Hi,

I use software which communicates with spectrum analyzer Agilent HP35665A.

Short description of communication between software and spectrum analyzer HP35665A:

The software sends commands SENS:FREQ:smileyfrustrated:PAN x (for example SENS:FREQ.SPAN 25 specifies the

frequency bandwidth to be measured, in this case 25 Hz) and SENS:FREQ:CENT x (for example

SENS:FREQ:CENT 50,25 specifies the center frequency for the current measurement, in this

case 50,25 Hz) to the analyzer. Analyzer does FFT analysis with 400 lines (the frequency

resolution of an FFT analyzer is stated in number of lines, default value is 400 lines) and

keeps center frequency. Then, returns start (lowest) frequency (in this example 37,75 Hz

because 50,25-(25/2)=37,75), frequency bandwidth (in this example 25 Hz) and results of the

FFT analysis (401 values because the number of lines is 400). Frequency resolution is equal

to the frequency span/number of FFT lines, in this example the frequency resolution is

0,0625 Hz because 25/400 = 0,0625.

Problem:

In some cases the start frequency is not multiple of frequency resolution. For example, the

software sends SPAN 400 Hz and CENT 2343,5 Hz. Analyzer returns start frequency 2143,5 Hz

and bandwidth 400 Hz. Frequency resolution = 400/400 = 1 Hz. 2143,5 Hz is NOT multiple of 1

Hz. According to the theory, the starting frequency should be multiple of frequency

resolution. Is it possible that the analyzer automatically set longer data record length

(in this example, the start frequency would be multiple of frequency resolution with

frequency resolution = 0,5 Hz, then the data record length of input signal should be 2

seconds)?

I use software which communicates with spectrum analyzer Agilent HP35665A.

Short description of communication between software and spectrum analyzer HP35665A:

The software sends commands SENS:FREQ:smileyfrustrated:PAN x (for example SENS:FREQ.SPAN 25 specifies the

frequency bandwidth to be measured, in this case 25 Hz) and SENS:FREQ:CENT x (for example

SENS:FREQ:CENT 50,25 specifies the center frequency for the current measurement, in this

case 50,25 Hz) to the analyzer. Analyzer does FFT analysis with 400 lines (the frequency

resolution of an FFT analyzer is stated in number of lines, default value is 400 lines) and

keeps center frequency. Then, returns start (lowest) frequency (in this example 37,75 Hz

because 50,25-(25/2)=37,75), frequency bandwidth (in this example 25 Hz) and results of the

FFT analysis (401 values because the number of lines is 400). Frequency resolution is equal

to the frequency span/number of FFT lines, in this example the frequency resolution is

0,0625 Hz because 25/400 = 0,0625.

Problem:

In some cases the start frequency is not multiple of frequency resolution. For example, the

software sends SPAN 400 Hz and CENT 2343,5 Hz. Analyzer returns start frequency 2143,5 Hz

and bandwidth 400 Hz. Frequency resolution = 400/400 = 1 Hz. 2143,5 Hz is NOT multiple of 1

Hz. According to the theory, the starting frequency should be multiple of frequency

resolution. Is it possible that the analyzer automatically set longer data record length

(in this example, the start frequency would be multiple of frequency resolution with

frequency resolution = 0,5 Hz, then the data record length of input signal should be 2

seconds)?

The start frequency and the stop frequency are strictly determined by the center frequency setting and the span setting. It is not dependent on the # of resolution.

In your exammple, the center frequency of 2343.5Hz and a span setting of 400 Hz, sets the start frequency at 2343.5 - 200 = 2143.5Hz and the stop frequency at 2343.5 + 200 = 2543.5 Hz. The # of lines of resolution is not even considered.

Regards -