DocScrutinizer05 changed the topic of #qi-hardware to: Copyleft hardware - http://qi-hardware.com | hardware hackers join here to discuss Ben NanoNote, atben / atusb 802.15.4 wireless, and other community driven hw projects | public logging at http://en.qi-hardware.com/irclogs and http://irclog.whitequark.org/qi-hardware
_whitelogger has joined #qi-hardware
_whitelogger has joined #qi-hardware
_whitelogger has joined #qi-hardware
_whitelogger has joined #qi-hardware
_whitelogger has joined #qi-hardware
_whitelogger has joined #qi-hardware
_whitelogger has joined #qi-hardware
<whitequark> wpwrak: can you check whether my idea is sane or not?
<whitequark> so here it goes: I've found out that my machine apparently misses some steps, because it's not as accurate in repeat positioning as I wish
<whitequark> however, how exactly does one measure whether it's off 0.05mm by x or y axis?
<whitequark> it's not like I can place a ruler and read it
<whitequark> so the idea is, I should get a conical cutter, with as thin end as possible, and then engrave a vernier scale
<whitequark> assuming it is accurate with (slow) feeds and inaccurate with rapids, that should allow me to measure the runoff very precisely at a negligible cost
apelete_ has quit [Ping timeout: 252 seconds]
wolfspraul has quit [Remote host closed the connection]
roh has quit [Write error: Broken pipe]
FDCX_ has quit [*.net *.split]
wolfspra1l has joined #qi-hardware
roh_ has joined #qi-hardware
FDCX_ has joined #qi-hardware
roh_ has quit [*.net *.split]
wolfspra1l has quit [*.net *.split]
ornotermes has quit [*.net *.split]
jow_laptop has quit [*.net *.split]
valhalla has quit [*.net *.split]
wpwrak has quit [*.net *.split]
rozzin has quit [*.net *.split]
Codora has quit [*.net *.split]
wpwrak has joined #qi-hardware
valhalla has joined #qi-hardware
roh_ has joined #qi-hardware
wolfspra1l has joined #qi-hardware
ornotermes has joined #qi-hardware
Codora has joined #qi-hardware
rozzin has joined #qi-hardware
jow_laptop has joined #qi-hardware
<whitequark> wpwrak was right, 100µm is fine but 50µm really isn't
<whitequark> (it detaches while etching, and in addition seems to be edge-etched)
<whitequark> though... that's not really 100µm. all other traces are roughly the right size, but 100µm got butchered by the printer again.
<whitequark> it's actually 150µm, same as the trace below it.
<shevek> whitequark: What sort of machine are you talking about? A cnc mill? Do you have full control over the firmware?
<shevek> What I did on my 3-D printer to detect missing steps, is to go somewhere to the middle of the axis; move a piece at a high speed to test, then move back to the limit switch and see if it still is in the position where I left it from. If not, I missed steps.
<whitequark> shevek: a CNC router (spindle moves rather than table), and I use EMC2
<whitequark> well, I'm already sure I do have missing steps. I want to know how much exactly and I want to have a way to reliably check for their presence in order to determine the optimal feedrate
<shevek> Yes, that's what my method was meant for.
<whitequark> hm, how did you determine 'if it's still in the position where you left it from' ?
<shevek> When going at low speed to the limit switch, it very reliably is always in the same position.
<whitequark> ohh, limit switches. don't have any here.
<shevek> Ah, then it won't work.
<whitequark> but I could very well mount one.
<shevek> How do you home without one? Just bump into the wall for a while?
<whitequark> nah, that would likely damage the ballscrew if done a lot
<shevek> Yes, but how else can you do it without a limit switch?
<whitequark> I just jogged it manually to, roughly, (0.5, 0.5, 0.5) and made it my pivot
<shevek> Or do you manually position it before each job?
<whitequark> then each time before turning it off, I tell it to go back to origin
<whitequark> so I won't have to jog it again
<shevek> Ah yes.
<shevek> Does that work well enough for Z though? I would expect that to require high accuracy.
<whitequark> I have to set Z coordinate system for each workpiece, otherwise it can be off about ±0.5mm
<shevek> Because of variations in pcb thickness?
<whitequark> this is what annoys me; each rapid in Z throws accuracy out of the window
<whitequark> no, due to missing steps. PCBs are really flat, and while it's mounted not perfectly horizontally on the table, that's within ±0.1mm, I'd say
<shevek> Oh; missing steps should be avoidable, I would think.
<whitequark> same thought here
<whitequark> I set it up from the accompanying "documentation", so in theory the values should work
<whitequark> I can see my power supply not providing enough peak current for steppers, or the values just being wrong.
<shevek> If you have stepper driver controllers with a current adjustment, turning that up may help as well. But I would just make it go slower.
<whitequark> mine are TB6560AHQ
<whitequark> cnczone.com almost unanimously considers those crap
<whitequark> hm, seems to have a bunch of configuration pins. excitation mode, current decay, torque setting
<whitequark> shevek: any of that ringing a bell?
<shevek> It didn't, I'm just looking at the datasheet now.
<shevek> The pin I'm talking about is 1 and 2 (TQ2 and TQ1). Looks like it's a digital setting though; you probably already have both low, which just gives maximum torque.
<shevek> Ah, but wait; there's an analog reference resistor on NFA/NFB.
<shevek> IOUT (A) = 0.5 (V) / RNF (Ω)
<shevek> If you have a potmeter there, you can change the value to whatever is best for your system.
<shevek> There must be some resistor already, otherwise it wouldn't work.
<whitequark> hmmm
<whitequark> I didn't yet trace the whole board...
<whitequark> seems like they're pulled down indeed
<whitequark> resistor seems to be 510 ohm
<whitequark> no, that can't be right
<whitequark> 0.510 ohm
<shevek> Why not? That would set the current to about 1A; that sounds reasonable.
<whitequark> 0.5 / 510 = 0.5 mA
<whitequark> whereas 0.510 is right one
<shevek> Yes. :-)
<whitequark> so... my adjustable PSU says all three steppers at max speed consume about 2A
<whitequark> which is weird, but in the same ballpark
eintopf_ is now known as eintopf
viric has joined #qi-hardware
viric has quit [Changing host]
viric has joined #qi-hardware
<wpwrak> whitequark: i think the net split ate my comment about missed steps ... here we go again:
<wpwrak> sounds overly complicated. if you have a "good" and a "bad" mode, just go back and forth enough times. "good" in one direction, "bad" in the other. eventually, the missed steps will add up
<wpwrak> you can of course also use a ruler :) like this: go to (0, 0). then go to (max, 0), engrave a little line in y+ direction, then go back to (0, 0). then to (max, step), engrave another one in y+, then (0, 0) and (max, 2*step), and so on
<wpwrak> with time, you should have a line of "dashes" should then be easy to check with a ruler whether it veers off
<whitequark> wpwrak: ohhh, you're entirely right
<whitequark> should be even possible to write that entirely in gcode
<whitequark> parametric gcode is the best idea ever. I wonder if it's possible to export from a parametric CAD to parametric gcode
<wpwrak> ;-))
<whitequark> of course it's not if the CAD uses something like a SAT-solver
<whitequark> like freecad
<whitequark> does fped have a solver?
<wpwrak> fped doesn't need one - all calculations are "procedural"
mth has joined #qi-hardware
<whitequark> so it's possible to make that with fped
<wpwrak> it does hunt around for variables a bit, though. so you have dynamic scoping instead of static scoping.
<whitequark> you could resolve it at export time
<whitequark> I mean, you won't add any new variables or elements *after* gcode export, just vary parameters
<whitequark> and variable resolution isn't dependent on that
<wpwrak> yup, you could turn each variable into <frame>_<var-name>
<whitequark> by the way, I now think it's possible to make an USB-LPT adapter for CNC
<whitequark> that is, without embedding a full interpreter or any really clever stuff for that matter
<whitequark> the thing is, CNC contol software doesn't really use LPT interactively, or at least at level where USB latency becomes significant
<wpwrak> are delays between writes constant ?
<whitequark> only thing it needs is precise waveform generation. you can simply ensure you always have, say, a second of them buffered, and then refill the USB dongle via bulk transfers
<whitequark> it doesn't really matter too
<whitequark> it's absolutely OK if the dongle slightly "lags" behind the actual control software--there's pretty much no realtime feedback
<wpwrak> hmm, you could basically generate your own timing. so the dongle would have to know max speed, acceleration, and so on
<whitequark> no, the whole idea is to avoid putting *any* EMC2 logic in dongle
<whitequark> since that's means I'm making a poor clone of EMC2, which I'd like to avoid
<whitequark> the only inputs EMC2 knows are... home/limit switches, emergency stop, spindle phase (for speed measurement)
<wpwrak> so the timing resolution is much more coarse than 1 ms ?
<whitequark> wpwrak: no, it's much finer
<wpwrak> so usb will lose the timing information
<wpwrak> and you have to reconstruct it somehow
<whitequark> I want emc2 to generate a waveform; timing+levels chart. then transmit that chart to dongle.
<wpwrak> ah, you want to tweak the format. i see.
<wpwrak> yes, that could of course be done.
