wej has quit [Read error: Connection reset by peer]
wej has joined #qi-hardware
dos1 has joined #qi-hardware
lekernel has joined #qi-hardware
michael_lee has joined #qi-hardware
fire has joined #qi-hardware
<DocScrutinizer05>
>>Shipping: $399.00<< WTF?
<whitequark>
DocScrutinizer05: doesn't matter, there are others selling the same machine with sane shipping price
<DocScrutinizer05>
0.05mm precision sounds good, if that's true
<DocScrutinizer05>
particularly spindle motor axial often is a problem
<DocScrutinizer05>
when the mill axis has 0.1mm room to move up when under pressure and come down again while slowly milling too deep into the workpiece, you got a pretty nasty problem
<DocScrutinizer05>
you *might* be able to cure this by simply rotating the whole contraption upside down
<DocScrutinizer05>
but obviously you mustn't have slackness in any other of the parts then
<DocScrutinizer05>
well, in any of the parts that get pushed up by mill engaging to the workpiece
<DocScrutinizer05>
the pricetag for that thing sounds reasonable, when it's not a flimsy scam
<DocScrutinizer05>
however please ask roh who seems to have some RL experience with those sorts of machines
<whitequark>
DocScrutinizer05: I've seen several people on the net who bought this machine
<whitequark>
the general consensus seems to be that its mechanics are up to the task
<DocScrutinizer05>
maybe a good idea to check by google what they are doing, and then asking them what they think about it
<DocScrutinizer05>
for an artist interested in milling tiny angel figurines from wood, the machine may be perfect while still completely unsuited for e.g. PCB milling
<DocScrutinizer05>
(not saying it actually _is_ not suited for PCB)
<whitequark>
main two complaints seem to be shitty wiring (I can cope with that), and the fact it requires a hardware parallel port
<DocScrutinizer05>
I don't know the machine
<whitequark>
(the controller is *incredibly* dumb)
<DocScrutinizer05>
no problem either :-P
<whitequark>
yup
<whitequark>
either buy an old PC or use a micro to reimplement half of the control logic
<DocScrutinizer05>
actually that's better than the controller being incredibly fubar
<DocScrutinizer05>
huh? use a usb-parallel adapter
<whitequark>
I suspect the controller is done entirely of 74HC series or something like that
<whitequark>
DocScrutinizer05: doesn't cut it. timings are off
<DocScrutinizer05>
ummm
<whitequark>
with HW LPT port it even requires Linux RT, the regular one won't work
<DocScrutinizer05>
errrr- ummmm
<DocScrutinizer05>
prolly a job for a beagleboard or sth like that then
<whitequark>
an atmega would suffice :p
<DocScrutinizer05>
yeah
<whitequark>
but, yes.
<whitequark>
it's not a problem for me either
<DocScrutinizer05>
or simply abuse an old injet printer's control logic ;-P
<DocScrutinizer05>
inkjet*
<whitequark>
that actually sounds much more complicated
<DocScrutinizer05>
yeah, it is
<DocScrutinizer05>
but it shows what grunt in MCU you need
<DocScrutinizer05>
I can't think how a USB is too slow
<whitequark>
I think it's more the fact that it doesn't have guaranteed latency that's a problem
<whitequark>
well, not with interrupt transfer, which USB-LPT adapters use
<whitequark>
btw:
<whitequark>
"You also need to be aware that there are not limit switches, and the system is open loop in design. The spindle speed control is manual, which is not too bad for a tool this small."
<DocScrutinizer05>
even when you control the steppers PWM by bitbanging the parallel port's IOs, it still should easily do the maybe 5kHz with a precision of 8bit timing
<whitequark>
I'm not sure how bad is that
<DocScrutinizer05>
aka phase modulation
<DocScrutinizer05>
and requirement for RT-linux is basically nonsense. The developer just never heard of kernel drivers
<DocScrutinizer05>
and high precision timers
<DocScrutinizer05>
and IRQ
<whitequark>
it uses linuxcnc, not its own software
<DocScrutinizer05>
yeah, and that got developed by a noob I guess
<DocScrutinizer05>
of course you can't implement a PWM phase controlled stepper controller as a userland process
<DocScrutinizer05>
but honestly, 480Mb/s USB2.0 speed should suffice to control a USB<->parallel converter pretty tightly
<whitequark>
that's really more about latency than bandwidth
<DocScrutinizer05>
latency is a function of bandwidth
<DocScrutinizer05>
and actually latency here is irrelevant unless this whole concept depends on meaningless IRQ events generated by the CNC-controller
<DocScrutinizer05>
you could introduce a delay in propagation of control signals from PC to parallel port, in the range of hours or days, and nothing would change
<whitequark>
ah, that's just the big stop button. so... no idea why it needs realtime
<DocScrutinizer05>
OT: Cree has really awesome LED products
<whitequark>
as well as several types of photoresist... I just want to go and properly test and document all the convenient (absolutely no screwing with toner) ways of making PCBs I know
<whitequark>
and figure out which ones work best
<DocScrutinizer05>
a white HighVoltage LED with Ufwd=23V
<whitequark>
huh, 23V
<DocScrutinizer05>
yeah! :-D
<whitequark>
do they have several junctions on the same crystal?
