#homecmos IRC log for Thursday, 2011-12-22

--- Thu Dec 22 201100:00
azonenbergHi guys14:52
azonenbergJust built myself a new toy, http://i.imgur.com/TvI3s.jpg14:53
azonenbergexperiments will be resuming shortly14:53
soul-dnice whats the cabin material made off14:54
azonenberg1/8" PVC from Mcmaster-Carr, held together with L-brackets14:54
azonenberg1/8" clear acrylic on the sash14:54
azonenberg240cfm duct fan14:55
azonenbergwhcih on the 1x2 foot opening gives 120 linear FPM, minus a bit for the cracks (joints arent caulked etc) and you get 80-10014:55
azonenbergwell within the safe range14:55
azonenbergi tested with IPA and acetone and it kept concentrations well below the odor threshold at all times14:55
azonenbergwhcih, depending on who you ask, is 40-100 PPM for acetone and the safe exposure limit is 500-1000 depending on the regulators you ask14:56
azonenbergSo i'm well within safe limits14:56
azonenbergIOW, the thing actually works14:56
soul-dnice  had 500m3 somthing  fan deliverd today :P  probably overkill for that ;)14:57
soul-di take those numbers are from woerking area14:57
azonenbergYou actually dont want too high or you get fortices14:57
soul-degg 8 hour work day ?14:57
azonenbergwhich have a habit of leaking nasties out of the hood on the edges14:57
azonenberghecne the 120fpm cap14:57
azonenbergand yes, the PEL i quoted was 8-hour time weighted14:58
azonenbergthe MSDS listed three limits depending on agency but they were all 500-1000 PPM range for acetone14:58
azonenbergiirc all of the chemicals i work with except HF have the PEL significantly above the odor threshold14:58
azonenbergMeaning if you cant smell it, its safe14:58
azonenbergI wanted to test first with relatively safe substances like IPA14:59
azonenbergThe next test will use something stronger like concentrated HCl15:00
azonenbergif i cant smell that at all from outside the hood i'm good :)15:00
azonenbergOdor threshold 0.25 to 10 PPM15:01
azonenbergpersonally i think the 10 is high, that stuff is pretty potent15:01
soul-ddonno if i smelled that before id probably try to convert some amonia fertilizer to KNO3  since  ive done that before15:02
azonenbergConcentrated hydrochloric acid is not something you are going to forget the smell of lol15:02
soul-dbut then again chemistry is on hold still15:02
soul-donly have 30% and 10%15:02
soul-dbut never opened the 30%  yet15:03
azonenbergMine is 32% i think15:03
azonenbergAnd do so with caution15:03
azonenbergIt will most likely fume strongly15:03
soul-dit's 5 liters to :P  havent touched it since i bought i lol  donno even why :D15:03
azonenbergFive liters??15:03
azonenbergMine is a 1L bottle  and i thought that was a lot15:03
soul-dstron concetrations only came in this package15:03
azonenbergi had trouble buying any less except in very high purity15:04
soul-dlol ?15:04
azonenbergnot as in concentration15:04
azonenbergbut like ACS reagent grade or trace metals basis15:04
azonenbergyou can get in half-liter bottles15:04
azonenbergTechnical grade the smallest i could get was 1L15:04
soul-di had trouble finding it :)15:04
azonenbergwhere are you located again?15:04
soul-doh still is wall cleaning grade15:04
azonenbergWell it's technical, but pretty high grade15:04
azonenbergas far as technical goes15:04
azonenbergI intend to buy some ACS grade when this runs out15:05
azonenbergbut it's lasting me forever15:05
azonenbergjust doing PCB etching etc with it15:05
soul-dstill use  ferric chloride for that15:05
azonenbergi use one part conc HCl to 6 parts 3% H2O215:05
azonenberggently heated15:05
azonenbergit's stronger and also transparent15:06
azonenbergso you can watch the board etch15:06
soul-dprobably be easyer to follow along true  but i assumed  that  stuff was bit more nasty then   this brown stuff15:07
azonenbergYou do need better ventilation15:08
azonenbergBut if you have it, the etch quality is superior15:08
azonenbergit also doesnt get rust stains on your glassware15:08
azonenbergcleans off nicely, all of the waste products are water soluble15:08
soul-dalso have the sodium perchlorate stuff for etching somewhere15:08
azonenbergpersulfate you mean?15:08
azonenbergperchlorate would probably blow up, not etch :p15:08
soul-dyes that srry :)15:08
azonenbergAnd when i'm done etching i just store the CuxClx contaminated waste in a jar until the next scheduled hazmat pickup15:09
azonenbergthe county has one or two a year15:09
