ping timeout: 240 seconds]
lekernel has joined #qi-hardware
xiangfu has joined #qi-hardware
<DocScrutinizer05>
that's unfortunate
<DocScrutinizer05>
shouldn't the mill motor be mounted with >dehnungsmessstreifen< which instantly cut power to the stepper motors (or otherwise stop X/Y/Z movement) as soon as force exceeds a certain threshold?
<wpwrak>
it did eventually stop. but it was already too late. the bit isn't very strong, some 0.8 mm diameter.
<DocScrutinizer05>
hmm, the question is how sensible such sensor and how elastic the bit in comparison
<wpwrak>
i did some more experiments. it happened again. something is wrong with the toolpath. the weird bit is that it didn't really change ...
<DocScrutinizer05>
and how much inertia in steppers
<wpwrak>
dunno how it senses an overload. of course, if the load is strong enough to snap the bit, then it means that it's still within what the mill itself can handle. so in a way it's correct not to stop ...
<wpwrak>
(the mill doesn't know the strength of the bit)
<DocScrutinizer05>
sure
<DocScrutinizer05>
won't "work2 that way
<DocScrutinizer05>
I think dehnungsmessstreifen is the way to go
<wpwrak>
i'm not going to redesign my mill ;-)
<DocScrutinizer05>
well, at least I know how I'm going to design mine now ;-)
<wpwrak>
now ... i wonder what went so horribly wrong with my toolpaths ...
<DocScrutinizer05>
btw the redesign for dehnungsmessstreifen is usually exactly zilch
<DocScrutinizer05>
in your case however it actually could mean to weaken some lever or something so it has a defined location for elastic deformation
<wpwrak>
compared to, say, building a 40 GW matter-antimatter reactor, i'm sure it's pretty easy ;-)
<DocScrutinizer05>
meh
<DocScrutinizer05>
do you know at all how dehnungsmessstreifen usually are mounted?
<wpwrak>
no, but i'm reasonably sure there are no easy places to put them in my mill :)
<DocScrutinizer05>
that's the nice part about those critters, you simply glue them to the surface of some steel that deforms a few nanometers
<wpwrak>
in any case, there would be no reference data for setting a threshold
<wpwrak>
i would have to literally break dozens of bits to find the limits
<DocScrutinizer05>
or concrete or whatever you like
<DocScrutinizer05>
and no, you simply had to apply a defined force radially to the axis of the tool and calibrate the DMS output
<DocScrutinizer05>
the rest is math and you setting the upper limit you want for a certain tool
<wpwrak>
yes, but it's still need to determine the maximum allowed load
<wpwrak>
and besides, the things that deform most likely are the wires. and i doubt the dehnungsmesstreifen would survive doing around a wheel :)
<DocScrutinizer05>
well, yes, you got a storage scope, to watch the sawtooth output of the DMS while breaking ONE tool
<wpwrak>
also, don't forget that vibrations and load changes are normal. so you need to factor in all that, too.
<DocScrutinizer05>
yeah, I see you're in arguing mood onnce more
<DocScrutinizer05>
so nevermind, maybe eventually I show you a working design
<wpwrak>
looking forward to see the joergmill
<wpwrak>
... and then read you curse as it breaks bit after bit despite all the precautions ;-)
<DocScrutinizer05>
now you're starting trolling
<DocScrutinizer05>
find someone else please to bash, to compensate your frustration
<wpwrak>
the unfortunate truth is that bits do break. they're considered consumables for that reason. and a bad toolpath is pretty much guaranteed to kill your bits. also with "industrial" mills.
<DocScrutinizer05>
yeah, since even with industrial mills you can have an inapt programmer that runs the tool into the wall with full X speed
<DocScrutinizer05>
SCNR
<DocScrutinizer05>
cya
<wpwrak>
hehe ;-)
<wpwrak>
good that my mill isn't all that fast. so at least i don't have to worry about it punching through walls :)
* whitequark
switches to #qi-hw
<whitequark>
"13:09 < wpwrak> compared to, say, building a 40 GW matter-antimatter reactor, i'm sure it's pretty easy ;-)
<whitequark>
... yeah
<wpwrak>
oh. and i see how i broke the drill. the same problem. bit was too low. low enough to try to "drill" sideways. not the best idea with a 0.35 mm bit. i was very lucky that the other one survived, since it had also been slithering over the surface.
