< Return to Video

35C3 - LibreSilicon

  • 0:00 - 0:19
    35c3 pre-roll music
  • 0:19 - 0:26
    Herald: The next talk is the talk on
    LibreSilicon project that's meant to
  • 0:26 - 0:31
    create a free and open silicon
    manufacturing process. And our speakers
  • 0:31 - 0:37
    today are leviathan, chipforge and
    Andreas Westerwick, creators of
  • 0:37 - 0:45
    LibreSilicon. So let's give them a firm
    round of applause and please welcome them.
  • 0:45 - 0:50
    applause
  • 0:50 - 1:00
    David: Oh, it works, ok. That's problem.
    That's what essentially is all this fuss
  • 1:00 - 1:09
    about is actually a description of how
    we... what this waver means and where we
  • 1:09 - 1:20
    will go with it. And yeah, I give now
    already over to Hagen which already starts
  • 1:20 - 1:28
    elaborating on the basic conceptional
    things.
  • 1:28 - 1:40
    Hagen: OK. Hello everybody. Hope you have
    a fresh mind. It could be heavy. OK. Let's
  • 1:40 - 1:53
    start. What we are. Last year David was
    involved at the project to looking for
  • 1:53 - 2:00
    free silicon, just a way to manufacture
    his own chips and figured out it's
  • 2:00 - 2:06
    difficult. You need a lot of contracts for
    that, NDAs (non-disclosure agreements). So
  • 2:06 - 2:10
    he looked around and find a clean room. We
    had to come in and say, OK, we can rent
  • 2:10 - 2:17
    it. Then he entered a scene on the last
    Congress - a lightning talk - and said, I
  • 2:17 - 2:25
    like to do that. And I wasn't in the
    auditorium there, but a guy told me later,
  • 2:25 - 2:30
    OK, look at this lightning talk. It's very
    interesting. You'd already doing chips. So
  • 2:30 - 2:38
    I entered in sees it or seen the talk
    recording and see it. Nice idea, I will do
  • 2:38 - 2:49
    that too. And the whole year we meet us by
    mumble. It's just a thing of distance you
  • 2:49 - 2:57
    know. David is located in Hong Kong. The
    clean room is there. And I worked from
  • 2:57 - 3:06
    Germany. So we exchanged e-mails. We
    talked on a mailing list. We built up a
  • 3:06 - 3:14
    small community for that and we had a
    first hackathon just to figure it out,
  • 3:14 - 3:18
    what we are doing with the tools, which
    tools are available how we can use them,
  • 3:18 - 3:27
    are they usable at all or not. And this
    was in May and during the process the
  • 3:27 - 3:34
    group wised up and already two of us got
    their qualification to enter the clean
  • 3:34 - 3:40
    room. The Hong Kong University is a
    little bit strict in that. You have to sit
  • 3:40 - 3:46
    there in the courses, you have to do exams
    and if you're fine with the exam then you
  • 3:46 - 3:52
    get the permissions to go in. So Victor
    which is on the most left and David on
  • 3:52 - 3:56
    the most right, they have the
    qualification for that and they
  • 3:56 - 4:03
    manufacture our wafer which you have seen
    there. It's a small one, but it's the
  • 4:03 - 4:12
    first stuff we have, right? OK. The basic
    points, what we are doing. We are using
  • 4:12 - 4:20
    a quite, let's see, old technologies from
    the 80s. It's one micrometer feature size.
  • 4:20 - 4:26
    It means a gate length of the transistor
    has one micron. It's not comparable with
  • 4:26 - 4:32
    all the processors you can buy now. It's
    quite old, it's really stuff from the 80s.
  • 4:32 - 4:38
    But we do it in a new way. We don't use
    the technology from the 80s. We do it with
  • 4:38 - 4:46
    the knowledge and all the experience from
    newer technology. Doing it again and using
  • 4:46 - 4:53
    some steps which are not so common, well,
    it wasn't common in the 80s. So why one
  • 4:53 - 5:02
    micron? One micron also means that the
    transistors are very robust against five
  • 5:02 - 5:12
    volt. Five volt was a usual supply voltage
    in the 80s, 90s and something like that.
  • 5:12 - 5:19
    Now the supply voltage is going down,
    down to less than one volt but for
  • 5:19 - 5:28
    tinkerers, for hobbyists, for makers, it's
    a nice value because older stuff, many
  • 5:28 - 5:36
    boards are still working with five volts
    and we're able to handle this voltage. So
  • 5:36 - 5:44
    we have a twin-well process; usually in
    the 80s there was just one well. OK, we
  • 5:44 - 5:50
    have to hurry up. We have three metal
    layers. We have interesting additions and
  • 5:50 - 5:57
    we are suitable for low tech. Ghetto tech,
    I would say. You can use it without
  • 5:57 - 6:03
    sophisticated equipment. We can analog
    stuff and so on and analog stuff means you
  • 6:03 - 6:12
    don't need small structures. OK. Areas
    where we have to work on. First, the
  • 6:12 - 6:19
    process. It's almost done. You have
    figured out it works with measuring. OK,
  • 6:19 - 6:24
    the next stuff: we need the tools. But the
    tools are also very old and mostly not
  • 6:24 - 6:31
    usable. We have to deal with that stuff.
    We have to rethink the tooling for that
  • 6:31 - 6:38
    and we need standard cells. That's my
    task. OK, so a couple of thoughts about
  • 6:38 - 6:45
    standard cells. They are very common.
    Usually, if you have a need to translate
  • 6:45 - 6:55
    your Verilog or VHDL and to bring it on a
    silicon you need small gates. NAND gates,
  • 6:55 - 7:05
    OR gates and so on. But these gates need a
    lot of representation, the combinatorial
  • 7:05 - 7:09
    sequencing. So OK. These are typical
    cells. Just a couple of them. But imagine,
  • 7:09 - 7:16
    we need much much more and there's design
    goals for the standard cells as we need
  • 7:16 - 7:24
    almost complete possibilities. If you have
    just this small selection of cells, the
  • 7:24 - 7:33
    netlist becomes huge and every gate in the
    netlist also means a dedicated delay. If
  • 7:33 - 7:39
    you have long chains we have a long delay
    so that our operating frequency goes down.