<whitequark>
or what?
<DocScrutinizer05>
a 10W 5*5mm LED
<DocScrutinizer05>
don't ask me how they do it, they say SC^3[TM] technology
<DocScrutinizer05>
I found a few really nice things at cree.com
<DocScrutinizer05>
the 10W LED is ~5EUR at digikey
<whitequark>
seems I could make it distributed (spatially) better than with lamps
<whitequark>
more even
<whitequark>
that strip eats 24W, at 30% efficiency it's 8W of UV... more than enough for quick exposure
pcercuei has quit [Quit: leaving]
<DocScrutinizer05>
my comment? generally you prefer a small concentrated light source for contact exposures of any kind
<DocScrutinizer05>
ideally you'd have uniform parallel beams of light, from a light source infinitely far away
<DocScrutinizer05>
the next best thing is a huge uniformly shining surface that has no minima and maxima
<DocScrutinizer05>
so you achieve uniform exposure across the complete PCB even when the distance between light source and PCB is relatively short, like in the sub-meter range
<whitequark>
DocScrutinizer05: well, my thoughts were that 300 leds evenly distributed over surface is better than 2-4 lamps
<DocScrutinizer05>
I'd dare to give a rule of thumb which is: the distance between light source and exposed object needs to be 10 times plus the size of the largest structure found in the lightsource itself
<DocScrutinizer05>
so when your LED's are spaced at 1 inch, you should move away the lamp at least 10 inch from the PCB
<whitequark>
makes sense
<DocScrutinizer05>
and of course your area of observation in the lamp is at least large enough so a point at edge of PCB "sees" the same when looking up, like a point in center of PCB
<DocScrutinizer05>
means your lamp area has to be significantly *larger* than the PCB
<whitequark>
yes, that was my issue with my previous lamp. it was pretty tiny, about 15cmx4cm
<whitequark>
maybe 20cm
<DocScrutinizer05>
or much much smaller ;-)
<whitequark>
mhm
<DocScrutinizer05>
anyway, the size of the complete lamp also counts when checking for "largest structure seen inside the light source"
<DocScrutinizer05>
unless the lamp#s size is way larger than the PCB
<DocScrutinizer05>
it's all about unifirm exposure aka brightness the lamp creates on PCB surface
<DocScrutinizer05>
and human eye is a pretty poor probing tool here, it doesn't notice */ 2
<DocScrutinizer05>
photo resistive damn sure does
<whitequark>
I wonder if I could verify it with the photoresist itself
<whitequark>
like, expose it for five seconds, then develop... will it shed uniformly or not
<DocScrutinizer05>
probably
<DocScrutinizer05>
yes
<DocScrutinizer05>
sounds feasible
<DocScrutinizer05>
I'm just thinking of 4 mirrors forming an endless mirroring of original lamp
<whitequark>
that sounds weird
<DocScrutinizer05>
but they need to be pretty good
<DocScrutinizer05>
nah, that's not weird, that's simply a 4 walls enclosing the PCB and lamp setup
<DocScrutinizer05>
so close to the lamp area from each of the 4 sides that when looking up to the lamp the PCB 'sees' and enless uniform area of one true and multiple mirrored lamps in all directions
<DocScrutinizer05>
ideally gapless
<whitequark>
thanks, I'll consider this
<DocScrutinizer05>
get 4 mirrors 30*80cm, build a square tube from them. then place your LEDs evenly spaced all over the one open end of the tube. Place your PCB into center of the plane of the other open end
<DocScrutinizer05>
sounds complex but actually is pretty trivial
<whitequark>
yeah I got it
<DocScrutinizer05>
except for the mirrors themselves which need to reflect UV light
<DocScrutinizer05>
usual mirrors don't
<whitequark>
metal foil?
