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/
chris_99 has quit [Quit: Ex-Chat]
<whitequark> ++
<nmz787> _Sync_: 3 bux!!!
<nmz787> I thought the SEM for $200 was a good deal
<nmz787> (the one I just got)
<nmz787> I too thought turbos were more than $150/250
<_Sync_> yeah, I'm also still looking for a sem
<_Sync_> but I don't really have the space
kanzure has quit [Ping timeout: 245 seconds]
nmz787 has quit [Ping timeout: 255 seconds]
kanzure has joined #homecmos
nmz787 has joined #homecmos
nmz787 has quit [Client Quit]
nmz787 has joined #homecmos
<nmz787> is there a 'best' motor controller over on fusor.net? I see one that uses some NAND gates, and then a few pages later someone posted an improvement using a full H-bridge chip... but while they have inputs for the hall sensor, I can't quite tell how they slowly ramp up the motor speed then hold it there once the max RPM is reached.
<_Sync_> I never looked into that because I have controllers for all my pumps
<whitequark> nmz787: does your pump have a hall sensor?
<whitequark> or does it rely on back-EMF?
<nmz787> whitequark: haven't bought one yet... currently I have a diffusion pump, but after looking on ebay it /does/ seem possible to get something with a comparable pump rate
<_Sync_> it's not hard
<_Sync_> and turbos are so much easier to deal with
<nmz787> the diffusion pump says 400l/s but I read someone else replaced a diffusion pump with a lower pump rate turbo and they actually pumped down /faster/
<nmz787> i know a friend made a centrifuge once using a BLDC and back-EMF so I am going to ask him for tips
<nmz787> but the multiphase seem like a bit more of a challenge
<_Sync_> I just ran the leybold from a simple fixed frequency driver
<_Sync_> but yeah turbos usually pump faster
<nmz787> SpeedEvil: are you going to try and replicate some MOSFET (just based on your interest in those die MOSFETs)?
<nmz787> _Sync_: are you the one who got the varian starcell IGP?
<_Sync_> yeah
<nmz787> did you say it stopped working?
<whitequark> nmz787: my advice is to use a dedicated BLDC driver
<whitequark> there's a whole lot of them these days and they're very easy to use
<_Sync_> no, there just was an itsy bitsy spider in there
<nmz787> whitequark: looked through your lab notes last night... some cool stuff you've been doing!
<_Sync_> they want quite a lot of current starting up
<whitequark> nmz787: fairly trivial so far
<nmz787> motor starting capacitor the trick?
<_Sync_> on a 3phase turbo?
<_Sync_> nah
<_Sync_> they want a high drive frequency
<whitequark> insides of the exdc80 controller i have
<whitequark> nmz787: given that exdcs cost about $60 on ebay i think putting time into making your own controller is absurd
<nmz787> m
<whitequark> but maybe it's different for leybold or pfeiffer
<nmz787> hmm
<nmz787> looks like 'doped with europium' and I have the same o-scope
<_Sync_> yeah pfeiffer has a 2 phase
<whitequark> nmz787: ds1074z
<whitequark> firmware hacked to... something slightly better
<whitequark> 1100?
<nmz787> yeah
<_Sync_> haha whitequark
<nmz787> I did the software package upgrade
<nmz787> 'upgrade'
<_Sync_> thanks for dropping the quote
<whitequark> _Sync_: hm?
<whitequark> oh
<nmz787> anyone in here in Oregon by chance?
<_Sync_> also nmz787 I dunno if the starcell still works
<_Sync_> the magnets are still magneting
<_Sync_> and the cell looks clean, apart from the spider
<whitequark> you really like that spider don't you
<_Sync_> not particularily
<_Sync_> I just think it's funny
<whitequark> I can see that ebay description: SPIDER FOR SALE, INCLUDED
<whitequark> SCALE* well either works
<whitequark> hm. if i ever sell anything on ebay, i'll add a spider for scale. it is decided.
<nmz787> heheh
<nmz787> whitequark: you give me hope for interfacing HV circuits with microcontrollers/etc
<nmz787> whitequark: I'm not sure what you were thinking about the decoupling caps being so far away in the first place though... :) that's the first rule in all the datasheets 'C1,C2,C3 should be as close to the IC as possible' :D
<nmz787> etc
<_Sync_> well, there is no problem in interfacing them
<_Sync_> you just gotta be aware where what is
<_Sync_> but yeah bad choice of the inductor
<_Sync_> but I'll test the ion pump with one of my power supplies I have floating around
<_Sync_> oh whitequark
<_Sync_> can you please go fuck yourself?
