--- Day changed Sat Sep 19 2009
00:14 < kanzure> plot 10 * (exp(-((((x-.1)/(1/(.01 * sqrt(2 * pi))))**2)/2)))
00:14 < kanzure> the two parameters are actually a,b where a and b are the two positional coords of the user's mouse
00:15 < kanzure> and, er, 10 and .01 were supposed t obe the same, and .1 was the other variable
00:42 -!- any24868271 is now known as katsmeow
01:54 -!- any26456771 [n=someone@75-121-60-107.dyn.centurytel.net] has joined #hplusroadmap
02:12 -!- katsmeow [n=someone@75-120-26-200.dyn.centurytel.net] has quit [Connection timed out]
02:39 -!- any99285391 [n=someone@75-120-34-164.dyn.centurytel.net] has joined #hplusroadmap
02:42 -!- any68022525 [n=someone@75-120-17-111.dyn.centurytel.net] has joined #hplusroadmap
02:53 -!- any84150457 [n=someone@75.120.32.240] has joined #hplusroadmap
02:56 -!- any26456771 [n=someone@75-121-60-107.dyn.centurytel.net] has quit [Read error: 110 (Connection timed out)]
02:57 -!- any91034331 [n=someone@75-120-47-47.dyn.centurytel.net] has joined #hplusroadmap
02:59 -!- any99285391 [n=someone@75-120-34-164.dyn.centurytel.net] has quit [Read error: 110 (Connection timed out)]
03:03 -!- any84150457 [n=someone@75.120.32.240] has quit [Read error: 145 (Connection timed out)]
03:05 -!- any68022525 [n=someone@75-120-17-111.dyn.centurytel.net] has quit [Read error: 110 (Connection timed out)]
03:16 -!- any91034331 [n=someone@75-120-47-47.dyn.centurytel.net] has quit [Read error: 110 (Connection timed out)]
06:11 -!- dira [n=chatzill@de2-as1165.alshamil.net.ae] has joined #hplusroadmap
07:37 -!- katsmeow-afk [n=someone@75-120-41-37.dyn.centurytel.net] has joined #hplusroadmap
08:32 -!- genehacker [i=genehack@wireless-128-62-58-99.public.utexas.edu] has joined #hplusroadmap
08:51 -!- genehacker [i=genehack@wireless-128-62-58-99.public.utexas.edu] has quit [Read error: 110 (Connection timed out)]
09:01 -!- Phreedom_ [n=quassel@195.216.211.175] has joined #hplusroadmap
09:02 -!- Phreedom [n=quassel@195.216.211.175] has quit [Read error: 104 (Connection reset by peer)]
09:13 -!- Phreedom_ is now known as Phreedom
09:50 -!- genehacker [i=genehack@wireless-128-62-40-12.public.utexas.edu] has joined #hplusroadmap
10:02 < genehacker> kanzure you seen the taq-purification thread?
10:04 -!- genehacker [i=genehack@wireless-128-62-40-12.public.utexas.edu] has quit []
10:04 -!- genehacker [i=genehack@wireless-128-62-40-12.public.utexas.edu] has joined #hplusroadmap
10:05 < genehacker> kanzure you seen the taq-purification thread in diybio?
10:14 < drazak> genehacker: the one from like 3 months ago? or the restriction enzyme one currently?
10:17  * katsmeow-afk wants to slap someone across the face with flowers, whoever it was that mentioned "actionary" caused me to dl gigabytes of data last nite
10:18 < katsmeow-afk> luckily, i avoided dl'ing it in 28 difefrent languages!
10:20 < genehacker> the one currently
10:24 < drazak> making your own taq or restriction enzymes isn't hard
10:24 < drazak> purifying them is
10:25 < genehacker> CORRECT
10:48 -!- davidnunez__ [n=davidnun@209-6-203-217.c3-0.smr-ubr1.sbo-smr.ma.cable.rcn.com] has joined #hplusroadmap
10:54 < kanzure>  just so that it doesn't happen again:
10:54 < kanzure> http://adl.serveftp.org/lab/researchers-to-keep-track-of
10:54 < kanzure> i'm missing a lot
10:56 < fenn> richard feynman is dead..
10:57 < kanzure> :(
10:57 < kanzure> fine. i'll relabel.
10:58 < fenn> also you shouldn't post peoples' email in plain text, it's not nice
10:58 < kanzure> it's on their website
10:58 < kanzure> but ok
10:58 < fenn> there's a war on, man
11:01 < genehacker> drexler?
11:01 < genehacker> hmmm...
11:01 < genehacker> I don't know about this one
11:01 < fenn> perhaps you'd prefer to delete drexler and add bolonkin
11:02 < kanzure> i saw the most awesome name today. 'mackinaw'
11:02 < fenn> yea bro
11:02 < genehacker> I think there was a bridge named that
11:02 < genehacker> it collapsed horribly
11:02 < fenn> gallopin girdie
11:03 < fenn> no i guess that was something else
11:04 < genehacker> yeah I guess so
11:13 < genehacker> it's talk like a pirate day
12:08 < fenn> as per recent discussion on desktop chemical synthesizers: http://fennetic.net/irc/Peter%20F.%20Hamilton%20-%20Misspent%20youth.pdf
12:38 < strages> ybit is at the makers local 256 hackerspace!
12:39 < fenn> has he made a fool of himself yet?
12:40 < genehacker> what page fenn
12:40 < genehacker> that's HUGE
12:40 -!- strages is now known as swabbie_strages
12:40 < fenn> genehacker: hmm?
12:41 < fenn> it's a novel that quantumg mentioned
12:41 < genehacker> huh?
12:41 < fenn> nevermind
12:41 < genehacker> does it have a desktop chemical synthesizer in it?
12:42 < fenn> praps
12:43 < genehacker> ok
12:48 < genehacker> dang
12:48 < genehacker> I WANT ONE
12:48 < swabbie_strages> fenn: nah he's cool
12:49 < genehacker> 256 hacker meetings?