<wpwrak>
all of which makes me wonder what the heck is happening with those toolpaths ..
<whitequark>
that looks hugely complicated, not to mention my mill doesn't really allow to set spindle speed in predictable way...
jvalleroy has joined #qi-hardware
<wpwrak>
it's all trial and error ...
xiangfu has joined #qi-hardware
<whitequark>
nicksydney: (bitcoin) I'm using it moderately often
<DocScrutinizer05>
yet another probably easily installable sensor design consists of two lasers precisely monitoring the position and thus bending of the tool shaft itself
<DocScrutinizer05>
BOM for electronic components: ~20$
<DocScrutinizer05>
ask whitequark to share knowledge he acquired with his maglev
<DocScrutinizer05>
ooh, oops. That been with a hallsensor, not with lasers
<DocScrutinizer05>
iirc
<whitequark>
yeah hall
<lekernel>
DocScrutinizer05, what's that plasma thing?
<DocScrutinizer05>
really interesting footage starts at ~32:00
<whitequark>
DocScrutinizer05: it's not youtube
<DocScrutinizer05>
the tool in that "plasma" looks like a wrench or sth like that
<whitequark>
jwplayer rather, and video is hosted likely by same guy who owns the website
<DocScrutinizer05>
or a part of a diesel motor maybe
<DocScrutinizer05>
it looks like quite massive steel, which I thought was great to demonstrate a few particular aspects of DMS usage
<DocScrutinizer05>
with DMS you theoretically can probe the streching of brooklin bridge steel cables when a car drives over the bridge
<DocScrutinizer05>
I admit that for this usecase the instrumentation amp you need to read out the DMS will cost a multitaude of wpwrak's whole CNC mill
dos1 has quit [Quit: Kabum!]
<DocScrutinizer05>
anyway the longer I think about it the more I'm in favor of the laser probes directly controlling the tool bit shaft
<wpwrak>
an environment full of dust and often oil is certainly ideal for precision optical measurements ;-)
<DocScrutinizer05>
aaah right, good point
<DocScrutinizer05>
so a strictly mechanical sensor is the method of choice then. an Alu lever like L with a exactly fitting hole for the bit shaft in the foot of the L, and the vertical part equipped with DMS
<DocScrutinizer05>
or even make that brass, since steel on brass doesn't need any lubrification
<DocScrutinizer05>
as long as your tool bits all have same shaft diameter...
<DocScrutinizer05>
for the stronger ones you can simply swivel the sensor L out of the way
<DocScrutinizer05>
you could even auto-adjust the advance then, so the mill slows down when the material to mill is harder and takes longer to grind away
<DocScrutinizer05>
and the control software could define where to expect force load on the tooling bit so the advance simply slows down, and where no force is expected so any force detected would be considered a major error and causes immediate emergency stop
<DocScrutinizer05>
obviously the maximum advance speed has to be limited to allow for the advance done during maximum steep ramp-down of steppers (plural! sqrt(X^2 + Y^2) ) needs no longer and advance distance than the maximum the tool bit may get bent before damage kicks in
<DocScrutinizer05>
s/and /any /
<qi-bot>
DocScrutinizer05 meant: "obviously the maximum advance speed has to be limited to allow for the advance done during maximum steep ramp-down of steppers (plural! sqrt(X^2 + Y^2) ) needs no longer any advance distance than the maximum the tool bit may get bent before damage kicks in"
<whitequark>
or you could perform software verification
<whitequark>
which is what high-end CAM software is doing
<DocScrutinizer05>
you always do that
<whitequark>
no, I mean simulate the entire process in 3D
<whitequark>
with precise models of the mill and workpiece
<DocScrutinizer05>
the problems arrise when your verification isn't in line with reality
<whitequark>
sure, it's different approaches with different flaws
<DocScrutinizer05>
ooh, you say that software does NOT do any simulation of where's material and where's not?