<_Sync_> your twitter account reminded me of the solar physics lecture last semester
<_Sync_> where we had to calculate the amount of generated charge carriers in a solar cell by hand.
<_Sync_> doing so is no fun at all.
<whitequark> nmz787: I did place them 'as close as possible'!
<whitequark> except I did not have a very good idea of what is possible
<whitequark> (seriously)
<whitequark> like I went over it in a few days and went "lol. wtf did I do here"
<whitequark> nmz787: generally keep tuned, I want to make a whole bunch of these, with a standardized STM32-based controller, all bound in a single ethernet network and controlled via UDPv6
<whitequark> _Sync_: "doped with europium" actually refers to strontium aluminate
<whitequark> glow ceramic
<_Sync_> I meant your tweets about how fets and bjts
<whitequark> oh
<whitequark> oh yeah I can't imagine doing like 80% of undergrad physics by hand
<whitequark> why bother with symmetric approximations if you have FEM?
<whitequark> like it's not even pedagogic, the only thing it teaches you is "physics is boring"
<_Sync_> because you need to understand what synopsys does before you mess with it
<whitequark> I'm aware of this argument but not convinced by it
<whitequark> like.. by all means, do tell me to write this integral
<whitequark> then let me compute it numerically or something
<whitequark> fuck analytical solutions that just make you waste paper
<_Sync_> okay, I'll give you a session of synopsys and a few parameters and your task is to optimize it for a certain efficiency (solar cells)
<_Sync_> you can just #yolo and have it simulate *all* differences within reason
<_Sync_> but then you are waiting for it to converge in the next century
<whitequark> let me rephrase: you absolutely should understand the underlying effect. but I am not convinced that doing dozens of plug-and-chug type problems does anything to help understanding
<whitequark> or rather, I know it doesn't for me
<_Sync_> well, it is the same for me
<_Sync_> but I can understand the teachers
<whitequark> oh, I *can* understand the teachers. I see where they're coming: a place where the most advanced computational tool is a slide rule
<_Sync_> because it is easier to have people chew on a hard example to work through than to explain it to death
<_Sync_> haha no
<whitequark> well
<whitequark> that is more malicious :p
<_Sync_> not really
<_Sync_> it helped me to do it on my own
<_Sync_> to see what parameter does what
<_Sync_> and that is all simple maths
<_Sync_> should be in your toolkit
<whitequark> full disclosure: I dropped out after failing diffequations eight times
<nmz787> whitequark: sounds like you did your due diligence with trying to pass at least!
<whitequark> I sort of like the concept behind, but in the time I wasted trying to learn doing them by hand, I could probably write a computer algebra system or two
<nmz787> the last calc class i took last fall to get my degree, I used sagemath cloud and it made the class a lot nicer IMP
<nmz787> IMO
<_Sync_> although if you are on the level of the wave functions it starts to get over my head
<nmz787> and I started looking into the different solver software out there... coding helps me organize complex tasks with mundane portions of work for sure
<whitequark> exactly
<_Sync_> and you then need to understand the grand canocial ensemble to actually understand pauli
<_Sync_> I was like F that then
<whitequark> oh ha I was just reading about that yesterday
<_Sync_> but what do I know about solid state physics
<_Sync_> I'm just an EE
<_Sync_> and thank god I'm done with all that semiconductor bsnz
* whitequark looks at _Sync_ then at channel name
<nmz787> heh heh
<_Sync_> well, from an education standpoint
<_Sync_> I might take some more of it if I finally decide to study further
<_Sync_> whitequark: it's not that I don't do semiconductors http://sync-hv.de/~tmp/moswafer.jpg
<whitequark> I much prefer studying physics and math now that there's no authoritarian asshole to tell me how I should do it
<whitequark> neat, how was that made?
<_Sync_> wet oxide, litho, gate oxide, back/frontside metal, litho, post temper
<whitequark> some kind of DIY?
<whitequark> or a commercial / uni system?
<_Sync_> I did all of that myself
<_Sync_> but in the universities clean room
<whitequark> I see
<_Sync_> in the end it's not hard
<whitequark> *nod*
<_Sync_> the main kicker is cleanliness
<whitequark> hrm
<whitequark> _dead meaning what exactly?
<_Sync_> look at it
<_Sync_> and think what it should do
<whitequark> how about labeling axes
<_Sync_> overrated.