12:50 < swabbie_strages> Hackerspaces Meetup happening this weekend at our space.  5 hackerspaces represented so far
12:50 < swabbie_strages> more on the way
12:52 -!- mason-l [n=x@202-89-188-136.static.dsl.amnet.net.au] has quit [Read error: 110 (Connection timed out)]
12:53 < genehacker> oops
12:53 < genehacker> thought you were talking about 2600 meetings
13:07 < genehacker> now if we only had programmable molecular seives
13:09 < fenn> like a membrane transport protein?
13:09 < fenn> i know it's not diamondoid cogs and wheels but it does the job
13:14 < genehacker> yeah
13:14 < genehacker> if you had a programmable molecular seive it'd work
13:14 < genehacker> thanks to the equilibrium principle
13:15 < genehacker> heck viagra can spontaneously form and unform in solution, but if you remove it you can get more
13:24 -!- any22980092 [n=someone@174.124.129.94] has joined #hplusroadmap
13:42 -!- katsmeow-afk [n=someone@75-120-41-37.dyn.centurytel.net] has quit [Read error: 110 (Connection timed out)]
14:54 -!- dira [n=chatzill@de2-as1165.alshamil.net.ae] has quit ["ChatZilla 0.9.85 [Firefox 3.5.3/20090824101458]"]
15:28 -!- wrldpc2_ [n=benny@ool-ad03fe34.dyn.optonline.net] has quit [Read error: 104 (Connection reset by peer)]
15:37 -!- any22980092 is now known as katsmeow
16:11 -!- wrldpc2 [n=benny@ool-ad03fe34.dyn.optonline.net] has joined #hplusroadmap
16:25 -!- Phreedom [n=quassel@195.216.211.175] has quit ["http://quassel-irc.org - ????????????? ??????. ?????."]
16:25 -!- Phreedom [n=quassel@195.216.211.175] has joined #hplusroadmap
16:25 -!- Phreedom [n=quassel@195.216.211.175] has quit [Client Quit]
16:25 -!- Phreedom [n=quassel@195.216.211.175] has joined #hplusroadmap
16:26 -!- Phreedom [n=quassel@195.216.211.175] has quit [Client Quit]
16:41 -!- Phreedom [n=quassel@195.216.211.175] has joined #hplusroadmap
16:51 -!- katsmeow [n=someone@174.124.129.94] has quit ["Silent gratitude isn't very much use to anyone."]
17:09 -!- dira [n=chatzill@de2-as1165.alshamil.net.ae] has joined #hplusroadmap
18:08 -!- mason-l [n=x@202-89-188-136.static.dsl.amnet.net.au] has joined #hplusroadmap
18:14 -!- Phreedom [n=quassel@195.216.211.175] has quit ["http://quassel-irc.org - ????????????? ??????. ?????."]
18:16 -!- Phreedom [n=quassel@h86-62-96-212.ln.rinet.ru] has joined #hplusroadmap
18:16 -!- Phreedom [n=quassel@h86-62-96-212.ln.rinet.ru] has quit [K-lined]
18:17 -!- Phreedom [n=quassel@195.216.211.175] has joined #hplusroadmap
18:55 < bct> hello all.
18:55 < bct> i think skdb has a lot of potential and is a very interesting idea so don't take this questtion the wrong way:
18:55 < bct> is it useful for anything right now?
19:11 < fenn> no. go away.
19:11 < fenn> :)
19:11 < kanzure> he's being mean
19:11 < kanzure> the a/c is out again
19:11 < fenn> rawr
19:12  * kanzure feeds fenn some chocolate
19:12 < fenn> i think i've had enough candy already
19:12 < kanzure> bct: it can plug stuff together
19:13 < fenn> really we just need people to start putting data into computer parseable formats.. once that's done we can massage the data however we want
19:13 < kanzure> but it has to be data
19:13 < kanzure> not a picture of a datum
19:13 < fenn> that first step is incredibly time consuming if you're trying to do everything in the universe
19:13 < bct> assuming there's data though, what can be done with it?
19:14 < fenn> then you can draw conclusions about what can be done with various sets of tools and materials
19:14 < fenn> or conversely build up a list of things needed to do some task
19:14 < fenn> like build an electric car or something
19:15 < bct> right, but that's not actually useful unless everything needed to build an electric car is readily available
19:15 < kanzure> no it tells you what you need to get
19:15 < fenn> everything needed to build a car is readily available
19:15 < bct> is there a more immediate goal than "can specify the parts for an electric car"?
19:15 < kanzure> bct: have you read the wiki page?
19:15 < fenn> yes
19:16 < fenn> specifically i want to do inventory management for fab lab activities
19:16 < bct> i think so, but let me take another look
19:16 < fenn> http://adl.serveftp.org/dokuwiki/skdb
19:16 < kanzure> i'm not really interested in making an electric car right now, fyi
19:16 < kanzure> but that should be obvious
19:17 < bct> yeah, that's why i'm asking about a more immediate goal. an car isn't realistic right now.
19:17 < fenn> i'm just tired of having to reinvent the wheel every time i try to replicate someone's results
19:18 < fenn> why isn't specifying how to build a car a realistic goal?
19:18 < fenn> ok how about a DIY CNC machine
19:18 < fenn> is that realistic/immediate enough for you?
19:19 < bct> because the current bottlenecks in diy cars aren't something that can be solved by skdb (i suspect)
19:19 < fenn> you'd be surprised
19:19 < fenn> a lot of building an electric car is sourcing components and materials
19:20 < bct> aah, does skdb handle sourcing?
19:20 < bct> because that would be pretty immediately useful.
19:20 < fenn> well, it's supposed to
19:20 < fenn> but so far we only have like two packages
19:21 < fenn> basically whatever you need to know to build something
19:21 < bct> ok, thanks. i think that helps :)
19:21 < kanzure> how about you help us out?
19:22 < bct> do you have anything in mind?
