Hi folks,
I did a noise analysis with SPICE3 the other day and got confusing results. Actually I wanted to twiddle around with the JFET pair we've been discussing and let SPICE do the math for me. SPICE and I have never been a good team because of mutual misunderstanding, of which this matter is just another instance.
My circuit has a gain of about 9. After running the sim, plot sqrt(onoise_spectrum) gives me a noise voltage density of about
15nV/rtHz, but the inoise is 1.7 uV/rtHz!I re-checked with LT's version of SPICE, and sure enough, I get
1.65nV/rtHz which is about 1/9 of the output noise as I expected.The SPICE netlist is below. E1 (Gain 1) was just added to reference the circuit's output to GND. To exclude any circuit discrepancy between LT's SPICE and the SPICE3 on my linux box I first ran the sim on LT, then exported the netlist and sicked the Linux version on it.
-----
- Z:\home\bl\SWCAD\Noise1.asc J1 N002 N004 N005 2N5912 J2 N003 0 N005 2N5912 R1 N005 N006 2000 R2 N001 N002 1000 R3 N001 N003 1000 V1 0 N006 12 V2 N001 0 9 V3 N004 0 0 AC 1e-3 0 E1 1 0 N002 N003 1 .noise V(1) v3 dec 10 1 1e4 .model 2n5912 njf (vto=-2 beta=6e-3 lambda=70e-6 cgs=3e-12 cgd=1e-12) .end