<_Sync_> magic over more magic
<_Sync_> current over voltage
<_Sync_> interestingly you can see the ideality factor 1 and 2 regions of the diodes
<_Sync_> the difference are a few metal ions
<whitequark> gotcha
<whitequark> fascinating
<_Sync_> well
<_Sync_> it's kinda the killer of actually working devices at home
<_Sync_> because the amount of work you need to put in to keep everything as clean is huge
<_Sync_> and is a cost factor
<whitequark> what I was thinking is trying to push as many processes as possible into the chamber. use sputtering and RIE instead of wet processes
<whitequark> since it seems like that would simplify keeping it clean
<whitequark> I don't know how to do lithography though
<nmz787> whitequark: this is a board I recently designed, I basically have the caps almost touching the pads of the IC https://644db4de3505c40a0444-327723bce298e3ff5813fb42baeefbaa.ssl.cf1.rackcdn.com/uploads/project/top_image/Vs4zWf41/i.png
<whitequark> nmz787: yeah I know how you're supposed to do it 'right'
<whitequark> I just fucked up QA on that board
<nmz787> :D
<_Sync_> how do you clean the wafers at first whitequark?
<nmz787> no worries, I swapped rx and tx on that first board, and didn't even add power decoupling caps (those present are for charge pumps)
<_Sync_> that's one of the bigger issues
<_Sync_> because even though they are metal free there can be crap on there
<nmz787> and on the second link I used an incorrect footprint for the LDO
<_Sync_> (yes, the mylar vacuum bagging sheds organics)
<nmz787> bye!
<_Sync_> also the fluroware will outgass
<whitequark> _Sync_: good question
<_Sync_> you will need to do pirahna and RCA
<whitequark> ah, that's fine
<_Sync_> *cough*
<whitequark> I'm aware
<whitequark> I'm mentally comparing it to something like "hot HF vapor"
<_Sync_> enjoy buying jugs of chemicals that should cost dollars for hundereds of dollars
<whitequark> hahaha
<whitequark> i'm from RU
<_Sync_> so what?
<whitequark> well, i'm planning to fab in RU
<whitequark> chemicals are very cheap here
<whitequark> and very accessible, past a few oddities
<_Sync_> it's not the chemical itself
<_Sync_> gotta buy certipur VLSI or better
<whitequark> hm
<_Sync_> otherwise rip wafers
<whitequark> not clean enough
<whitequark> ?
<_Sync_> yes
<whitequark> ok that's a pain
<_Sync_> don't have a special quartz glass basin?
<_Sync_> enjoy your metal ions
<_Sync_> it's a pain.
<_Sync_> :/
<whitequark> special as in, special kind of quartz glass?
<_Sync_> yeah
<_Sync_> accidentally touched the tip of your plastic tweezers with your glove?
<_Sync_> might as well etch all of your wetware gain
<_Sync_> +a
<whitequark> that sounds fairly extreme
<whitequark> did the first ICs have the same fab requirements? it seems unlikely
<_Sync_> they did
<_Sync_> but they did not always follow that procedures
<_Sync_> so their yield was low
<whitequark> I see
* whitequark ponders whether his plan to do czochralski will give any remotely viable result whatsoever
<_Sync_> no
<_Sync_> there also is no reason to go to all that hassle
<_Sync_> when you can just buy wafers
<whitequark> I know I can buy wafers
<whitequark> I'm interested in the process though
<whitequark> not even necessarily for silicon, but for silicon as well
<_Sync_> well, you have to buy highly pure Si in the first place
<_Sync_> which is not easy
<whitequark> *nod*
<SpeedEvil> shots
<SpeedEvil> In about 15 seconds, she would be sweating heavily from the IR
<whitequark> how about purifying low-grade Si by recrystallization
<SpeedEvil> whitequark: It's all quite achievable.
<_Sync_> sure, if you can buy low purity Si
<_Sync_> which again is not easy
<_Sync_> but the thing is
<SpeedEvil> whitequark: But almost certainly insane
<SpeedEvil> _Sync_: broken solar cells?
<_Sync_> you do *not* want to deal with the chlorine chemistry
<whitequark> SpeedEvil: I never laid a claim to sanity!
<_Sync_> SpeedEvil: impossible
<_Sync_> gotta fish all the dopants out and the lead
* SpeedEvil imagines unbreakable solar cells.
<whitequark> _Sync_: that's simple. Step 1: take a bucket of sand
<whitequark> ;D
<_Sync_> ...
