I have simulated simple RC circuit to calculate 
NOISE CURRENT in HSPICE. In one of 2 cases, RESULTS are 
not matching with my hand calculated values. Circuit is simulated with 2 cases (figure 1 and 3) which are shown in the pdf which is attached to this post. Hand calculations along with spice code are shown in the same PDF.
Case 1 (Figure 1 in the PDF): When 
VOLTAGE source is connected at the 
INPUT: 
In this case answers are very well matching. First with the statement .NOISE      
V(3)      V1      5 I could calculate 
NOISE VOLTAGE DESNISTY which resulted in 
4.57pV/rt(Hz) which is well matched with my calculated values. 
For calculating 
Noise CURRENT DESNSITY, I replaced earlier statement by the following one i.e. is .NOISE       
I(C1)      V1      5  where only V(3) is replaced by I(C1). With this even I am getting the expected/calculated CURRENT DESNSITY (through CL) that is 
574f A/rt(Hz) .
Case 2 (Figure 3 in the PDF): When 
CURRENT source is connected at the 
INPUT:
In this case, I am using the statement .NOISE      
V(3)      I1      5 and .NOISE      
I(C1)      I1      5 for calculating 
NOISE VOLTAGE DENSITY and 
Noise CURRENT DESNSITY respectively. Here both statements yielding a different results i.e. are 
3.22pV/rt(Hz) and 
405.571fA/rt(Hz) compared to the expected/calculated results of 
4.57pV/rt(Hz) and 
574fA/rt(Hz) respectively for VOLTAGE and CURRENT density. 
Question is 
WHY NOISE HAS TO DEPEND ON THE TYPE OF SOURCES CONNECTED AT THE INPUT ? Actually it must not depends on the source because both cases consists of same resistors.  Or Am I doing anything wrong in this noise analysis?
Note : This circuit is taken from the website:       
http://www.ecircuitcenter.com/Circuits/Noise/Noise_Analysis/res_noise.htm