Connect with us

parasitic substrate bjt, how do you make it go away?

Discussion in 'Electronic Design' started by Roger Bourne, Mar 6, 2007.

Scroll to continue with content
  1. Roger Bourne

    Roger Bourne Guest

    Hello all,

    I will have a n-well/p-substrate photodiode (PD) on a asic.
    I am currently in the midst in attempting to quantify the impact of
    the parasitic substrate bjt that will be formed with the nearby n-
    wells. (E=cathode of PD, B=anode of PD, a.k.a grounded (0V) substrate,
    C=nearby nwells @ +1.5V). In order to minimize the [collector-base
    current contribution to the emitter] AND the [emitter-base current
    contribution to the collector], I plan to have the photodiode
    surrounded by a massive ring of substrate contacts. Thus the parasitic
    bjt aplha parameter will be as near as zero as possible, tending to
    mimic the bjt behavior to 2 diodes. However, I do not know how to
    evaluate how destructive the bjt will be to the photodiode's
    photocurrent resolution. In other words, how much of the collector
    current from the reverse-biased CB junction actually makes to the

    The photodiode needs to be able to measure photocurrents in the range
    of (less than 100fA), and as such needs no additional currents to be
    injected from the collector (even though they are ~Is).

    Does anyone have any data/info/advice concerning the parasitic
    substrate bjts of nwell (photo)diodes and their anullment w.r.t to a
    ring of substrate contacts?

    Any help will be appreciated
  2. colin

    colin Guest

    How about a guard ring or two driven at the same potential as the photodiode

    Colin =^.^=
  3. Robert Baer

    Robert Baer Guest

    Farichild used a "wraparound" technique to kill the lateral NPN beta
    from the output transistors (on one end of the opamp die) to the input
    transistors on the other end of the die.
    That beta was verrry small, maybe in the region of 10e-6 but it was a
    It has been over 30 years and i was never into design, but i think
    the trick was to have the power NPN surrounded with an N epi region
    surroundes with a P well - which added a lateral NPN. That added N epi
    )collector) was then tied to the substrate, therby collecting most of
    the (substrate) injected carriers.
    Do your modelling with all of the silicon - not just the photodiode
    (and other) parts.
    A lateral NPN with the collector (the ring) tied to the substrate
    should be modelled and connected to the photodiode, where extra
    collectors go to the other transistors / diodes; betas based on
    collector areas as well as "base" (ie: substrate) recombination losses.

    I hope that i have stated this correctly, or at least well enough to
    give you an idea as what to try.
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.
Electronics Point Logo
Continue to site
Quote of the day