PADS/ESD cells in TSMC & UMC 0.18u kits?

Discussion in 'Cadence' started by spectrallypure, Jun 9, 2008.

  1. Hello everybody! Sorry for this not-so-on-the-topic question...

    My research group has just been granted access to the 0.18u mixed
    signal Cadence design kits for both TSMC and UMC foundries, which we
    plan to use in order to implement a full-custom design. But to our
    surprise, there are no PAD or ESD protection cells at all in these
    design kits!! I have installed and browsed the libraries provided in
    these design kits and all haven't been able to find a single periphery
    cell. Now I wonder if these kits provide any of these cells at all, or
    if the designer has to design and layout all them manually!!! :S
    Does anybody know where are these cells located/supplied?

    I have noticed in the www that many people mention the use of ARM/
    Artisan IO cells... is this the only resource for IO/ESD cells for the
    aforementioned processes? Are there any non-commercial alternatives?

    Thanks in advance for any help/ideas!

    Regards,

    Jorge.

    P.S. By the way, other than the obvious, I also need the PAD/ESD cells
    because I need to take into account the parasitic loading of these
    structures in the early stages of our design.
     
    spectrallypure, Jun 9, 2008
    #1
  2. spectrallypure

    Riad KACED Guest

    Hi Jorge,

    I have no experience with the UMC design platforms but the other
    Foundry (at the mentioned technology node) does not provide any of
    your target libraries as far as I know. So it is worth looking at the
    two other companies you've mentioned above unless you're fit enough (I
    mean you or your team/company) to design that IOs/Pads. Well It's not
    that easy and needs to be qualified for the given process and
    application which is not straight forward really.

    Don't ask me my opinion about foundries that does not supply a full
    EDA design package (PDK + Digital Corlibs + IOs + Memories ...) I
    could be very rude ...

    Well, It's may be worth asking those foundries in case they have got
    something up sleeves ...

    Good luck anyway !

    Riad.
     
    Riad KACED, Jun 10, 2008
    #2
  3. spectrallypure

    Ozgur.Ates Guest

    Europractice grant access to Analog and Digital IO Pads from Faraday
    IP provider. Faraday Tech has included Analog ESD cells to
    Europractice portfolio recently. But be informed that this cells has
    neither schematic views for simulation nor physical layout views. You
    got only lef, def files with net layers and they replace IO cells with
    physical ones when you send the chip for fabrication. I used these
    cells but I can not simulate any cells for example I would like to
    know the leakage values of Analog ESD cells.

    Ozgur
     
    Ozgur.Ates, Jun 10, 2008
    #3
  4. spectrallypure

    S. Badel Guest

    I have noticed in the www that many people mention the use of ARM/
    Both foundries are supported by many different IP providers. ARM, Faraday, Virage Logic to name a few.

    ARM and Faraday both provide libraries free-of-charge, see
    http://www.arm.com/products/physicalip/product_overview.html and
    http://freelibrary.faraday-tech.com. Foundries also have in-house and sponsored libraries. You
    should get in touch with them for these things.

    Unless you're in a university, which seems to be the case, then they'd rather let you deal with your
    intermediary (MOSIS, Europractice, CMP or whoever).


    For non-commercial alternatives, there's a library available from Virginia Tech for TSMC 0.18 and
    0.25 processes. See
    http://www.vtvt.ece.vt.edu/vlsidesign/cell.php.

    There's also Oklahoma state university at http://avatar.ecen.okstate.edu/projects/scells/download.php

    Not sure on the quality of these cells free cells, though. And I'm not sure if they include I/O
    cells, never used them myself.


    Cheers,
    Stéphane
     
    S. Badel, Jun 10, 2008
    #4
  5. Thanks so much for the information. I contacted Europractice (which is
    our EDA & foundry provider) with regard to the Artisan cells for TMSC,
    but unfortunately they don't support them. So I contacted ARM
    directly, and they told me they don't directly provide those to
    universities, and that I should refer to an intermediary distributor
    like MOSIS, with whom we don't have any account/membership :(

    So, before going through the painful amount of paperwork and time that
    would mean affiliating to another foundry intermediary, I would
    appreciate if someone can comment on the type of IO/ESD cells
    available from Artisan for TSMC processes... do they allow for
    parasitics annotation / simulation in Analog Environment? I checked
    Faraday cells for UMC's processes, and as Ozgur said they don't
    provide this functionality. I would really like to know if Artisan
    cells are of the same type or if they have better simulation
    capabilities.

    In the meantime I'll give a look to those free academic cells...
    thanks again for the help.

    Regards,

    Jorge.
     
    spectrallypure, Jun 10, 2008
    #5
  6. spectrallypure

    S. Badel Guest

    http://www.mosis.com/Technical/Designsupport/artisan-university.html

    As a general rule, you'll always get frontend packages from those distributors. These're are ok for
    digital, but it can be a pain for mixed designs.

    If you motivate your request, you should probably be able to get the full layouts after signing
    proper NDAs if applicable. Try first with your intermediary ; others will automatically bounce you
    back to them anyway.

    Of course, this can take time...



    Stéphane
     
    S. Badel, Jun 11, 2008
    #6
  7. Hello Stéphane! Thanks for the link. I read a bit around it, but at
    http://www.mosis.com/Technical/Designsupport/artisan-instantiate.html,
    they finally state:

    "ADDITIONAL RESTRICTIONS:
    -You are not permitted to see the resulting complete GDS. MOSIS will
    not provide you with the GDS for your instantiated layout.
    -MOSIS cannot make any meaningful DRC results available, as the DRC
    polygons reveal details of the ARM cell layouts.
    -You cannot obtain cell layouts from MOSIS, nor will MOSIS refer you
    to ARM to obtain them."

    I think this is kind of conclusive, so I believe it might be wiser to
    point to other foundries providing full design kits. The problem is
    that so far the only one we know that provides fully-accessible IO/ESD
    cells is Austriamicrosystems, which unfortunately doesn't have any
    0.18um or smaller processes :( It's a real pity, because their design
    kits are the most comprehensive, well documented and instructive ones
    I've seen so far; other foundries' kits assume you hold a PhD in solid
    state process engineering, have taped out designs since 1980, and have
    12-inch wafers for breakfast, by the way! ;)

    ...According to your experience, do you know of any (university-
    accessible) foundries that work in very deep submicron, and provide
    fully-accessible IO/ESD cells on their design kits?

    Thanks again for any help.

    Regards,

    Jorge.
     
    spectrallypure, Jun 12, 2008
    #7
  8. spectrallypure

    S. Badel Guest

    ...According to your experience, do you know of any (university-
    Nope...

    Well I still think that you can get them. Now you've seen that it's not easy though... The real
    question I guess is do you *really* need them ?



    Stéphane
     
    S. Badel, Jun 13, 2008
    #8
  9. question I guess is do you *really* need them ?

    Yes, unfortunately. I am working on new type of output buffer for
    pulsed communications, and the prototype I did in AMS' 0.35um worked
    fine until I added the analog IO pads... then the capacitive loading
    of the pads corrupted (sucked away) the pulses significantly, and the
    ESD clamps between VDD and VSS generated a stray capacitive path that
    resonated with the package/pin parasitic inductances... a whole
    mess! :O

    Seems that I will have to stick to this kit anyway...
     
    spectrallypure, Jun 13, 2008
    #9
  10. Kholdoun TORKI, Jun 22, 2008
    #10
Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.