azonenberg changed the topic of #homecmos to: Homebrew CMOS and MEMS foundry design | Wiki: http://homecmos.drawersteak.com/wiki/Main_Page | Repository: http://code.google.com/p/homecmos/ | Logs: http://en.qi-hardware.com/homecmos-logs/
<glowplug> Went to mcdonalds. =)
<glowplug> Agreed this is not easy stuff. It will be very painful and slow.
<glowplug> The only way to know if 1um is possible with this design is to try it. =)
<Sync_> I'll see tomorrow if it is feasable
<Sync_> I have no small angular contact bearings but that should be no problem
<glowplug> You have some way of making them?
<Sync_> I'll just order some
<Sync_> making bearings is no fun
<Sync_> and 7200 angular contact bearings are like $4 each
<glowplug> Is that like a rollerscate bearing?
<Sync_> no
<Sync_> a rollerskate bearing has no angular contact
<Sync_> which is important here :D
<glowplug> Back!
<glowplug> I did not know about these. They are just more robust in radial and axial directions.
<glowplug> Do you plan on trying to copy the design I linked earlier (the aluminum one)?
<Sync_> it's not about being more robust
<Sync_> you can preload them in O or X configuration to get a precise spindle constraint
<glowplug> So this bearing is going to hold the end of the threaded rod.
<glowplug> I was going to buy a rollerscate bearing. Good thing you said something. Haha
<glowplug> I've never done preloading so I have to figure all of that out. The design says "anti backlash nut" I'm assuming thats where I put my normal m3 hex nut?
<Sync_> yes
<Sync_> the problem is with a regular bearing that while it supports your piece radially and axially it does so with play
<glowplug> Alright. I'm not sure that m3 antibacklash nuts exist.
<glowplug> Which is insanely important. So preloading it keeps everything level.
<Sync_> huh. there seems to be standards which specify m8x0.25
<Sync_> well the preload forces the assembly in a certain position
<glowplug> .25mm pitch????
<Sync_> yes
<glowplug> I wonder how expensive that is per meter. o.O
<Sync_> I guess you are not going to get threaded rod in that size
<glowplug> How do you get m8x.25 then?
<Sync_> by making your own spindle?
<glowplug> I'm good on that. Haha
<glowplug> I think that 10031AB is an m3 antibacklash nut but its $43. O.o
<glowplug> This is interesting.
<Sync_> hm
<Sync_> my dumb lathe only goes to 0.45 pitch
<Sync_> it doesn't need to be that fancy
<Sync_> you can just use a split nut with two orings
<Sync_> or a clampable nut
<glowplug> If thats true then I should have no problem putting that mechanism in. I can just design the cuts for that mechanism into the stage.
<glowplug> I don't have a co2 cutter yet I'm just going to have a local shop cut my acrylic parts for now.
<glowplug> I might go into i3 detroit do it for free.
<Sync_> oh detroit
<Sync_> I was there once
<glowplug> You poor soul...
<glowplug> Just kidding its not that bad here. There are worse places on Earth. =)
<Sync_> well it was in 99 or something
<glowplug> I see. 8)
<glowplug> I'm going to do some more updates to the wiki now then. We have a good amount of info at add.
<glowplug> *to add
<Sync_> visited some relatives in grand rapids
soul-d has quit [Remote host closed the connection]
<glowplug> Ahh I see. Grand Rapids is actually really nice. I was born in Lansing which is right next to it. =)
<Sync_> bah. conflict of interest
<Sync_> I'm tired but also hungry
<glowplug> Same here. My solution was McDonalds.
<glowplug> Conflict... Resolved...
<Sync_> well I either go to bed hungry or eat something and go to bed later
xiangfu has joined #homecmos
<glowplug> Option #2. 8)
<glowplug> Where did you find the angular contact bearings?
<glowplug> They are $28 on mcmaster. O.o
<Sync_> my local bearing shop
<glowplug> Ahh
<Sync_> interestingly bearings die pretty fast in vacuum applications
<glowplug> Ok so the 7200 has a 10mm inner diameter and my threaded rod is 3mm.
<Sync_> yes that is one of the reasons I want a bigger spindle
<Sync_> :D
<glowplug> I will succeed....