  • 7:39 - 7:46
    So if you have more complex gates we are
    better but doing all this stuff is heavy,
  • 7:46 - 7:55
    but we like to be lower power. That means
    our cells have to be consumption less
  • 7:55 - 8:01
    power than usual. We want to be fast but
    yes, of course, it doesn't fit all
  • 8:01 - 8:07
    together. OK. So we need it for
    simulation. We need it for synthesis. We
  • 8:07 - 8:15
    need it for timing. As you can see
    everywhere on the slides and, of course,
  • 8:15 - 8:22
    documentation. That's a lot of work. We
    are a small team. I am the only guy who is
  • 8:22 - 8:26
    dealing with the standard cells it's
    usually our teams also are doing that. So
  • 8:26 - 8:34
    OK, we need a tool for that which does all
    the stuff for us. And this cell generator,
  • 8:34 - 8:42
    I called it popcorn, because I put in some
    corn and it rised up with the heat. So we
  • 8:42 - 8:54
    can get all the representation. So
    currently I have this tool on the
  • 8:54 - 9:02
    repository which is Tcl which does some
    stuff I like, I need, but not all. But it
  • 9:02 - 9:08
    already seems very ugly. So for me I like
    to rewrite that but I don't figure out
  • 9:08 - 9:13
    currently which language I like to use for
    that. Next time it could be rust, it could
  • 9:13 - 9:18
    be scheme or something like that. We need
    another language for that. So if someone
  • 9:18 - 9:23
    would help - please, but that's the next
    task, if you have the wafer done
  • 9:23 - 9:31
    completely and measured. OK. That's a link
    for the repository where you can look at
  • 9:31 - 9:37
    the current status and there's a wiki
    where I like to describe why I'm doing
  • 9:37 - 9:46
    what in which way. But yes, we have to do
    a lot more. OK.
  • 9:46 - 9:55
    Andreas: OK. Hi. I take a look at the
    current tooling that exists like
  • 9:55 - 10:02
    layouting, place and route to minimize the
    yield on the wafer. And obviously because
  • 10:02 - 10:09
    this is the LibreSilicon project we look
    at open source tools. So we have Yosys and
  • 10:09 - 10:17
    graywolf, qrouter and several other FPGA
    routers that exist. Yosys is pretty good.
  • 10:17 - 10:24
    We can probably use this for the
    synthesis. But the other tools, they lack
  • 10:24 - 10:31
    very critical qualities for this... for
    silicon because they, for example, they
  • 10:31 - 10:41
    are part of qflow which is an FPGA
    workflow. So the graywolf tool, it
  • 10:41 - 10:50
    originates in academia. It's, like, it's
    many decades old. It comes with some very
  • 10:50 - 10:55
    good ideas, for example simulated
    annealing which is a meta-heuristic
  • 10:55 - 11:02
    you can use to solve NP-hard problems, but
    it's only one of the many choices you can
  • 11:02 - 11:11
    make to solve the extra hard problems. But
    it also comes with bad implementation, for
  • 11:11 - 11:17
    example inline syscalls is a very bad
    idea. And it's also written in C and blah
  • 11:17 - 11:25
    blah, OK. Qrouter is actually... it's
    pretty good. It started in 2011 by Tim
  • 11:25 - 11:34
    Edwards. It's widely used for, by hobbyists
    and enthusiasts to route for FPGAs. But
  • 11:34 - 11:38
    it's not ready for silicon and it's
    especially not ready for our LibreSilicon
  • 11:38 - 11:49
    process which would require us to to write
    a lot of C code for Qrouter. Also
  • 11:49 - 11:56
    parallelism apparently is not in scope, so
    I mean if we want to scale up, for example
  • 11:56 - 12:04
    place and route in the cloud or whatever or
    use modern CPU architectures, we are stuck
  • 12:04 - 12:12
    with sequential routing which is pretty
    bad. Also it lacks a very important
  • 12:12 - 12:18
    aspect, in my opinion, which is formal
    correctness. So when we produce wafers in
  • 12:18 - 12:25
    the fab we want to make sure that they
    don't blow up in our faces. This is why we
  • 12:25 - 12:31
    need some form of proof that our
    algorithms are correct and therefore the
  • 12:31 - 12:41
    result is correct. There are also other
    productive tools that are proprietary
  • 12:41 - 12:45
    where we can look at, but we cannot use it
    or fork it or whatever but we can learn
  • 12:45 - 12:52
    from the research that has been done, for
    example BonnRoute. BonnRoute is used by
  • 12:52 - 13:01
    IBM. The Cadence suite, I believe, is used
    by Intel and the Alliance tools is French
  • 13:01 - 13:08
    academia. Very UNIXy, I mean it's a very
    it's a very large set of small tools that
  • 13:08 - 13:13
    convert different file formats to another.
    I mean, maybe you encountered this problem
  • 13:13 - 13:18
    before when you did some hardware design;
    you have many different file formats that
  • 13:18 - 13:25
    all don't play together very well. So you
    have tools like X to Y which convert file
  • 13:25 - 13:33
    format x to y. And you see when you want
    to place and route and layout a very very
  • 13:33 - 13:40
    large chip, like a Very Large Silicon
    Integration, then this isn't even done,
  • 13:40 - 13:45
    like, automatically by tools. This is done
    with manpower. When you look at a very
  • 13:45 - 13:54
    large chip done by Intel or IBM. So this
    is an example of a very very large chip as
  • 13:54 - 14:03
    you can see. I mean do you think this has
    been done by automation like industry 5.0?
  • 14:03 - 14:09
    No. This is all manpower and a lot
    of manpower. Which we don't have, The
  • 14:09 - 14:17
    LibreSilicon project at the moment. So
    this is the state of the art is like
  • 14:17 - 14:25
    okay the manpower thing is one aspect but
    the other thing is so what you do is
  • 14:25 - 14:36
    you do placing and routing at different
    steps at the design process so you do
  • 14:36 - 14:45
    placing for a very large chip, floor
    planning and then you do a global routing
  • 14:45 - 14:52
    which is you /can imagine it like
    routing along a rough chessboard. And
  • 14:52 - 14:56
    after that you do a very detailed
    routing where all the different
  • 14:56 - 15:04
    constraints regarding your technology come
    into play and so again the formal
  • 15:04 - 15:10
    correctness aspect. So you have some
    imperative algorithm that you cannot prove
  • 15:10 - 15:16
    will blow up. And it's also not a very
    parallel code. So you're still stuck with
  • 15:16 - 15:29
    the sequential nature of the code and you
    have no parallelism. What we propose is to
  • 15:29 - 15:35
    not place and route for large chip but
    to decompose the large chip into much
  • 15:35 - 15:43
    smaller units like a component hierarchy
    or a sub cell hierarchy and then place and
  • 15:43 - 15:51
    route the small chips at the same time and
    then reuse the small units in larger
  • 15:51 - 15:58
    units. So you get an evaluation tree you
    can work on and compile just the
  • 15:58 - 16:10
    components you need. Also we propose
    satisfiability modulo theory solvers so we
  • 16:10 - 16:18
    can have some first order logic where we
    can have constraints on the components,
  • 16:18 - 16:24
    how they are placed for example they
    don't - they must not overlap.