19:22 < kanzure> data
19:22 < fenn> well there's supposed to be this wishlist page
19:22 < fenn> but right now it's a bunch of crap
19:22 < bct> i'm not sure i have any useful data atm.
19:23 < kanzure> http://adl.serveftp.org/dokuwiki/skdb/wishlist
19:23 < bct> i'll take a look though, and maybe i can contribute a simple electronics project or two.
19:24 < kanzure> that would be nice
19:35 -!- ve [n=a@94-193-95-252.zone7.bethere.co.uk] has quit [Read error: 113 (No route to host)]
19:36 -!- branstrom [n=branstro@c-0f1ce055.438-1-64736c10.cust.bredbandsbolaget.se] has quit [Read error: 110 (Connection timed out)]
19:36 -!- Irssi: #hplusroadmap: Total of 30 nicks [0 ops, 0 halfops, 0 voices, 30 normal]
19:38 -!- branstrom [n=branstro@c-0f1ce055.438-1-64736c10.cust.bredbandsbolaget.se] has joined #hplusroadmap
20:02 -!- dira [n=chatzill@de2-as1165.alshamil.net.ae] has quit ["ChatZilla 0.9.85 [Firefox 3.5.3/20090824101458]"]
20:24 -!- branstrom [n=branstro@c-0f1ce055.438-1-64736c10.cust.bredbandsbolaget.se] has quit [Read error: 110 (Connection timed out)]
20:25 -!- branstrom [n=branstro@c-0f1ce055.438-1-64736c10.cust.bredbandsbolaget.se] has joined #hplusroadmap
20:25 < kanzure> "Probability minimizes the system failure probability whereas possibility maximizes the normalized deviation of the uncertain variables from their nominal values that the system can tolerate without failure."
20:39 -!- jonathan__ [n=jonathan@66-90-167-249.dyn.grandenetworks.net] has joined #hplusroadmap
20:40 < jonathan__> no one should synthesize their own oligos.   oligos are a commodity item.   price is rapidly dropping while quality is rapidly improving.  
20:41 < jonathan__> diy dna synthesis is a ridiculous idea
20:41 < jonathan__> better to work on the bigger issues, like making re-usable dna parts
20:41 < jonathan__> leave the synthesis to the commodity vendors.
20:41 < jonathan__> so there :-P
20:42 -!- kardan [n=kardan@p54BE4E32.dip.t-dialin.net] has joined #hplusroadmap
20:46 < bct> i guess what i meant to ask is: how close is skdb to being able to describe something like an arduino or a ronja?
20:50 < kanzure> bct: arduinos have .sch files so you should be able to just drop them into an skdb package
20:50 < kanzure> jonathan__: maybe you should stop your body from synthesizing dna while you're not looking
20:51 -!- branstrom_ [n=branstro@c-0f1ce055.438-1-64736c10.cust.bredbandsbolaget.se] has joined #hplusroadmap
20:51 < jonathan__> in that context, my body is a commodity vendor
20:51 < kanzure> jonathan__: hey, do you know of any good papers on transfer functions and interface compatibility and probability of (catastrophic) failure between two connecting components?
20:52 -!- branstrom [n=branstro@c-0f1ce055.438-1-64736c10.cust.bredbandsbolaget.se] has quit [Read error: 113 (No route to host)]
20:52 < jonathan__> tho only for world class dna.   arbitrary dna it refuses to make
20:52 < kanzure> making microliters of the premium every day
20:52 < jonathan__> good papers on what kinds of transfer functions
20:52  * kanzure drums his chest
20:52 < jonathan__> H(s) = X(s) / Y(s)
20:53 < kanzure> jonathan__: well there seems to be a transfer function that can be used to represent the component. then, there's some probability that the output of the component will exceed the acceptable "range" for the next connected component
20:53 < jonathan__> probability of failure would depend on the type of system
20:53 < kanzure> but it's not really a "range"
20:53 < kanzure> i wanted to originally specify it as a range between two magnitudes of some unit in skdb
20:53 < kanzure> so a pipe could take 50..100 psi or something
20:53 < kanzure> but that's not really how it works
20:54 < jonathan__> transfer functions are very complex polynomials, often dozens of vars.    so range of any 1 of those vars will affect the output correspoding to the transfer function itself
20:54 < kanzure> sure. let's assume the polynomial is or has been computed for some range of inputs.
20:54 < kanzure> given some output of the system, is it "probably going to make this interface/connection fail" ?
20:54 < jonathan__> yes, transfer functions are modelled
20:55 < kanzure> i have this whole list of search terms but i'm getting absolutely nowhere
20:55 < jonathan__> that is a very very tough q and likely there is no real answer.   because, transfer functions have very complex behavior.
20:55 < kanzure> http://adl.serveftp.org/~bryan/2009-09-18.txt to see my notes
20:55 < jonathan__> systems are designed to have self-normalizing parts.
20:56 < jonathan__> i.e. a stereo amplifier for example
20:56 < jonathan__> let's say a resistor in the sytem, of the dozens of resistors and transistors, has a crazy range
20:56 < kanzure> sure
20:57 < jonathan__> the resistor results in change in current, which will pump more current thru the transistors
20:57 < jonathan__> so in an amateur system, perhaps the system could go into overload (results in loud noise) or damping (no output)
20:57 < kanzure> hm
20:57 < kanzure> well i guess i just mean for "holy shit it will catch on fire" scenarios
20:57 < jonathan__> however....  a lot of work is done so that, i.e. transistors are self-biasing based on temperature
20:58 < kanzure> not for "everything works but the overall output is wrong"
20:58 < jonathan__> the point is that the transistor will automatically change beta (amplification) because the current is too high and the temperature changes too hot.   so the system re-normalizes
20:58 < jonathan__> it is fail safe
20:58 < jonathan__> that's the whole point of engineering
20:58 < kanzure> huh i may need to design some shitty noise generator circuits
20:59 -!- kardan| [n=kardan@p54BE5A4F.dip.t-dialin.net] has quit [Read error: 110 (Connection timed out)]
20:59 < jonathan__> amplifers physically mount transistors to the same heat sink, so all transistors will act in unison, etc. for this purpose (among other purposes)
20:59 < jonathan__> if one transistor goes into overload, the others will damp.  etc.
