#homecmos IRC log for Sunday, 2013-03-24

B0101Hi, I was looking at the logs and there was this thing about interferometry. I would like to know, what is it all about?06:12
Sync_B0101: the interferometers are for direct laser litography position feedback09:29
B0101so you guys are experimenting writing to the photoresist directly?12:07
glowplugDid you read through the wiki page I'm working on for the litho?15:43
glowplugI'm going to work on the interferometer section today hopefully.15:56
glowplugIt might be awhile before there are any pictures.  Maybe I will toss up some drawings soon.15:57
glowplugB0101 what is your area of expertise?15:58
B0101I am no expert at anything related to semiconductors. I only do experiments and research for the learning and the understanding of the science15:59
glowplugWhich parts are you most interested in?  If you want we need a lot of help with the wiki.  I think Andrew is working on the spincoater.  We have nothing for the magnetron, the furnace, or the developer station.16:04
glowplugAlso lists of supplies and suppliers with prices ect.16:05
B0101idk, is extreme DIY on your list?16:07
B0101I could also do the list of suppliers16:09
glowplugI plan on making a microscope with PVC pipe and an x,y stage out of acrylic and homemade leadscrews.  Infact I think everything on the direct write litho page is ghetto.  Haha16:09
glowplugAs long as there is a decent chance the extreme DIY option can meet some basic requirements then its best to use those.  In many countries people dont have access to retail machinery at all.  And many times not enough funds.16:10
glowplugIn an ideal world everything would be custom made from common materials.16:11
B0101some people have the cash, but denied access even to the basic of lab stuff16:14
glowplugExactly.  And ideally anyone would be able to contribute.16:15
glowplugEven if they are 2,000 miles from a university.16:15
glowplugOr large equipment manufacturers wont ship there.16:15
glowplugThat goal may not be met immediately.  Kits might be possible to help with that in the future.16:16
B0101thats why I have no choice but to DIY a lot of stuff, and even obtain some of the basic things like chemicals from the black market16:16
glowplugEven if we DIY many things there will still be some things that are hard without some access to the outside world.16:16
glowplugWhere you do you live?  (if you dont mind me asking)16:16
glowplugAhh I see.  Extreme DIY it is!16:17
glowplugSo you did read this wiki?16:17
glowplugIt needs a lot of work still.  Are you interested in the litho or some other steps?16:18
glowplugActually the litho is a very small part of the whole picture.  Although important.16:19
glowplugIt's just the part that I know the most about.16:20
glowplugWhich still isn't much.  O.o16:20
B0101I work with things not commonly done in the semiconductor industry16:20
B0101maybe only used for special applications16:21
B0101I do software, but I research on things like transmutation of atoms using neutrons and such16:22
glowplugI see.  Well we actually need quite a lot of software surprisingly.  8)16:22
glowplugI'm learning Python and Verilog right now but I's a very slow process.16:23
B0101well, for software, I work with hardware and operating systems16:24
glowplugAssembly?  C?16:24
B0101I use both of them16:24
glowplugIn automation and motion control Verilog and C are king. =)16:25
B0101i see16:26
glowplugWe need software to control half a dozen machines give or take.  Some of it could be done manually but it would be an extremely bad idea to do so.16:27
B0101machines like?16:27
B0101Direct write exposure system?16:28
glowplugMagnetron (sputter coater/evap).  Spincoater. Litho.  Mask alignment/developer station.  Furnace.16:28
glowplugThose are more or less the basic tools.16:29
glowplugEach one needs some kind of software.  The spincoater is just a BLDC motor and a timer that one is easy.  The litho is LinuxCNC but we will need custom HAL scripts and GUI elements (they are Python).16:30
glowplugThe sample alignment optics will need to be totally custom software.  Nothing off the shelf exists for that.16:31
B0101ahhh, i see16:31
glowplugHave you done some reading on how the process works more or less?16:31
B0101for mask alignment? not really :P16:33
glowplugI mean for the entire process.  =)16:33
glowplugStarting with a mask all the way to a semiconductor.16:34
B0101oh, I know most of the basics for the entire fab process16:34
glowplugAwesome!  You are probably ahead of me then.  Haha16:35