  • 16:24 - 16:33
    Let's take the most simple example I will
    show you like later. And also we want to
  • 16:33 - 16:44
    achieve parallel or declarative code. So
    as you can see we have some, we have many
  • 16:44 - 16:51
    disagreements with academia and industry
    which work very well together for example
  • 16:51 - 16:56
    when you want to study semiconductor
    design you have to sign some NDAs with IBM
  • 16:56 - 17:06
    or Intel to do that. So, they say
    placement and routing or floor planning
  • 17:06 - 17:11
    and routing are different problems and
    they need to be solved at different times
  • 17:11 - 17:19
    in the process. And then all the
    components can be registers or NAND gates
  • 17:19 - 17:23
    it does matter they all treated the same.
    No it only matters that uh the floor
  • 17:23 - 17:27
    planning stand first and then the routing
    the closed routing then a detailed
  • 17:27 - 17:35
    routing. What we propose is that place
    and route is actually the same problem and
  • 17:35 - 17:45
    that registers are different from full
    adders. Okay. So the geographical
  • 17:45 - 17:54
    partitioning of a wafer is called floor
    planning or the placing step. And this
  • 17:54 - 18:01
    results in a cut tree. So this is how they
    do routing hierarchies. They just divide
  • 18:01 - 18:09
    the wafer into smaller pieces and then do
    the following steps based on this
  • 18:09 - 18:16
    placement. What we want to do is have
    subcell hierarchies and those sub cells
  • 18:16 - 18:23
    they are either explicit like they are
    explicitly developed for example the
  • 18:23 - 18:33
    rocket ship is very modular and it has
    many explicit verilog modules you can use
  • 18:33 - 18:40
    and place and route that and then reuse
    it. And it also has implicit sub cells like
  • 18:40 - 18:46
    for example most of the time. For example
    you have a full adder it obviously is
  • 18:46 - 18:52
    composed of one bit adders so you can
    place and route one bit adder and then
  • 18:52 - 18:57
    place and route based on the one bit
    adders that you artfully placed and routed.
  • 18:57 - 19:01
    And as a result you get a full adder.
    That's just one example but I will show
  • 19:01 - 19:11
    you a tree, a few slides. So there you see
    parallelism. There's something very
  • 19:11 - 19:20
    important for us. BonnRoute allocates a
    lot of research to have some mathematical
  • 19:20 - 19:28
    model for concurrency and shared memory
    models. qrouter, which is the open source
  • 19:28 - 19:35
    alternative, has none. I mean that's
    apparently not in scope. And what I
  • 19:35 - 19:47
    propose for the LibreSilicon compiler is
    the map and reduce approach. And as I've
  • 19:47 - 19:52
    mentioned you get explicit subcell
    hierarchies through high modularization.
  • 19:52 - 19:57
    That is done by the developers and
    you also get implicit subcell hierarchies
  • 19:57 - 20:06
    by compression-like algorithms that exline
    as opposed to inline the registers or one
  • 20:06 - 20:14
    bit adders. And this is also about
    preserving these newfound hierarchies in
  • 20:14 - 20:20
    the compiler interfaces so you don't end
    up inlining them again because this is
  • 20:20 - 20:25
    not a Von Neumann architecture where it
    would make sense to inline a lot of code.
  • 20:25 - 20:31
    So the code runs on the stack and the
    level 1 cache. This is about reusing
  • 20:31 - 20:42
    components. Okay. So this is a part of the
    rocket ship, the system bus is one
  • 20:42 - 20:50
    component of a very modular chip rocket
    ship. And as you can see it is composed of
  • 20:50 - 20:57
    several simple lazy modules and those
    simple modules are again composed of other
  • 20:57 - 21:02
    components. And then you have a lot of
    queues and this number on the left says
  • 21:02 - 21:08
    how many times it's been used. For example
    queue 15 is used 5 times in the
  • 21:08 - 21:16
    AXI4Deinterleaver and this is only the
    explicit hierarchy that is declared by the
  • 21:16 - 21:24
    developer. Okay. Now when you apply some
    compression-like algorithms you can
  • 21:24 - 21:32
    actually gain, you can get more leaves so
    you can be even more modular. For example
  • 21:32 - 21:40
    queue 1 is composed of several implicit
    modules and you can see one queue is even
  • 21:40 - 21:47
    reused seven times. So you just route,
    place and route these green leaves like
  • 21:47 - 21:52
    once and then you can reuse it in the
    queue 1 and everywhere where queue 1 is
  • 21:52 - 22:02
    reused some at some other point in the
    chip. Now I want to state a very simple
  • 22:02 - 22:10
    optimization problem. What we need for
    the process is to have components and
  • 22:10 - 22:16
    wires that connect the components or nets
    and these nets and components are actually
  • 22:16 - 22:24
    rectilinear geometries, the components
    shall not overlap and the nets shall
  • 22:24 - 22:32
    overlap with the respective pins they are
    supposed to connect. The minimizing goals
  • 22:32 - 22:36
    of this optimization problem is layout
    area, which is the most critical one
  • 22:36 - 22:43
    because this is what maximizes yield, the
    maximum wire length because it's about
  • 22:43 - 22:50
    resistance, the wire count you want to
    keep very small but you want to allow for
  • 22:50 - 22:57
    wires. The crossing number is a
    computational thing. It doesn't really
  • 22:57 - 23:01
    matter for the implementation on the
    silicon and you also want maybe you want
  • 23:01 - 23:12
    to minimize the wire jogs which is bends
    in the wire. So to to solve optimization
  • 23:12 - 23:19
    problems in 2018 maybe you want to use an
    abstraction from the SAT solvers. You used
  • 23:19 - 23:26
    to know academia came up with some pretty
    neat theory is called satisfiability
  • 23:26 - 23:35
    modulo theories and you can just put some
    first order logic and give it to a solver.