21:00 < kanzure> this seems to be something you design into the system?
21:00 < jonathan__> the same goes for bridge construction....  alternating cables damp the others, etc
21:00 < jonathan__> yes and it takes a lot of work
21:00 < jonathan__> now look at a s/w algorithm, lets say a feedback control loop, motor control PID, etc, those have big crazy H(s)
21:01 < jonathan__> and likewise they have self-compensating behavior
21:01 < jonathan__> for example a jolt to the system which makes it start oscillating crazy, will re-normalize to expected behavior..   auto-damping
21:01 < kanzure> sure
21:01 < kanzure> but this is before that's designed into the system
21:01 < jonathan__> negative & positive feedback
21:01 < kanzure> at the component selection stage
21:01 < jonathan__> no it's part of the algorithm
21:01 < jonathan__> it's in the math
21:02 < drazak> jonathan__: aka jonathon cline/
21:02 < kanzure> hm
21:02 < jonathan__> when systems are connected together, they use buffers in electronics
21:02 < jonathan__> so one system fails, it does not affect the next
21:03 < jonathan__> of course the input may be bad... but the crazy oscillation or whatever, does not carry thru
21:03 < jonathan__> in IC design there is some ridiculous % of buffers compared to the actual blocks
21:03 < kanzure> "blocks"?
21:04 < kanzure> ok
21:04 < jonathan__> consider that 50% of all s/w is error checking, sometimes more..  (or user arg checking)..  the actual code that does the "algorithm" is quite small usually
21:04 < jonathan__> compare to a pentium..   some ridiculous % of the chip in terms of the transistors, is all just for connecting one small system to another small system
21:04 < jonathan__> buffering
21:05 < jonathan__> so if one system gets a funny clock impulse, the entire pentium clock doesn't go crazy
21:05 < jonathan__> clock lines are horrible sources for noise
21:05 < drazak> IC's still have oscillation problems
21:05 < drazak> if you're making something with opamps, oscillation could cause the whole thing to fail
21:06 < jonathan__> if it is designed well, then it should never happen
21:06 < jonathan__> opamps are negative feedback
21:06 < drazak> if the system or the opamps?
21:06 < drazak> certain opamps are more susceptive than others
21:06 < jonathan__> for ex. the op amp will have a temperature compensating element or voltage ref.
21:07 < jonathan__> that's because the gain is so high
21:07 < drazak> right, it needs to be unity gain stabilized
21:08 < jonathan__> designers know not to use certain components for that reason, or if they are used, then they have to be biased for run-away
21:08 < jonathan__> electronics is ridiculous in the amount of fool-proofing included
21:08 < jonathan__> amazing actually
21:09 < jonathan__> of course a lot of regulation forces that as well
21:11 < jonathan__> fault tolerance is the last thing to add on to a system
21:11 < drazak> yes, but certain opamps, such as the lm617x often oscilates for a variety of reasons, but it's a very distortion free opamp and one of the best out there
21:11 < jonathan__> thats different than suggesting a component tolerance problem willl create that behavior
21:12 < jonathan__> I assume you mean will oscillate due to external fields or such
21:12 < jonathan__> electircal noise
21:12 < jonathan__> whereas bryan is saying, what happens if the 1% resistor is actually substituted for a 20%
21:13 < jonathan__> mostly though, transfer functions are for academics
21:13 < drazak> you might have issues
21:13 < drazak> that's why you should test your resistors with a watergate circuit
21:14 < jonathan__> uh, I havent heard of manufacturers testing their resistors
21:14 < drazak> er, watergate bridge, that is
21:14 < jonathan__> they build the system, then test the major input & output, that's it.  
21:15 < jonathan__> the system works or it's thrown into the recycle bin.
21:15 < jonathan__> even bridge builders, for final test, drive really really heavy trucks across the bridge.l.  that's the "test"
21:16 < drazak> well if you buy 100 resistors, you can safley assume they're from the same batch so their differences in resistance are spread out over a gaussian distribution, so if you check 10 of them and they are all within 1% of your target value then you can fairly safley assume all your resistors are right
21:16 < drazak> that's what good electronics people do
21:16 < fenn> i hear modern EDA systems actually alternate through all the high and low end of the error range for every component
21:16 < jonathan__> if it were really possible to do such analysis like "what happens to this system, if X subsystem fails" etc...    that is likely impossible, because if it were solvable, then NASA would have solved it already, rather than ditching $100 million satellites into the ocean
21:16 < drazak> they test their caps with the capacitor version of a watergate bridge
21:16 < fenn> and do a simulation to see if it still works
21:17 < fenn> "if it were possible NASA would have done it"
21:17 < fenn> i want to smack you for that
21:17 < jonathan__> instead what teams focus on, is unit test the hell out of each component
21:18 < jonathan__> then ensure the buffering systems are also good, then hook everything together.
21:18 < jonathan__> it's true
21:18 < jonathan__> they probably spend the most effort on such stuff.   who else has a system that must work otherwise all $$$ is lost?
21:18 < fenn> one of the profs working on VOICED was actually doing failure analysis for NASA
21:18 < jonathan__> without any chance of fixing it later?
21:19 < fenn> what happens to X if Y fails (and some info about how X and Y are connected and what their functions are)
21:19 < jonathan__> so, did he find a generalizable solution?   
21:19 < fenn> they were just doing data mining
21:19 < jonathan__> or was it case-by-case basis
21:20 < fenn> i don't think there was really enough data though
21:20 < fenn> NASA still uses inches and foot pounds ffs
21:21 < jonathan__> you kow, the couple japanese companies I have worked for, have been the best at testing.  they test every single little tiny thing.  and if it doesn't work, they want it fixed, and want to know why it was broken.    they focus a lot, a lot, a lot, on unit tests
21:21 < fenn> anyway i don't see why it should be impossible or even hard to model failure propagation
21:21 < jonathan__> I have heard this is part of their engineer culture 
21:22 < jonathan__> uh, because you assume there is a model that resembles the real system well enough to even matter?