<whitequark> the important point is, there's no requirement to reproduce that chart *immediately*. the only requirement is to reproduce it *accurately*
<wpwrak> of course. you could even pass handling of the switches to the dongle
<whitequark> ESTOP and spindle phase inputs won't suffer too from several (dozen) ms of delay
apelete has joined #qi-hardware
<whitequark> switches... it may be more complicated. I never actually used them in EMC2 so I wouldn't know the requirements
<wpwrak> while (!SWITCH) STEP;
<whitequark> nah, that ignores deceleration profile
<whitequark> would mean missed steps.
<wpwrak> if it's just for finding (0, 0), that's about what you'd want, no ?
<whitequark> ignoring deceleration profile would mean arriving at home quickly and slowly would actually set it to different positions
<whitequark> I mean, deceleration profile is there to combat machine inertia
<whitequark> and see above, I don't want to duplicate EMC2 logic in dongle
<wpwrak> so (0, 0) is not a real limit. well, you could extend this. while (!SWITCH) STEP; DECELERATE;
<whitequark> already answered :p
<wpwrak> STEP and DECELERATE would be waveforms :)
<whitequark> that implies EMC2 would need to stop treating dongle as "weird LPT port" and actually know to which waveforms to upload there
<whitequark> so, much more fundamental changes to EMC2. don't like that.
<whitequark> I don't even want dongle to know what it controls. I want it to be kind of
<whitequark> logic analyzer in reverse
<wpwrak> with this simple algorithm, you'd have to run X and Y separately. doing X and Y in parallel would be a little more complicated. i.e., you'd need two concurrent "channels"
<wpwrak> well, the "weird lpr" kinda disappears when you send (t, pattern) tuples instead of just patterns :)
<whitequark> (t, pattern) tuples still require insignificant changes to EMC2 backend
<wpwrak> and your 1 s buffer is far too large for properly responding to a switch
<wpwrak> so you have to cut that loop
<whitequark> I mean, I'm sure EMC2 internally just generates those tuples
<whitequark> (too large) yep. that's what I'm sad about.
<wpwrak> seems plausible
<whitequark> possible solution: tell EMC to find home at relatively low speed
<whitequark> then cut the step waveform ignoring deceleration profile
<whitequark> another possible solution, dependent on how exactly EMC is built: ask EMC for tuples in "advance", so I could send data for t+1s while EMC still thinks it's t c
<whitequark> *t s
<whitequark> then I could poll switches at, say, 1000Hz, typical USB interrupt freq
<whitequark> that ought to be enough
<whitequark> it all hinges on whether EMC is able to give me waveform for t+1s while receiving input for ts
<whitequark> buut if that works, I really like it. very simple dongle, no LPT or rt kernel needed, works with existing proven gcode interpreter
<whitequark> DocScrutinizer51: ^ relevant to your interests
<wpwrak> wave(t+1s) where wave(x) is a function of wave(t_switch) would require clairvoyance. if it can do that, it may just send the right profile to come to a precise stop at (0, 0) ;-)
<DocScrutinizer05> whitequark: eek, how many lines back?
<wpwrak> all of them :)
<whitequark> DocScrutinizer05: about 40 minutes
<whitequark> 30 even
<DocScrutinizer05> 1: (( just go back and forth enough times.)) sorry, doesn't work this way. "back & forth" would mean steep ramp-up and slow ramp-down in positive direction, while slow ram-up and steep ramp-down in the other direction. Otherwise the error compensates, no matter how many km you run forth & back
<whitequark> DocScrutinizer05: if it compensated, I wouldn't be having any problems! I specifically notice error when I do feed down (plunge into material), then rapid up, move, and feed down again -- which makes a hole in the table because Z missed steps
<whitequark> ramp up and down for any single coordinated movement are identical
<wpwrak> DocScrutinizer05: i mean slow & safe on one direction, fast and jumpy in the other. any skipped step, no matter in which phase of the movement, would therefore add up in the same direction
<whitequark> ^
<wpwrak> ah, so the problem is mainly Z, not X/Y ?
<whitequark> well, on Z it's immediately problematic because poor table. I *think* it would manifest itself on X/Y, because on X/Y settings allow it to travel even faster
<whitequark> but I have not specifically noticed it on X/Y yet
<wpwrak> well, you could run the aforementioned test for X/Y
<whitequark> sure
<DocScrutinizer05> wpwrak: no it won't
<DocScrutinizer05> missed steps are during ramp-up. During ramp-down you see bogus additional steps
<wpwrak> for Z, you can do the same. you can test height very easily by placing some object on the table, moving the head over it, lowering the bit manually until it touches, lift off and move elsewhere, do the fast up / slow down ritual, then come back and see if you're still at the same height
<whitequark> DocScrutinizer05: you sure it's ramp-up and not linear motion?
<wpwrak> ah, if bogus steps get inserted, then that wouldn't work. but are they ?
<DocScrutinizer05> I guess it simply is missing steps on move up due to much higher load (force)
<whitequark> yes, exactly my thoughts
<wpwrak> i concur
<DocScrutinizer05> and of course it could add steps particularly when motor gets "pushed", when moving down
<DocScrutinizer05> depends on your gear
<whitequark> right, it seems to creep downwards
<DocScrutinizer05> as a rule of thumb: when you apply AC to the stepper, it should run relatively free. When it starts moving then, you have a problem with force pushing motor
<DocScrutinizer05> but I guess you have a threat spindle, and that isn't known to feed back forward force to motor
<whitequark> threat spindle ?
<DocScrutinizer05> basically a long screw
<whitequark> ballscrew, yes
<DocScrutinizer05> thread, I guess
<DocScrutinizer05> coffee deficit
<wpwrak> a threatening spindle, its weight exceeding the crush force of your skull :)
<DocScrutinizer05> reduce your Z-speed particularly up
<DocScrutinizer05> increase voltage to motor
<DocScrutinizer05> change coil config on 6/8-wire stepper
<whitequark> 4-wire
<DocScrutinizer05> :nod:
<DocScrutinizer05> thought as much
<whitequark> voltage, hmm... it's 6 volts below driver IC limit
<whitequark> given my previous, ahem, experience with AMRs on this CNC, I'd rather not check whether they are manufactured to spec :S
<wpwrak> voltage is at the designed level ? or below that, too ?
<whitequark> from factory the voltage was at 24V
<whitequark> I'm currently driving them with 29V because I need same PSU for steppers and spindle
<whitequark> driver IC AMR is 36V
<DocScrutinizer05> you seen last c't hacks issue already?
<wpwrak> hmm, i wonder what +20% voltage does to a stepper ...
<whitequark> not entirely
<wpwrak> (with identical timing)
<whitequark> wpwrak: they're not rated to voltage. rather to current
<wpwrak> so the driver regulates the current ?
<whitequark> yep
<whitequark> currently it's set up to drive them at roughly 1A, though my PSU reports lower values
<DocScrutinizer05> sure, since that's maximum current.
<DocScrutinizer05> limit
<wpwrak> did you have to change the "programming" resistor ? they're sometimes a function of the supply voltage.
<DocScrutinizer05> the true current is a function of voltage/speed
<whitequark> wpwrak: from the datasheet it appears that the driver regulates current regardless of supply voltage
<whitequark> since it's not in equation
<wpwrak> good
<DocScrutinizer05> the higher the stepping speed, the lower the current
<nicksydney> wpwrak: normally what footprint you use for Balun ?
<wpwrak> oh, another thing: you can detect table contact by mounting a needle on the head, grounding the table, and running a wire from needle to a suitable input
<wpwrak> if the table isn't isolated, add battery + opto-coupler
<whitequark> wpwrak: already thought of it, the CNC even has a suitable probe input
<whitequark> I only need to run another two wires and change EMC config to make that work
<whitequark> however I couldn't use that for "production" work, because I can't reliably insert tool to a predefined depth
<whitequark> the collet can't do that
<DocScrutinizer05> tools are conductive too
<whitequark> DocScrutinizer05: right
<whitequark> but there's another my concern, what if the contact is broken somehow? EMC is going to run tool into workpiece
<wpwrak> (depth) the idea is to use this for determining the tools' Z position :)
<whitequark> wpwrak: meaning: I can't set coordinate system based on readings off a needle and then change that to a tool
<whitequark> because I can't ensure tool has well-known stickout
<wpwrak> raise the contact. then you can just stop if you must have missed it
<wpwrak> well, for adjusting tools, you'd use the tool itself, instead of a needle
<wpwrak> like DocScrutinizer05 said
<DocScrutinizer05> make the "table2 the conductive lever of a microswitch
<whitequark> hmmm
<DocScrutinizer05> when switch operates then you got a problem
<wpwrak> if you just want to debug your motor, use a needle since if anything goes wrong, it can't do much damage
<whitequark> wpwrak: yeah, for debugging I'd probably do that
<whitequark> another thought, how to ensure electrical contact between moving parts at 20000rpm
<whitequark> I've checked and I can connect to the tool from any exposed part of axis assembly (but not motor case)
<wpwrak> you may also be able to test your connections: for each, have two signals. then check for connectivity between A1-A2 and B1-B2, before checking A-B
<whitequark> not sure I follow
<DocScrutinizer05> you don't calibrate with spindle running o.O
<wpwrak> if you're worried one cable may get disconnected, have two cables and test whether they're connected to each other
<whitequark> DocScrutinizer05: hm, I thought of making some kind of permanent mod
<DocScrutinizer05> so what?