  • 23:35 - 23:41
    I've listed a few. For example ABC is used
    by Yosys and Z3 from Microsoft, also very
  • 23:41 - 23:47
    promising product, but you can obviously
    choose from many products by academia and
  • 23:47 - 23:55
    industry. Just a quick reminder what
    boolean satisfiability is: find
  • 23:55 - 24:00
    assignments for all these six variables
    which are boolean so that the whole term
  • 24:00 - 24:07
    is true. And now with SMT or
    satisfiability modulo theories you can do
  • 24:07 - 24:14
    the same thing but now with integers and
    also more complex data types but integers
  • 24:14 - 24:23
    are the most interesting. So let's do
    something with SMT. For example we have a
  • 24:23 - 24:29
    component that is rectangular. And now you
    can see this is like a Cartesian
  • 24:29 - 24:35
    coordinate system and you have the left
    bottom point which is x and y and then you
  • 24:35 - 24:41
    have the right and the top point. And now
    if you for example have this problem that
  • 24:41 - 24:47
    you don't want to have overlapping
    rectangles you can have a rectangle A and
  • 24:47 - 24:57
    rectangle B and declare these coordinates
    and then have some proposition that shall
  • 24:57 - 25:04
    be true and to have a proposition that
    says they shall not overlap is to say
  • 25:04 - 25:10
    this. I mean it's actually the lower half
    that makes sure that they don't overlap
  • 25:10 - 25:15
    and the upper half makes sure that the
    components actually have the right
  • 25:15 - 25:20
    dimensions. Well in this example they
    obviously have the same dimensions the
  • 25:20 - 25:26
    same components. And so you make sure
    that the left point of the second
  • 25:26 - 25:44
    rectangle is right of the... Okay no,
    never mind. One last important point I
  • 25:44 - 25:50
    want to make is that this this framework
    we want to create, it's not based on the
  • 25:50 - 25:58
    inheritance model that we've seen in the
    process steps right now. But we want to
  • 25:58 - 26:02
    combine the problems. For example the
    overlapping problem, the pin connect
  • 26:02 - 26:05
    problem, and then arbitrary constraints
    that come up during the process
  • 26:05 - 26:10
    development that Dave and Hagen will
    supply me with and I will formulate that
  • 26:10 - 26:15
    in first order logic. And then this
    makes sure it's formally correct and it
  • 26:15 - 26:24
    doesn't blow up. And as you can tell I
    mean I've combined many NP-hard
  • 26:24 - 26:29
    problems at the same time but I think we
    can manage that if we have very small
  • 26:29 - 26:37
    cells so I'd suggest we just stay here and
    don't do all this for very large chips but
  • 26:37 - 26:46
    reuse small chips and then reuse the small
    chips in other small chips. The silicon
  • 26:46 - 26:54
    compiler is one half of maximizing
    yields. And the other half is to get the
  • 26:54 - 27:03
    process right so to get the process right,
    we have David and Victor. So please.
  • 27:03 - 27:14
    David: So thanks for the handover. So very
    first. There's a lot of questions why Hong
  • 27:14 - 27:20
    Kong. So one thing why this is a
    really suitable place to do that is
  • 27:20 - 27:27
    because of history like the epic Commodore
    64 has been made in Hong Kong. Then the
  • 27:27 - 27:33
    chips in the first Macintosh have been
    made in Hong Kong and all of these
  • 27:33 - 27:43
    manufacturing lines. Some of them at least
    one is still available. So also there is a
  • 27:43 - 27:51
    very advanced laboratory. That's the NFF,
    Nano Fabrication Facility in
  • 27:51 - 27:59
    Clearwater Bay and they let us kindly use
    their equipment to develop this process.
  • 27:59 - 28:07
    Also one of the sectors I mentioned
    before, RCL semiconductors, they're really
  • 28:07 - 28:12
    open to introduce LibreSilicon in their
    mass-manufacturing lines: one in Shenzen,
  • 28:12 - 28:28
    one in Tai Po. So in conclusion of that we
    have advanced R&D labs there. There is
  • 28:28 - 28:36
    factories available. We can easily export
    it to here over channels which already
  • 28:36 - 28:44
    exist. Right. And also in general it's
    just more relaxed over there. And I don't
  • 28:44 - 28:59
    like minus degrees. So our process is a
    little bit of a monster. So it makes sense
  • 28:59 - 29:05
    to tackle that one by one so we are right
    now feeling ourselves upwards to get the
  • 29:05 - 29:14
    standard CMOS debugged, final with
    optimized frequencies there. But we
  • 29:14 - 29:19
    already have on the Pearl River, I've shown
    you, we already have test structures for
  • 29:19 - 29:27
    high voltage MOSFETS, B junction
    transistors, Zener diodes, even flash,
  • 29:27 - 29:35
    resistors, and caps. So it's only a
    question of effort I guess in the next few
  • 29:35 - 29:46
    months to get that working. When we
    designed the process like, how it usually
  • 29:46 - 29:51
    works when you make a process, you look at
    the machines you have availlable, what can
  • 29:51 - 29:57
    these machines do, optimum operation range
    and then you look what substrate, what
  • 29:57 - 30:01
    material you have available and then you
    start tinkering you own little proprietary
  • 30:01 - 30:09
    process. That's how fabs do that. And we
    said, OK, well, to the point where we look
  • 30:09 - 30:16
    at the machines - what can they do? We
    do the same, but afterwards we look that
  • 30:16 - 30:24
    it's portable. Not specific to the
    equipment. So just because we have certain
  • 30:24 - 30:30
    machines which can do awesome things, but
    are really exotic, doesn't mean we have to
  • 30:30 - 30:38
    use them. So we avoid exalting machines so
    that it's as portable as possible. And we
  • 30:38 - 30:43
    also try to use wet etching whenever
    possible in order to make sure that you
  • 30:43 - 30:55
    even can build it in a basement. And here
    Evan Heisenberg may be interested now in,
  • 30:55 - 31:03
    you know, changing business into a less
    dangerous business. And, yeah, they can't
  • 31:03 - 31:07
    be leading the innovation hub Hamburg I've
    seen, like this improvised clean room with
  • 31:07 - 31:16
    just a diffusion furnace. So, that's a
    cross-section of the... it's not
  • 31:16 - 31:22
    finalised, but you see a cross section
    theoretical that's... by the way, you can
  • 31:22 - 31:29
    find it on GitHub as well. It's all in the
    publications, everything we develop, all
  • 31:29 - 31:38
    the measurement data, all this on GitHub.
    So that's actually the layout of these
  • 31:38 - 31:48
    little squares here on the wafer. You
    see the apple in the middle. It's just in
  • 31:48 - 31:56
    this year. That's, uh, it's nice. I have a
    Python script in the GDS2 generator
  • 31:56 - 32:01
    tool folder for Python and you
    can take any png or anything and just
  • 32:01 - 32:07
    convert it into layout format, so you can
    put your own pictures onto the metal free
  • 32:07 - 32:15
    layer. So in case you already have
    interest into making little trips also.