21:22 < fenn> yes because the "units" are manufactured by different entities in the zaibatsu organism
21:22 < jonathan__> again, the transfer functions are for academics
21:23 < jonathan__> at a certain point, it has to be built, and big trucks drive across it
21:23 < fenn> jonathan__: what's so hard about "if i give this 5 watt resistor 20V*A it will explode"
21:23 < fenn> i'm not disagreeing that you have to actually do a physical test
21:23 < fenn> but most things are just braindead simple
21:24 < jonathan__> nothing is hard about that, it's not a transfer function either
21:24 < fenn> please explain what you think a transfer function is
21:24 < jonathan__> dont talk in circles
21:25 < fenn> well, the way i understand it, a transfer function is a model of what happens to the output, based on the input
21:25 -!- branstrom [n=branstro@c-ed1ce055.438-1-64736c10.cust.bredbandsbolaget.se] has joined #hplusroadmap
21:25 < jonathan__> ok, i'm talking about systems, and you're talking about components
21:25 < fenn> every system is a component
21:25 -!- Phreedom [n=quassel@195.216.211.175] has quit ["http://quassel-irc.org - ????????????? ??????. ?????."]
21:25 -!- Phreedom [n=quassel@195.216.211.175] has joined #hplusroadmap
21:25 -!- Phreedom [n=quassel@195.216.211.175] has quit [Client Quit]
21:26 < fenn> ok so whether a resistor will explode is not a transfer function
21:26 < fenn> but you can calculate the power dissipation and see if it exceeds the ratings
21:26 < jonathan__> when someone says transfer function, usually I envision an equation with about 20 polynomial factors, as the simplest case
21:26 < fenn> why?
21:27 < fenn> you can model things to whatever level of detail you want
21:27 < jonathan__> you're talking about a component that has a simple linear response
21:27 < fenn> oh i suppose i can add in some temperature variation
21:27 < jonathan__> sure, but the model wont be correct, and many times won't even be worth the time spent modeling it
21:27 < fenn> or some terms to represent arcing across the leads at high voltage
21:27 -!- Phreedom [n=quassel@195.216.211.175] has joined #hplusroadmap
21:28 < fenn> but it won't be worth the time spent modeling those details so i'll just use V=IR
21:28 < jonathan__> ok simplest case.   the manufacturer can not tell you anything about whether a resistor will short, or will open, if tolerance is exceeded
21:28 < jonathan__> already at the component level, the model fails
21:29 < fenn> huh?
21:29 < fenn> they always give you a power rating at some temperature
21:29 < jonathan__> "if i give this 5 watt resistor 20V*A it will explode"  =>  there are 2 choices;  it might explode, or it may short
21:29 < jonathan__> R goes to infinity or R goes to zero
21:29 < fenn> it will fail, either way
21:29 -!- branstrom_ [n=branstro@c-0f1ce055.438-1-64736c10.cust.bredbandsbolaget.se] has quit [Read error: 110 (Connection timed out)]
21:29 < jonathan__> these are very different failure modes
21:30 < jonathan__> and - the manufacturer can not provide any math that tells which will happen
21:30 < fenn> "can not" or "will not"
21:30 < jonathan__> so what happens to "2 components connected together"  already gets more complex, just with 1 resistor
21:30 < jonathan__> they can not
21:31 < jonathan__> there is no such physics
21:31 < fenn> bah
21:31 < fenn> anyway i'm not after comprehensive failure analysis
21:31 < jonathan__> you can buy special "will always explode" resistors at extra cost, that are manufactured in a way that exceeding current will guarantee "open" at some probability like 99.91% of the time or something
21:32 < fenn> i just want to know if the thing should work correctly if put together
21:32 < jonathan__> then?  just use PSPICE
21:32 < fenn> sure, spice works fine for EE
21:32 < jonathan__> however some circuits are not modellable in PSPICE
21:32 < fenn> nobody has bothered to write spice for legos or biobricks or whatever
21:32 < jonathan__> the simple voltage doubler I posted on my blog, "can't be modelled"
21:32 < fenn> why not?
21:33 < jonathan__> because it depends on inherent inductance in the leads of the diodes and such
21:33 < fenn> i can do it in falstad.com circuit simulator java applet..
21:33 < fenn> oh pff
21:33 < jonathan__> it depends on "real world factors" not the transfer function
21:33 < fenn> you're ignoring reality now
21:34 < jonathan__> try it in spice and see for yourself
21:34 < fenn> i'll take your word for it
21:34 < jonathan__> how about some ME stuff..  like fermentors
21:34 < fenn> um, a big steel tank?
21:34 < jonathan__> why cant bio guys grow biodiesel in bulk enough to feed all the worlds cars
21:35 < jonathan__> because they can't model the fermentors
21:35 < kanzure> no
21:35 < fenn> what fermentors?
21:35 < jonathan__> yeast in big steel tank
21:35 < kanzure> (the real problem is that they don't model anything. i've dropped off my local biofuel team.)(
21:35 < jonathan__> their "simple transfer functions" dont scale
21:35 < fenn> you have to feed the yeast something, that's the real problem
21:35 < jonathan__> no, its not that simple "you just need to feed them more"
21:35 < fenn> growing corn to turn to alcohol to burn in a petrol engine is what won't scale
21:36 < jonathan__> even if it did, they wouldnt be able to, today
21:36 < jonathan__> however they dont frequently admit this because they want $$
21:36 < fenn> forgive me but i find it hard to believe
21:36 < jonathan__> and, they want to further the modeling research
21:37 < fenn> people have been fermenting mash for thousands of years
21:37 < fenn> scientifically for hundreds of years
21:37 < jonathan__> sure, and at what scale?
21:37 < fenn> you're telling me there are no good models of fermentation?
