BJT hybrid-pi model parameter determination

Started by sert November 18, 2008
I'm using SwCAD and trying to determine the parameters of the 
high-frequency pi-model for the BJT. For convenience, I've 
combined the circuit with the BJT and the model in one circuit 
so I can combine the graphs as well:

http://i35.tinypic.com/vxf1ad.png

First I used small dc signals to determine the values for rpi, 
gm, ro and then I tried to use small ac signals to determine the 
values for the capacitors (the values in the model are just 
placeholders.) I wasn't able to do so. When I write the nodal 
equation for the collector of the BJT and solve for cu using the 
small-signal values obtained from the simulation, the value will 
always have an imaginary part which is obviously unacceptable.

Is it some limitation of the way SwCAD models the BJT? I don't 
think it is a limitation of the model because it is cited in 
many books in electronics as a reliable model up to ~100 MHz.
"sert" <jerry@hotmail.com> schrieb im Newsbeitrag 
news:Xns9B5AD36F45694jtjdfjdfjnbj@147.102.222.230...
> I'm using SwCAD and trying to determine the parameters of the > high-frequency pi-model for the BJT. For convenience, I've > combined the circuit with the BJT and the model in one circuit > so I can combine the graphs as well: > > http://i35.tinypic.com/vxf1ad.png > > First I used small dc signals to determine the values for rpi, > gm, ro and then I tried to use small ac signals to determine the > values for the capacitors (the values in the model are just > placeholders.) I wasn't able to do so. When I write the nodal > equation for the collector of the BJT and solve for cu using the > small-signal values obtained from the simulation, the value will > always have an imaginary part which is obviously unacceptable. > > Is it some limitation of the way SwCAD models the BJT? I don't > think it is a limitation of the model because it is cited in > many books in electronics as a reliable model up to ~100 MHz.
Hello, There are no limitations in LTspice, but in your circuit. You try to compare apples and oranges. 1. Your BCW60A SPICE-model model has a base resistance RbOhm and some other parameters you don't have in the pi-model. 2. To be honest, using any transistor with Vbe=2V is useless. Best regards, Helmut
On 18 nov, 19:47, sert <je...@hotmail.com> wrote:
> I'm using SwCAD and trying to determine the parameters of the > high-frequency pi-model for the BJT. For convenience, I've > combined the circuit with the BJT and the model in one circuit > so I can combine the graphs as well: > > http://i35.tinypic.com/vxf1ad.png > > First I used small dc signals to determine the values for rpi, > gm, ro and then I tried to use small ac signals to determine the > values for the capacitors (the values in the model are just > placeholders.) I wasn't able to do so. When I write the nodal > equation for the collector of the BJT and solve for cu using the > small-signal values obtained from the simulation, the value will > always have an imaginary part which is obviously unacceptable. > > Is it some limitation of the way SwCAD models the BJT? I don't > think it is a limitation of the model because it is cited in > many books in electronics as a reliable model up to ~100 MHz.
Hello Sert, As Helmut said, you should add an Rbb', otherwise your BJT will have extremely high maximum oscillation frequency (or very high gain at RF). Another point is the diffusion capacitance of the forward biased BE junction (do not confuse this with the space charge "varicap" capacitance"). I don't know the DC collector current, but when it is around 10mA, total Cbe will be higher then 60pF. In the Gummel Poon model, the "ideal forward transit time (TF)" models the diffusion capacitance. It is roughly proportional with the collector current and inversely proportional with the transistor's ft (Total Cbe = (40*Ic)/(2*pi*ft)) You might also add Ree when you run the transistor at relative high current. Try to play with it in the transistor's model to see the influence on phase shift between base voltage and collector current. You might add some low pass action between the base and the input of the VCCS to model frequency depended behavior when you also want to put Rbb' and Ree' into your pi-model. Best regards, Wim PA3DJS www.tetech.nl without abc you can use the PM address.
On 18 nov, 19:47, sert <je...@hotmail.com> wrote:
> I'm using SwCAD and trying to determine the parameters of the > high-frequency pi-model for the BJT. For convenience, I've > combined the circuit with the BJT and the model in one circuit > so I can combine the graphs as well: > > http://i35.tinypic.com/vxf1ad.png > > First I used small dc signals to determine the values for rpi, > gm, ro and then I tried to use small ac signals to determine the > values for the capacitors (the values in the model are just > placeholders.) I wasn't able to do so. When I write the nodal > equation for the collector of the BJT and solve for cu using the > small-signal values obtained from the simulation, the value will > always have an imaginary part which is obviously unacceptable. > > Is it some limitation of the way SwCAD models the BJT? I don't > think it is a limitation of the model because it is cited in > many books in electronics as a reliable model up to ~100 MHz.
Hello Sert, As Helmut said, you should add an Rbb', otherwise your BJT will have extremely high maximum oscillation frequency (or very high gain at RF). Another point is the diffusion capacitance of the forward biased BE junction (do not confuse this with the space charge "varicap" capacitance"). I don't know the DC collector current, but when it is around 10mA, total Cbe will be higher then 60pF. In the Gummel Poon model, the "ideal forward transit time (TF)" models the diffusion capacitance. It is roughly proportional with the collector current and inversely proportional with the transistor's ft (Total Cbe = (40*Ic)/(2*pi*ft)) You might also add Ree when you run the transistor at relative high current. Try to play with it in the transistor's model to see the influence on phase shift between base voltage and collector current. You might add some low pass action between the base and the input of the VCCS to model frequency depended behavior when you also want to put Rbb' and Ree' into your pi-model. Best regards, Wim PA3DJS www.tetech.nl without abc you can use the PM address.
"sert" <jerry@hotmail.com> schrieb im Newsbeitrag 
news:Xns9B5AD36F45694jtjdfjdfjnbj@147.102.222.230...
> I'm using SwCAD and trying to determine the parameters of the > high-frequency pi-model for the BJT. For convenience, I've > combined the circuit with the BJT and the model in one circuit > so I can combine the graphs as well: > > http://i35.tinypic.com/vxf1ad.png > > First I used small dc signals to determine the values for rpi, > gm, ro and then I tried to use small ac signals to determine the > values for the capacitors (the values in the model are just > placeholders.) I wasn't able to do so. When I write the nodal > equation for the collector of the BJT and solve for cu using the > small-signal values obtained from the simulation, the value will > always have an imaginary part which is obviously unacceptable. > > Is it some limitation of the way SwCAD models the BJT? I don't > think it is a limitation of the model because it is cited in > many books in electronics as a reliable model up to ~100 MHz.
Hello, There are no limitations in LTspice, but in your circuit. You try to compare apples and oranges. 1. Your BCW60A SPICE-model model has a base resistance RbOhm and some other parameters you don't have in the pi-model. 2. To be honest, using any transistor with Vbe=2V is useless. Best regards, Helmut
I'm using SwCAD and trying to determine the parameters of the 
high-frequency pi-model for the BJT. For convenience, I've 
combined the circuit with the BJT and the model in one circuit 
so I can combine the graphs as well:

http://i35.tinypic.com/vxf1ad.png

First I used small dc signals to determine the values for rpi, 
gm, ro and then I tried to use small ac signals to determine the 
values for the capacitors (the values in the model are just 
placeholders.) I wasn't able to do so. When I write the nodal 
equation for the collector of the BJT and solve for cu using the 
small-signal values obtained from the simulation, the value will 
always have an imaginary part which is obviously unacceptable.

Is it some limitation of the way SwCAD models the BJT? I don't 
think it is a limitation of the model because it is cited in 
many books in electronics as a reliable model up to ~100 MHz.