soul-dwe do have a disposal thingy for chemicals15:09
azonenbergYeah, I'm trying to run my lab as by-the-book as I can15:10
azonenbergeven if the volumes involved probably arent large enough to matter15:10
azonenbergrather not take chances15:10
azonenbergAnd its a good habit to get into15:10
azonenbergfor if i end up working in a commercial lab or handling larger volumes of stuff15:11
soul-dyeah thats why  i stopped until  reasonable buget/safe enviroment  since im pretty clueless15:11
soul-dproblem is hard to do self study15:11
azonenbergI figured an open window was probably good enough for what i was doing, not handling anything too nasty15:11
azonenbergBut better engineering controls never hurt15:11
azonenbergSo i decided to build the hood15:11
azonenbergNext step is going to be to use the aluminum stock i just ordered to machine a little jig15:12
azonenbergto hold my wafer over the evaporator basket so i can do nickel deposition15:12
azonenbergit sublimes so i cant use the tool's normal under-basket moun15:12
azonenbergwhich expects a material that melts before evaporating15:12
azonenbergAt that point i'll be good to go for through-wafer etch testing15:13
azonenbergIn parallel i have some ideas for improving lithography resolution15:13
azonenbergtheoretically down to 330nm15:13
azonenbergbut realistically i think 500-750 will be a limit with my optics15:13
soul-dyou where at 22um now ?15:14
azonenbergIt was nominally the 20um node but i think the exact half-pitch was 21 and change15:14
azonenberggenerally overetch meant features were a little smaller than that and spaces a little larger15:14
azonenbergThis is doing 10x reduction of a 200um mask15:14
azonenbergI'm going to buy 12.5um masks from a real mask shop15:15
azonenbergthen try 4/10/40x reduction of them as well as contact printing15:15
azonenbergI dont know if i can project 12.5um features so i'm going to put some larger features on the mask too15:15
azonenbergprobably 25 and 5015:15
azonenbergi'm quite certain i can resolve 50 on the mask, times 10x reduction is 5um15:15
azonenbergso i can hit that as a minimum15:15
azonenbergbut if i use the 40x objective (which has proven flaky in the past) i could potentially go 12.5um / 40x = ~330nm15:16
azonenbergagain thats a theoretical number and i think 500-750 will be a practical limit15:16
azonenbergBut submicron would be very nice15:16
soul-dtrue, spend bit to much on electronics so most projects came to grinding halt  also since  pc is getting old and buggy don't want to code to much on it15:20
soul-dmostly scavenging stuff that  could be used15:20
soul-dwas even looking back at  amature telescope making  for fun  want to do some mirrors myself15:21
azonenbergWell, if you decide to build an evaporation or sputtering rig15:25
azonenberglet me know lol15:25
azonenbergwe can collaborate on design15:25
soul-dlong way to go still but shall see what 2012 will bring  i do want to build more stuff15:28
soul-dboth of those where with a vacume right   only idea i had for evap is  a basic  high temp oven schematic 40 years old :P carbon rods and lots of power15:29
soul-dbut i want allot  like  a ballanced  buget book, get piece of community land to grow own food   budget for rehabilitation  need to gain lots of (healty ) weight15:41
soul-danything that is left over is for learning purposes and projects :)15:45
azonenbergNo, for evap there's a bit more involved15:59
azonenbergYou want to use a tungsten filament coiled into a basket15:59
azonenbergYou can buy them from Ted Pella among other vendors15:59
azonenbergPut pellets or wire of your material inside it15:59
azonenbergThen pump the chamber down to the 1E-6 torr range15:59
azonenbergand run current trough the filament16:00
azonenbergGenerally speaking you'll want a couple of volts at pretty high current16:00
azonenbergtens to hundreds of amps16:00
azonenberggeneral control method is a variac off of wall power to control voltage, followed by a step-down transformer16:00
soul-dk yeah still have my eyes out on  vacume sture  thats probably the hardest part to get right16:09
azonenberga large enough bell jar will cost $$, yes16:09
azonenbergthen you'd need a roughing pump and a high-vac pump, probably a diffusion pump16:09
azonenbergyou cant get deep enough to evaporate with only a mechanical pump16:09
soul-d2 stage or somthing i thought you said ?16:09
azonenbergI have a 2-stage mechanical pupm but that only hits 4E-2 torr16:10