<DocScrutinizer05> can't you control your spindle?
<whitequark> so that CNC wouldn't need by attention each time it needs to probe something
<whitequark> I can
<whitequark> but to be permanent, it needs to always be connected to tool/axis
<DocScrutinizer05> no, only when not spinning
<whitequark> hm... add some mechanical contraption to make contact before it needs measurement?
<wpwrak> you could have a lateral pogo pin to connect to the head. that way, there would be no permanent connection there
<whitequark> wpwrak: ooooh, pogo pin, an excellent idea. I even have a bunch
<wpwrak> (nuhc) good. since you'll want two of them :)
<wpwrak> #s/nuhc/bunch/
<DocScrutinizer05> wow, what a typo
<DocScrutinizer05> cat?
<DocScrutinizer05> ;)
<whitequark> okay, idea: spindle (at the bottom) has this spinning plastic thingy with axis end exposed
<DocScrutinizer05> UGH what's that?
<whitequark> I could make a fixture, place it on the stepper near, and place both pogo pins for contact and IR LED/transistor for speed measurement
<whitequark> DocScrutinizer05: it's apparently some kind of cooling fan
<DocScrutinizer05> mhm
<wpwrak> DocScrutinizer05: lack of morning caffeine after being rudely awakened by a gnat
<DocScrutinizer05> I guess the axis is straight to other side
<whitequark> actually no, it's not
<DocScrutinizer05> gnat?
<whitequark> it's just a plastic handle for you to spin the axis manually, I think
<whitequark> like that handle on a stepper nearly
<DocScrutinizer05> ooh
<DocScrutinizer05> ding lnows gnat
<DocScrutinizer05> knows
<DocScrutinizer05> ok, that's absolutely fine to place ONE pogo in center
<DocScrutinizer05> absolutely in center
<DocScrutinizer05> ang give it some graphite
<DocScrutinizer05> or find a ball bearing, they make fine contact to moving parts
<DocScrutinizer05> the motor shall have at least two of them
<whitequark> not exposed
<DocScrutinizer05> indirectly exposed
<DocScrutinizer05> the ball bearings are mounted to motor's steel case
<whitequark> told you: there's no contact between axis and case
<DocScrutinizer05> you can't tell me your spindle axis is partially hold by plastic contraptions
<whitequark> that was the first thing I checked
<whitequark> the plastic contraption above has no structural function
<whitequark> I dunno what's inside motor
<DocScrutinizer05> now that's extremely weird then
pcercuei has joined #qi-hardware
<DocScrutinizer05> go for the pogopin
<whitequark> right
<DocScrutinizer05> check if at least axis is a conductor along full length ;-)
<whitequark> verified that, it is
<DocScrutinizer05> I've actually seen industrial designs that did exactly this for grounding of motor axis (in tape recorder for example)
<whitequark> pogo?
<DocScrutinizer05> yep
<whitequark> cool
<DocScrutinizer05> spring lever with a coal
<DocScrutinizer05> or brass on steel
<DocScrutinizer05> brass+steel combo needs no lubrification
<whitequark> axis looks like steel for sure
<whitequark> dunno what pogo pins are made of
<DocScrutinizer05> gold?
<whitequark> no
<DocScrutinizer05> ;-D
<DocScrutinizer05> no?
<DocScrutinizer05> eeek, the extral long weapon
<DocScrutinizer05> don't use that one, it will give you a lot of headache
<whitequark> oh?
<DocScrutinizer05> you can't use it to point at a spinning axis' center
<DocScrutinizer05> it will move avay in no time
<whitequark> hm
<DocScrutinizer05> use a brass spring, like a hand of clock
<whitequark> hand of clock?
<DocScrutinizer05> L
<whitequark> not sure I follow
<DocScrutinizer05> short thing: axis. Long thing: lever
<DocScrutinizer05> I _
<DocScrutinizer05> /_
<DocScrutinizer05> \_
<whitequark> I think I get it
<DocScrutinizer05> (side view)
<DocScrutinizer05> |_[motor]____<toolbit>
<DocScrutinizer05> you might use a contact spring from a slaughtered relay
<DocScrutinizer05> (though that usually isn't brass)
<DocScrutinizer05> well, sometimes it actually is
<DocScrutinizer05> phosphor brass
<DocScrutinizer05> hey, how about mounting an optical mouse against arbitrary flat moving_up/down surface of your spindle unit?
<DocScrutinizer05> 1200dpi resolution? sounds fine
<DocScrutinizer05> a decent mouse also shouldn't 'lose steps'
<DocScrutinizer05> sure no absolute value ;-) only relative. But then, so what?
<DocScrutinizer05> if you had a GTA04-jr-edition, this would be integrated in a convenient way on left small side, to mount the whole thing to CNC and directly read Z coord from display ;-D
apelete_ has joined #qi-hardware
jow_lapt1p has joined #qi-hardware
<whitequark> hm
wolfspraul has joined #qi-hardware
ornoterm1s has joined #qi-hardware
valhalla_ has joined #qi-hardware
<DocScrutinizer05> 6.Optical mouse sensor as X/Y scroll­”wheel”
<DocScrutinizer05> Additional
<DocScrutinizer05> use cases: ultra­precise detection of velocity, direction, and distance of movement of objects
<DocScrutinizer05> in relation to device, e.g. robotics, digitizing roads off paper maps, electronic “folding meter
<DocScrutinizer05> rule”...
<whitequark> right, pretty cool
Coyo has joined #qi-hardware
Coyo is now known as Guest9509
apelete has quit [*.net *.split]
mth has quit [*.net *.split]
roh_ has quit [*.net *.split]
wolfspra1l has quit [*.net *.split]
ornotermes has quit [*.net *.split]
jow_laptop has quit [*.net *.split]
valhalla has quit [*.net *.split]
wpwrak has quit [*.net *.split]
rozzin has quit [*.net *.split]
Codora has quit [*.net *.split]
<DocScrutinizer05> [2014-02-23 12:42:31] <-- wpwrak has left this server (*.net *.split).
<DocScrutinizer05> dman friggin DDoSers
<DocScrutinizer05> I really wonder if that's the GCHQ or whatitscalled
<DocScrutinizer05> or just a skiddie that found out how to write a NTP exploit
<DocScrutinizer05> and now practicing
<whitequark> prolly skiddies
<DocScrutinizer05> I also like to kick some IETF butts thru all the town
<DocScrutinizer05> for doing an incredibly poor job when specifying some RFCs (see NTP etc)
<DocScrutinizer05> connectionless echo with volume multiplier >1 should be punishable
<larsc> is this how the ddos is done?
<DocScrutinizer05> in RFC speak: a machine MUST NOT answer to any inbound message that's not authenticated regarding its origin with an answer that's larger in size than the query
<DocScrutinizer05> larsc: prolly
<larsc> DocScrutinizer05: as in you assume this, or people in general assume this?
<DocScrutinizer05> NTP listsomething UDP, with fake source IP
<DocScrutinizer05> people generally assume this, it's a known exploit seen in the wild during last few weeks aiui
<whitequark> there's also routers which don't check source IP
<DocScrutinizer05> send a UDP to vulnerable server: "protocol: NTP. command: tell me about your list of foobar! from: <IP of DDoS target>"
<DocScrutinizer05> multiplier factor: up to several 100
<DocScrutinizer05> depending on size of 2list of foobar"
rozzin has joined #qi-hardware
<DocScrutinizer05> [2014-02-22 23:57:11] <alamar> net4all: it's amplification using ntp (probably monlist) probably with requests sourced from networks not enforcing bcp-38 (and thus enabling the reflection)
<DocScrutinizer05> [2014-02-22 23:42:52] <xordern> The current attack seems to use NTP reflection.
<DocScrutinizer05> [2014-02-22 23:44:13] <xordern> ... and this produces currently traffic up to 35000 Mbit per seconds.
roh has joined #qi-hardware
wpwrak has joined #qi-hardware
mth_ has joined #qi-hardware
<DocScrutinizer05> [2014-02-22 23:58:25] <alamar> Rubas: your server can get ddos (and the traffic type does not matter)
<DocScrutinizer05> [2014-02-22 23:58:36] <alamar> if it's dns, ntp or snmp
<DocScrutinizer05> [2014-02-22 23:59:02] <alamar> by "filling" the available bandwidht with "any kind of traffic" the legitimate traffic will eventually get lost
<DocScrutinizer05> [2014-02-23 00:00:59] <alamar> with the current trend it's open ntp servers replying to monlist requests
<DocScrutinizer05> [2014-02-23 00:04:08] <alamar> the original requests may be sourced from infected servers/pcs/whatever but are using spoofed IP addresses (using freenode server addresses) they contact lists of ntp servers asking for their monlist. ntp using udp which is a connectionless protocol not requiring any real interaction answers to the requests sending the data to the freenode servers (which did not ask for them). ntp monlists can be quite
<DocScrutinizer05> large (so if your request takes 10bytes and the answer is a 500byte long list you have an amplficiation factor of 50)
<DocScrutinizer05> larsc: HTH
<DocScrutinizer05> note that there seem to be other vulnerable protocols (snmp, dns, ...), thus my rant about IETF and their terribly poor protocol specs
<DocScrutinizer05> hell, you could argue UDP never whould've been invented
<whitequark> I'd argue networks must obey bcp-38 rather
<whitequark> DNS works because it's stateless
<DocScrutinizer05> TCP SYN/ACK is a minimalistic "authentication" layer at least
<whitequark> no fucking way you can answer billions of req/s with state tracking
<DocScrutinizer05> right, SYN flooding is a problem in itself
<whitequark> yep
<DocScrutinizer05> but a problem we know what's the possible cures
<DocScrutinizer05> you never can defeat *all* kinds of DoS
<DocScrutinizer05> but you can mitigate the impact
<DocScrutinizer05> and make it expensive for the attacker
<DocScrutinizer05> it's all a question of cost on either side
<DocScrutinizer05> when the cost on attacker's side is magnitudes higher than the effort/cost to reject stuff at victim's side, a DoS attack becomes unattractive
dos1 has joined #qi-hardware
<DocScrutinizer05> hi dos1
<dos1> hello!