  • 32:15 - 32:21
    It's also possible to make, like, ear
    rings also with ... We don't care as long
  • 32:21 - 32:27
    as there are 4 more millimeters on the
    silicon. You can put pictures on the
  • 32:27 - 32:37
    silicon. So that was the Pearl River
    right. And the Pearl River fulfills the
  • 32:37 - 32:45
    function for us at the moment to debug all
    the features of this LibreSilicon process.
  • 32:45 - 32:51
    Then the next thing we have to use it to
    calibrate new foundries so now, we
  • 32:51 - 32:56
    developed it at the NFF in Clearwater Bay
    right. And afterwards we go over to HQ
  • 32:56 - 33:03
    with, to the RCL guys in Tai Po, and they
    have the machines and then we have to pipe
  • 33:03 - 33:09
    the Pearl River layout through there as
    well and repeat that process over and over
  • 33:09 - 33:15
    again until the measurement data, like
    the frequent, the you know the Beta
  • 33:15 - 33:22
    depending on Omega of the transistors and
    the resistance of the wires and everything
  • 33:22 - 33:28
    kind of is the same as at NFF so that you
    can basically, as I mentioned before one
  • 33:28 - 33:33
    of the design concerns is portability
    that you can basically prototype a chip
  • 33:33 - 33:40
    at the NFF and then produce it in RCL or
    in maybe some other fab in Shenzhen or
  • 33:40 - 33:49
    whatever. And so and if there are new
    features coming out which also make a new
  • 33:49 - 34:00
    release of the Pearl River test waver and
    we give that around they push it to GitHub
  • 34:00 - 34:09
    and people can introduce and calibrate the
    process to support the new feature. And so
  • 34:09 - 34:13
    that's how does that work. So usually,
    typically you have something like a photo
  • 34:13 - 34:20
    mask like here. I didn't bring that one
    because it's in a clean room there and the
  • 34:20 - 34:27
    dust might scratch my micro structures on
    there. So also afterwards I have to clean
  • 34:27 - 34:32
    it for half an hour and when I come back
    to Hong Kong from here I'm so jetlagged I
  • 34:32 - 34:35
    just want to get started again, not wait
    for the mask.
  • 34:35 - 34:41
    But there's a picture.
    And these masks,
  • 34:41 - 34:49
    usually a stepper/aligner specific. If you
    don't have a stepper then you need to make
  • 34:49 - 34:56
    a direct transfer that means you actually
    have to put the chips in the size you
  • 34:56 - 35:00
    want to expose them directly onto the
    mask. Then press the mask onto the
  • 35:00 - 35:06
    photoresist, expose and develop. That's
    messy because you have to clean the mask
  • 35:06 - 35:10
    all the time. And it really depends. So
    actually you can do exposure even without
  • 35:10 - 35:15
    a stepper. So we actually really could do
    it also there in this university lab in
  • 35:15 - 35:24
    Hamburg. So all you need is a new UV
    light. laugs So we have a little bit more
  • 35:24 - 35:33
    advanced tech in Hong Kong. So we have
    here an SVG coater, this baby dispenses
  • 35:33 - 35:41
    automatically HPR 504, a resist. So we
    actually just have to put in the left, you
  • 35:41 - 35:46
    see the cassette slot. So you put there
    like twenty five wavers or so and then you
  • 35:46 - 35:51
    have a receive slot and put another
    cassette there and it just starts sucking
  • 35:51 - 36:02
    in the wafers one by one, puts primer on
    it, soft bakes it, and easy. Then you
  • 36:02 - 36:10
    expose it, develop it, hard bake it,
    chilled. We have two types of resist actually
  • 36:10 - 36:19
    and the 6400L for the implantation
    unfortunately has to be put in manually.
  • 36:19 - 36:24
    So it comes and it gives you 10 seconds
    to open the chamber and put the resist on
  • 36:24 - 36:31
    it. In both cases however it doesn't
    really matter so much because the
  • 36:31 - 36:38
    thickness of the resist is depending on
    the RPMs of the spin coating unit. So you
  • 36:38 - 36:45
    just have to kind of put two thirds of the
    waver should be somehow covered with the
  • 36:45 - 36:54
    resist and the excess resist goes away.
    But you have to control the RPMs because
  • 36:54 - 37:03
    depending on when you do wet etching for
    instance and HPR 504 has to be enough
  • 37:03 - 37:11
    thick because of selectivity, so that you
    don't etch and consume the polymer,
  • 37:11 - 37:15
    the resist. So you have to make it thick enough
    that you don't have,
  • 37:15 - 37:18
    you haven't consumed all the polymer before
  • 37:18 - 37:24
    you have etched your structures. And the
    same goes for the implantation because you
  • 37:24 - 37:42
    need 6400L, this one can sustain higher
    temperatures so you can use an implanter.
  • 37:42 - 37:49
    Now afterwards after exposure development
    it looks like that. That's an alignment
  • 37:49 - 37:57
    cross for our optical stepper and for
    instance that's our ring oscillator. So
  • 37:57 - 38:07
    it's one of the structures on our Pearl
    River actually. So N well, P well. I have
  • 38:07 - 38:12
    to hurry up, only 10 minutes or so. So
    that's a picture of the developing we have
  • 38:12 - 38:16
    some P well mask developed so we have
    everywhere resist except in this little
  • 38:16 - 38:23
    crosses and stripes there. That's there
    below is the silicon where we implant. The
  • 38:23 - 38:33
    recipe is easy, first coat, expose the
    implant and then resist strip. Same for
  • 38:33 - 38:41
    the P well and after the resist strip you
    can put it into a diffusion furnace in
  • 38:41 - 38:49
    the atmosphere for like four hours. So
    where does the four hours come from? So
  • 38:49 - 38:54
    we have the Fick's equation. And the
    Fick's equation is essentially in a
  • 38:54 - 39:01
    similar shape like the laplace heat
    conduction equation, so to solve, there
  • 39:01 - 39:07
    are already nice solutions for it. So for
    instance if you use boron or phosphorus
  • 39:07 - 39:13
    which has the nice property that they have
    the same constants for this Dₑ. So if you
  • 39:13 - 39:19
    have the same temperature you basically
    have the same Dₑ for phosphorus and boron
  • 39:19 - 39:24
    so you can implant them next to each
    other and then put them at once into the
  • 39:24 - 39:30
    diffusion furnace and the wells are the
    same depth. So that's why these two
  • 39:30 - 39:37
    materials are usually used for diffusion.