azonenbergyou have to be able to hit the -6 range which means a diffusion or turbopump16:10
azonenbergthe second are very expensive so for hobbyists it basically means you need a diffusion16:10
azonenbergSputtering uses much less deep vacuum, you can hit that with a mechanical pump16:10
azonenbergBut you need a source of inert gas (usually argon) and high voltage16:10
azonenbergSo it poses its own set of problems16:10
soul-dhigh voltage is easy  depending of current16:11
soul-dgot 5 kv  transformers :P16:11
azonenbergDepends on what you're sputtering, even 500 - 1kV might be enough16:11
azonenbergBut you need tens to low hundreds of watts16:11
azonenbergto power the plasma16:11
azonenbergThis assumes DC sputtering, not RF16:11
azonenbergalso, if you are doing a deposition of more than a small amount of material you need to cool the target16:12
azonenbergso it wont metl16:12
azonenbergsmall meaning <50nm probably16:12
soul-dmy kv transformers are 400watts16:12
soul-d3-4,5 kv16:13
azonenbergYou might not want that much voltage depending on the material, i am more familiar with the basics of the process and not specific parameters so i could be wrong16:13
soul-dbut thats for lighting L(16:13
azonenbergBut the operating principle is similar to that of a neon sign16:13
azonenbergyou strike a plasma through low-pressure inert gas16:13
azonenbergin fact, the plasma will glow very much like a neon sign16:14
soul-dthere are plasma light's there day's16:14
soul-dbut expensive to steal the transformer from16:14
azonenbergThe only difference is that you tune the parameters for removing of material from the target (and depositing on your sample) and not for maximum brightness16:14
soul-dbut will be along way before i start playing with high voltage though :)  light bulb is bit easyers  then above :P16:17
azonenbergYeah, i dont trust my HV skills yet16:19
soul-dbut this high rating  tranformers start at 70watt16:19
soul-dfor lights16:19
azonenbergi want to start slow and make sure i'm not going to electrocute myself :p16:19
soul-dhps  and mh16:19
soul-dhpresure sodium and metalic halide or somthing16:20
azonenbergi see16:20
soul-dand 2 kinds electrincs ones i think done  by som swithcing stuff or simple  bloks of copper types ( like for TL  but but bigger )16:23
soul-dmaybe look at what type of componets they use in electric types16:23
azonenbergI'm not too familiar with HV in general so i cant say16:25
soul-dme neither just hook up the light to grow peppers under it :P16:25
soul-dand glad i don't get shocked bu the system lol16:26
soul-dbut still turn on all high voltage stuff on from a distance16:31
azonenbergLol, yeah16:32
azonenbergGood practice16:32
soul-dyeah ever since i had to quickly  try  laserprinter 's stepper motor and plugged it in  wrong way around   chips  do bang loud :)16:34
soul-don  the printers main board *16:34
soul-dbut i should make a picture of my shed :P16:40
soul-dyou probably  call it chemical hazard lol16:40
soul-dnow pots and pan's with   kno3 and sugar   to :)   as smoke stuff to play around on new year a bit    don't have a ball mill otherwise would made some candles16:43
azonenberglol, ok thats pretty bad compared to my setup :p16:44
soul-dlol  no it's worse16:44
soul-dmaybe ill  blur an image :P16:45
soul-doh on picture it looks ever worse then in real :P16:47
soul-dbut it will show you why im waiting on buget to fix the shed :P16:47
soul-dim so glad im not my neighbour -> http://i.imgur.com/QPs8F.jpg16:48
soul-dmost of it is junk anyhow :P  need to bring it away  some day16:49
soul-dnow ofcourse you where wondering what the oven is for :P16:50
soul-di used that  to heat kno3 and sugar :P16:50
azonenberglol, ok16:50
azonenbergyeah, my place looks nowhere near that bad16:51
azonenbergi'm trying hard to make sure i'm as by-the-book as possible16:51
soul-dusaly  before i do somthing i first need half a day  to clean16:51
azonenberglol wow16:52
soul-dif i knew winter was gonna be this hot16:53
soul-di would have had it clean though16:53
azonenbergI'm actually in the process of tidying up the whole apt, roommates are gone and the semester is over16:53
azonenbergSo now i can bring the place up to my usual standards16:54
azonenbergi'll post some pics including all of the new lab equipment once the cleaning is done16:54
soul-dyeah did  home partly  especialy since bought gazillion electrionics i need to get stuf sorted16:54
