Does anybody know how ADS calculates the distribution of Cbcx and Ccox in the HiCUM BJT model?

It seems that the Cbcx modeling in ADS is not consistent with the equations in the HiCUM online manual:

http://www.iee.et.tu-dresden.de/iee/eb/comp_mod.html

Thanks

It seems that the Cbcx modeling in ADS is not consistent with the equations in the HiCUM online manual:

http://www.iee.et.tu-dresden.de/iee/eb/comp_mod.html

Thanks

Actually I did some simulations to check that.

I turn off (or minize) all other terms and play with Cjcx, Rbx and fbc parameters. I set Ccox=0, and then set Cjcx=1pF, rbx=100Ohm.

From the equations given in the document, the external capacitance should always be zero no matter fbc=0 or 1, because Cbcx>=0=Ccox. But from the simulation I get that external capacitance equal to Cjcx when fbc=0, and 0 when fbc=1.

The results simulated using Cadence and HPads are same. However, there is a tiny difference between Cadence and HPads, I turn off all other terms and set Ccox=1pf, Cjcx=0, rbx=0, fbc=0. Then if I look at base it should be a pure capacitor, the real part of the conductance should be zero. In Cadence, I get the value about 4e-12 (sigma) from 1-100GHz, which can be considered as a numerical offset.

But in ads, I get several tens of micro-sigma (0-60uS) at 1-100GHz, and it is increases as a function of frequency. I also set Ccox=0 and put an external 1p F capacitor between the base and ground. This time the real part of the conductance is about 4e-12 (sigma), which agrees with the Cadence simulation.

It seems that there should be something different in ADS in the HiCUM model.

Thanks again.

>From the equations given in the document, the external capacitance should always be zero no matter fbc=0 or 1, because Cbcx>=0=Ccox. But from the simulation I get that external capacitance equal to Cjcx when fbc=0, and 0 when fbc=1.

No, the external capacitance should not be 0.0, instead the internal capacitance should be 0.0.

Please see equations at page 40 of Schroter's document, where C'jcx0 = C'bcx0 -Ccox != 0

and C"jcx0 = Cjcx0 - C'jcx0 = 0 for the given case.

>The results simulated using Cadence and HPads are same. However, there is a tiny difference between Cadence and HPads, I turn off all other terms and set Ccox=1pf, Cjcx=0, rbx=0, fbc=0. Then if I look at base it should be a pure capacitor, the real part of the conductance should be zero. In Cadence, I get the value about 4e-12 (sigma) from 1-100GHz, which can be considered as a numerical offset.

I don't know what user was looking at, If Y parameters, then the base capacitance will also include Cds, Ceox, Cjep as well as diode capacitances. The base conductance will include the conductances associated with diodes, its, ibet as well as the leakage conductance (1e-12) associated with each diode. It will be clearer if user look at device dc operating point results.

Again, ADS equations are the same as the equations given at page 40 of Schroter's document.

However, back to the difference between the HPADS and cadence simulation results, what I did is set the model parameters to minimize all other terms including the current (ibet, favl,...) and parasitics capacitors (ceox, cjcx,...). Only left one capacitor Ccox at the base. And I checked the Y-parameter when base, collector and emitter are biasing at 0V. Then I got a non-negligilbe real part of the y-parameter in ADS. I used the same model parameters in cadence and I got a pure capacitor.

Thanks