21:37 < jonathan__> i'm saying it doesn't scale to the size needed
21:37 < jonathan__> also known as "it works great in the lab but not in real world"
21:38 < fenn> fermentation doesn't scale? have you been to milwaukee?
21:39 < jonathan__> i'm talking about a different defiition of scaling.
21:39 < jonathan__> "scaling" does not mean:   it works for 1 tank so use 1,000,000 tanks
21:39 < jonathan__> "scaling" means:  it works for 1 tank so use a tank 1,000,000 bigger
21:39 < jonathan__> which is what is required to be cost competitive
21:40 < jonathan__> anyway the point is:  they have a transfer function that works for a tiny lab beaker "model".  then they go to a bigger tank and suddenly the model is different.  etc.
21:40 < jonathan__> classic ME problem
21:41 < jonathan__> "(the real problem is that they don't model anything. i've dropped off my local biofuel team.)("   ==> they would model things if they could, most likely, if it was worth the time
21:42 < jonathan__> laziness aside, likely they choose not to model, because it's a waste of time, and easier to build it & measure directly
21:42 < kanzure> i'd rather not go into the horror stories of yet-another-biology-lab
21:42 < kanzure> measurement doesn't mean it will be the same measurement in the future ..
21:42 < kanzure> if you have no model how do you know what is a local effect and what is going to probably happen again?
21:42 < fenn> so why is it we can model combustion with CFD but not fermentation?
21:43 < jonathan__> physical size
21:43 < jonathan__> what's the size of the combustion chamber?
21:43 < fenn> a gazillion teracubits cubed
21:44 < fenn> seriously what does it matter?
21:44 < jonathan__> thermodynamics people will be the first to admit they have a very hard time modeling
21:44 < jonathan__> because they can't model it!  thats why 
21:44 < jonathan__> tiny = sometimes much easier to model
21:44 < fenn> the nuclear bomb simulations looked pretty convincing
21:44 < jonathan__> bigger = too many variables to model, system behaves other than expected
21:45 < jonathan__> it depends on the system and the reaction obvioulsy
21:45 < jonathan__> microfluidics = tough to model apparently
21:45 < fenn> fluids tend to behave differently at different scales
21:45 < fenn> but why would a fermenter?
21:45 < jonathan__> something about the coefficients and volume vs. size
21:46 < jonathan__> the most recent talk on biofermenters by the top guys on bio fuels, the guy basically said:  "we don't know what we don't know yet"
21:46 < fenn> btw i don't get why this is mechanical engineering
21:46 < jonathan__> a lot of it has to do with mixing 
21:46 < jonathan__> the stuff in the fermentor mixes easily at small scale, but not at big
21:47 < kanzure> they certainly don't teach me about this in the ME classes
21:47 < fenn> well how are they mixing it? one big propeller blade?
21:47 < jonathan__> that's as much as I know though...  same thing about volume vs size
21:47 < fenn> i mean obviously you have to pay attention to the reynold's number
21:48 < jonathan__> there are a lot of different types and shapes etc of blades, that in itself is apparentlyquite a science
21:48 < fenn> not much of a science if they don't have a model of it
21:48 < jonathan__> they have "a" model...   just not a 'very good' model
21:48 < jonathan__> combustion engine is a good example too, there are not good models as far as I know
21:49 < fenn> ok so how does this relate to transfer functions and whether stuff will nominally work or not?
21:49 < jonathan__> "good" meaning, something > 2nd order effects
21:50 < jonathan__> it's all over drew endy's talks, ha, he loves saying "to a first order approximation"
21:50 < jonathan__> sure, it works, to a first order approximation
21:50 < jonathan__> oh yea, I can turn myself into a tomato, to a first order approximation
21:52 < jonathan__> it relates because, how can a modeling system work, if you cant even model if the resistor will short, or will open, when it fails?
21:52 < jonathan__> how can it work when the system is scaled up or down?
21:52 < fenn> because i don't care what happens after it fails
21:52 < fenn> i just want to know if it will fail or not
21:52 < fenn> "failure is not an option"
21:52 < jonathan__> "the answer is probably yes but no one really knows"
21:53 < jonathan__> "the system is designed to withstand simple component failure"
21:53 < fenn> i'm fine with that answer
21:53 < fenn> because it's better than no answer
21:53 < jonathan__> also the exampe you said about voltage arcing?   well no one knows about voltage arcs
21:53 < fenn> bullshit
21:53 < fenn> put 200kV through a standard 1/4 watt resistor and it'll arc
21:53 < jonathan__> maybe if you factor in humidity, then you can get some "first order approximation"
21:54 < jonathan__> the examples you are giving are way outside 6 sigma
21:54 < jonathan__> they are not normal falure
21:54 < fenn> how precise is precise enough?
21:54 < jonathan__> and I'm quite sure 200kV can be put thru a 1/4 watt resistor as long as the 200kV is only carried on 1pA current
21:54 < jonathan__> I thought you guys were being realistic
21:55 < fenn> well sure but what is "realistic"
21:55 < jonathan__> i mean, real world
21:55 < jonathan__> suppose a component is within 1 sigma tolerance
21:55 < jonathan__> then the general failure would be to put that component at 3 sigma
21:56 < jonathan__> not way out at 6
21:56 < fenn> er. what does failure have to do with the component's manufacturing error?
21:56 < fenn> i mean you don't manufacture a resistor with a built-in voltage
21:56 < fenn> it has a resistance
21:57 < fenn> that resistance is +- 5% usually
21:57 < fenn> so sigma = 5% right?
21:57 < fenn> ok let's pretend it's a 1 watt 1 ohm resistor
21:58 < fenn> i'm putting 1V through it so i should get 1W, right on the borderline
21:58 < jonathan__> sigma refers to the probability that it will be within the stated value
21:58 < fenn> if resistance is too high, it works, too low and it fails
21:59 < fenn> so if i say "valid voltage range is 0 .. 0.95V" is that "realistic"?
