R3_CMC Report - WCM2004 Presentation

Document Sample
R3_CMC Report - WCM2004 Presentation Powered By Docstoc
					       r3_cmc Model Feedback
• Bugs in model code reported, notably
  by Analog Devices
  – “old” syntax @(condition) for ADMS
  – “uoff” not passed to `r3Ibody macro
  – opPoint variables declared but not set
  – duplicated definition of variables in module
    and analog block scope
  – “df” used before it is calculated
• Did not affect results, but are now all
  fixed
                   Resistor Update December 2006 - 1
           r3_cmc Extraction
• Uncertainty about how to extract
  parameters
  – getting unphysical values
• Must have data over “gate” bias to
  allow parameters to be determined
  – and low Vds so velocity saturation and self-
    heating are negligible
• Defined direct (non-iterative) extraction
  procedure
                   Resistor Update December 2006 - 2
   Depletion Model Parameter Extraction
                                   
   I ds  gVds , g  gf 1  df dp  Vi , Vi  Vds  2Vsg                   
               1            g i2                     2g i
                                                                   dp  Vi
             df 2        df 2 gf 2               df 2 gf
 g1  g 2
   2     2
             2( g 2  g1 )  p1  V1  V2                                        p1  1 df 2 gf 2
 2                          p   V  V 

 g1  g 3
         2
             2( g 3  g1 )   2   1 3 
                                                                                   p 2  1 df 2 gf


      gf  p 2 p1                                                           2
                                                       1                   g1            2 g1
                                   dp                                                         Vi
                                                           2               2    2         2
                                                   df                df gf              df gf
      df  1       p 2 gf
                                       Resistor Update December 2006 - 3
                                    long, >2 Vgs




Resistor Update December 2006 - 4
                                    long, only 2 Vgs




Resistor Update December 2006 - 5
                                       short, >2 Vgs




                                    velocity saturation
                                     and self-heating




Resistor Update December 2006 - 6
                                    short, only 2 Vgs




                  velocity saturation
                   and self-heating



Resistor Update December 2006 - 7
         r3_cmc Model Feedback
• Further Analog Devices evaluation
  – self-heating is done in a “sneaky” way
     • actually nicely allows RTH to go to zero or sw_et to turn
       self-heating off
  – (aside) should be thermal discipline (true)
  – can mess up thermal coupling
  – end resistors have no noise
     • neither do parasitic diodes
  – excellent analysis of possible numerical issues,
    uncovered mandatory ecorn<ecrit constraint
  – detected other possible numerical issue for rp or
    rm going to zero
  – multi-section model results differ from single
    section
                         Resistor Update December 2006 - 8
      Thermal Model Formulation
                                                                         dt_et

                                                                                 RTH
                ddtdt _ et  CTH 
       dt _ et
 pwr                                                        I TH  VI
                                                                                       CTH
        RTH


sw _ et  RTH  pwr  dt _ et  ddtRTH  dt _ et  CTH 
    • allows thermal resistance to go to zero nicely
    • allows simple toggling from ET to IT via sw_at
    • RTH has to be temperature independent
    • scales thermal quantities by RTH
    • also turns off thermal coupling if RTH or sw_et is zero

                         Resistor Update December 2006 - 9
                    Poly Resistors
• Highly linear in Vgs
            
    g  gf 1  df dp  df V sg                      dp     
• Really only two independent parameters
                
    g 0  gf 1  df dp                                   s   gf df   dp

• Extraction is underdetermined, cannot
  get all 3 parameters
• Physical analysis must be used
              Si Tox
                    2
    dp 
            0 X s  ox
                      2
                           Resistor Update December 2006 - 10
                     Needs
• Companies to check fits to technologies
  – diffused resistors
  – poly resistors
• Not aware of any additional issues, but
  needs more evaluation
• Verify poly resistor modeling
  – negative TC devices
• Finish documentation
• Want to finalize and vote yes or no as a
  CMC standard
                   Resistor Update December 2006 - 11

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:85
posted:5/8/2010
language:English
pages:11