glowplugThe mask alignment is four crosshair patterns.  Two patterned on the sample (silicon wafer) and two patterned on the mask.  You have two high power microscopes above each set of crosshairs.  Basically you need to align the sample to the mask perfectly.  Then you develop the resist using UV.16:37
B0101i see, i see. so how much of an error margin is acceptable in alignment?16:38
glowplugNone.  Haha16:38
glowplugThe goal is to make devices with 10um detail as a starting point.  Even though that is very large for semiconductors you still dont have any margin for error in alignment.  Each additional layer needs to align perfectly with the previous or the devices just wont work.16:40
glowplugThis is why having software assistance would be nice.  =)16:41
B0101hmmm, I was thinking of using red colored lasers to aid in alignment16:42
glowplugThe patterned crosshairs do a good job in terms of visualizing where the device needs to be moved.16:43
glowplugThe stage that moves around under the mask will be exactly the same as the litho stage actually.16:44
glowplugTwo dials, one for X and one for Y.  Each turn moves the sample a very small distance (~.1mm).16:45
glowplugThat comes out to about 200nm per degree.  But we wont have that much accuracy anyways.16:46
glowplugSync_ has a great PDF explaining the process I don't have the link though.  =(16:47
B0101i see16:50
glowplugBe right back.16:52
B0101sigh, I wonder what's up with all these random router resets...17:30
glowplugGet one of these and put openwrt on it.  =)18:02
B0101i am using a router same brand, and quite close to that18:06
B0101strange thing is, the resets only happen at night18:07
glowplugDid you put OpenWRT on it?18:08
B0101but the thing is, why only at night?18:10
B0101why couldn18:10
B0101*why couldn't the thing reset itself at any other time of the day?18:10
Sync_B0101: NTD is done quite often in semi18:13
B0101it is? I thought they rarely use it, except in special applications18:14
Sync_well it is not done in huge industrial processes but for experiments it is done quite often18:25
glowplugI think his router keeps resetting.  =(18:54
glowplugNTD = Negative Tone Development?18:54
Sync_neutron transmutation doping18:54
glowplugIs that possible to DIY?18:55
glowplugIt looks like its mostly for n-type wafers arent most modern semiconductors p-type? (risc cpus ect.)18:56
Sync_if you like to have a high flux neutron oven in your cellar, sure18:57
Sync_(also have fun getting the license to have one)18:57
glowplugGood enough for me. I like keeping it simple anyways.  =)18:58
Sync_getting neutrons is actually not that hard19:07
Sync_you could either build a fusor or just get some ionizing smoke detectors and beryllium foil19:07
glowplugStill for only n-type I think there are better alternatives.19:08
Sync_it produces very very uniform dopant concenctrations19:21
Sync_which implantation cannot do19:21
glowplugFrom what I understand n-type is used for modern chips though for idle power consumption reasons.19:24
glowplugI made a super basic logo just so we can toss something up.19:27
glowplugHere is the tiny one.19:32
glowplugUnfortunately I can't set the logo.  Andrew has to do that19:33
glowplugIt's an Ivy Bridge wafter.  8)19:35
glowplug*wafer (cant type today).19:35
Sync_hmm just won two vernier micrometers20:20
glowplugDo not fall into temptation.  The world needs your leadscrews!20:23
Sync_well, they are quite neat, the vernier allows for instant 1µ adjustment20:55
glowplugThat is pretty epic.  8)20:58
Sync_hmm now I need undersize reamers20:59
glowplugFor making the threaded rods?21:00
glowplugI give up.  Haha21:04
Sync_I'll draw something21:10
Sync_note: somehow the spindle must run through the middle21:26
Sync_but the problem is that the red bushings need to be adjusted to be free of play21:26
Sync_I'm debating how to do that21:26
Sync_the easy solution would be to use the same idea as I used with the nut but probably the friction will be too high21:27
Sync_I'm favoring something like diamants moglice which is an epoxy with very low friction and good shape conformity21:28
glowplugAren't we trying to copy the design I linked before?21:34
Sync_that is very close to the design you linked21:36
glowplugIt is actually.  Haha21:36
glowplugLets just copy his exactly.  Seems good enough to me.21:36
Sync_I wont :D21:37
Sync_it is just very flimsy21:37
glowplugThe center "carriage" is just a block with four guiderods and two leadscrews.21:37
glowplugI doubt that its flimsy.  But even if it was its only due to travel.21:37
Sync_get a 1/10000" dti and play around a bit, you'll get a feel for what is strong and what is not21:37