    So that's one of the solutions that you
  • 39:37 - 39:45
    get, the surface for doping for the
    threshold equation which I also will rush
  • 39:45 - 39:53
    through in a moment as well. The
    equations you see here with background
  • 39:53 - 40:02
    doping it's a little bit much. As you have
    here this natural logarithm inside. But
  • 40:02 - 40:08
    besides that you see this jump and that's
    how you essentially build a well, you have
  • 40:08 - 40:14
    the background doping and you compensate
    the donors and acceptors with each other
  • 40:14 - 40:22
    so that's what this absolute value of the
    difference means. So the threshold
  • 40:22 - 40:28
    equation is pretty easy. And like
    basically mirrored for PMOS and NMOS that
  • 40:28 - 40:33
    just like mirrored in the sense that one
    of the transistors as PMOS is built on a N
  • 40:33 - 40:47
    well and NMOS is built on a P well. Right.
    And what essentially controls the
  • 40:47 - 40:51
    threshold voltage, so the operational
    voltage, which usually in the standard
  • 40:51 - 40:59
    CMOS is around 0.8 respectively minus
    0.8. That's doping here like the donars
  • 40:59 - 41:08
    respectively acceptors and the q as
    usually that's the oxide charge. This is
  • 41:08 - 41:16
    usually a process specific constant but
    that can change. And then you get
  • 41:16 - 41:22
    flash, it can change Q_SS and then
    it's flash. That's what you use in SONOS
  • 41:22 - 41:29
    flash, stands for silicon oxide nitride
    oxide silicon. So there you have a
  • 41:29 - 41:39
    standard again, NMOS in this case but you
    have a sandwich instead of a normal oxide
  • 41:39 - 41:45
    layer and for the gate oxide you have a
    nitride and oxide. These oxide layers
  • 41:45 - 41:53
    above and below the nitrate are called
    tunnel oxides. And the trick is that with
  • 41:53 - 41:59
    high enough energy you tunnel electrons
    into the, through the oxide into the
  • 41:59 - 42:04
    nitride where it's trapped and then you
    shift the operation voltage, the threshold
  • 42:04 - 42:12
    of the transistor. And when you then put
    one at it it doesn't turn on anymore and
  • 42:12 - 42:19
    that's essentially how the most used flash
    solution besides normal floating gate
  • 42:19 - 42:28
    works. It's really simple. So. And after
    you get your wells out of the furnace, so
  • 42:28 - 42:36
    I did a little detour. You want to make
    sure that the lateral diodes which got
  • 42:36 - 42:41
    into existence after diffusion don't
    create unwanted short circuits. So we use
  • 42:41 - 42:47
    the technology actually developed much
    later after one micron already has been
  • 42:47 - 42:53
    out. It's called STI shallow trench
    isolation. It's from the ULSI technology
  • 42:53 - 42:59
    as well as the silicide we use to reduce
    the resistance of the polysilicate.
  • 42:59 - 43:09
    Here are some pictures, we did etch this
    one in the lab. That's the islands so that
  • 43:09 - 43:14
    around everything going down that's the
    trenches in between the gates and between
  • 43:14 - 43:24
    the wells. So we isolate them from each
    other. So the recipe is pretty easy.
  • 43:24 - 43:27
    So either you have a plasma
    etcher around or if you're not
  • 43:27 - 43:32
    rich and don't have money to buy a plasma
    etcher from eBay you can also get this
  • 43:32 - 43:43
    tetramethylammonium hydroxide. And it's
    not even the german name, so cool, and
  • 43:43 - 43:52
    dilute it with deionized water 3:1 and
    this 25% TMAH solution you heat
  • 43:52 - 43:58
    it up to 80°C, dip your
    wafer in for six minutes and then you
  • 43:58 - 44:05
    would get your structures. Metal is
    easier. So we did here the metal
  • 44:05 - 44:12
    interconnect for the ring oscillator.
    They're etching it, also you make a
  • 44:12 - 44:19
    vacuum, deposit 100 nanometres aluminum,
    30 nanometers titanium for passivation.
  • 44:19 - 44:24
    Take the vacuum away dip it into HF until
    you don't see streaks on the titanium,
  • 44:24 - 44:28
    then into aluminum etchant until you don't
    see streaks from the aluminum. And then
  • 44:28 - 44:36
    you have your wires. I'll skip
    that one. That's just really interconnect.
  • 44:36 - 44:45
    But I plan to make videos soon where I go
    through the you know like daily video blog
  • 44:45 - 44:50
    of results but just that you see that you
    see the oxide depending on the angle it
  • 44:50 - 44:58
    has different colors. So that's L2 the
    isolation. And then you see the
  • 44:58 - 45:04
    topological measurement device. You see
    this one micron because we only deposited
  • 45:04 - 45:13
    a micron for now. You'll see the heights
    the differences and we see that one micron
  • 45:13 - 45:18
    is not enough. So we'd still have these
    sharp edges which we don't want. So we have
  • 45:18 - 45:24
    back in Hong Kong have to deposit another
    2 microns. And if you want a follow up you
  • 45:24 - 45:31
    go to my Github. OK? So Victor that's him
    and I have done that so far. It's only
  • 45:31 - 45:37
    like two weeks because it took a lot of
    time to get all the masks manufactured and
  • 45:37 - 45:42
    so a lot of bureaucracy. We already have
    that much and just stay tuned. We already
  • 45:42 - 45:47
    have figured out so much in the last two
    weeks that it shouldn't be long before we
  • 45:47 - 45:58
    can well finish all the features of Pearl
    River. Create models with Hawkins popcorn
  • 45:58 - 46:03
    and start figuring out all the analog
    stuff for our MCU and then we make
  • 46:03 - 46:07
    an MCU. That's the first thing we want to
    do as soon as we have the features figured
  • 46:07 - 46:17
    out of Pearl River. If the Goddess is nice
    to us. Yeah it's a discordia figurine,
  • 46:17 - 46:23
    it's really cheap on ebay. laugs
    So yeah. And that's like an overall
  • 46:23 - 46:31
    of the features. And we want them build
    this microcontroller, and yes because all
  • 46:31 - 46:34
    the folks don't believe that there are
    people who want to buy such items you
  • 46:34 - 46:45
    please fill out the survey. That one
    is from Hagens trip, i skipped it but
  • 46:45 - 46:50
    yeah. So yeah. Thanks. I'm done. And too
    late but sorry.
  • 46:50 - 47:01
    applause
  • 47:01 - 47:06
    Herald: Thank you for the talk. No, but if
    you wait we have time for questions. So
  • 47:06 - 47:11
    there are two microphones. One is in the
    middle and one is on the left side of the
  • 47:11 - 47:16
    stage. Line up and we're going to take
    some questions and there is already one
  • 47:16 - 47:23
    question from Microphone number two.