<Sync_> one of the issues with the spinde so thin is that it will bend and flex
<Sync_> even under the small load
<glowplug> I have a solution for that.
<glowplug> Tons of nuts. =)
<glowplug> Actually wait there shouldn't be any flex.
<glowplug> Not if the threaded rod is only ~100mm long.
<Sync_> hm probably, still larger is better
<Sync_> and huh 1/4-32 UNS is expensive
<glowplug> What is that?
<Sync_> a thread
<glowplug> Oh I see.
<glowplug> Well I found a 3mm to 5mm coupler they are $4 on Ebay.
<Sync_> hm, I really think you are better off using a larger spindle
<Sync_> even if it is only less fiddly to work with
<glowplug> Ok I didn't notice this before.
<glowplug> There is .5mm "low-strength steel" rod on mcmaster.
<glowplug> In m5
<glowplug> These are interesting too. 50mm long .5mm pitch bolts.
<glowplug> Not quite long enough I suppose.
B0101 has joined #homecmos
<glowplug> 8)
<glowplug> I guess the new wiki needs a logo. O.o
<Sync_> I thought about that too
<glowplug> I'll try to toss something together tomorrow.
<glowplug> Still working on the litho wiki.
<glowplug> Do you know how to measure changes in X and Y with a single interferometer?
kanzure has quit [Ping timeout: 276 seconds]
<Sync_> beamsplitter glowplug
kanzure has joined #homecmos
<glowplug> Yes but how do you position the sensor optics?
<Sync_> after it
<Sync_> you just put in two interferometers
<Sync_> gaaaah
<Sync_> die google
<glowplug> Haha!
<Sync_> there
<glowplug> Alright so the "Remote Detector and Interferometer" boxes.
<Sync_> yes
<glowplug> They take in one beam, offset it, and read the reflection with some optics?
<Sync_> yes
<Sync_> you get a fringe pattern
<Sync_> and the intensity varies
<Sync_> which you can read with a photodiode
<glowplug> I know how interferometers work. =)
<glowplug> I figured out what those mysterious boxes are. They are "linear interferometers".
<glowplug> A system like that might be more difficult and expensive (due to need for increased optics quality) than just having two seperate interferometers.
<glowplug> Which this really is two seperate interferometers operating off of a single beam source.
<glowplug> Yeah I'm going to try it the other way. Lasers are cheap optics are expensive. Screw that. Haha
kristianpaul has quit [Ping timeout: 245 seconds]
<Sync_> well you are only going to save one beamsplitter
<B0101> yikes... chlorine gas is never fun to play with, more so when you have a gas leak in your test chamber
<glowplug> You save all of the internal components that comprise a "linear interferometer" which is tons of mirrors, a beam splitter, and a retroreflector. The system is also very very sensitive.
<glowplug> You have to be careful dawg. O_O
<Sync_> huh how are you going to make a interferometer with just a laser?
<glowplug> They are called "linear interferometers" but its confusing.
<glowplug> They are really black boxes with tons of integrated components. Two of those are added to the configuration to make it work.
<glowplug> So you have all of the necessary components of a single interferometer. Plus the two black boxes which are extremely complex. Plus massively increased system sensitivity.
<glowplug> It's getting late though going to bed. 8)
B0101 has quit [Quit: Ex-Chat]
<Sync_> glowplug: the integrated linear ones seem to be just michelsons with retroreflectors
<Sync_> yeah two prisms and a beamsplitter will do
soul-d has joined #homecmos
<nmz787> there's a video online of a $20 interferometer
<nmz787> using a CD case, laser pointer, mirrors and a cheap lens
fpgaminer has quit [Ping timeout: 258 seconds]
<Sync_> I just ordered some stuff from EO
<Sync_> you can make one for under $200, that's not too bad
<nmz787> if you just need micron-ish resolution, you probably just need something like the $20 one and a photodiode and comparator or schmitt trigger
<Sync_> prisms are cheap and beamsplitter glass too
<nmz787> if you want better res then you need to keep the sinosoid pretty noise-free, and get an ADC instead of the comparator/schmitt
<Sync_> well, if it only has to have micron resolution I'd use glass scales all day long
<nmz787> that's not consistent as interference though, so I guess it depends on your goal
<nmz787> how do you read glass scales?
<Sync_> rs422 output
<nmz787> oh so it's not roll-your-own system?
