Problem with 1/f bipolar noise in gummel-poon model

[btoner]

1/f Bipolar noise

Hia,
I have a problem when I simulate the low frequency noise for a gummel-poon model. I find that the model correctly calculates the base noise from the model equations. However when I look at the ratio of the input to output noise spectrum, it is not beta squared. Indeed the collector noise is actually lower than the base noise, at low biases anwyway. The Beta is still near 100. Any ideas?

Brendan

 

[flatulent]

Re: 1/f Bipolar noise

Does the gain for an AC signal decline at the frequencies where the output noise declines?

 

[rfsystem]

Re: 1/f Bipolar noise

There are:

1. Thermal noise from RB and RE
2. Shot-Noise from IC and IB
3. 1/f-Shot-Noise prop. IB and KF

But it depend on your circuit embedding what is the actual current of the base current to the collector current. Remember a emitter follower. Base current source have little impact on collector current.

 

[dsjomo]

1/f Bipolar noise

What's your simulation tool and how do you terminate the transistor loading port? Would you mind uploading your schematic?

 

[btoner]

Re: 1/f Bipolar noise

hi guys,

thanks for the replies.
I have included my ADS circuit and simulation result. I am running an AC simulation to find the noise voltage at the base and collector. I then divide these voltages by the base and collector resistances to find the noise current. I then square this to get the noise current spectral density.

thanks

Brendan

 

[dsjomo]

1/f Bipolar noise

The device is nearly off, are you trying to proof the model in that operating region??

The 100ohm load might dominate the output noise under this situation. I guess this is why the blue curve is flat over 1KHz.....

 

[btoner]

Re: 1/f Bipolar noise

Yes, it is low bias but I would expect the relationship to hold. The base and collector resistances have their noise switched off so they will not affect the result.
I notice that many people do their comparisons at the base because that is where the model equations apply and not at the collector. I have not seen this comparison before. The input noise is modelled correctly according to the model equations but the model does not relate it to the collector correctly.

thanks

Brendan

 

[btoner]

Re: 1/f Bipolar noise

Hi Flatulent,

I checked the AC gain of the circuit and it seems to be OK. However the noise current is still not being multiplied by this gain. Mmm, strange?

B

 

[lastdance]

Re: 1/f Bipolar noise

at what frequency did you take the readings?
how about the capacitances and inductances?
If you make inductance suffciently large, you may ignore it. But at high enough frequency, the impedance of the capacitors can affect your result.