<DocScrutinizer05>
yup
<whitequark>
eg thin Al foil
<DocScrutinizer05>
alu foild maybe
<whitequark>
over say plexiglas
<DocScrutinizer05>
:nod:
<whitequark>
afaik plexiglas is UV-transparent
<whitequark>
atleast it very much seemed to be transparent to me, seeing as the resist got exposed through it
<DocScrutinizer05>
why not expose the alu foil directly
<whitequark>
it's not mechanically stable
<DocScrutinizer05>
place it on top of some stable surface, like plexi or wood or steel
<DocScrutinizer05>
use sticky or glue to fix it
<whitequark>
ah, right, I could just flip the plexiglas base
<whitequark>
so the foil would reflect UV directly
<DocScrutinizer05>
yep
<whitequark>
as a bonus point it will shield whatever else there is in my room from UV
<whitequark>
not sure how much degradation would there be from occasional firing of ~10W lamp, but still
<whitequark>
right
<DocScrutinizer05>
s/chamber\tower
<DocScrutinizer05>
;-)
<DocScrutinizer05>
you could place a small hole on top between all the UV LEDs to peep inside, and a few red or maybe yellow LEDs for convenient illumination for human inspection
<DocScrutinizer05>
and when you anna go fancy, you can even have a small string pulling up a mask that covers the PCB and protects it from any ambient UV while handling
<DocScrutinizer05>
wanna*
<DocScrutinizer05>
ooh, nope, as that would pretty much spil the mirror cabinet#s properties
<DocScrutinizer05>
spoil*
<DocScrutinizer05>
you could use a black plastic foil sheet to cover the PCB, and you pull up and out of the tower this whole plastic sheet with a string, through a hole (same hole used for peeping)
<DocScrutinizer05>
sheet like "cloth"
<DocScrutinizer05>
easily pull it out through a tiny hole
<DocScrutinizer05>
since it folds up and slips through the hole
<wpwrak_>
it's always interesting what the body craves after a night of drinking. today, not just water but sugared water
<DocScrutinizer05>
then for 'calibrating' you prepare a mask that has ... "40" "20" "10" "5" ..., place it in your chamber and cover it by a black sheet that comes out under the morror so you can pull it away step by step. Then you pull the cover until 40 is seen and uncovered and expose for 40s, then pull cover wurther until 20 is seen and expose for 20s. Then develop and see which number looks best. Let's ay it's the "10" then your optimum exposure
<DocScrutinizer05>
time is 10+5+2+1
<DocScrutinizer05>
when your numbers been like 1 2 5 10 20 50 100 200 500
<DocScrutinizer05>
or you use 1 2 4 8 16 32, and expose "1" for one second, and everything N >1 for N/2 seconds
<wpwrak_>
whitequark: (cnc mill) looks nice. having the table split is a little unfriendly but the rest looks good
<wpwrak_>
that was about the second one. details are little hard to see for the first one
<wpwrak_>
DocScrutinizer05: USB imposes its frame timing. it's completely unsuitable for this type of basically GPIO control
lekernel has quit [Ping timeout: 272 seconds]
<wpwrak_>
(10 W LED) it seems to be more like 2 W for continuous use, 2.9 V * 700 mA. the 3 A seem to be just pulse current. e.g., for a camera flash
<DocScrutinizer05>
wpwrak_: I think that statement is a tad too oversimplified
<DocScrutinizer05>
re USB
<DocScrutinizer05>
it would largely depend on timing of that frames
lekernel has joined #qi-hardware
<DocScrutinizer05>
given those USB2parallel coverters easily handle GDI printers which often are no less stupid regarding in-device embedded intelligence and the amount of timing the host PC is supposed to provide for driving the GDI printer...
<DocScrutinizer05>
(10W) nope, 3A/10W is coninuous operation
<DocScrutinizer05>
700mA is just the test current
<DocScrutinizer05>
for which color, efficientcy, forward voltage etc is rated
<DocScrutinizer05>
you however need to take care about proper heatsink
<DocScrutinizer05>
and I'm a tad lost there since I wonder how much of the power you feed to the LED is converted to light and which fraction actually dissipated for heating up the chip
pcercuei has joined #qi-hardware
<wpwrak_>
(usb) frame timing is 1 ms at least up to full-speed. they have slightly finer granularity at higher speeds but i don't know how much finer. maybe 4x or so on high-speed.
