Signet
 
 Home
 Introduction 
 Release Information 
 Events 
 Source Code 
 Parameter  Determination  
 Documentation 
 Model Testing  &  Results 
 Verilog-A  &  Compiler  Information 
 FAQ 
 Contact Us 

Model Testing & Results

Model testing is an integral part of model implementation. The HICUM group tests each code in a rigorous manner before a release. The test is briefly outlined below. The test results for different model versions are also provided.

Test results including CMC QA suite for each model version

Any model code incorporates some default parameters into it unless and until it is forced to use some other specific model parameters. The set of default parameters guarantees the convergence of the code without showing any effect of HICUM model specific behavior. On the other hand, the test parameters are employed to test the transistor in different conditions to see the effects of HICUM model specific phenomena.

Typical test strategy:

  1. Intrinsic transistor: One dimensional treatment: no series resistance, no perimeter/external elements


    1. Temperature: T=300K
      DC Test: Forward/Reverse Gummel
        VB-/IB-forced forward output characteristics
        VB-forced reverse output characteristics

      AC Test: fT versus IC plots
        Y-parameter versus IC plots
        Y-parameters versus Frequency plots


    2. Temperature: T=200K, 400K, 600K
      DC Test: Forward/Reverse Gummel
        VB-/IB-forced forward output characteristics
        VB-forced reverse output characteristics

      AC Test: fT versus IC plots
        Y-parameter versus IC plots


    3. Electrothermal effect in the intrinsic transistor: (with non-zero RTH & CTH)

      As in 1.


    4. collector current spreading in intrinsic transistor: (with non-zero LATL & LATB)

      As in 1.

  2. Internal transistor: Added internal base resistance (RBI), tunnelling current source


    1. Temperature: T=300K
      DC Test: Forward/Reverse Gummel
        VB-/IB-forced forward output characteristics
        VB-forced reverse output characteristics

      AC Test: fT versus IC plots
        Y-parameter versus IC plots
        Y-parameters versus Frequency plots


    2. Temperature: T=200K, 400K, 600K
      DC Test: Forward/Reverse Gummel
        VB-/IB-forced forward output characteristics
        VB-forced reverse output characteristics

      AC Test: fT versus IC plots
        Y-parameter versus IC plots
  3. Complete transistor: Added external series resistances, oxide/metal capacitances

    1. Without substrate transistor and substrate network elements

      1. Temperature: T=300K
        DC Test: Forward/Reverse Gummel
          VB-/IB-forced forward output characteristics
          VB-forced reverse output characteristics

        AC Test: fT versus IC plots
          Y-parameter versus IC plots
          Y-parameters versus Frequency plots

      2. Temperature: T=200K, 400K, 600K
        DC Test: Forward/Reverse Gummel
          VB-/IB-forced forward output characteristics
          VB-forced reverse output characteristics

        AC Test: fT versus IC plots
          Y-parameter versus IC plots

      3. Electrothermal effect in the complete transistor: (with non-zero RTH & CTH)

        As in 1.


      4. collector current spreading in complete transistor: (with non-zero LATL & LATB)

        As in 1.

    2. With substrate transistor, but without substrate network

      Temperature: T=300K
      DC Test:Forward/Reverse Gummel along with substrate current

      AC Test: fT versus IC plots
       Y-parameter versus IC plots

    3. With substrate transistor and substrate network

      Temperature: T=300K
      AC Test: Y-parameter versus Frequency plots

Noise modeling can only be tested properly against the noise measurements of actual transistors.


Test Results: In the following, few test results are provided for the Verilog-A HICUM models. The corresponding codes, netlists and test parameters can be found in the Source Code section.

  • HICUM Level2 Version 2.4.0: QA setup:  ZIP file
  • HICUM Level2 Version 2.4.0: QA results:  ZIP file
  • HICUM Level2 Version 2.34: QA setup:  ZIP file
  • HICUM Level2 Version 2.34: QA results:  ZIP file
  • HICUM Level2 Version 2.33: QA setup:  ZIP file
  • HICUM Level2 Version 2.33: QA results:  ZIP file
  • HICUM Level2 Version 2.32: QA setup:  ZIP file
  • HICUM Level2 Version 2.32: QA setup documentation:  PDF file
  • HICUM Level2 Version 2.32: QA results:  ZIP file
  • HICUM Level2 Version 2.31: QA setup:  <ZIP file
  • HICUM Level2 Version 2.31: Comments on QA setup and results:  <Text
  • HICUM Level2 Version 2.31: QA results:  <ZIP file
  • Plot results for HICUM Level2 Version 2.31:  PDF file
  • HICUM Level2 Version2.30: QA results with Spectre:  <ZIP file
  • HICUM Level2 Version2.30: QA results with ADS:  ZIP file
  • Plot results for HICUM Level2 Version 2.30:  PDF file
  • HICUM Level2 Version2.24: QA results with Spectre:  ZIP file
  • HICUM Level2 Version2.24: Comparison of results with previous model HICUM Level2 Version2.23:  PDF file
  • Plot results for HICUM Level2 Version 2.23:  PDF file
  • Plot results for HICUM Level2 Version 2.22:  PDF file
  • HICUM/L2V2.22: Comparison with Analytical and DEVICE results:  PDF file
  • Plot results for HICUM Level2 Version 2.21:  PDF file and  ASCII files
  • Results showing compatibility between HICUM Level2 Version 2.21 and Version 2.11:  PDF file
  • Plot results for HICUM Level2 Version 2.11:  PDF file and  ASCII files
  • Plot results for HICUM Level0 Version 1.11:  PDF file and  ASCII files
Model verification with actual device measurements are the ultimate test of a model's quality and capability. So far HICUM has been widely tested and qualified as a state-of-the-art bipolar transistor model applicable to modern bipolar/BiCMOS technologies. In the publication archive, one can find various published documents showing the corresponding results. Here few representative results from the early model evaluation are given for a quick understanding:



A.Mukherjee, L.Hofmann
20 June 2014