<DocScrutinizer05>
pathetic
<whitequark>
not usually, no
<whitequark>
unless you pay a lot, and I mean a LOT of $$$, the software and machine just do as they're told
<DocScrutinizer05>
incredible
<whitequark>
to be fair, it is not an easy problem
<whitequark>
but I agree
<whitequark>
kinda sucks, especially when the result of an error is not a segfault but a broken mill :]
<DocScrutinizer05>
yes
<DocScrutinizer05>
exactly
<DocScrutinizer05>
and tbh the basic problem seems to be simpler than e.g. map navigation
<whitequark>
and open-source CNC software seems to be basically in its infancy
<whitequark>
EMC2, which is one of the best specimens, is basically a stepper driver with gcode interpreter
<DocScrutinizer05>
yeah, sounds about correct then
<wpwrak>
the 2nd toolpath also produces a different cut
<DocScrutinizer05>
seems like an approach with voxels and an intersection volume that mustn't exceed a certain threshold per time unit was pretty trivial to code, eh?
<wpwrak>
if both are considered equivalent, then there's something wrong with the model already
<DocScrutinizer05>
well, or sth wrong with the mechanical parameters like tool shape you imply
<wpwrak>
you'd also have to make sure you don't move along existing walls. else you could get very different results depending whether you just touch the wall or not.
<DocScrutinizer05>
obviously
<wpwrak>
precisely predicting tool loads may be just too hard. then there's also contact with things that have been cut loose. sometimes you end up milling through them. they shouldn't provide a lot of resistance, but they'd certainly create some load.
<DocScrutinizer05>
tool load is a pretty simple function of material grinded away per time unit, no?
<DocScrutinizer05>
first approach
<DocScrutinizer05>
and there's not supposed to be loose material usually
<wpwrak>
yes, as a first approximation
<DocScrutinizer05>
that's not already dust and removed by vacuum
<wpwrak>
then you have of course material characteristics. also, tools wear out with time, so the load increases. not sure how significant that effect is.
<DocScrutinizer05>
well, fine, then your software moves the model of the tool through the voxel space until it intersects with a number of populated/filled voxels that exceeds resp approaches the threshold constant you defined for allowed tool load, then the sw moves the tool to new position during the timespan your threshold been defined for, and marks the according voxels as empty/removed. Then go to start and do next loop
<DocScrutinizer05>
and you should stay well below any such parameter range like caused by tool wear when defining your threshold
<DocScrutinizer05>
sounds to me like a marginally modified floodfill algo
<DocScrutinizer05>
should I register a sw patent? ;-P
<DocScrutinizer05>
((...make sure you don't move along existing walls)) you simply implement that by switching your model of tool bit to a oversized one, oversized just as much as the guard distance you want to keep to your walls, while same time you set threshold to zero
<DocScrutinizer05>
so tool bit stays out of areas with material
<DocScrutinizer05>
when your tool moved to new starting point of an active vector, you switch back to the working toolshape and threshold, and follow the vector while observing the above sketched algorithm
<DocScrutinizer05>
given your voxels << accuracy of your CNC, I don't see how it could fail
<DocScrutinizer05>
and obviously it creates the boolen var about "load expected" / "no load expected" en passant, for controlling the sensor operation mode between speed-reduction and emergency-full-stop
<DocScrutinizer05>
for the sensor this would not only switch action taken when "overload" detected, but obviously also should reduce the threshold to ~0 for "no load expected" mode
<wpwrak>
phew. so much effort. it's probably a lot cheaper to just buy a hundred spare bits ;-)
<wpwrak>
(they're kinda expensive, USD 10-20 is not uncommon)
arielenter has joined #qi-hardware
wej has quit [Ping timeout: 252 seconds]
wej has joined #qi-hardware
<wpwrak>
btw, there are a lot of other more interesting improvements to mills. like a z sensor for automatic tool position calculation.
xiangfu has quit [Remote host closed the connection]
<wpwrak>
another things that would be useful for pcbs would be an automatic tool changer. that's hard, though.
<DocScrutinizer05>
hmm, seems feasible
<wpwrak>
well, you need another motor, for the mandril. if you have a fixed-size shaft then it's probably easier since you just have a screw to hold the bit in place
<DocScrutinizer05>
actually I'd think it can't be hard, given you use a small permanent magnet in your drill chuck
<wpwrak>
i think they use air. you may also need to blow it out since it may get a little stuck
<DocScrutinizer05>
I'd move the chuck over the bit shaft and insert it - should stay in chuck thanks to permanent magnet. Then move chuck plus bit to a fixed rubber clutch. press chuck into clutch and operate the drill motor to close chuck
<wpwrak>
you'd need a fairly precise spindle motor. but yes, why not.