<wpwrak_>
so this means that, e.g., if you do something like read a register (not the case in a cnc mill, but just to give an example), it takes 1-2 ms: 0-1 ms to send the read command and 1 ms to get the result.
<DocScrutinizer05>
on highspeed that should be faster
<wpwrak_>
yes, it's faster because you have, i think they're called microframes
<DocScrutinizer05>
the whole communication probably can be done streaming data words to the USB->parallel latch
<DocScrutinizer05>
and even when not, 4k/s new values appearing on parallel output seem fine grained enough a timimg to drive a stepper motor, even by PWM
<wpwrak_>
(microframe) it's even 8x, so 8 kHz
<DocScrutinizer05>
I think for driving stepper motors you need Ft of some 100Hz
<DocScrutinizer05>
so we have a headroom of at least factor 20 here
<DocScrutinizer05>
for multiplexing 3 motors via one 8bit (10bit) parallel
<wpwrak_>
lemme calculate ... 2500 mm/min, hmm, doesn't say what resolution. let's assume 0.05 / 16
<wpwrak_>
that would be 13.3 kHz. that may be close enough
<DocScrutinizer05>
where from you take that factor 16? for PWM?
<wpwrak_>
not sure if pulse timing also matters, though. it might. e.g., it's not just stop and go but you try to accelerate the motor gradually. so you'd vary the frequency.
<wpwrak_>
from "16 subdivided driving down"
<DocScrutinizer05>
that's however on a *completely* different scale
<wpwrak_>
and knowing that for example my mill has a 62 um resolution.
<DocScrutinizer05>
we're talking about that CNC3020 though, and that has precision of 0.05mm
<DocScrutinizer05>
I guess precision >= resolution here
<wpwrak_>
yeah, could be. 30 um would be 500 um/16
<wpwrak_>
dear china, please make a USD 500 mill like that ;-)
<wpwrak_>
(steps) yeah, not sure how that fractional steps vs. full steps thing works
<wpwrak_>
phew. 36 C. getting a little hot today ...
<wpwrak_>
may be the hottest day the second half of this year
michael_lee has quit [Quit: Ex-Chat]
<DocScrutinizer05>
the fractional step thing works by giving (e.g.) 20% of current to phase-1 and 80% to phase-2, so the axis magnet positions on 1/5 of the angle between that phase-1 and phase-2 position
<DocScrutinizer05>
the sequence is kinda like: 1; 1&2; 2; 2&-1; -1; -1&-2; -2; -2&1; 1...
<DocScrutinizer05>
you can position between any of those 8 full steps by applying fractions of 100% current to either of both phases
<DocScrutinizer05>
aka coils
<DocScrutinizer05>
aka electromagnets
<wpwrak_>
oh, even analog regulation. fancy.
<DocScrutinizer05>
PWM usually, yeah
<DocScrutinizer05>
obvously you usually got a multiple of 8 steps per 360° rotation on most stepper motors
<DocScrutinizer05>
up to 360 or more
<DocScrutinizer05>
for full speed you often even skip intermediate steps and do "full step" 1, 2, -1, -2, 1...
<DocScrutinizer05>
while the 8 step mode is calling the additional 4 steps half steps iirc
<DocScrutinizer05>
sorry it's like 10 years since I last looked into that stuff
<DocScrutinizer05>
>>After installing the Mach 3, on the desktop, there will occur 3 icon as follows: XXXXXX Notice: do not open them, please restart your PC at first (if you open the software not restart PC,it need uninstall fully and reinstall it) <<
<DocScrutinizer05>
LOL
<DocScrutinizer05>
see "Engine Configuration... Prts & Pins"
<DocScrutinizer05>
>>Kernel: 25000Hz<<
<DocScrutinizer05>
according to 2nd window "Ports & Pins" Mach3 doesn't do half steps, and particularly no step fractions
<DocScrutinizer05>
looks like lowest level interface for a pretty standards hardware stepper controller: direction and step signal
<DocScrutinizer05>
so the controller box of the 3020 has 3 stepper controller chips I'd guess
<DocScrutinizer05>
via USB according to what you wrote, we can make the motor do 8k steps per second
<DocScrutinizer05>
and that's it. the only problem with this limitation is that stepper motor can't go faster than that, and the 'noise' that's particularly relevant when moving two axis concurrently is determined by the jitter in that 8k timeframe
<DocScrutinizer05>
