Measuring Frequency Response with FFT

[danilocunha]

Hello, I´m not sure this is the right place for this post. But I believe it is.

I need to measure the attenuation of a filter (low pass) along the frequency (0 to 100KHz). I´m trying to do that with a FFT Analyzer from Stanford model SR770 to measure the frequency response o my circuit.

This analyzer has an source output and two inputs. I connected the source output to the input of my circuit and the output of my circuit to one of the inputs of the analyzer. And then I configured the source output to be a Chirp source. As I understood, the chirp source is equivalent to several senoids (or should I say sine?), one for each frequency bin. I setted the amplitude of the chirp output to be 1000mV.

So, if I´m not wrong, the measurement I´m doing in the output of my circuit is the gain of my filter. Right? And since what I want is the attenuation, I should invert this values (Att = 1/Gain). Right?

Another concern is related to impedance matching. The SR770 has an output impedance described in the specs as < 5 ohms. Should I include a 5 ohms resistor after my circuit? So it would have the same impedance in both sides.

Thanks. Any help and suggestions are welcome.

 

[danilocunha]

I´m defenetly doing something wrong. The attenuation I´m measuring is too high. To make sure it isn´t my circuit, I bypassed it and the attenuation is more than 40dB

Someone? Any hints?

 

[ark5230]

You can test this without any filter. If the response is flat, you can test it with the filter.
Inclusion of additional 5 ohms may contribute to modification of characteristics of your filter.
Handle and attend to attenuation independently.

 

[danilocunha]

You can test this without any filter. If the response is flat, you can test it with the filter.
Inclusion of additional 5 ohms may contribute to modification of characteristics of your filter.
Handle and attend to attenuation independently.

Hi, I tested without any filter. The response is flat but not in 0dB. It´s a line and it´s around -40dB.

I will test with the 5 ohms resistor.

Thanks.

 

[ark5230]

This could be normalization issue related to the settings.

 

[danilocunha]

5 ohms resistor didn´t help

What do you mean by normalization? Something like an offset? Like if the equipment needed to be calibrated?

 

[ark5230]

Most of the cases the DB is with respect to some standard, bringing the reference and test at par (to identical scale) is what is normalization conceptually. In practice there could be may situations and case dependent explanations.