<dos1> what's up?
<DocScrutinizer05> NTP exploit ;-P
<DocScrutinizer05> send a UDP to vulnerable server: "protocol: NTP. command: tell me about your monlist! from: <IP of DDoS target>"
<dos1> oh yeah, DDoS multiplication
<DocScrutinizer05> server sends monlist (which can be quite a bunch of data traffic) to <IP of DDoS target>
<DocScrutinizer05> dos1: [2014-02-22 23:44:13] <xordern> ... and this produces currently traffic up to 35000 Mbit per seconds.
<DocScrutinizer05> :-O
<DocScrutinizer05> poor freenode
<dos1> funny how such things become widely used after so much time
<dos1> makes you wonder how many similar vectors exist in other commonly used software, just waiting for aha moment in someones mind :P
<dos1> previously they used DNS, but monlist turned out to be much more effective
<DocScrutinizer05> snmp
<DocScrutinizer05> and there are a few others *known* to be vulnerable, aiui
<DocScrutinizer05> they even exploited ICMP iirc
<whitequark> consider you can scan 0.0.0.0/32 in a day or two *on commodity servers*
<DocScrutinizer05> funny: ICMP package got sent to invalid IP, so the machine got a NAK and that created further ICMP to invalid IPs
<DocScrutinizer05> maybe google for icmp-flood or sth like that
<DocScrutinizer05> wpwrak: wb!
<DocScrutinizer05> I wonder if eventually IETF will come up with a protocol RFC that allows banning of machines at POS level, in expression a inter-carrier protocol where destination end asks origin end to ban certain IPs resp their users physically from internet
<DocScrutinizer05> ? s/POS/POP/ ?
<DocScrutinizer05> POP, Point Of Presence - iirc
<DocScrutinizer05> an ancient obsolete term I guess
<whitequark> GEMA and MPAA would be so pleased
<DocScrutinizer05> sure
<DocScrutinizer05> but they already enforced/pushed this, though not on a protocol level
<whitequark> what you need is wider deployment of bcp-38
<DocScrutinizer05> tbh nfc about bcp-38
<whitequark> and it would solve pretty much all UDP amplification attacks and some other
<whitequark> well
<whitequark> source ip checking by routers
<DocScrutinizer05> DPI
<DocScrutinizer05> ?
<whitequark> no, not really, I mean routers already compare destination ip with their tables
<whitequark> could run one more lookup per udp packet
<DocScrutinizer05> well, in which table shall a router at my ISP keep a record for that UDP packet I want to send to my SIP far end, to my NTP server stratum-4711, or whatever?
<DocScrutinizer05> and where from originates that record in that table?
<whitequark> ISP should reject all packets with source IP not originating from its network
<whitequark> it's a very basic rule
<DocScrutinizer05> does UDP have source IP?
<whitequark> how exactly could you answer to UDP otherwise?!
<DocScrutinizer05> well, you can't
<whitequark> lol what
<DocScrutinizer05> aiui
<whitequark> all IP packets have src and dest IP
<DocScrutinizer05> okok
<DocScrutinizer05> more coffee needed
<DocScrutinizer05> yeah, basically NAT and other crap is what kills the internet
<DocScrutinizer05> ;-P
<whitequark> well, no NAT in IPv6
<DocScrutinizer05> generally all sorts of proxies that change source IP
<whitequark> except for dumb people who think they need to "hide their network structure"
<whitequark> which is frankly absurd but so awfully prevalent, or maybe they're just goddamn loud
xiangfu has joined #qi-hardware
<DocScrutinizer05> :nod:
mth_ is now known as mth
<DocScrutinizer05> >>Similar to a DNS amplification attack, an attacker sends a small command to the NTP server (typically the MON_GETLIST command) with a spoofed return address.<<
<dos1> yep, it was actually "fixed" long ago in ntpd
* DocScrutinizer05 glares at "[x] Fritzbox router serves as NTP server time normal for LAN"
<dos1> but attackers now realized how many old, vulnerable versions are still out there
<dos1> monlist got obsoleted and afaik it's not supported anymore at all in latest versions
<DocScrutinizer05> fucking fritzboxen had a vuln in all builds and versions, opening up the admin web-UI to $world, without authentication needed
<DocScrutinizer05> of course via admin UI you could flash arbitrary new frimware
<DocScrutinizer05> :-o
<dos1> yep, "NTP version 4.2.7, which removes the monlist command entirely"
<DocScrutinizer05> I'd absolute not be surprised to find a 27.4 mio routers worldwide provide a vuln NTP implementation
<dos1> who would? ;)
<DocScrutinizer05> and particularly snmp implementation
<DocScrutinizer05> you seen the story where some dude at a university had rooted all routers worldwide which had that backdoor vulnerability ?
<DocScrutinizer05> by simply scanning THE INTERNET aka /32
<dos1> the internet scan?
<DocScrutinizer05> err /0 ?
<DocScrutinizer05> yup
<dos1> I wonder how many of them are now mining some cryptocurrencies... :D
<DocScrutinizer05> I wonder how many of them are now used in botnets for spam
<DocScrutinizer05> and DDoS
<dos1> the answer for that one is obvious: "lots of them" :P
<dos1> even in that census it was mentioned that many of scanned targets already had some malware installed
<DocScrutinizer05> when that's true, it's fucking ridiculous that (E)TLA agencies use illegal DDoS attacks to "fight" Anonymous and other groups that they think need to get extinguished because they do... DDoS attcks
<whitequark> let's just get rid of TLA jerks
<DocScrutinizer05> hell, all the anti-western terrorists won the battle, really
<whitequark> I don't know, only the USA population seems to be actually moved by the "terrorist threat"
<whitequark> others just use it as a convenient excuse
<DocScrutinizer05> they managed to turn all our authorities into exactly the enemies of our western lifestyle
<whitequark> did they do it? or did we do it to ourselves, ultimately?
<DocScrutinizer05> western moral assets abolish themselves, during fighting threats put up to them by terrorists
<whitequark> weren't that strong then
<whitequark> picasso seems to be tailored for covert agents, not surveillance
valhalla_ is now known as valhalla
xiangfu has quit [Remote host closed the connection]
<DocScrutinizer05> meh, not exactly
<whitequark> hmm?
<DocScrutinizer05> it's just doing what can get done
<dos1> speaking of surveillance - amazing (and kinda scaring) game: https://nothing-to-hide-demo.s3.amazonaws.com/index.html
<DocScrutinizer05> and this type of malfirmware is available commercially for a few 1000 bucks to everybody nowadays
<DocScrutinizer05> (`unrelated`)
apelete_ is now known as apelete
<nicksydney> wpwrak: ping
<whitequark> apparently this happens if you get NaOH in your eye: http://i.imgur.com/U7dyr4L.jpg
* whitequark looks suspiciously on his bowl with 5% NaOH solution
<whitequark> in unrelated news I want a tshirt with this sign: http://3.bp.blogspot.com/-IrZOLGJB8rI/UrhOGlRO_3I/AAAAAAAAJO0/hOHbTc_cmeM/s1600/5220-1.jpg
<whitequark> then it is very easy to respond to allegiations of being annoying: here, you've been warned
<larsc> If only there was a way to order customized t-shirts
<whitequark> larsc: eh? I know it's simple
<larsc> you said that as if it is still 1990
<whitequark> aren't we talking via irc?
<larsc> the communication technology of the future ;)
<whitequark> wow, I'm really impressed with this photoresist
<whitequark> it requires several *seconds* to develop
<whitequark> not even minutes. this time I developed it for 10 seconds and it was well damn ready at 5
<whitequark> this is so much better than the USSR-era crap I used in the past
kristianpaul has quit [Ping timeout: 264 seconds]
kristianpaul has joined #qi-hardware
<DocScrutinizer05> whitequark: new photoresist?
<DocScrutinizer05> ooh, I just notice you're logging the channel :-)
<DocScrutinizer05> on unrelated sidenote: seems Nik and me sorted our organizational woes regarding Neo900 management
<DocScrutinizer05> \o/
<whitequark> DocScrutinizer05: (photoresist) yeah, imported one: http://www.amazon.com/gp/product/B00B0Z8AZ6. very good.
<whitequark> I was using it for a long time though, but I just noticed how fast it actually develops
<whitequark> it kinda evaded my sight before, I just kept brushing it around, thinking "it could not develop that fast"
<whitequark> and actually damaging the exposed layer
<DocScrutinizer05> so you used that PhRes for your recent studies, or a crappy old one?