<nmz787> link?
<nmz787> shars.com says $520
<nmz787> "(the smallest are just 5 m apart)"
<nmz787> m==micron
<Sync_> that's not true
<Sync_> you can get them with 100nm resolution
<nmz787> heh, I don't like suppliers that don't list prices
<Sync_> I don't care, I pay what they want
_franck_ has left #homecmos ["Goog bye !"]
<Sync_> but actually it is not that bad nmz787, if you get chinese ones with 500nm resolution they are like 200$ for 70mm length
<Sync_> and not much more for longer ones
xiangfu has quit [Ping timeout: 260 seconds]
xiangfu has joined #homecmos
xiangfu has quit [Ping timeout: 246 seconds]
xiangfu has joined #homecmos
B0101 has joined #homecmos
xiangfu has quit [Read error: Connection reset by peer]
xiangfu has joined #homecmos
diginet has joined #homecmos
xiangfu has quit [Read error: Connection reset by peer]
xiangfu has joined #homecmos
xiangfu has quit [Ping timeout: 252 seconds]
xiangfu has joined #homecmos
xiangfu has quit [Quit: Reconnecting]
xiangfu has joined #homecmos
diginet has quit [Quit: Leaving]
xiangfu has quit [Ping timeout: 276 seconds]
xiangfu has joined #homecmos
kristianpaul has joined #homecmos
kristianpaul has quit [Changing host]
kristianpaul has joined #homecmos
kristianpaul has quit [Client Quit]
Prottey has joined #homecmos
B0101 has quit [Quit: Ex-Chat]
<glowplug> Good morning. 8)
<glowplug> I agree with the glass scales being very accurate but we already have *some* amount of mechanical accuracy built into the system via the leadscrew assembly. The interferometers will be a reliable way to know if something else in our system is broken instead of guessing all day.
<glowplug> nmz I agree they can be built for very cheaply. My current plan (need to update the wiki) is to buy a giant piece of 50/50 glass and CNC hundreds of tiny 10x10mm squares out of it. That gives you $1 beamsplitters with extremely good optical quality.
<glowplug> In terms of replicating the linear box units lets call them. As you increase the amount of sensor positions in the beam network you multiply the system sensitivity by many times.
<glowplug> I still think that two beam sources will give the best price/performance/reliability ratio.
<Sync_> well if you have alot of 50/50 splitters a second source is a waste
xiangfu has quit [Quit: leaving]
xiangfu has joined #homecmos
<glowplug> That is technically correct. But aligning the system with a second beam is many many times more difficult.
<glowplug> The alignment instruments need to be many times more precise. And I'm not sure if the quality of the optics I mentioned is high enough for a setup like that.
<glowplug> The optics they are using in the PDF you linked are probably $10,000+
<Sync_> I just ordered stuff from EO to try that
<Sync_> I like the idea of using prisms for the beam return
<Sync_> that way you can shift the receiver diode a bit
<glowplug> Good luck. Haha
<glowplug> The prisms can be fashioned from multiple standard mirrors correct?
<Sync_> yes but it'll be annoying to align that
<Sync_> and they are only $29
<Sync_> :D
<glowplug> $29 each! Ouch..
<glowplug> You can get another HeNe tube for $50.
<glowplug> This is why I abandoned the single beam idea.
<Sync_> wat, you need the prisms anyway
<glowplug> For two standard michealsons you only need beamsplitters and mirrors.
<glowplug> And two beam sources.
<Sync_> yes but that is not as compact
<Sync_> with two prisms and a beamsplitter you can move the receiver to the side of the input beam
<Sync_> which is cool
<Sync_> and $29 per piece, come on
<glowplug> That is extremely cool. But I have plenty of desk space and not very much money. HAha
<Sync_> that's not too bad
<glowplug> I'm shooting for $29 for the entire assembly. 8)
<glowplug> I wonder is it possible to put the entire thing under vacuum. That would make the lasers more accurate and reduce the chance of contamination.
<glowplug> Everything is electronically controlled anyways.
<Sync_> then you cannot make it out of acrylic
diginet1 has joined #homecmos
<glowplug> Why not?