<DocScrutinizer05>
then to fasten, lift a 2mm from clutch and speed up rotation, then contact clutch with spinning chuck to fasten by momentum
<wpwrak>
i smell burnt rubber ;-)
<DocScrutinizer05>
open chuck by reverse operation of whole moves and procedure
<wpwrak>
and how do you unto the permanent magnet ?
<wpwrak>
#s/unto/undo/
<DocScrutinizer05>
with a stronger electromagnet in toolbit magazine
<DocScrutinizer05>
thanks for asking, that saved me a half sentence on next post ;-)
<apelete>
larsc: tried to enable dma bit in MSC_CMDAT register with:
<wpwrak>
heh, i thought you may consider that approach :)
<wpwrak>
well, you could make it moveable and use the mill to shift it. that way you don't need one for each bit - or a huge monster that covers al the bits
<apelete>
larsc: will come back and look into it later
<wpwrak>
still, so much magnetism may not be great. especially since you also have magnetic dirt around your mill. (dunno where it comes from. just noticed recently when i used a magnetized tool in the area.)
<DocScrutinizer05>
even easier: use a bit holder magazine that holds the bits from side and is open to one side. Like a magnetig knife holder board in kitchen
<DocScrutinizer05>
attach bit: down and left. Detach bit: right and up
<wpwrak>
ah yes, good idea
<wpwrak>
and you don't need a magnetic magazine. it can just use springs.
<wpwrak>
you get the same kind also as tool holders and such
<DocScrutinizer05>
yeah, probably also "just works"
<DocScrutinizer05>
so, tool changer isn't THAT impossible, eh? :-D
<wpwrak>
i'm still a bit dubious about the mandril
<wpwrak>
but at least the holder looks quite feasible indeed
<DocScrutinizer05>
if you don't like the smell of rubber, you use a ring wrench alike thing to grip into a "nut" between your chuck and motor, and you rotate that thing to support/turn the motor axis while you keep the chuck fixed in some rubber clutch or other stuff. An old powertool may be the right motor for that
<DocScrutinizer05>
what's a mandril? isn't it a sort of monkey?
<DocScrutinizer05>
do you refer to the drill chuck or the tool bit shaft?
<wpwrak>
hmm, if there was an AAAAA battery size, it would just about be perfect for this :) AAAA may be close enough, though. 7.7 mm diameter while those bits have rings of 7.5 mm
<DocScrutinizer05>
should mount on your motor axis or into your existing mandril/chuck
<DocScrutinizer05>
I can't see how to fasten a screw automatically during tool change
<DocScrutinizer05>
o.O
<wpwrak>
i have that one on my dremel. the mechanical interface is quite different from my mill, though. e.g., the machine side of that mandril is threaded on the inside
<DocScrutinizer05>
weird scary design
<wpwrak>
;-)
<DocScrutinizer05>
yeah, that's how chucks usually are mounted
<DocScrutinizer05>
even for turn-right/turn-left powertools
<wpwrak>
you could grab it with a U-shaped structure on that part below the double ring
<DocScrutinizer05>
yeah, probably the best way to carefully insert the chuck base into such fixed wrench tool by moving mill sideways while carefully and slooooowly spinning the drill motor
<wpwrak>
but you still need to have very fine control of the spindle motor. also of torque because you don't want to tighten it more than you can loosen it after
<wpwrak>
or use a stepper motor for the spindle ;-)
<DocScrutinizer05>
then attach the rubber clutch from below by pressing down the whole mill head and then rotate the rubber clutch with a slow powertool motor
<wpwrak>
you and your rubber clutch ;-)
<wpwrak>
if you have that you don't need it anymore
<DocScrutinizer05>
(fine control) the motor spindle usually should run very easily, when operated by PWM controlled electric power
<wpwrak>
let's hope whitequark reads all this. he'll be very motivated for a while to make his mill work better ;-)
<wpwrak>