soul-dclean sorted work area does work  better16:55
soul-dand less stress full16:55
soul-dbut i need a reasonable buget16:56
soul-dnot gonan fix that  with few hunderd bucks16:56
soul-dsince its single walled shed  you cant realy store much17:02
soul-dmetals rust17:02
soul-dread the vacume pdf  for a bit  so if i understand that system only reaches  the wanted  vacume if chilled (probably by liquid nitrogen ) ?17:41
superkuhYes. You need to cool the zeolite (13x).17:42
superkuhThe conductance between the zeolite and main chamber should be as large as possible as well.17:43
soul-d13x stands for some industrie standard form of the zeolite stuff ?17:44
superkuhYes. If I recall correctly it absorbs nitrogen and oxygen fairly well within the structure. I cannot recall the exact width of the cavities, some tens of nm.17:46
superkuhEr, wrong.17:47
superkuhSorry. angstroms.17:47
soul-di saw 8 passing by on google17:47
soul-dk then makes sense why azonenberg  looks at sputtering :)17:51
soul-dbtw you have glassblowers near you that  make chemisty labware on design?  cause i actually do have a few around me17:54
superkuhMe? I could probably find one at the state university that's 100m away from where I am sitting. But it's too expensive.18:03
soul-dheh true it would be expensive for custom stuff   but this shops build chemistry labware and   gassious lighting  so they probably  could make half sputtering thingy from the store18:08
azonenbergsoul-d: A sputtering apparatus would only need a bell jar and a baseplate, both of which can be bought premade18:20
azonenbergthe only custom work would be drilling holes in the base plate and inserting high-voltage feedthroughs18:20
azonenbergand sealing appropriately to hold vacuum18:20
soul-dmmm, don't forget im in tiny country only jar bell i found is  cheap bottle looking one with no information18:28
soul-dhttp://www.labstuff.nl/contents/media/1g.jpg  <- is large picture  and looks just like a bottle to me :)18:29
smedingheh, labstuff18:35
smedinglove that place though i do not have the money to buy stuff there18:35
smedingor enough interest, really, a lot of the time18:35
soul-dneed some food  bbl18:39
berndjis the "snooker ball" model for calculating mean free path sorta sane for estimating the MFP of an electron in a gas?21:21
azonenbergHmm, maybe? I'm not too familiar with electrostatic interactions21:22
berndjjust thinking of how to bootstrap a high vacuum vessel22:01
berndjsince apparently welds are really, really hard to make properly gas-tight22:02
berndjbut if electrons can get far enough, why not use a big beam of those to suture up the crude weld from the inside?22:02
azonenbergHmm, no idea22:11
azonenbergNever tried building a pressure vessel22:11
azonenbergalways bought them22:11
berndji'm pretty sure i can't afford one :)23:08
soul-di thought id download macguyver to inspire me  but it was dubbed in french :(23:09
soul-dotherwise id probably make one with an elastic band some tape and a coke can23:09
berndjthat was my favourite show when i was a kid. i DON'T want to see reruns now and ruin the memory23:09
soul-dtrue :)23:10
soul-dbut din't see them all :P23:10
azonenbergWell right now I'm busy working on something unrelated to IC fab (yes, I do other things too :P)23:10
azonenbergFPGA switching fabric design :D23:10
soul-dalthough i have studied fpga a bit it doesn't tell me anything :P   you mean the underlying tech to make the general logic ?23:11
azonenbergi mean a 16 port 12.8 Gbps-per-port switching fabric23:12
azonenbergto connect various elements on my SoC23:13
soul-dk, not that deep in designing yet  so can't visualize it :P23:19
azonenbergIt's basically the equivalent of a 16-port 10gbe switch23:19
azonenbergExcept each packet is, instead of an ethernet frame, my own interconnect format23:20
azonenbergwhich includes source/dest port numbers (only one device allowed per port so no MAC addressing etc), 32 bits of address, a read/write bit, and 64 bits of data if it's a write23:21
smedingsounds fun23:22
smedingespecially compared to what i'm doing23:22
azonenbergwhats that?23:22
smedinghorribly simplistic, horribly outdated ASIC design23:22
azonenbergoh, yeah you mentioned that23:22
smedingwell, although we test it on an FPGA23:22
azonenbergthe 2um SoG process?23:22
soul-dgot some money  think gonna check out the  cheap  pcb chemical23:23
smeding1.6u but yeah23:23