<Sync_> outgassing properties
<glowplug> There are retail acrylic chambers that pull 30"
<Sync_> yes, but 30" is not really a vacuum
<Sync_> at least from my vacuum perspective
<glowplug> I knew that was coming. Haha
<glowplug> I agree that is not 0 torr but it would still greatly increase laser accuracy and greatly reduce contamination for a very low cost.
<diginet1> the best way a DIYer can get a vacuum is a bell jar
<Sync_> and then do what? suck on it?
<Sync_> ;)
<diginet1> acrylic is just about the worst possible thing I can imagine for vacuum stuff
<diginet1> Sync_: well you need a pump too :P
<diginet1> but I more meant for the chamber
<glowplug> I wonder why they sell $2000 acrylic vacuum chambers then.
<Sync_> to outgas resin
<diginet1> yeah, calling that a "vacuum" is a bit of a stretch
<glowplug> So the acrylic chamber wont hold any amount of vacuum that is useful for increasing beam stability and reducing contamination.
<Sync_> I mean, yeah it is lower than atmosphere but oh well
<glowplug> That just doesn't seem realistic.
<glowplug> Remember it's a $500 litho device. No need for a $1500 bell jar.
<Sync_> I'd just put some shields
<glowplug> It basically is a shield with a low vacuum.
<glowplug> I mean... maybe its not worth the extra ~$50.
<glowplug> But something that both improves beam stabilit and reduces contamination for such a low cost is at least worth examining
<glowplug> *stability
diginet1 has quit [Quit: Leaving]
xiangfu has quit [Remote host closed the connection]
Prottey has quit [Quit: Leaving.]
<Sync_> nice
<Sync_> that came out pretty well
<glowplug> ?
<Sync_> just made a spindle
<Sync_> rather amazed how backlash free it is
<Sync_> only the material pairing isn't the best
<Sync_> yeah I'll need oil but then it'll work fine
<Sync_> pitch error is very low
<Sync_> under 1/100mm in 5cm
<Sync_> backlash when properly adjusted maybe 200nm
<Sync_> but that is probably my setup
<Sync_> because it cannot really have backlash
<Sync_> the next one will have a brass nut and case hardened spindle
<Sync_> yeah :) I like it
<glowplug> Holy crap that is great!
<glowplug> What material did you use for the stage?
<Sync_> I did not build a stage yet, just the spindles
<Sync_> I'll probably use aluminium
<glowplug> How did you test the spindle accuracy?
<Sync_> mahr extramess 2001 and gage blocks
<glowplug> Aha! I see!
<glowplug> And your leadscrew is standard .5 pitch stainless?
<Sync_> no
<Sync_> aluminium, cut it myself
<glowplug> Your joking? That is seriously fantastic!
<glowplug> I would buy that from you in a heartbeat! How much for 200mm? Haha
<Sync_> I'll verify that next week but it looks quite acceptable
<glowplug> I was actually going to look at alu I saw it mentioned. I would totally buy some as soon as you think they are ready. You made them from M8 rod?
<Sync_> no, I took some stock, turned it to 7.90mm and cut a thread
<Sync_> bb in 15
<glowplug> So it is 7.9mm in diameter. That is A LOT better than the m3 rod from mcmaster.
<Sync_> no shit
<Sync_> that's why I wanted to do it that way
<glowplug> Easy there! Haha
<glowplug> I know it's better thats pretty obvious. I just didn't realize you had a precision lathe and mad skills. =)
<Sync_> I'm more or less amazed that the pitch error is that good
<Sync_> gotta verify that after my exams
<glowplug> Me too thats awesome if it works. =)
<Sync_> but in reality it does not matter
<Sync_> :D
<glowplug> Why is that?
<glowplug> It matters for me I'm the first customer of Sync_'s Custom Aluminum Threaded Rods Inc.
<Sync_> because if you use closed loop positioning with the interferometers only backlash will hurt you
<Sync_> but for that to work you need a very rigidly coupled system
<Sync_> and no backlash
<Sync_> but the next ones will be made from c45 steel and brass nuts
<Sync_> the Al-Al pairing likes to bind
<glowplug> Ahh so you will be using steel. Can your lathe cut steel?
<glowplug> I think we can get rid of backlash pretty easily using the DIY spring method.
<glowplug> And I agree about positioning otherwise. As long as we have good absolute feedback the accuracy of the leadscrews is not super important.