    Microphone 2: OK. So thank you for that
  • 47:23 - 47:29
    interesting talk and all the development
    that you're doing. I was wondering have
  • 47:29 - 47:38
    you had any time to test your transistors
    yet. And then later on do you plan to
  • 47:38 - 47:43
    release some sort of analog simulation
    capabilities.
  • 47:43 - 47:49
    David: Yes. Thats the plan for the next
    few weeks after I'm back in Hongkong. We
  • 47:49 - 47:53
    did go back to the cleanroom. We actually
    wanted to provide already something for the
  • 47:53 - 48:01
    Congress. Unfortunately we were noticed,
    short noticed that Thursday and Friday
  • 48:01 - 48:06
    they take the wet stations and the
    machines offline for maintenance of the
  • 48:06 - 48:13
    AC. So we have already like, the wafer, we
    have the isolation oxides but we didn't
  • 48:13 - 48:19
    have any time left to actually test the
    the you know only having polysilicon is
  • 48:19 - 48:23
    not enough. You have to also have metal to
    go with probes there, that stuff is
  • 48:23 - 48:27
    micron size.
    Hagen: Okay. So your question as I
  • 48:27 - 48:32
    understand was in the direction of
    simulation right? We like to measure all
  • 48:32 - 48:38
    the structures we have to produce and with
    the values we get we like to feed in spice
  • 48:38 - 48:46
    models. So you can do analog simulations.
    And yes we like to use this technology for
  • 48:46 - 48:50
    analog stuff because as I already
    mentioned one micron size is enough for
  • 48:50 - 48:56
    analog. You don't need smaller structures.
    Analog all this having huge transistor
  • 48:56 - 49:04
    size from 20 or 50 Microns. So they
    are huge, you don't need this small
  • 49:04 - 49:11
    technology. So they are quite feasible for
    analog stuff but let's say in this way if
  • 49:11 - 49:16
    you're doing analog stuff in a
    conventional way you have to sign all the
  • 49:16 - 49:21
    NDAs and you're stuck on this technology
    you're using. You can't transfer your
  • 49:21 - 49:27
    design to the next fab because in the next
    fab the PDKs are different. You have to
  • 49:27 - 49:31
    transfer or to translate all the
    structures there for a rebuild again for
  • 49:31 - 49:36
    the new technology if you have a
    technology which you can take from one fab
  • 49:36 - 49:43
    to another like our one. You are quite
    happy because the analog stuff you
  • 49:43 - 49:49
    designed once also fits for the next fab.
    So yes of course we like to support analog
  • 49:49 - 49:55
    stuff. We need help for that of course we
    have to measure, we are currently
  • 49:55 - 49:58
    developing the wafer, we are currently
    working on the documents how to measure,
  • 49:58 - 50:03
    what we like to measure and then we have
    to transfer the values to spice. But we
  • 50:03 - 50:09
    have documented how we are doing that.
    And so everyone can use the knowledge.
  • 50:09 - 50:12
    Mic 2: Thank you.
    Herald: Thank you. Mike one please.
  • 50:12 - 50:17
    Mic 1: Do you have any plans for
    open source mask production like.
  • 50:17 - 50:26
    David: Yes. Actually the problem is only
    that as I mentioned before. If you want to
  • 50:26 - 50:30
    have an opto mask for steppers that's
    always manufacturer specific. If you want
  • 50:30 - 50:38
    to have a direct transfer mask not a
    problem. So I guess so Sam is really
  • 50:38 - 50:45
    helpful in the lab. He runs the laser
    scriber. We could talk with the folks at
  • 50:45 - 50:51
    NFF. They were really lovely helpful
    really. They really like to really help us
  • 50:51 - 50:59
    a lot. And now that we talk with RCL.
    They also have laser scribers that we could
  • 50:59 - 51:05
    actually also start producing masks in the
    long run. So yes that's certainly one of
  • 51:05 - 51:13
    the things I intend to do is providing
    optical masks for exposure. Um yeah.
  • 51:13 - 51:17
    Herald: Thank you. Uh one more question
    from microphone two.
  • 51:17 - 51:25
    Mic 2: Great talk thanks. I'm really
    interested in the - what it would take to
  • 51:25 - 51:31
    build the fab. What's the minimum set
    of tools. We've already seen a couple of
  • 51:31 - 51:38
    orders of cost reduction in, through DIY
    bio hacking by making the tooling a lot
  • 51:38 - 51:44
    cheaper. Do you see that happening within
    the nearest decades and your sort of work?
  • 51:44 - 51:51
    David: Yes. So for instance I made my
    process by purpose this way that you can
  • 51:51 - 51:57
    actually improvise most of it like the LTL
    growing and deposition and everything with
  • 51:57 - 52:02
    a furnace. So what you need is a wet
    etcher like some wet etch station. You can
  • 52:02 - 52:08
    actually there is a video from Jeri
    Ellsworth called "making microchips
  • 52:08 - 52:16
    at home cooking with Jeri" and he does
    microchips in the kitchen so it's
  • 52:16 - 52:21
    not, you get scared like HFS, it dissolves
    your bones and so on and then you see the
  • 52:21 - 52:25
    guys who already have qualified, are
    qualified or employed there: they just
  • 52:25 - 52:34
    without any PPE, nothing just grab into
    the HF. That's just the skill to scare
  • 52:34 - 52:41
    folks from generating insurance problems.
    In general it's not really that dangerous
  • 52:41 - 52:48
    right. You can do the stuff at home. No
    problem. Yeah. So we intend. So this
  • 52:48 - 52:56
    process I made is so trivial. So we have
    also a branch called super low tech. We
  • 52:56 - 53:02
    just shall essentially but it's more RnD.
    But you could actually help there for
  • 53:02 - 53:08
    instance figure out the last details,
    get a furnace from eBay put it onto your
  • 53:08 - 53:17
    kitchen table start RnD-ing make some git
    pull requests and we're super happy. Okay.
  • 53:17 - 53:22
    So it's doable and the furnace you get on
    ebay. So no problem.
  • 53:22 - 53:30
    Herald: Thank you. Microphone 1 again.
    Mic 1: So you just said about the
  • 53:30 - 53:34
    analog stuff that a lot of that is usually
    under NDA from the fab. So have you
  • 53:34 - 53:39
    encountered any problems with the fab and
    that you're currently using in that you're
  • 53:39 - 53:44
    actually trying to discover these
    processes for yourself like you're
  • 53:44 - 53:48
    generating competition that they might not
    like, have you had any problems with that.