22:00 < jonathan__> depending on temperature too
22:00 < jonathan__> and also depending on the thermal conductivity of surrounding environment, including the board it is resting on
22:00 < fenn> or does it have to be 0 .. 0.7V (six * 5%)
22:01 < jonathan__> and also, resistors will fail more often if they are mounted vertically due to heat, which is why they are mounted horizontally
22:02 < jonathan__> if you are making design rules, then you want "head room"
22:02 < jonathan__> the head room depends on the kind of engineering.   in power systems they want more head room since failure can be dangerous.  NASA or DOD also likes a lot of head room
22:03 < fenn> yeah safety factor
22:03 < jonathan__> so you might say "30% head room"
22:03 < jonathan__> that is on top of the component variance
22:03 < fenn> safety factor for missiles is only 2
22:04 < jonathan__> safety factor for freeway overpass is only 30%
22:04 < fenn> really?
22:04 < jonathan__> I dont care about missles beause I dont deal with them every day...   freeway overpass however...
22:04 < fenn> i thought civil engineering was all about overkill
22:04 < jonathan__> I dunno if it's 30%.  standard ME factor is usually 30%
22:04 < fenn> no way
22:04 < fenn> it's like 7x
22:05 < fenn> 10x if catastrophic failure is possible
22:05 < jonathan__> 7x of what,  of my car?  etc
22:05 < fenn> 7x the design load
22:05 < fenn> vs breaking strength of materials
22:05 < fenn> nominal strength, whatever that means
22:05 < fenn> i don't know that much about how statistics is used in these guide books
22:06 < fenn> obviously not all batches of steel are the same
22:07 < jonathan__> hmmmmmm
22:08 < jonathan__> so if the speed limit is 65mph then I can go 7x right?
22:08 < jonathan__> i'll tell the hiway patrol man next time
22:09 < fenn> you can't always do 7x safety factor
22:09 < fenn> like speed limit is usually 40 .. 65 mph on freeways
22:10 < fenn> anyway it's a rule of thumb, probably based on the statistical variation in loads and material quality
22:10 < fenn> sometimes you just don't have a model though
22:39 < jonathan__> back to the original point, "well there seems to be a transfer function that can be used to represent the component. then, there's some probability that the output of the component will exceed the acceptable "range" for the next connected component"  ==>  the mode of failure is often unknown, so there is no way to know
22:40 < jonathan__> that's assuiming there is some given probability curve by the manufacturer
22:40 < fenn> you're saying we don't know the failure modes for a resistor?
22:40 < jonathan__> or that's assuming some given safety margin is acceptable or will have some desired effect
22:40 < jonathan__> uh, I already explained that
22:41 < jonathan__> 1) short, 2) open, maybe 3) some other value
22:41 < jonathan__> you do not know the probabilities of these events
22:42 < jonathan__> you could run the model for all 3 of course...   for all components in the system....   and then somehow group all the N answers together somehow..
22:43 < jonathan__> assuming that the system even fails (i.e. could not compensate for failure)
22:43 < genehacker> PROBABILISTIC FAILURE
22:44 < jonathan__> also, each component would have to be labelled as:  "required for operation, or just included for safety purposes"
22:44 < fenn> if you have a resistor inline with a signal that fails open, that will probably screw up the system
22:44 < fenn> guh.. this is all about failure modes and i don't care
22:45 < fenn> i just want to know if things are compatible
22:45 < jonathan__> sometimes you can pull parts out of circuits on purpose and they still work fine, ha
22:45 < fenn> sure the chinese do it all the time
22:45 < fenn> you can also run a GA and totally change the circuit to something else
22:45 < fenn> is that failure analysis? i don't think so
22:46 < genehacker> oh I know
22:46 < genehacker> why don't we say not intended for life-support or critical systems
22:46 < jonathan__> pfff
22:46 < jonathan__> that's all over every open source license ever
22:46 < jonathan__> no warranteee express or implied etc etc etc
22:46 < fenn> NO WARRANTY OR SUITABILITY FOR ANY PARTICULAR PURPOSE
22:47 < jonathan__> it was berkeley's #1 requirement for open source licensing way back when
22:47 < jonathan__> and mit's
22:47 < jonathan__> the lawyers insisted apparently
22:48 < jonathan__> suspiciously missing from biobrick legal discussions, too
22:48 < jonathan__> well not "suspicious" just "odd"
22:49 < fenn> i don't really understand the purpose in the first place
22:49 < jonathan__> I would like to know all the sizes & shapes of pcr tubes that people do pcr in.
22:49 < fenn> me too
22:49 < fenn> then i'd package it up in skdb
22:50 < fenn> then anyone else who wanted to know could look it up
22:50 < fenn> i'm doing that for nuts and bolts but after a week fighting with the geometry kernel i've put it aside in disgust
22:51 < fenn> apparently modeling a helix is difficult
22:52 < jonathan__> uh?   why isn't standard size sufficient?
22:52 < fenn> because it doesn't tell you anything about the actual geometry
22:53 < fenn> just a bunch of numbers, searle chinese room argument, etc
22:53 < jonathan__> isnt that all standardized though?
22:53 < fenn> yes, that's why it's interesting data
22:53 < fenn> if it were some random part that nobody uses, nobody would care would they?
22:53 < jonathan__> http://www.biocompare.com/ProductListings/2254/PCR-Tubes-%2802ml.html
22:54 < genehacker> now where do we get zinc nanoparticles?
22:55 < genehacker> besides kook medical products?
22:55 < jonathan__> I dunno what manufacturer is preferred in MBB
22:55 < fenn> what's MBB?
22:55 < jonathan__> building name
22:55 < kanzure> molecular biology building
22:55 < fenn> you should start saving samples of actual tubes
22:55 < fenn> then we can measure them
22:56 < jonathan__> compatibility chart http://www.nuncbrand.com/us/frame.aspx?ID=602
22:56 < jonathan__> b.s. measure them
22:56 < jonathan__> get the manufacturer data sheet
22:56 < jonathan__> mechanical drawing
22:57 < fenn> why does it say 96 well plate if it's talking about tubes?