azonenbergwanna bet your asic runs slower than the fpga version? :P23:23
azonenbergand lol, 1.6um is within the range of what i could be able to fab soon23:23
smedingno, that isn't a very fair bet23:23
smedinggot ion diffusion down yet? p23:23
azonenbergfair to the asic?23:23
azonenbergand i'm talking lithorgaphy23:23
azonenbergdoping is one of the processes i havent looked at yet23:23
smedinganyway, we were basically told "don't expect the ASIC to work"23:24
azonenbergthat bad?23:24
smedingbetween us being second-year EE undergrads and the people fabbing it being people who have never been in a clean room before23:24
smedingthere is not a lot of yield23:24
azonenbergwow lol23:25
smedingthey save money by having the clean room course consist of fabbing the ASIC for our project23:25
smedingi suppose23:25
azonenbergBut still, lol23:25
smedingyeah it's pretty silly23:26
smedingthe code some people put out is ridiculous though23:26
smedingi like to think what i wrote is reasonable, maybe even good VHDL23:26
azonenbergi'm working on verilog but i'm still a relative novice23:27
smedingwe have some dinky IR communications project, i wrote most of the IR subsystem and helped other groups23:27
azonenbergonly written a few tens of thousands of lines total23:27
smedingone group wrote a module that receives ASCII and determines where it goes on the little LCD we hook up23:28
smedingfrom the IR decoder and the keyboard interface23:28
azonenbergDoes it work on the FPGA?23:28
azonenbergand you expect it to work on the asic? :P23:29
smedingi've given up hope23:29
smedingthey claim they have fixed things though23:29
azonenbergThe fab lab class in my school uses a single mask set that's probably ten years old23:29
smedingbut looking at the design i don't see how anything short of a redesign could have fixed it23:29
azonenberg(i havent taken it)23:29
azonenbergBut eevryone makes the same basic design23:29
azonenbergjust some caps and transistors hooked up to probe pads23:29
azonenbergon a 2" wafer23:30
smedingsounds reasonable23:30
smedinganyway, you should see the design plans for this module... they 'designed' a D flip-flop, used it to implement a register, used two of those and a multiplexer, and it's controlled by an icky half-state-machine23:31
smeding(the library comes with standard blocks and the synthesis engine is fairly intelligent)23:31
smedingin their place i would just have made a big Moore machine and let the synth handle it23:32
smedingat least at first, maybe optimisation would require me to split it up23:32
azonenbergoptimization? How slow is the current code?23:32
smedingoh, it's not slow enough to matter23:33
smedingthe main limitation is chip size23:33
azonenbergoh lol23:33
smedingwe can do 80k transistors max, and on SoG that's not that much23:33
azonenbergi'm used to optimizing for speed on a giant chip23:33
azonenbergand lol, 80k transistors? That's like what, 20k gates not even counting  barrier transistors?23:33
smedingless, probably23:34
azonenbergSo maybe 10k usable?23:34
azonenbergand i thought my 50k gate fpga was small23:35
soul-di might have 50k if i add up all my fpga's :P23:36
smedingi still can't find where the 80k limit comes from though23:36
soul-dor  difrent from le's ? :P23:37
azonenberglol, yes23:37
smedingthe manual says 200k but we were assured by the tutor the limit is 80k23:37
azonenberga LUT is several gate equivalents23:37
azonenbergand a LE is usually several LUTs23:37
smedinganyway an inverter occupies 4 transistor positions23:38
azonenbergwhat about a 2NAND?23:38
azonenbergthats usually the definiteion of a gate-equiv23:38
azonenbergi see23:38
smedingthen you add barrier transistors and wiring room23:39
smedingand you get efficiencies of something like 30%23:39
smedingin terms of transistors used for switching vs transistor positions occupied23:39
smedingso that's like, 6k gates?23:40
smedingthe only groups who have a full design working on the FPGA have given up on trying to fit it onto the ASIC i think23:41
azonenbergyou could like barely fit a UART in that23:42
smedinghm, then i might be exagerrating23:42
smedingbecause we have a sort of UART, and more23:42
smedingwell, it's not an UART... the receive portion is ugly, from what i've seen23:43
azonenbergwell ok, you could fit a uart and change23:55
azonenbergBut not too much23:55
--- Fri Dec 23 201100:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!