- Hi Sven,

That's exactly as the maths dictate: 65536 point FFT, at 11025 samples per second:

-> each FFT (- window) is approxiately 5.9 seconds long.

In other words, an enormous overlap between the FFT windows.

This is the price to be paid for a good frequency resolution.

Besides using a shorter FFT (to improve the time resolution), you could try the *REASSIGNED* spectrogram display.

It uses the phase between consecutive FFTs to improve the resolutions.

Details:

http://www.qsl.net/dl4yhf/speclab/ra_spectrogram.htm

73,

Wolf DL4YHF

Am 16.04.2013 17:33, schrieb sm5le:Hi all

How to explain this:

A 10mS pulse is feeded into EM-U 0202 and then in to SpecLab.

On the time domain scope the pulse is OK BUT on watherfall

it is broadened up to 5 second. Soundcard 11025 and sample 16 Bit and FFT 65536 scroll 150 mS. Picture in SM5LE folder/files.

Tnx Sven

- With FFT you can either have accurate time resolution, or accurate frequency resolution. But not both at the same time.

The mathematical formula does not allow this.

Regards,

Mark McCarron

To: SpectrumLabUsers@yahoogroups.com

From: dl4yhf@...

Date: Tue, 16 Apr 2013 18:27:27 +0200

Subject: Re: [SpectrumLabUsers] Pulse broadened

Hi Sven,

That's exactly as the maths dictate: 65536 point FFT, at 11025 samples per second:

-> each FFT (- window) is approxiately 5.9 seconds long.

In other words, an enormous overlap between the FFT windows.

This is the price to be paid for a good frequency resolution.

Besides using a shorter FFT (to improve the time resolution), you could try the *REASSIGNED* spectrogram display.

It uses the phase between consecutive FFTs to improve the resolutions.

Details:

http://www.qsl.net/dl4yhf/speclab/ra_spectrogram.htm

73,

Wolf DL4YHF

Am 16.04.2013 17:33, schrieb sm5le:Hi all

How to explain this:

A 10mS pulse is feeded into EM-U 0202 and then in to SpecLab.

On the time domain scope the pulse is OK BUT on watherfall

it is broadened up to 5 second. Soundcard 11025 and sample 16 Bit and FFT 65536 scroll 150 mS. Picture in SM5LE folder/files.

Tnx Sven

- Thank you Mark and Wolf.

So it dont matter how thin or broad the puls is; the FFT will show the same. So maybe I will use one A/D - SpecLab for frequncy (high FFT) and then one A/D - Speclab for the waterfall (lov FFT).

It is Schumann 3-30 Hz my interest. I have alot of pulse noise at my sight.

Sven

--- In SpectrumLabUsers@yahoogroups.com, Mark McCarron <mark.mccarron@...> wrote:

>

> With FFT you can either have accurate time resolution, or accurate frequency resolution. But not both at the same time.

>

> The mathematical formula does not allow this.

>

> Regards,

>

> Mark McCarron

>

> To: SpectrumLabUsers@yahoogroups.com

> From: dl4yhf@...

> Date: Tue, 16 Apr 2013 18:27:27 +0200

> Subject: Re: [SpectrumLabUsers] Pulse broadened

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

>

> Hi Sven,

>

>

>

> That's exactly as the maths dictate: 65536 point FFT, at 11025

> samples per second:

>

> -> each FFT (- window) is approxiately 5.9 seconds long.

>

> In other words, an enormous overlap between the FFT windows.

>

>

>

> This is the price to be paid for a good frequency resolution.

>

> Besides using a shorter FFT (to improve the time resolution), you

> could try the *REASSIGNED* spectrogram display.

>

> It uses the phase between consecutive FFTs to improve the

> resolutions.

>

> Details:

>

>

>

> http://www.qsl.net/dl4yhf/speclab/ra_spectrogram.htm

>

>

>

>

>

> 73,

>

> Wolf DL4YHF

>

>

>

>

>

>

>

> Am 16.04.2013 17:33, schrieb sm5le:

>

>

>

>

>

>

> Hi all

>

>

>

> How to explain this:

>

> A 10mS pulse is feeded into EM-U 0202 and then in to

> SpecLab.

>

> On the time domain scope the pulse is OK BUT on watherfall

>

> it is broadened up to 5 second. Soundcard 11025 and sample

> 16 Bit and FFT 65536 scroll 150 mS. Picture in SM5LE

> folder/files.

>

>

>

> Tnx Sven

>