Understanding FFT

Forum for discussing PicoScope version 6 (non-automotive version)
Post Reply
Ax2_
Newbie
Posts: 0
Joined: Mon Sep 01, 2014 6:23 am

Understanding FFT

Post by Ax2_ »

Hello. Could you help me to understand FFT and Spectrum analyzer. See below some issues I can't understand.
Environment: [See attached image Environment]

Setup: Generator set to white noise which is about 800mV RMS. So far so good. [Image Setup]
A jumper from signal generator output to Channel 2 input. FFT Display mode: Average.

Changed to Spectrum mode and wave looks ok [Image Wave1]... however: I have set dBu to Y axis and 800mV RMS should be around 0dBu. Here it is somewhere -43dBu... why?

Lets change Spectrum bins few times between 512, 1024 and 2048. After 3-5 change the result looks like Wave2. This must be a bug. It is repeatable using low bin values.

Changed Y-axis to linear. Level of white noise is below 5mV. Hmm... I think I should see 800mV Linear level using 800mV RMS signal (of white noise in this case).

Ok. Back to logarithm Y axis. Lets change bin values 2048, 4096, 8192. dBu levels also changes (-57dBu, -60dBu, -63dBu). How come Bin width changes the resulting dB level? (wave3 = 8192 bins)

Axel.
Attachments
wave3.PNG
wave2.PNG
Wave1.PNG
Setup.PNG
Environment.PNG
Environment.PNG (3.02 KiB) Viewed 6735 times

Ax2_
Newbie
Posts: 0
Joined: Mon Sep 01, 2014 6:23 am

Re: Understanding FFT

Post by Ax2_ »

Spectrum range also affects to signal level. All kinds of weird issues when using White noise. Cannot get much out of White noise -option... but...

...I switched to Sine wave+Sweep mode and everything started to work properly. Levels are shown correctly and waveforms do not get crazy etc etc.

Axel.

Martyn
Site Admin
Site Admin
Posts: 4491
Joined: Fri Jun 10, 2011 8:15 am
Location: St. Neots

Re: Understanding FFT

Post by Martyn »

I am out of the office and on a mobile device so can't see the pictures clearly but I think what you are seeing is:-

With a sine wave the time domain amplitude will map to a frequency domain level at a specific frequency.

White noise is constant power across the spectrum so the time domain amplitude is spread across all the frequency components hence the lower levels.
Martyn
Technical Support Manager

Ax2_
Newbie
Posts: 0
Joined: Mon Sep 01, 2014 6:23 am

Re: Understanding FFT

Post by Ax2_ »

Hello,

Thanks for the reply. This sounds like an explanation. However I need to know an exact amplitude so white noise is not an option in this case then. It is ok since preferable methods is available.

Axel.

Martyn
Site Admin
Site Admin
Posts: 4491
Joined: Fri Jun 10, 2011 8:15 am
Location: St. Neots

Re: Understanding FFT

Post by Martyn »

If you can explain exactly what you are trying to achieve then hopefully we can help.
Martyn
Technical Support Manager

Ax2_
Newbie
Posts: 0
Joined: Mon Sep 01, 2014 6:23 am

Re: Understanding FFT

Post by Ax2_ »

Hello,

Basically it was all in the initial message, but in general I'm doing an output analysis of a filter. I thought that I can do it easily with White noise.
One of the critical parameter is an exact input level and that requirement is met when sweep-method is being used (rather than White noise-method). At least it worked for me for now.

Axel.

Martyn
Site Admin
Site Admin
Posts: 4491
Joined: Fri Jun 10, 2011 8:15 am
Location: St. Neots

Re: Understanding FFT

Post by Martyn »

Using a swept sine is the correct approach for this.
Martyn
Technical Support Manager

Post Reply