<whitequark> on a serious note, ebay has plenty of low-purity Si
<whitequark> anyway, I wanna try a few things with it when I get a massspec working. SIMS should be enough to confirm the purity of samples
<whitequark> if it's intractable within my lifetime, so it will be
<_Sync_> just buy a quadrupole
<_Sync_> those are the things that are sexy to DIY but just really not possible
<_Sync_> because you want them to just work
<whitequark> a QMA is something I'm definitely making from scratch and publishing as OSHW
<whitequark> well, QMA, detector, and control electronics
<SpeedEvil> TOF is interesting too
<whitequark> TOF is a real pain
<whitequark> TOF seems conceptually simple but the engineering is nasty
<_Sync_> so is a QMA
<whitequark> QMA involves shit like rf+dc fields but the engineering is quite tractable if not simple
<_Sync_> if you want any sensible resolution
<whitequark> _Sync_: what would you say be my worst problem with a QMA
<_Sync_> and requires really annoying fixturing
<_Sync_> because the alignment of the quadrupoles is critical
<whitequark> yes, it needs precision machining
<whitequark> I've considered using a commercial head but I think designing one could prove useful
<whitequark> maybe just a commercial rod assembly, if I turn out incapable of making a good enough one
<_Sync_> I mean I reversed most of my spectra mass
<_Sync_> the thing is not even the rods
<_Sync_> you can buy them
<_Sync_> because you don't really need profiled ones
<_Sync_> but alignment is tricky
<whitequark> sure, hence "assembly"
<_Sync_> and you need a good MS to get that teaked
<_Sync_> +w
* whitequark grumbles
<whitequark> *hopefully*, by the end of next week, i will have the chamber that i designed
<whitequark> *finally*
<whitequark> it took fucking weeks to convince some of the machine shop to take my money
<SpeedEvil> :)
<SpeedEvil> How large a chamber? Bakeable?
<whitequark> two NW160 flanges at the ends and 200mm long
<whitequark> should give you an idea of size
<whitequark> should be bakeable.
<whitequark> it's fully SS+viton
<whitequark> so... 150 or so
<whitequark> the pump is connected via a long bellows so that should take care of it
<whitequark> it's mainly to prove that I can do a viable chamber before moving on to something more complex, but it should serve me well as it is
<_Sync_> should be pretty easy, just gotta bore out two flanges and weld the pipe in
<whitequark> mmm actually nope
<whitequark> that's not at all how I went at it
<whitequark> I had the flanges turned--not even from rod stock since that's too large diameter, the machinist got it cut from 2cm SS sheet
<whitequark> and the pipe was roll-formed, or it WILL be roll-formed on monday assuming the transfer clears, with holes for two additional KF40 and KF25 laser cut before forming
<whitequark> all in all it the entire custom chamber will cost me $400 or slightly more
<whitequark> well, no
<_Sync_> I'd buy the flanges and just bore the extra holes
<_Sync_> there is no need to roll form the pipe
<whitequark> the chamber, the viewport, and a NW160 to 2*KF16+2*KF25+KF40 adapter
<_Sync_> which only adds a leak point
<whitequark> eh, it's going to be welded anyway
<whitequark> flanges are too expensive to buy rather than machine
<whitequark> and they also have nearly their own cost in shipping
<_Sync_> DN160 flange is around 44€
<whitequark> anyway, what's the point, if I'm machining these anyway?
<whitequark> I don't have my own lathe or anything
<_Sync_> the point is that it is cheaper to buy them rather than to machine them
<_Sync_> because the stainless costs about 2/3 of that
<_Sync_> and is annoying because you waste a lot
<whitequark> machining a single KF25 is $10 with labor and material included
<whitequark> it's not cheaper
<whitequark> *KF25 nipple
<_Sync_> and all the usual steel suppliers will charge for a 180mm plate
<whitequark> and i don't pay for what's bored out anywya
<_Sync_> kf25 flange with pipe 304 stainless, 11€
<_Sync_> can't pay someone to do it for cheaper
<whitequark> 316L here but yes
<whitequark> it's about same
<_Sync_> 316L has no benefit
<whitequark> oh? I've seen conflicting opinions on this
<whitequark> from what I gather the main difference is carbide precipitation in welds, which doesn't really matter for vacuum
<whitequark> or for the small thicknesses i would have
<_Sync_> well, the thing is 304 contains sulfur
<whitequark> but i went for 316L to be safe
<_Sync_> but for everything besides UHV 304 is preferred because it machienes way easier
<whitequark> ah i see
<_Sync_> then your machine shop is very nice to you, because if I'd be making the flanges for you you'd be paying for the slug that comes out
<_Sync_> too small to be really useful
<_Sync_> but too much material to waste
<whitequark> dunno maybe they recycle it
<whitequark> it seems like they should?
<_Sync_> well, you can calculate about 5mm per side wasted, so you get a disc maybe 140mm in diameter 12mm thick
<_Sync_> depending on the shop that might be okay to recycle
<_Sync_> but usually too small to keep on stock