  • 53:48 - 53:55
    David: Oh no I had a nice phone calls,
    e-mails with the owner of the fab over in
  • 53:55 - 54:02
    Tai Po who also has a second branch in
    Shenzhen that's RCL. I actually asked him
  • 54:02 - 54:08
    recently "Hey is it okay when I use your
    logo in the presentation and implicitly
  • 54:08 - 54:13
    make an advertisement for your fab here?"
    No prob go ahead. That is like...
  • 54:13 - 54:19
    He's really eager to, LibreSilicon
    is what they need because every fab
  • 54:19 - 54:24
    usually has to invest money in to develop
    it. First they develop a proprietary
  • 54:24 - 54:30
    process right, or they license some
    proprietary process from another company
  • 54:30 - 54:38
    and then they have to invest RnD costs to
    develop IP cores for their setup. With
  • 54:38 - 54:44
    LibreSilicon the problem is solved for
    the companies because these foundry is
  • 54:44 - 54:49
    using LibreSilicon everything the
    community develops is on github. And
  • 54:49 - 54:56
    that's the IP catalog essentially.
    So they don't have to invest any
  • 54:56 - 55:00
    additional money into RnD-ing IP cores
    that's in the nature of open source that
  • 55:00 - 55:06
    there are IP cores popping into existence
    all the time. They can focus on the thing
  • 55:06 - 55:11
    they're best at: making silicon, right? So
    it's actually positive but only for the
  • 55:11 - 55:16
    small foundries that are really interested
    especially Shenzhen and now some in India
  • 55:16 - 55:22
    and some of the big foundries and they
    will not, they are anyway the big
  • 55:22 - 55:27
    companies have the tendency to be as
    mobile as a cargo ship. So it will take at
  • 55:27 - 55:32
    least like two years until they
    acknowledge that LibreSilicon exists and
  • 55:32 - 55:40
    then we might expect some legal you know
    bullying. But for now they won't even they
  • 55:40 - 55:46
    just laugh right. They just laugh at best.
  • 55:46 - 55:50
    Herald: We're going to have two more
    questions before we're out of time.
  • 55:50 - 55:54
    Microphone 2.
    Mic 2: Why did you go for the twin well
  • 55:54 - 55:58
    process as opposed to the simpler single
    well?
  • 55:58 - 56:02
    David: Uhm that's a good point. That's
    also something with portability and if you
  • 56:02 - 56:07
    have different events or different
    supplier for substrate it might be that in
  • 56:07 - 56:14
    n-doped or un-doped substrate. So with
    twin well architecture and actually we
  • 56:14 - 56:18
    have on the n-well we also built p-bases
    and in these n-bases, so we have actually
  • 56:18 - 56:28
    like stacked wells in the n-wells and
    p-wells. So actually it's a one two. Um
  • 56:28 - 56:36
    Pentagon Well I don't know. Um and it's
    just that you can shift to
  • 56:36 - 56:43
    the doping of the n- and the p-substrate.
    According that you fit LibreSilicon
  • 56:43 - 56:48
    requirements to still have the physical
    properties ensured by LibreSilicon. No
  • 56:48 - 56:53
    matter whether you get your substrate from
    somewhere from Great Britain or from
  • 56:53 - 57:01
    TaoBao.
    Hagen: Okay. The thing is we looked before
  • 57:01 - 57:09
    at eBay which wafer we can get. Currently
    NFF is supporting us with wafers. But if
  • 57:09 - 57:15
    you're looking on eBay or Alibaba. What
    else. We get different wafers with
  • 57:15 - 57:18
    different dope agents.
    And if you have something with say OK
  • 57:18 - 57:25
    we're just building an n-well we have to
    verify or lie on the p-base right, or on
  • 57:25 - 57:31
    the p-substrate. And to avoid the obstacle
    the difficulty is: we're doing twin-wells.
  • 57:31 - 57:36
    We can just regulate our own dopant inside
    and we are fine. We don't want to have to
  • 57:36 - 57:43
    rely on the wafer or substrate itself.
    What was the basic point.
  • 57:43 - 57:47
    Herald: Thank you. And the last question
    from microphone 2.
  • 57:47 - 57:52
    Mic 2: So once you have your complete die
    how about packaging and bonding because
  • 57:52 - 57:57
    if you want to use it you have to place it
    somehow on the PCB.
  • 57:57 - 58:05
    David: Yes. So um. We have a bonding setup
    at Tai Po already. That's what still is
  • 58:05 - 58:10
    being used at the moment in Hong Kong is
    to bond a packaging. Then we have some
  • 58:10 - 58:15
    guys in HK SDP with packaging set up they
    have and can make nice tape reels and they
  • 58:15 - 58:21
    have also like uh after packaging tests
    like: did the bonding work, is it damaged
  • 58:21 - 58:27
    by the bonding, and so on. Hagen and I
    have figured out some nice bonding pad
  • 58:27 - 58:33
    design which didn't fit at all anymore
    into the talk I already over talk like
  • 58:33 - 58:43
    that. And but it absorbs the physical
    stress from bonding. So we think that
  • 58:43 - 58:48
    it's aluminum covered with titanium so you
    don't have to sweat away any oxides right
  • 58:48 - 58:55
    you have better bonding capability, better
    bonding properties. So it shouldn't be
  • 58:55 - 58:59
    such a problem. And we have plenty of
    bonding and packaging labs which have
  • 58:59 - 59:05
    already promised to help us. So it's
    really like small like to choose which one
  • 59:05 - 59:10
    we take.
    Hagen: Just an annotation if you like a
  • 59:10 - 59:15
    dedicated package please mail us. Right.
    We are fixed now on the dual in-line
  • 59:15 - 59:21
    package. We are thinking about flip chip
    BGA but if you have other package which is
  • 59:21 - 59:26
    more common for tinkerer or something like
    that please mail us.
  • 59:26 - 59:34
    Herald: Thank you. Thank you for the talk.
    That was the talk on LibreSilicon,
  • 59:34 - 59:40
    leviathan, chipforge, Andreas Westerwick
    and Victor. Thank you. Thank you.
  • 59:40 - 59:50
    applause
  • 59:50 - 59:55
    postroll music
  • 59:55 - 60:13
    Subtitles created by c3subtitles.de
    in the year 2020. Join, and help us!
Title:
35C3 - LibreSilicon
Description:

more » « less
Video Language:
English
Duration:
01:00:13

English subtitles

Revisions