22:57 < genehacker> current topic?
22:58 < jonathan__> it says at the top, "all tubes are compatible with all...."
22:58 < jonathan__> plates however, seem to vary +/- several mm's
22:58 < fenn> jonathan__: didn't you mention that needles were something like +-50% the stated diameter? 
22:59 < jonathan__> according to some lab's paper, yes for needles
22:59 < jonathan__> but those are metal
22:59 < jonathan__> I am sure the plastic parts do not have such variance
22:59 < fenn> well i'm not so certain
22:59 < jonathan__> who knows how the metal is made..  
22:59 < fenn> and without actual samples i'll never know
23:00 < jonathan__> I am pretty sure the plates have low variance
23:00 < jonathan__> the robot arm has resolution steps of 0.5mm
23:00 < fenn> actually i was intrigued by how to make hypodermic needle tubing so i searched and finally found it
23:00 < jonathan__> and between plates it is consistently correct
23:00 < fenn> they use this little double conical piece that floats inside the tubing as they draw it down like wire
23:01 < fenn> the angles are very important and it wears out after a while so it's made of carbide
23:01 < fenn> it's called a floating mandrel
23:01 < genehacker> I think the needles are made by some cold or hot work process
23:02 < genehacker> if you could send me a sample I might be able to do some analysis on them in the metal lab
23:02 < fenn> yeah, right
23:02 < jonathan__> ha, I am not going to mail needles thru USPS
23:02 < fenn> he's on campus
23:02 < genehacker> yup
23:02 < fenn> i'm just a cynical bastard, that's all
23:02 < genehacker> not far from the mol bio building
23:03 < genehacker> I might be able too
23:03 < jonathan__> well if you want to, I will round some up
23:03 < genehacker> I'd probably have to schedule time on the sem
23:03 < genehacker> and I don't know what etches whatever the needles are made of so...
23:03 < jonathan__> just make sure that you dispose them in some biohazard waste
23:04 < genehacker> I'd have to cut the needle
23:04 < jonathan__> or I suppose cut them up into bits
23:04 < genehacker> polish it
23:04 < fenn> BURN THEM
23:04 < jonathan__> at what temp?
23:04 < fenn> is "hell" a temperature?
23:04 < genehacker> put it phenolic resin
23:04 < genehacker> put it in phenolic then polishit
23:04 < jonathan__> I thik it would melt prior to burn lol
23:04 < genehacker> room temperature
23:05 < genehacker> needles are pretty small that'd be hard
23:05 < jonathan__> hm, you could tr different temps if you want, see how much it expands
23:05 < fenn> weird, this wikipedia page on tube drawing references the book i've been using
23:05 < jonathan__> "hell" is actually cold, so the temp is low anyway
23:05 < genehacker> give me the metal and it's thermal expansion coefficient and it's easy to figure out
23:06 < fenn> is there a unit conversion for the different levels of cold?
23:06 < fenn> erm. s/cold/hell/
23:06 < jonathan__> only for solid metals tho right?
23:06 < fenn> units tempH(7) tempC
23:06 < genehacker> what did hell freeze over or something?
23:07 < jonathan__> the units for hell are something like...   "hell", "damn hell", "f*ing hell", "m*f*ing hell", "god damn m*f*ing hell", .....
23:07 < jonathan__> read dante
23:07 < fenn> no those are prefixes
23:07 < genehacker> ok
23:07 < jonathan__> there are levels of hell
23:07 < jonathan__> circles actually
23:07 < kanzure> then where's the pi?
23:07 < fenn> abbreviated H, dH, fH, mfH, GdmfH
23:08 < drazak> man it's a bitch to add a bunch of parts to mouser
23:08 < genehacker> ok back to topic
23:08 < drazak> (you guys reminded me about a headphone amp project I started thinking about like 2 years ago)
23:08 < fenn> you and your topices
23:08 < jonathan__> ummm I think the radius of hell can't exceed the earth's radius, because the inner levels of hell are explicitly at the center of the earth
23:09 < fenn> but the earth is flat so its radius is essentially infinite
23:09 < drazak> I need to figure out all these resistor values too
23:09 < fenn> (essentially because only God is infinite, of course)
23:09 < jonathan__> ummmmm  i dont think earth is flat in dante
23:09 < fenn> no?
23:10 < fenn> does the bible say anything about this?
23:10 < genehacker> no hell is underground and had nuclear fusion reactors
23:10 < genehacker> but got invaded and moved to somewhere else
23:10 < jonathan__> well, to get to hell, you need to pass the undead boat man, and everyone knows that boat men are sailors, and sailors all know the earth is round
23:11 < jonathan__> anyway, how could it be a circle of hell on a flat plane, hmm
23:12 < fenn> i think you're thinking of pellucidar
23:13 < genehacker> hey don't you mean go down a well and fight demons and stuff?
23:13 < jonathan__> this isnt lord of the rings
23:13 < fenn> a well? that only goes down a couple hundred feet at most
23:13 < genehacker> a well that is incredibly deep
23:13 < fenn> like mel's hole
23:14 < fenn> or that borehole in russia
23:14 < fenn> that turned out to be a hoax
23:15  * fenn returns to his book
23:15 < jonathan__> well according to journey to the center of the earth.....
23:16 < jonathan__> oh here's wikipedia to the rescue.  "The Purgatorio is notable for demonstrating the medieval knowledge of a spherical Earth."
23:17 < genehacker> no this is a different reference it's a joke
23:17 < jonathan__> $self->sleep( 60*60*8)
23:17 < genehacker> sleep is for slackers
23:18 < genehacker> what are the needles made of?
23:19 < genehacker> or do I need to put it in the sem and have it analyzed?
23:20 < fenn> sounds like something you should know how to look up
23:20 < fenn> (i actually know the answer, i'm just not telling)
23:27 < genehacker> stainless steel