<whitequark> new photoresist for all recent stuff
<whitequark> crappy old one was like five years ago
<DocScrutinizer05> aah
<whitequark> by the way I'm logging this channel from, I think, late 2011
<whitequark> or late 2012
<DocScrutinizer05> I just got aware of that
<DocScrutinizer05> never looked closely to the URL before
<whitequark> right, late 2011. so 2.5 years already
<whitequark> (org woes) \o.
<whitequark> *\o/
<DocScrutinizer05> yeah, will still create a certain amount of annoying mess, but should pan out fine in the end
<whitequark> wpwrak: what do you do if you get copper 'bridges' between closely adjacent traces?
<DocScrutinizer05> unless I get hit by a bus, in which case the project was severely on peril (but hey, maybe it would be anyway in that case?)
<whitequark> oh let me guess, you don't with your devil mixture
<DocScrutinizer05> hehe
<DocScrutinizer05> usually you use a sharp knife
<DocScrutinizer05> though I sometimes prefer a 10A lab PSU ;-)
<dos1> DocScrutinizer05: stay away from buses!
<DocScrutinizer05> dos1: sure! :-D
<whitequark> dos1: I don't think he could work on electronics then
<DocScrutinizer05> hehehehe
<dos1> :D
<DocScrutinizer05> "PCI bus ran over me! :-S"
<DocScrutinizer05> wait, PCI isn't a bus actually
<DocScrutinizer05> make that I2C then
<DocScrutinizer05> for simplicity ;-)
<whitequark> PCI isn't a bus?
<DocScrutinizer05> ooh, obviously still lack of coffee
<DocScrutinizer05> PCIx or whatsitcalled
<whitequark> PCIe is p2p, right
<DocScrutinizer05> :nod:
<whitequark> probably can't be called a bus. you don't have an 'UART bus'
* whitequark carefully peels photoresist traces off board
<DocScrutinizer05> you also don't really have a ethernet nnnBt bus
<whitequark> it's oddly satisfying, like peeling sunburnt skin
<DocScrutinizer05> lol
<DocScrutinizer05> I can't recall when I last time did that
<DocScrutinizer05> sun? what's that?
<whitequark> lol
<DocScrutinizer05> peeling off dead skin from burn injuries, very common
<DocScrutinizer05> but those burns are usually from hot objects, rarely from flames
<DocScrutinizer05> never from sun
<whitequark> I think I grew bored of handling soldering iron with bare fingers a few years ago
<whitequark> well... sunburn is caused by UV
<DocScrutinizer05> yeah, sure
<whitequark> so if you decide to reproduce my PCB fab and use a 1000W gas discharge lamp for exposure, you can get one
<DocScrutinizer05> and I guess my last "sun"burn been actually a welding burn
<whitequark> ha, I guess that works too!
<DocScrutinizer05> I'm pondering to get a UV-B lamp of some sorts, to get some ozone and UV bleaching into my cave
<whitequark> so this board: http://i.imgur.com/3XRXKgg.jpg
<whitequark> it's kind of shitty actually. I'm wholly dissatisfied by it.
<whitequark> oh by the way: I drilled the holes before etching
<DocScrutinizer05> just wondering what type of lamp precisely I should get. Maybe get a simple HQI and break the UV protector glas?
<whitequark> (HQI) mercury, isn't it forbidden in germany? RoHS and all that
<DocScrutinizer05> not in lamps
<DocScrutinizer05> actually every crappy energy-saver lamp comes with Hg
<whitequark> well, not led ones
<DocScrutinizer05> FL lamps
<whitequark> also those contain about 2-3mg of Hg in them
<whitequark> and a single HQI lamp contains a whole drop of 1-2g probably
<DocScrutinizer05> nah!
<DocScrutinizer05> 0.1g maybe
<whitequark> though, considering how much of them there are sold... HQIs are probably less bad for environment at the end
<whitequark> oh fuck this, I'm going to get some H₂O₂. let's see if persulphate+H2O2 even works
<DocScrutinizer05> (PCB) yeah. looks nice from distance, terrible in close-up
<whitequark> perhaps I got too excited and underdeveloped it
<whitequark> some invisible traces of resist
<DocScrutinizer05> (persu + perox) you're the chemist
<whitequark> honestly it would be very hard to say from a purely theoretical standpoint how H2O2 would affect the reaction
<DocScrutinizer05> you should do a proper development test series, find when it starts "developing" aka the stuff that shall go does go, and when it starts "over developing" means when the stuff that shouldn't starts to go
<whitequark> that shit's really complex to model. in uni they just show you whatever cases are conveniently explained
<whitequark> hmmm
<whitequark> well, it starts developing immediately. as I've said it's nearly complete in 5 seconds
<DocScrutinizer05> guessing from both are oxidizers and based on oxygen, I doubt anything will happen
<whitequark> uhhh
<DocScrutinizer05> (immediately) then you got $l for your formula ($l + $h)/2 seconds
<whitequark> DocScrutinizer05: $h is really hard to figure out, because it doesn't just start to eat exposed areas
<whitequark> no, rather it digs under them and then they detach sometime during etching.
<DocScrutinizer05> when $l is 5s and $h is maybe 90s, then you should develop stuff for ~40s
<whitequark> I'll try to look for tiny bits of unexposed unremoved resist. I can't see them with eye but maybe my macro lens would help
<DocScrutinizer05> yeah, that's why I said you should do a test series
<whitequark> hm, MSDS for persulphate:
<whitequark> NOTE: Decomposes slowly in water to produce
<whitequark> oxygen ozone
<DocScrutinizer05> pretty simple to dunk a exposed PCB into NaOH solution, 1cm per 10s
<whitequark> DocScrutinizer05: it's not developed with NaOH, rather K2CO3
<DocScrutinizer05> ooh, whatever
<whitequark> NaOH immediately removes all exposed and unexposed resist
<DocScrutinizer05> what's K2CO3? backing soda?
<DocScrutinizer05> not exactly
<whitequark> no, that's KHCO3
<whitequark> or NaHCO3
<DocScrutinizer05> Na, I guess
<DocScrutinizer05> so, has that stuff a name?
<whitequark> which stuff? K2CO3? potash I think
<DocScrutinizer05> ooh
<whitequark> btw I'm looking up mechanism of reaction of S2O8 anion
<whitequark> it's very interesting. in some environments it actually *produces* H2O2
<DocScrutinizer05> yep
<DocScrutinizer05> and I seem to recall it gets produced using H2O2?
<whitequark> no clue about production
<whitequark> now, I wonder how can I cause it to produce H2O2
<whitequark> since the latter is apparently "kinetically faster" which is not surprising at all
<whitequark> ohhh, pH 0.3 to 3
<DocScrutinizer05> umm, catalyst?
<whitequark> gotta make that crap slightly acidic
<whitequark> grmbl. wonder if I can buy battery acid here.
<DocScrutinizer05> a few drops of an acid will prolly help, yes
<whitequark> well not "a few drops", a carefully measured amount :p
<DocScrutinizer05> "funny" sidenote: ammonia persulfate is rated here in Germany. Not exactly simple to purchase
<whitequark> oh? why?
<DocScrutinizer05> dangerous?
<DocScrutinizer05> dual use?
<whitequark> "dangerous" is bullshit, you can kill yourself easily with just about anything
<whitequark> dual use... none I'm aware of
<whitequark> oh also, persulphate is not even particularly toxic. your toilet cleaner is much much more dangerous
<DocScrutinizer05> maybe even because of H2O2?
<DocScrutinizer05> well, persulfate seems a semi-decent oxidizer for blackpowder-alike explosives
<whitequark> do you ever not think of explosives? :D
<DocScrutinizer05> actually... no
<whitequark> wasn't even a doubt on that
<DocScrutinizer05> a bad habit from my youth I never could get rid of. Started at the age of ~7
<DocScrutinizer05> when I hear ".*per.*ate" I get excited ;-)
<whitequark> haha
<DocScrutinizer05> actually potasium chlorate doesn't even have a "per" in it
<DocScrutinizer05> still wondering where to get tetranitromethane
rz2k has joined #qi-hardware
<whitequark> usually perchlorate is used for these purposes
<whitequark> our chemistry teacher in highschool mixed that with red phosphorus. he was *really* excited while doing this.
<whitequark> I think I couldn't hear anything for about five minutes afterwards
<whitequark> another nice episode happened when he cooked Al2C3. he put prolly less than a gram of reagents into a crucible and started to heat it. I was filming the whole thing on my phone
<whitequark> then at some time it, um, reacted. the crucible cover immediately flew up 2.5m and left an imprint on the ceiling
<whitequark> I dropped the phone. the record was kinda cool: first it cooks on a torch, then it's all bright white, then a loud bang and it ends with a black screen :p
<DocScrutinizer05> hehehe, where's the youtube link? and... Al2C3?
<DocScrutinizer05> aluminiumcarbide?