glowplugWe only need to move ~100mm in each direction.21:37
glowplugHis moves like 2000mm in each direction.21:38
Sync_then download roarks stress and strain21:38
Sync_and play a bit with the formulas21:38
glowplugHe designed this based on stress analysis ect. ect.21:38
glowplug"A reference design was created and 3D modeled using SolidWorks, along with a corresponding set of engineering calculations predicting properties such table stiffness based on component sizing. These equations were then inverted to calculate the smallest components necessary to achieve stiffness goals based on the desired lengths of travel."21:39
Sync_his isn't so large as you may think21:39
Sync_overall length is 14"21:39
glowplugI was exxagerating.  =)21:39
Sync_yes he may have done that21:39
glowplugThe point is that I dont think the design requires modification.  Just shrinking.21:40
Sync_build one, put the dti on it, scratch head, repeat21:40
glowplugI mean this is what commercial units look like more or less.21:40
glowplugI can't imagine that the design is bad.21:40
glowplugAlso the rate of movement of our stage is very very slow.21:41
Sync_well, for a cnc mill it is, if you actually want to achieve something21:41
Sync_and more stability is always better21:41
Sync_especially because one stage has to carry the other one21:42
glowplugIn all x,y stages one is ontop of the other.21:42
Sync_commercial stages often use dovetail ways21:42
Sync_which is something I'd prefer but you cannot do that without a machine shop21:43
glowplugAnd at these extremely slow speeds I just dont think theres any advantage.21:43
glowplugBut I suppose thats what open standards are for.21:43
glowplugIf we each build a stage our own way we can compare or even merge the designs.21:44
Sync_the problem with the second stage is the momentum, when it moves the CG will wander and twist the rods21:44
glowplugAt a movement rate of 10um per second?21:44
glowplugI'm not sure there will be any twisting.  O.O21:44
Sync_speed is not the important thing21:44
glowplugTwisting by forces applied by momentum over time.21:45
glowplugSpeed matters for sure.21:45
Sync_time is not important.21:45
Sync_we do not fast enough for dynamics21:45
glowplugCertainly twisting is a time dependant event.  =)21:45
Sync_this is a completely static system from an engineering point21:45
Sync_but only the maximum deflection is interesting21:45
glowplugSo backlash?21:45
Sync_because that is the number you engineer for21:45
glowplugDeflection where?21:46
Sync_imagine his XY table21:46
Sync_lets say the X is the top one21:46
glowplugAll x,y stages have a top but I'm with you.  Haha21:46
Sync_so the Y moves back and forth and the X right to left21:46
Sync_imagine the Y is in the middle of it's travel21:47
Sync_and now the X is at the position where the stepper is the furthest away from the Y21:47
Sync_in his pictures to the rightmost21:47
glowplugThe carriage?21:48
glowplugOk so what causes the deflection?21:48
Sync_now we can think of our X stage as a long rod21:48
Sync_with a function of weight21:49
Sync_that has a center of gravity somewhere21:49
glowplugAgreed.  Which in my design is .5 ounches including the sample.21:49
glowplugBecause the stepper has to move a nearly weightless sample and an acrylic carriage.21:50
Sync_well your stepper has a certain weight21:50
Sync_and when the CG is out of the second Y rail it'll try to bend the right one down and the other one up21:50
glowplugIt does.  But it would likely be bolted to the table.21:50
Sync_then you lock everything up21:50
glowplugSo the part of the design that you dislike is that the stepper is elevated?21:50
Sync_you cannot have both steppers fixed to the table21:51
Sync_no, that's not the problem but just the issue why this thing needs to be stronger21:51
glowplugRight.  So how do you mitigate the problem at all?21:51
Sync_making the rods stiffer by making them larger21:51
Sync_or limiting travel (or both)21:51
Sync_(and making everything lighter)21:52
glowplugBut we reduced travel by an order of magnitude and kept the same size rods.21:52
glowplugSeems legit to me.  Haha21:52
Sync_I'll have to run the calculations but chances are that it'll be fine21:52
Sync_the order of magnitude is quite the overstatement21:52
Sync_we have something between a third or half his travel21:53
Sync_but yes I'll need to do some simulations21:53
glowplugWe will be working on 7" diameter wafers?21:53
Sync_attention: 14" span does not mean 14" travel21:53
glowplugI thought you said he had 14" travel.21:54
Sync_I said overall length21:55