eventually he'll find a workflow he can live with and his enthusiasm for improving the tool itself will die down
<DocScrutinizer05>
the motor is too fast and too weak, the clutch seems pretty much exactly what you'd need
<wpwrak>
i kinda doubt they're designed for significant stress
<DocScrutinizer05>
they are designed for RC gasoline motors up to 10ccm
<DocScrutinizer05>
thoes need quite some force to turn them
<DocScrutinizer05>
those*
<DocScrutinizer05>
problem with that exact build of clutch: no space for the tool bit
<whitequark>
wpwrak: (for a while) hehehe, exactly
<DocScrutinizer05>
you probably want to build your own steel pipe and just insert the rubber ring
<whitequark>
reading the backlog now
<DocScrutinizer05>
or you find or modify a chuck so it allows another wrench on head nut to turn it
<DocScrutinizer05>
you could glue a nut "around" it, that the chuck fits into
<whitequark>
(automated tool changer) don't like the idea
<DocScrutinizer05>
then you can use a ring wrnch type of tool to turn the chuck nut
<whitequark>
sounds like waaay too much work and too fragile unless you're really good at designing this kind of things
<whitequark>
and I'm surely not
<whitequark>
the peak of my mechanical engineering capability is LEGO, and I'm not even very good at that
<whitequark>
I'm more interested in alternative heads. extrusor (yay, 3D printing, with way better mechanics than 99% of consumer shit), solder paste, maybe placement
<whitequark>
that's the right balance of complexity and benefit for me :)
<whitequark>
DocScrutinizer05: btw, check out OpenSCAM and HeeksCAD
<whitequark>
the former one is promising but not there yet
<whitequark>
the latter looks quite usable
<wpwrak>
has heekscad come back from the dead ?
<wpwrak>
he had abandoned it a while ago and told people to go to freecad instead
<DocScrutinizer05>
and who gave you all those weird ideas like solderpaste and 3D printer?? dang, must be a geek
<whitequark>
DocScrutinizer05: either you or wpwrak
<wpwrak>
(DMS video) veeery pedantic. but their soldering it some of the worst i've seen lately ;-)
<whitequark>
solderpaste was definitely yours
<DocScrutinizer05>
;-D
<whitequark>
anyway, I got the autotransformer
<DocScrutinizer05>
cool
<whitequark>
heading home now
<whitequark>
one thing, it's 127V and the control block is rated 110V-120V
<whitequark>
(would shouting "why can't they use switchmode power supplies?!" in the air again be too excessive?)
<wpwrak>
why can't those russians not order the 220 V version :)
<whitequark>
meh
<wpwrak>
after all, the voltage was indicating in big and clear hanzi ;-)
<whitequark>
it really wasn't, that's the issue
<whitequark>
they did not write anywhere that 110V and 220V control blocks are different!
<wpwrak>
small eel only connect. no grill large eel.
<whitequark>
it was sth like "110V or 220V mains input"
<whitequark>
what?
<wpwrak>
eel = electric fish. small eel = low voltage. etc.
<wpwrak>
well, there you see. 110V *OR* 220V. since you shipped to the US, they assumed you'd want 110V ...
<wpwrak>
my, those long noses are difficult ...
jekhor has joined #qi-hardware
<whitequark>
okay, plugged it in
<whitequark>
the magic smoke is still inside :p
<wpwrak>
have you tried turning it on ? :)
<whitequark>
sure
<whitequark>
at least the spindle board is still alive
<whitequark>
booting EMC2 now...
<whitequark>
ok, steppers don't step.
<whitequark>
ok, didn't connect the LPT cable. obviously.
<whitequark>
ooo it works
<whitequark>
of course the chinese docs are wrong in several places though
<DocScrutinizer05>
sure
<DocScrutinizer05>
127V is well in 5% range
<DocScrutinizer05>
and despite no switchmode PSU it still is pretty much capable of handline +-25% I'd guess
<DocScrutinizer05>
handling*
<DocScrutinizer05>
when they need e.g 12V= regulated, they will design the transformer secondary voltage before linear regulator to be 16V~ which in the end calculates to 22V=..