<qi-bot> [commit] Paul Cercueil: Fix launching non-OPK apps with files with special characters (master) http://qi-hw.com/p/gmenu2x/c19e789
<qi-bot> [commit] Paul Cercueil: Added key repeat on analog sticks and hats (master) http://qi-hw.com/p/gmenu2x/7777b9e
<whitequark> DocScrutinizer05: no youtube link sadly
<whitequark> I think youtube killed my old account because there was another clip with music from crazy frog overlaid
<whitequark> and some copyrast decided it had to die
* DocScrutinizer05 tries to recall that weird (at least back when I thought it's weird) story about an explosive created from heating a metal with err backing soda or sth similar
<whitequark> aluminium carbide, yes
<whitequark> baking soda? metal? exploside? weird
<DocScrutinizer05> I can't recall the details. How would you create Al-carbide?
<whitequark> (Al2C3) by reacting Al and C, duh
<DocScrutinizer05> nah, wait, I think it involved sulphur
<DocScrutinizer05> ooh, you mix Al and carbon and that explodes?
<whitequark> after heating to IIRC 600°C or something
<DocScrutinizer05> mhm
<whitequark> lol wikipedia
<whitequark> "Melting point2200 °C[2]
<whitequark> Boiling pointdecomposes at 1400 °C[3]"
<DocScrutinizer05> errr
<whitequark> wtf, why does google consider "carbide" and "nitride" equivalent and substitutes them in my queries?!
<whitequark> anyway I'm off to buy some peroxide
<whitequark> if that works, I can later add acid to solution, with hopefully same effect
<whitequark> but permanent
* DocScrutinizer05 wonders what opening hours have the shops in moscow
<DocScrutinizer05> whitequark: btw in http://i.imgur.com/3XRXKgg.jpg the footprints don't match the hole diameter
<DocScrutinizer05> s/footprint/pad/
<qi-bot> DocScrutinizer05 meant: "whitequark: btw in http://i.imgur.com/3XRXKgg.jpg the pads don't match the hole diameter"
<DocScrutinizer05> whitequark: and what's that circuit the PCB is for? a controller with 3 LEDs and a 2wire bus to $world?
<DocScrutinizer05> 4-wire?
<DocScrutinizer05> hmm, it seems to have no power supply on the connector
<DocScrutinizer05> only battery pads
<DocScrutinizer05> connector looks like GND, output, gpio, gpio, gpio
<DocScrutinizer05> prolly the connector is just for in-circuit-programming?
<DocScrutinizer05> \o/ http://trendwatching.com/briefing/
<wpwrak> buying peroxide on a sunday, at night. hmm. either it's weak peroxide and the shop is a pharmacy, or it's strong stuff and the shop is 24/7 supplies for the ... hmm ... pharmaceutical entertainment industry :)
<DocScrutinizer05> run by an individual that must avoid running into any police control
<DocScrutinizer05> ;-P
<whitequark> DocScrutinizer05: (opening hours) it depends. in center it/s 24/7 for a lot of shops, and 9-11 for others
<whitequark> (hole diameter) why you think so?
<whitequark> (PCB is for) just a test PCB with atmega8 and 3 leds
<whitequark> (connector) yes, just ICSP indeed
<whitequark> test pads are for power source
<whitequark> (peroxide) pharmacy obviously
<whitequark> I can order strong stuff completely legally, but it'll take a few days to ship it from St. Petersburg
<whitequark> (legally) I give them my complete ID
<whitequark> and I have to show same ID when receiving parcel
<DocScrutinizer05> LOL, I just found out my LG TV doesn't want to switch to videotext after audio got muted. It however allows muting audio after switched to videotext
<DocScrutinizer05> I own this TV for maybe 5 years now
<DocScrutinizer05> never noticed that bug
<DocScrutinizer05> (diameter) the pads are too small for the holes
<DocScrutinizer05> not enough copper around the hole, at sides
<whitequark> hrm
<whitequark> it should be 1.0 mm
<DocScrutinizer05> that trendwatching mentioning is actually pretty nice
<DocScrutinizer05> (though in direct neighborship to bloks nonsense :-S )
<whitequark> heh
<whitequark> DocScrutinizer05: hmm you seem to be right (maybe). my possibly inaccurate measurement indicates 1.115mm holes
<whitequark> perhaps spindle runout? .1mm seems in the right ballpark for that crappy chinese spindle
<DocScrutinizer05> what happened to the Phres covering the holes?
jow_lapt1p is now known as jow_laptop
<whitequark> it stayed on half of holes and broke away on others
<whitequark> however it never covered it from *other* side of this onesided PCB
<whitequark> so etchant could freely access copper though hole itself
<whitequark> >No more products, development cycles or campaigns built around planned obsolescence. Instead, smart brands will plan, build and sell for constant iteration.
<whitequark> cute. rather what happens is digitally locking you out of features to force you into a subscription model.
<whitequark> >A world of new information for consumers to track, share and then act on (such as the driving data made possible by Automatic, above).
<whitequark> consumers, yeah.
<whitequark> wow, MSDS on persulphate:
<whitequark> Persulfates decom-
<whitequark> pose to form solid sulfate salts and emit noxious fog or
<whitequark> fumes of SO x and NOx . This decomposition may form a
<whitequark> high temperature melt. The material will flow like magma
<whitequark> magma!
* wpwrak wonder how long until whitequark comes around and accepts that HCl is the only resonable choice :)
<whitequark> meeeh
arielenter has joined #qi-hardware
<DocScrutinizer05> vomit on your PCB X-P
* DocScrutinizer05 idly ponders electro-erosion
<DocScrutinizer05> high frequency AC current, and a tungsten spike for the arc
<DocScrutinizer05> AC for the "grounding" of the PCB: by a large area electrode hoovering 1mm above the PCB, that has a maybe 3..5mm hole in the midle, for the arc
<wpwrak> (self-made etchant) maybe eat something really hard to digest first so the stomach increases the acidity
<whitequark> it doesn't quite work like that
<DocScrutinizer05> or simply read poettering's blog or something like that ;-)
<whitequark> anyway, stomach pH is always around 1-2 so it'll fit if not for contaminants
<whitequark> (poettering) hehehe
<DocScrutinizer05> after reading that, I probably can cut the traces by glaring at the PCB
<whitequark> I'm afraid it'll delaminate and/or turn into toxic vapor
<whitequark> anyway, whitequark vs pcb round 2
<DocScrutinizer05> :-)
<DocScrutinizer05> s/round 2/sour/
<whitequark> sour?
<DocScrutinizer05> not?
<whitequark> oh, "sour" as "acidic"
<whitequark> well no, I bought H2O2 but no acid
<whitequark> *cough* one of the cases where lack of acid severely impedes progress *cough*
<DocScrutinizer05> hm?
<whitequark> the other one is apparently smartphones
<whitequark> (steve jobs really liked lsd)
<DocScrutinizer05> ooh
<DocScrutinizer05> didn't know that
<whitequark> quote, "one of the best things I did in my life"
<whitequark> s,things,decisions,
<DocScrutinizer05> to take acid?
<whitequark> yes
<DocScrutinizer05> hehe
<DocScrutinizer05> well, depending on when he did, I might concur
<whitequark> 'when' ?
<DocScrutinizer05> I think all that prychedelic stuff is fine while you are about to develop a mind, once you go productive it's more hindering than giving any benefit
<DocScrutinizer05> IOW it's a lerning tool maybe, but for sure no production resource
<wpwrak> you mean you'd only give it to kinds ? :)
<whitequark> yeah, sounded weird for me too
<DocScrutinizer05> wpwrak: not exactly to kids, but yes, basically
<whitequark> essentially since it appears that on kids even innocent stuff like lsd and pot can (even if not often) cause lasting negative consequences
<DocScrutinizer05> maybe the age between 16 and say 30 max is the right time for such experiments
<whitequark> interesting opinion
<DocScrutinizer05> I'd even say, 25 to 30
<wpwrak> 20-70 ? :)
<DocScrutinizer05> but then, how should I know, I dunno both alternatives
<whitequark> well, current consensus is that frontal lobe development finishes at about 25
<dos1> good, so I still have few years to develop it properly... :)
<wpwrak> well, now you have a project: find a bunch of nice girls, father a child with each, then administer LSD to them at different ages. study the results. publish either under the pseudonym Frankenmann oder Hoftstein
<whitequark> more like Dr. Mengele
<wpwrak> is he already a classic ?
<whitequark> plus the sample size is too small
<whitequark> well... he isn't?
<whitequark> (reminds me of https://xkcd.com/507/)
<DocScrutinizer05> the whole drug legislation isn't driven by scientific reasoning, that's for sure
<DocScrutinizer05> wpwrak: I thought you ask "is he already dead? I seen him last week"
<DocScrutinizer05> he might be your neighbour for all I can tell
<whitequark> ha
<whitequark> (drug legislation) well, not everywhere in the world... europe seems to be saner about this
<DocScrutinizer05> not really
<whitequark> in US it's a deeply religious topic for sure, both metaphorically and literally
<dos1> depends on which part of europe
<dos1> and it varies in time as well
<DocScrutinizer05> Europe in the err 30s of last century only agreed to place pot on the index when they were allowed to continue exporting some chemical that's needed for herion production
<DocScrutinizer05> sth along that line
<DocScrutinizer05> and s/Europe/Germany/
<DocScrutinizer05> meanwhile they bred a whole new scene of designer drugs that are not known by law and thus not forbidden, just because they had that insane politics regarding the "classical" drugs
<DocScrutinizer05> of course that new crap is extremely dangerous and not under any control or research
<whitequark> well, designer drugs were inevitable
<DocScrutinizer05> not really
<whitequark> and afaik at least in some countries laws are formulated in a way which extends to functionally equivalent drugs
<whitequark> I dunno about germany specifically
<DocScrutinizer05> nobody would've bothered creating them when the known drugs would've been less outlawed
<DocScrutinizer05> hah! "functionally equivalent" - where does that stop? at coca cola? coffee, alcolhol, jogging?
