Page 257 - Microsensors, MEMS and Smart Devices - Gardner Varadhan and Awadelkarim
P. 257

THERMAL  SENSORS     237

   the  voltage sensitivity S T  is  a constant depending on  the drive current:

                    k ET   //o    \          dV out  & B  .  (k  ,  A   to  im
               Vout = -  In  — -f 1  and S T =  -rtr-  = — In  — + 1 )  (8.10)
                      q    \h     J          dT     q    \I S  J

  The  overall temperature sensitivity  of  the  diode  depends  on  the  relative  size  of  the drive
   current  and  saturation.  When  the  drive  current is  set to a value well  above  the  saturation
   current, Equation (8.10)  becomes


                   r,  ^       IQ        % &B  /o  ,   ,  ^  j         /0  , , ,
                   y out % -  In — and ST  — In — when /o ^> 7 S         (8.11)
                          q    I*         q    I s
  Let us suppose that the forward current is 0.1  uA and about  100 times the diode  saturation
   current of approximately  1 nA; then, the expected  temperature sensitivity is  +0.2  mV/K.
   However,  in  practice,  the  temperature-dependence  of  a  diode  depends  on  the  strong
   temperature-dependence  of the  saturation current itself.  The  actual value can be  obtained
  experimentally  from  the  temperature-dependence  of  the  forward  junction  voltage  of  a
             10
   silicon  diode ,  that is,  -2  mV/°C, and therefore  V f0  oc T.
     In a similar way, a bipolar transistor can be used as a temperature sensor. For example,
   Figure  8.9(b)  shows  an n-p-n  transistor  in a common-emitter  configuration and  constant
   current circuit. From  our basic theory, the base-emitter voltage  V BE is proportional  to the
   absolute  temperature and simply related to the collector  current  Ic  by
                             k^T   /  /c \
                       VBE  =  —  In ( TH  where  7 co =  ^E-/S        (8.12)
                              q    Vco/
   where  A E  is  the  area  of  the  emitter,  J s  is  the  saturation  current  density,  and  I C0  is  the
  reverse  saturation  current.  More  accurate  models  can  be  developed  from,  for  example,
   those discussed previously for  a bipolar  transistor (Equation (4.20)),  but  the base-emitter
  current  approximates  well  in practice  to

                                V BE  ^  V BEO + AT                     (8.13)

  where  A.  is  an  empirical constant  that  depends  on  the  current density  and  process  param-
  eters  and  the  offset  voltage  V BEO  has  a  typical  value  of  1.3 V  when  the  base-collector
  voltage  V BC is  set  to  zero.
     To make a truly PTAT sensor, it is necessary to fabricate  two transistors -  one with an
  emitter  area A EI  and the other with A E2.  Then the difference in their base-emitter  voltages
  is  directly proportional  to  the  absolute temperature and  is  given  by


         A i/    ^7     ^r  \       i                   .   /A E2 \
         A VBE = (VBEI -  ^BE2) = -  In  - — — —  ^ -   In  -—          (8. 14)
                                 q    \Ic2J si  AEI/  q    \AEI/
  When  the two transistors  are identical, the collector  currents and saturation  current  densi-
  ties  are  equal, and  the  ratio of  the emitter areas  only determines  the  sensor's  response.

   10
     Typically 0.7  V at  25 °C for  silicon  (and  0.25  V for germanium).
   252   253   254   255   256   257   258   259   260   261   262