I can't see the motors mechanically following any such jitter
<DocScrutinizer05>
axis magnet's inertia is a pretty good lowpass filter
<DocScrutinizer05>
which also means the timing of the controller must not jitter so much that motor skips a step
<DocScrutinizer05>
all this can get handled by simply not moving the motors at maximum speed
<DocScrutinizer05>
I don't see any basic showstoppers for a USB<->parallel converter
<DocScrutinizer05>
I think on windows some of those (rather their drivers) can even fake a IO port number that the driver is listening to, to emulate a built-in parallel interface
rz2k has quit []
<dos1>
most of usb<->parallel converters (especially the cheaper ones) won't work - usually they're just emulating the printer interface
<whitequark>
I think I can manage to assemble that myself, even
<whitequark>
do they have any in stock? doesn't look like there's been made a lot of them
<whitequark>
ok, doesn't seem that anyone sells complete devices at all
<dos1>
DocScrutinizer05: I was looking for some usb<->lpt converter while working on that drawing robot - finally gave up and used some old laptop with integrated lpt to control it :D
<whitequark>
and the chip is EOL
<whitequark>
not stocked even
<DocScrutinizer05>
whitequark: it's a DIY thing
rz2k has joined #qi-hardware
<whitequark>
dos1: wow, a laptop with LPT. I don't think I've ever seen one
<DocScrutinizer05>
hehe, yeah, a funny relic in CF-29
<DocScrutinizer05>
dos1: yeah, quite impressive. I also seen videos of divers operating CF-29 under water :-)
<DocScrutinizer05>
though I don't recommend doing that "at home", the seals pretty often are _not_ 100% water tight
<DocScrutinizer05>
anyway I guess when you get that 3020 CNC, you should get a dedicated PC for it right away, and better get one with parallel interface
<DocScrutinizer05>
shouldn't cost you more than max 15% on top of the price for the CNC
<whitequark>
"when" already :p
<whitequark>
15%? pretty sure I can find one for free
<whitequark>
there's *tons* of old hardware lying around
<DocScrutinizer05>
yep
<DocScrutinizer05>
I really like that 3020 thing, for that pricetag
<DocScrutinizer05>
almost temped to get one for me
<whitequark>
huh! guess that's a good thing
* whitequark
opens local ebay clone and sorts by price, asc
<DocScrutinizer05>
swapping/adding the mill motor for/with a "printer" would make for a nice combined 3D-printer&mill, for really nifty hybrid stuff to build
<whitequark>
"pentium 100mhz, will give away for a liter of juice"
<whitequark>
DocScrutinizer05: hm, that is a very interesting idea
<DocScrutinizer05>
duh! using a fibreoptic and focusing lens I guess you could do UV-curing for 3D printing *inside* a resin bath
<whitequark>
can you even send resin by mail?
<whitequark>
I doubt I can buy it here
<DocScrutinizer05>
I have no clue which type of resin or other chemical would be used for conventional 3D light-curing printing
<whitequark>
apparently bisphenol-A based epoxy. eww.
<whitequark>
I'm not sure I want to touch that at all
<DocScrutinizer05>
you are not supposed to _eat_ it :-D
<whitequark>
oh, it's not nearly as bad as pure phenol. okay then
<whitequark>
still... I don't need a 55-gal drum of stuff
<DocScrutinizer05>
haha
<DocScrutinizer05>
I wonder if you could do useful printing with a modified hotglue gun as print head
<DocScrutinizer05>
could result in quite unique elasic objects
<whitequark>
I'm not sure they fit my definition of "useful"
<whitequark>
although they are sure unique. unique in their complete lack of applications?
<DocScrutinizer05>
well, there might be *some* applications
<DocScrutinizer05>
when usually you'd use silicon-rubber
<DocScrutinizer05>
e.g protective sleeves
<whitequark>
yeah you can't really buy uv-curing resin here. it's either super-expensive specific stuff (special glues, dental, etc) or gigantic amounts from china or wherever
<DocScrutinizer05>
contact the fabber scene
arielenter has joined #qi-hardware
<DocScrutinizer05>
they must have such stuff around
<whitequark>
fabber
<whitequark>
?
<DocScrutinizer05>
reprap?