glowplugIf the travel difference is not that great.  Then I agree we should probably make a few changes. =)21:55
glowplugFor starters we can use NEMA-14 steppers that are extremely thin, and tighten up that coupling joint.21:56
Sync_the other problem is that the bushings will have a clearance fit21:56
Sync_I guess nema 8 or 11 would do21:57
glowplugI couldn't find ones with 400 steps per rev that were smaller than 14.21:58
Sync_hmm true21:59
Sync_the smallest nema 14 that's in a useful format is 90g in a quick search22:00
Sync_that's still a log22:00
Sync_it's round22:02
glowplugDo you have something against roundness?  Haha22:02
Sync_yes, annoying mounting provisions22:02
Sync_squareness is bad from a force correct engineering thing but humans like square things22:03
Sync_because it makes problems easier to grasp22:03
glowplugSo round stepper it is then?  Haha22:03
Sync_I'd use these22:03
glowplugAgreed.  Not in stock though.  Im sure we could find that model somewhere else.22:04
Sync_hm those seem to be the same as the ones in old FDDs22:05
glowplugThe manufacturer is schrittmotor.22:05
glowplugNo way those are 400s/r....22:06
glowplugI don't think the FDD steppers are that good.  We might have to order directly from schrittmotor.22:11
Sync_they are pretty fine22:11
Sync_1.8° at least I do suppose they are .922:11
glowplugThe FDD steppers?22:12
glowplugIf they are .9 degrees I would be amazed.22:12
glowplugIn a good way.22:12
glowplugThey are the round type also that you dislike.22:13
Sync_not in all drives22:13
Sync_I have a fes that are like the nema14 one that is square22:13
Sync_they probably are 1.8° from what I feel22:14
Sync_but you can just do halfsteps22:14
glowplugOh you are talking about 5.25" floppy drives...22:14
soul-dbrowsing tea section online  supermarket for  plain mint  tea22:14
soul-dim at page 6  now22:15
Sync_I just get fresh mint out of the garden22:15
Sync_it's win22:15
soul-dyeah i think thats better ieda22:15
soul-dseeds are like 1 meter away22:15
Sync_you should be able to grow it insides22:15
Action: soul-d whistles22:16
soul-dknow nothing about growing indoor honestly22:16
glowplugThose are unipolar.  I haven't figured out its pole density.22:16
Sync_I also have some melissa and sage22:16
glowplugWhere would we reliably find 5.25" floppy drives?22:16
soul-dgrowing buth jolokia atm22:16
Sync_the local scrapyard?22:16
soul-danyhow  quite sad   all those flavored tea's  page afther page22:17
soul-dofcourse all face  aroma's  on black tea22:17
Sync_I import some out of india22:17
Sync_because all the stuff you can get here is meh if you had some that was the real deal22:18
soul-dwonder if i would reacorgnise  true  tea22:18
soul-dsame with appel juices22:18
soul-dall fruit juices22:18
Sync_we bring our apples to the juice place22:19
soul-dwaiting till my tiney tree  will bear  first fruits22:19
Sync_they now even have a service where you actually get your own juice and not a mix22:19
soul-d2 years in garden now22:19
soul-doh well super does have some funky machiene  but still then the fruits come from some weird country22:20
soul-dprobably pest sprayed22:20
Sync_did you buy one from a tree school?22:20
soul-dbought it in belgium  from some tree thingy   got send trough mail even :P22:20
soul-dhad to see the face of the delivery guy22:21
Sync_hmm no local tree school?22:21
Sync_it takes quite a while for them to bear fruits22:21
Sync_we actually have pear that produced TWO damn pears in 30 years22:22
soul-dit is alive and standing so thats fine  maybe this year well see22:22
Sync_I'm so close to cut it down22:22
soul-dll dont have that much patience i think22:22
Sync_we have some apples we planted 10 years ago and they took 3-4 years to be useful22:23
Sync_the cherry we planted kind of dies every year22:23
soul-d"sleepy time green lemon / jasmine "22:24
soul-di think im gonna give up on that mint tea22:24
soul-dcherrt tree we had in old garden did well as long you where  quicker then birds22:25
Sync_just buy some mint and grow it22:26
Sync_it is so much better22:26
soul-dgrabbed the seeds better start planting them   : field mint , spear mint , agastashe rugosa  to  sweeten it  stevia/sugarleaf22:32
soul-dand seemds  indianp paintbrush/nettle  leaves might be used to   but since ive  bought also  lots of  dye  plants  so better double check it  next to what bag say's22:34
Sync_bah stevia22:38
Sync_I'm not a fan22:38
soul-ddonno never had it grown22:39
Sync_huh, those values are intersting.22:47
--- Mon Mar 25 201300:00

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