<wpwrak> food !
<DocScrutinizer05> so you say weed is legal when I prepare a salad from it?
<DocScrutinizer05> :-P
<wpwrak> naw, i say food should be outlawed, too. after all, a nice meal makes you feel good. we can't have that, can we ?
<DocScrutinizer05> and: who are the testers to check the physiological effects of an unknown substance to decide if it's legal or not? bwahahaha!
<wpwrak> i suspect that a good portion of the obsity in the US can be traces to prohibition / stigmatization of all the other sources of "feel good" stuff
<DocScrutinizer05> yep
<whitequark> or to the fact that you have access to cheap nutritious food
<whitequark> I mean, the most obese country now is Mexico
<wpwrak> cultural contamination
<DocScrutinizer05> I guess two of the top 10 facotrs are: nutrisweet (and other sweeteners), and that hormone stuff in plastic
<whitequark> that's really unfounded claim :)
<DocScrutinizer05> no
<whitequark> well... such things can be confirmed by large-scale studies or metaanalyses
<whitequark> 10k+ people
<whitequark> do you have one?
<DocScrutinizer05> for both there is scientific evidence that they result in users getting fat
<whitequark> otherwise it's mostly picking the data points you like for an arbitrary p-value, unfortunately even in quite some scientific studies
<DocScrutinizer05> of course for both there are like 30 times as many studies that say those substances are perfectly safe and the best thing indistry ever invented
<whitequark> it's not as much quantity of studies as quality
<DocScrutinizer05> check out the history of nutrasweet, it will make you puke or pick your kalashnikov
<whitequark> and of course you have to watch affiliations of researchers
<whitequark> but still. lack of good evidence is a well-known problem in medicine today
<DocScrutinizer05> in some countries nestle(?) did a study three times, until the result was as they expected, to get the crap thru the food&drug-admin certification
<whitequark> yep, unpublished studies. another known problem
<whitequark> I think there's an initiative in UK to force pharma to disclose all such results
<DocScrutinizer05> ooh, those even were published
<whitequark> oh? then it's some lousy job on FDA side
<DocScrutinizer05> but then they just waited for a new leader of the local FDA and repeated the opproval process with a new study
<DocScrutinizer05> or even the same study
<DocScrutinizer05> it's incredible what is going on there
<whitequark> regardless, by picking just the results you like you're no better than pharma
<whitequark> except to the other extreme
<wpwrak> just wait until those who got sick have died, then add that given zero incidence of sickness, mortality has not been studied
<DocScrutinizer05> yep
shevek is now known as Guest55600
Guest55600 has quit [Killed (morgan.freenode.net (Nickname regained by services))]
shevek has joined #qi-hardware
<DocScrutinizer05> whitequark: I'm picking the results industry didn't like but found no better way than bribery to defeat them
<whitequark> so you admit you're biased
<DocScrutinizer05> if those results were based on false studies, it would've been a joy for indistry to defeat those studies on a scientific level
<whitequark> 18:46 < DocScrutinizer05> of course for both there are like 30 times as many studies that say those substances are perfectly safe and the best thing indistry ever invented
<whitequark> it appears they did, but you ignore it because of a prejudice
<whitequark> or is that not so?
<DocScrutinizer05> no, since those 30 studies been paid by industry, and most of them obviously faked
<DocScrutinizer05> [2014-02-23 19:55:10] <DocScrutinizer05> whitequark: I'm picking the results industry didn't like but found no better way than bribery to defeat them
<whitequark> how exactly industry can defeat studies on scientific level without sponsoring another one? you're contradicting yourself
<DocScrutinizer05> mhm, sure
<DocScrutinizer05> maybe discuss that with the author of that report in TV? instead of me
<whitequark> a good way would be to hire an independent researcher. but then you can just claim they were bribed
<whitequark> it's a catch-22
<DocScrutinizer05> yeah, that's what industry wants to make you believe
<whitequark> I don't give a fuck about industry. I give a fuck about better drugs
<whitequark> and having a good drug recalled because some random guy cries "bribery" loud enough is just as bad as the opposite
<whitequark> neither of those have any relation to science
<DocScrutinizer05> I'm not interested in that topic. I said nutrasweet makes you go fat like a whale
<DocScrutinizer05> and that been tested by universities with animals
<DocScrutinizer05> and the bisphenol case is even accepted by industry itself
<DocScrutinizer05> partially
<DocScrutinizer05> of course they still argue that the stuff doesn't escape from soda bottles into the soda
<DocScrutinizer05> but whole EU forbid that crap at least in baby bottles and other stuff that toddlers might suck on or eat from
<wpwrak> we don't want no pictures of fat babies. fat kids, that's okay.
<whitequark> except no state-level health agency ever found it toxic even to babies
<whitequark> WHO, FDA, EFSA, Health Canada, etc, etc
<DocScrutinizer05> that's exactly the nonsense rationale. Hormones are not toxic
<whitequark> this claim is just moronic
<whitequark> and it suggests you didn't even read the reports, because those don't play with words like you suggest
<whitequark> e.g. quote: "In vivo studies have not consistently demonstrated either androgenic or anti-androgenic activity of BPA."
<DocScrutinizer05> aha, moronic. I just answered your statement. So when my answer been moronic, what been your statement then?
<whitequark> no one ever claimed that BPA is safe because it acts like a hormone but hormones are not toxic
<whitequark> I don't know where could you source that answer
<DocScrutinizer05> and I wonder why bisphenol-A and a number of other softeners is forbidden for baby ,ilk bottles then
<whitequark> because one of 'environmental' organizations good mainly at self-promotion and raising money induced a public outcry. that's all.
<DocScrutinizer05> mhm
<whitequark> same reason as e.g. with illegal drugs you mentioned above. public policy in this case is far, far from rational
<whitequark> guess what, if 100k people phone their MP and tell "ban BPA", BPA will be banned
<whitequark> but that doesn't make them right
<DocScrutinizer05> guess what, for that gene manipulated corn it didn't work that way
<DocScrutinizer05> literally nobody here in Germany wants it, yet Germany didn't vote against it in EU
<DocScrutinizer05> not even the reports from MExico about that crap corn causing *more* instead of less pesticides getting used did change that
<DocScrutinizer05> nah, search for recent stuff
<DocScrutinizer05> like, this week
<whitequark> I'd also want to see some pools about GM corn in Germany
<whitequark> can't find sources for it, but in e.g. czech about half of residents are okay with it
<whitequark> okay, I can't find any polls except Greenpeace-sponsored ones
<whitequark> there's this: https://www.openpetition.de/petition/online/stoppt-gen-mais-1507 but seems like no one either knows or cares
<whitequark> 1k5 people
<larsc> I never quite understood why the general public is so against gm stuff
<whitequark> larsc: so far I don't even see general public being so against. rather a very vocal minority promoting itself via fads.
<whitequark> all articles I see by that query are either describing facts (there's a split in germany, etc) or is "greenpeace greenpeace"
motley has joined #qi-hardware
<whitequark> and polls in other parts of europe are 50/50 at worst, definitely not so polarized as they claim
<DocScrutinizer05> well, I don't need any polls in media. I can do my own polls in my neighborhood
<whitequark> it surely isn't representative of EU, Germany, even the city you're in
<DocScrutinizer05> larsc: maybe because it's totally useless, and they don't care about the genes "escaping" and irreversibly polluting "nature"
<DocScrutinizer05> whitequark: I'm not going to argue on that level with you. larsc also said "...why the general public *is* so against gm stuff"
<DocScrutinizer05> it's even stated in the agenda of our new fresh government, that we don't want GMO in Germany
<whitequark> yeah, I already see you'd happily pick data points to support your arbitrary claim
<DocScrutinizer05> nevertheless they voted contrary in recent EU decision
<whitequark> those news said "abstained"
<DocScrutinizer05> whitequark: and you do what? argue for the fun of it?
<whitequark> passing time while board cleans
<DocScrutinizer05> whitequark: I don't enjoy this silly argument with you. abstained means "voted against a ban" when the vote is set up so itz needs a certan number to vote FOR the ban or otherwise it will not happen
<larsc> it means we are against it, but we don't want to say we are against it ;)
<DocScrutinizer05> and I won't continue arguing about any such stuff with you, since I'm seeing *you* for sure are biased
<whitequark> I simply want to see science and not a witchhunt at work
<whitequark> guess that's futile.
ornotermes has joined #qi-hardware
<DocScrutinizer05> for sure the government added that phrase about GMO to their agenda just because they were picking arbitrary data points to support ... what exactly?
valhalla_ has joined #qi-hardware
<DocScrutinizer05> that's for sure futile
<whitequark> grrr. fucked up attaching resist.
valhalla has quit [Write error: Broken pipe]
ornoterm1s has quit [Write error: Broken pipe]
<wpwrak> DocScrutinizer05: btw, any news (hopefully good) from the neo900 peace talks ?