<DocScrutinizer05>
3D printer
<whitequark>
sure, there could be about three guys in entire RU doing that
<DocScrutinizer05>
o.O
<DocScrutinizer05>
duh! weird
<whitequark>
it's a very expensive hobby
<whitequark>
with $100/l resins? no way
<DocScrutinizer05>
I thought particularly russian guys should be very interested in that sort of "let's see what can get done"
<whitequark>
they are. but with docs mainly in english, expensive hard-to-obtain resins with possible customs complications... very few people are going to go through that
<DocScrutinizer05>
hmm, yet another weird idea: use a inkjet printhead to spit droplets of hardener into the 2 components resin bath
<whitequark>
1) afaik printerheads are very finely tuned to the specific consistency of ink
<whitequark>
2) you think it would harden fast enough? not sure
<whitequark>
(1) so the hardener wouldn't properly "boil out" or maybe will just flow through
<DocScrutinizer05>
instead of epoxy "resin" you could go experimental and see what soluble glass gets you
<whitequark>
soluble glass? O_o
<whitequark>
ohhhh sodium silicate. I think it's sold here as glue
<DocScrutinizer05>
that liquid based on silicon
<DocScrutinizer05>
yeah
<DocScrutinizer05>
sodium silcate
<whitequark>
you can add ethyl alcohol to it, it'll form a weird substance with nonlinear properties
<DocScrutinizer05>
it hardens with some other chemical
<DocScrutinizer05>
actually a number of other chemicals. Probably all metal salts
<whitequark>
at long timespans it flows. at medium it acts elastically (kinda like shitty rubber). at short it shatters in shards
<whitequark>
DocScrutinizer05: not sure about mechanical properties of the result
<DocScrutinizer05>
me neither
<whitequark>
sounds like it would be fragile and/or weak
<DocScrutinizer05>
well, that's a good start to do some experiments, no?
<whitequark>
yep, why not
<whitequark>
hm, it hardens because metal silicates form
<whitequark>
so that'll require a ton of metal salts
<DocScrutinizer05>
quite possible, though i know it also hardens in free air, that's why it's been used to write on bottles with it
<DocScrutinizer05>
maybe by the carbon from CO2 ?
<whitequark>
yup, also evaporation of water
<DocScrutinizer05>
well, it's not water soluble once it hardened
<whitequark>
oh, it's not? I misremember then
<DocScrutinizer05>
afaik not
<DocScrutinizer05>
I think I heard it's been used for permanent writing on glass bottles
<DocScrutinizer05>
mixed with chalk or some other pigment
<whitequark>
here it's used as a shitty paper glue
<whitequark>
like really shitty, it's really weak
<DocScrutinizer05>
yeah
<DocScrutinizer05>
it doesn't really glue at all
<DocScrutinizer05>
about as much as salt water
<DocScrutinizer05>
but it seems to even kinda mix with the glass of the bottles
<DocScrutinizer05>
I dunno, just wild guessing here
<whitequark>
eh? how'd that happen
<DocScrutinizer05>
I can't think of a substance that simply dries to a chrystaline form to really stick on glass bottles
<DocScrutinizer05>
there must be more to it than just some shit drying
<whitequark>
van der waals forces can be quite high with glass
<DocScrutinizer05>
and glass is also silica
<whitequark>
that uv curing epoxy is apparently used for gluing legs to glass tabletops
<whitequark>
and it sure as hell doesn't mix with glass... it's pure organic
<DocScrutinizer05>
:nod:
<whitequark>
maybe the similar composition helps it somehow. I don't think they mix though
<DocScrutinizer05>
the UV hardening acrylate glue I told wpwrak_ about
<whitequark>
yes, that's about the only thing I can find references to in russian
<DocScrutinizer05>
not mix literally, for sure
<DocScrutinizer05>
nothing _mixes_ with glass
<DocScrutinizer05>
but some substances act on the surface of glass. E.G alkali substances tend to etch glass suface smooth
<whitequark>
I believe alkalis do dissolve glass, very slowly
<DocScrutinizer05>
that's why it's dangerous to boil alkali stuff in glass pots, it may "explode" since there's no micro-scrs left in glas for the steam bubbles to build on it
<whitequark>
that's... 0.0022% solution of NaOH, if I remember anything from my chemistry courses
<DocScrutinizer05>
>>Molten NaOH eats glass, as will a 50% boiling solution of NaOH; though at a lower rate. Doesn't touch stainless steel though. This is a well tried process for de-enamelling sheet iron articles which cannot be shot-blasted<<
<whitequark>
I guess that would just work extemely slowly
<DocScrutinizer05>
when it's well tried, then I gather it works in finite amount of time
<whitequark>
ah, the old thermodynamics vs kinetics
<whitequark>
thermodynamically, you're a cloud of hot gas and a pinch of ashes