Hi All,

     I  have simulated a fully differential HBT amplifier with cascode  transistors  for avoiding Miller effect.  The simulator is Spectre 5.0  and the HBT models are VBIC 1.15. In order to find  the input swing range ,  I have done a dc sweep of  differential signal using the Kundert's testbench (http://www.designers-guide.org/Analysis/diff.pdf ).  And  
checked the operating points  of  the  NPN cascose transistors. When the vbc is  99.48 mV and vbe is 912.1 mV , the cascode transistor is in forward active region ( region 1).
But when the vbc is  99.62 mV and vbe is still 912.1 mV , the transistor moves into  saturation region ( region 3). This is quite amazing since  in the HBT  model file  ,the B-C  forward
voltage ( vbcfwd ) is set to 200mV.  

    It seems to me that Spectre does not use the "vbcfwd" parameter for giving warning about operating region. Am I right ?  For our current project , we wish to allow vbc as high as 400 mV and still the HBT transistor will be physically in the forward active region. How can we  pursuade the Spectre to think like us  ? I mean, which parameter in the VBIC model  should I change ?
             
  With Regards,

         Analog Aroma

vbcfwd is not a standard VBIC model parameter; it must be something that Cadence added when implementing the model.

Andrew: you'll need to find someone in Cadence who can look at how that parameter is handled to see if it is done properly.