<DocScrutinizer05> [2014-02-23 16:08:31] <DocScrutinizer05> on unrelated sidenote: seems Nik and me sorted our organizational woes regarding Neo900 management
<wpwrak> oh, great ! congratulations !
<DocScrutinizer05> prolly I'm gonna leading the project
<DocScrutinizer05> not really happy with that, but somebody needs to lead it
<wpwrak> as long as you can get him to keep the bookkeeping and such :)
<DocScrutinizer05> nope, that's my job
<wpwrak> ah, that sucks
<DocScrutinizer05> yep
<DocScrutinizer05> but that's life
<wpwrak> yeah, someone's gotta do it ...
<DocScrutinizer05> there has to be one person with the hat on, and that person also is holding the credit card
<wpwrak> and can see that you'll want to be that person ;-)
<DocScrutinizer05> I can't ask my butler to do the boring dealing with money for me
FDCX has joined #qi-hardware
<DocScrutinizer05> no, actually I would prefer when somebody else would lead the project and do all that stuff, but Nik prefers me doing it
<wpwrak> well, in many companies it works like that. finance has usually no authority to deny payments (unless somethings is badly wrong)
<DocScrutinizer05> sure, but that would require neo900 group hiring Nik as our finance officer
<DocScrutinizer05> and the bookkeeping is only one aspect
FDCX_ has quit [Ping timeout: 246 seconds]
<DocScrutinizer05> also the account is not on the person who does the finance, when you got that situation in a company
<DocScrutinizer05> which boils down to: when neo900 group (or me) is leading the project, the funds can't sit on GDC account
<wpwrak> usually not, but that may not be as much a requirement than simply being something that wouldn't make sense otherwise
<dos1> hah, just wanted to ask about the funds :)
<DocScrutinizer05> it's all just about the funds
<dos1> so, what happens to them after all?
<DocScrutinizer05> we probably are going to send out a mail to all donors, asking them to confirm/allow transfer of their donation from GDC to Neo900 group
<wpwrak> so you'll create a company or an eV and then transfer the funds there ?
<DocScrutinizer05> yep
<dos1> sounds reasonable
<DocScrutinizer05> those donors who don't allow that will not contribute to Neo900 development
<DocScrutinizer05> GDC can't use their donation for anything, and Neo900 group can't access it
<wpwrak> suckish. you'll lose a few in the process, simply because they don't see your mail, etc.
<DocScrutinizer05> yeah
<DocScrutinizer05> we obviously screwed that up from beginning
<dos1> I think only pledges under 100 EUR are at risk of being lost that way
<dos1> some smart part of them, that is
<DocScrutinizer05> I need to sort out with Nik about how users/donors send back answers - sending a fax isn't probably going to pan out in the end
<dos1> s/smart/small/ lol
<qi-bot> dos1 meant: "some small part of them, that is"
<DocScrutinizer05> o.O qi bot is weeeiiiiird
<dos1> DocScrutinizer05: won't e-mail be enough?
<DocScrutinizer05> that's a point to evaluate with Nik
<DocScrutinizer05> I dunno what his lawyer/clerk/tax-advicer thinks is needed
<apelete> wpwrak DocScrutinizer05: wrote a small piece to keep track of my soldering tools and record the advice you gave me:
<apelete> wpwrak DocScrutinizer05: here is a shot of the ridiculously heavy power supply brackets that came with the soldering station:
<apelete> wpwrak DocScrutinizer05: any idea what those are intended for ?
<DocScrutinizer05> :-o nope
<wpwrak> wall-mounting ? maybe they were supposed to put one into the box and someone typed "100" :)
<wpwrak> article is nice
<apelete> that's 3Kg of useless steel then :-)
<wpwrak> pretty much, yes
<apelete> wpwrak: I used the soldering station today, to solder a few wires on the gcw-zero:
<apelete> wpwrak: it's a fantastic tool :)
<DocScrutinizer05> sell the brackets on bay
<apelete> DocScrutinizer05: heavy like they are, shipping will be more expensive than the steel itself :)
<apelete> plus I don't even know what to say to the buyer, since I don't what those are
<wpwrak> you just copy the label and add the picture ;-)
<wpwrak> soldering looks almost good. you should expose less wire and also try to keep the strands from fanning out
<DocScrutinizer05> I bet some bored packer had a article number on his/her paper, seen same article number on the bag with clamps, and picked the bag to ship it
<DocScrutinizer05> so at least one ot two of the clamps should *somehow* fit to the tool
<wpwrak> you can do the latter by holding the strands between two fingers and rotating the wire. then tin (solder) the exposed end of the wire and cut off anything you don't need
<apelete> wpwrak: had a hard time soldering the black wire to the gnd pad, so I finally put it in the upper corner of the pcb
<apelete> but the strand fanning out is a mistake, my hand was shakky :-(
<DocScrutinizer05> looks abit like you're adding too much solder, and not adding it to object to solder but to soldering iron tip
wej has quit [Ping timeout: 245 seconds]
<wpwrak> DocScrutinizer05: with SMT, you usually don't have much of a choice ...
<wpwrak> not enough hands ;-)
<wpwrak> a prehensile tail (with fingers) would be useful for that. alas, evolution didn't think of that in time
FDCX_ has joined #qi-hardware
<wpwrak> apelete: e.g., the green wire looks as if it almost touches the capacitor. a little push and it's short RX. and you'll spend a bit of time searching the problem :)
<apelete> wpwrak: wow, didn't even see that one. was worried about how close rx and tx wires were
<apelete> didn't test the serial output yet, I haven't receive the serial-to-usb converter bought off ebay
<apelete> wpwrak: could a magnifying glass help for smt soldering ? was thinking about getting one of these:
wej has joined #qi-hardware
<DocScrutinizer05> apelete: anyway thanks for the nice attribution :-)
<DocScrutinizer05> for that type of wiring you want max 1mm blank wire
<DocScrutinizer05> abs max 2mm
<DocScrutinizer05> will suffice completely to make fine contact to the pad
<apelete> DocScrutinizer05: you're welcome, had a good laugh while learning a lot that night
arielenter has quit [Quit: Leaving.]
<DocScrutinizer05> then you wet that 1mm with solder. you add a tiny solder bead to the pad. Then after adding a bit of flux to both, you put the wire end on top of bead and heat both with your soldering iron chisel tip. Reflow and contact will come after some 2..3s the latest
<DocScrutinizer05> ooh, and don't forget to twist the wire strands as suggested by wpwrak, before you're wetting the wire with solder
<DocScrutinizer05> I love to use the 1..2mm cut-off end of plastic and turn that while pulling it off the wire
<DocScrutinizer05> but that depends on you cutting carefully through the plastic before you pull it off. Not exactly simply without cutting some of the strand wires as well
FDCX_ has quit [Ping timeout: 264 seconds]
<apelete> yes, judging from the pictures I definitely need some practice
<apelete> but it went way better than the first time, working on the ben nanonote
<apelete> I guess the better tools helped a lot :-)
<DocScrutinizer05> sure
<DocScrutinizer05> what did that other guy say? "with the right tools everybody can solder. With bad tools even experts hardly can solder"
<DocScrutinizer05> "it's eeeeeeasy!" ;-)
<DocScrutinizer05> highly recommended!
<DocScrutinizer05> that guy is not telling bullshit
<DocScrutinizer05> apelete: this is no sarcasm or kidding, just a honest friendly suggestion: get a raster PCB ( http://www.strippenstrolch.de/1-1-3-die-lochrasterplatine.html ) and practice a little, soldering wires to the holes
<apelete> DocScrutinizer05: that's a nice idea indeed. shouldn't be expensive on eBay (raster PCB + a few cheap components and wires)
<DocScrutinizer05> wow! https://github.com/mpue/blackboard
dandon has quit [Quit: .]
dandon has joined #qi-hardware
<wpwrak> apelete: (magnifying glasses) hmm, can't hurt to have something for examination. but for working they're probably too inconvenient because you have to adjust things, move them around, etc.
<wpwrak> but for looking at it when done, to see what you did, why not. or get any other good magnifying glass. or, if you get a lab lamp, these often have built-in magnifying glasses
<wpwrak> a strong light source often works wonders
<apelete> wpwrak: these are nice, and you're right about the inconvenience of having to adjust things while working. a lab lamp w/ magnifying glass should be perfect
<DocScrutinizer05> wpwrak: 160 bucks? ESD save 190 bucks???
<DocScrutinizer05> WTF?
<wpwrak> DocScrutinizer05: i didn't say he'd have to order from there :) i'm sure there are cheaper sources ...
pcercuei has quit [Quit: dodo]
<DocScrutinizer05> and particularly sources who not charge an extra 30 bucks for a wire
<wpwrak> DocScrutinizer05: i'm sure you can dig out something cheaper from pearl.de or such :)
<DocScrutinizer05> yeah, sure
<DocScrutinizer05> I was just stunned by their 30 bucks for "ESD safe"
<DocScrutinizer05> an excellent example why "when you find this cheaper anywhere, we will adjust our offer to beat them" actually means "we're perky in prices but who cares, we're selling nevertheless"
rz2k has quit []
<wpwrak> "ESD safe" may also have some conductive coating and such. or simply come from a more expensive company.
<wpwrak> or maybe it comes with a cool certificate. certificates are $$$ ;-)