<DocScrutinizer05>
we don't have 110 and 220V since years now, it's 115V/230V officially
<DocScrutinizer05>
and for sure worste than +-5%
<DocScrutinizer05>
worse
<DocScrutinizer05>
generally any 115V device should cope with 130V without any problems
<whitequark>
it's 134V as measured by my multimeter actually
<whitequark>
I guess it's fine
<DocScrutinizer05>
yes
<DocScrutinizer05>
unless they dimensioned something insanely tight it shall be OK
<DocScrutinizer05>
worst case the internal trafo gets hot or the regulator chips bail out for overvoltage or, more likely, overtemperature
<DocScrutinizer05>
occasional manual control of temperature of whole thing can't hurt, until you got familiar with it
<CYB3R>
whitequark: nice one, but the only interface is lpt
* DocScrutinizer05
ponders to stamp "for the record, FYI" into subject of every spam mail and forward it to nsa@whitehouse.gov
<whitequark>
DocScrutinizer05: Stallman kinda does that
<whitequark>
well, it depends on whether you consider his email spam, and I'm close to doing that
<whitequark>
CYB3R: yes, I've kinda looked down on it too, but now I realized this is the 'lesser evil'
<whitequark>
see, LPT allows you to poke >8 pins with nanosecond precision, and it's the only easy way to do so from a PC
<whitequark>
USB is not realtime, hence, in order to make an USB CNC, you need a controller with an RTOS and an interpreter of some intermediate language
<wpwrak>
plus you can always do better L(
<wpwrak>
s/L(/:)/
<whitequark>
now, what would that intermediate language would be? basically the only rational choice is... g-code
<whitequark>
so you arrive to a CNC machine with a proprietary gcode interpreter written by some chinese guys
<whitequark>
I'd take LPT over that at any time of day
<wpwrak>
or lpt and a usb-to-lpt adapter with whitequark's great open source gcode interpreter :)
<whitequark>
wpwrak: yep
<whitequark>
for this particular machine it would be likely better to make a new control block
<whitequark>
because frankly its features kinda suck
<whitequark>
(or well, maybe I can poke the unpopulated stuff on boards. but dunno whether it'll work)
<whitequark>
but as a general solution, yeah, USB->LPT configurable adapter would be neat
<whitequark>
plus gcode is a general-purpose language and I've implemented two general purpose PLs already so... not hard for me :)
<wpwrak>
;-)
<wpwrak>
the usb-to-lpr dongle would also be a well-defined project by itself. a controller with all the power electronics could be a 2nd step
<whitequark>
I completely agree, it's much easier to control complexity by splitting the project into several simpler ones
<whitequark>
and test it and so on.
<whitequark>
even if it means having LPT interface somewhere :p
<wpwrak>
the lpt itf is also ubiquitous while your power electronics may not be
<wpwrak>
did you get a good set of bits as well ? endmills, drills ?
<whitequark>
wpwrak: nope, no bits
<whitequark>
that was an oversight in hindsight
<wpwrak>
how does you mill hold bits ? can you accommodate various shaft sizes or just a fixed one ?
<wpwrak>
what's typically useful to have it a large endmill, around 1/8" / 3 mm in diameter for coarse work, cases and similar. by being large it can also go deep.
<wpwrak>
for pcbs you'll want a smaller endmill, in the 1 mm range. that can go deep enough to cut a pcb but is also small enough for fine structures. the one i use is about 0.8 mm, so it's small enough also for most of the larger holes / slots
<wpwrak>
and then, if you want to drill holes for vias, you need a drill that's matched with a wire you use. i use an 13.5 mil drill with 30 AWG wire
<wpwrak>
the big endmill will be nearly indestructible so you only need 1-2 of them
<wpwrak>
the smaller one can be broken but it takes a bit of an effort. you'll want at least 5 of them.
<wpwrak>
the drills are very fragile but if you handle them with care (and especially avoid sideways cutting movements) they don't die easily. so about 5 of them will do to get started.
<wpwrak>
of course, if you can get a bit more, that's never a bad idea
<wpwrak>
if you want to mill pcbs directly you need finer endmills, 12 mil for coarse boards, less for finer boards. e.g., for a board with 0.5 mm qfn you'll want at most 10 mil
<wpwrak>
this is also a use where you should expect to break a lot of endmills ...
<whitequark>
wpwrak: it has a 3mm fixed-size collet
<wpwrak>
3 mm metric ? or 1/8" imperial ?
<wpwrak>
(given their obsession with x/2^y inch for lengths, i wonder why the "mil" is 1/1000 in and not 1/1024 in ...)