[LTSPICE] Why input-capacitor degrades noise floor?

[railcannon]

Hi I'm newbie interested in audio electronics

I practice a simple basic non inverting opamp amplifer with input capacitor for prevent input dc offset with LT spice.

I designed 3db frequency of the high pass filter is about 10hz
(Its just not exact value but I think it is ok to block input dc offset)

This amp works very well but FFT seems very high noise floor at low frequency.

with input capacitor of 2.2uF/ 10kohm
the noise floor is 85db at 20Hz, 140db at 10kHz

without input capacitor
under 140db at audible frequency with plotwinsize 0, numdgt 7 options.

I know the capacitor makes thermal noise like resistor, I guess the impedance of capacitor is getting bigger at lower frequency so the input capacitor induced noise probably be bigger at low frequency.


However I think There are So May GOOD amps with 2.2uf input capactor that does NOT have these kind of noise floor degradation at low frequency problems.

I Want to solve these kind of problems. Please help me.

P.S I don't want to use DC servo, output capacitors.

 

[abraxalito]

I suspect what you're seeing isn't noise as such rather an artifact of the FFT windowing. There are a few 'gotchas' when using FFT in LTSpice.

 

[Mooly]

I suspect its just the way have the simulation set up. Time constants (caps) either need removing or making very large.

Post a shot of your results and or post the .asc file (put it in a zipped folder to attach to the forum)

 

[brig001]

Could the noise be in the 10k resistor and the opamp? At low frequencies, the capacitor's impedance is higher shorting less of the noise to the output of the signal generator and therefore to ground.

Brian

 

[railcannon]

Here my files

I think it may be characteristics of FFT
I don't know FFT well

 

[abraxalito]

You could do worse than have a mosey through this thread :http://www.diyaudio.com/forums/tubes-valves/105496-how-make-sense-ltspice-fft-analysis.html

 

[jcx]

most Spice device models don't give device noise in .TRAN - which is what you need for FFT

unless you have added Behavioral Source noise generators to your models

so you are almost certainly looking at settling time and windowing

try running the sim for 5-10x longer, just fft the last 4-5 periods (with Blackman window) of the test signals

 

[Mooly]

This is your circuit with 2.2uf cap run for 1 second and looking at the last 10ms.

 

[Mooly]

And noise with and without cap.

 

[railcannon]

I'm Very X4 APPRICIATE all of you. Thanks.

It probably a result of settling time I think.

Anyway I need to study much more. BYE!!!