Raw digital output from usb dongle

Main forum to discuss RTL-SDR related topics.
Post Reply
Hesam
Posts: 2
Joined: Wed Feb 14, 2018 7:02 pm

Raw digital output from usb dongle

Post by Hesam » Wed Feb 14, 2018 7:17 pm

Hi everyone,
I know a few about radio's and sdr's
I have an Rtl-sdr dongle (24MHz-1.8Ghz bandwidth =2MHz)
Does it mean it is possible to see the spectrum in 2MHz at the same time?
What if I wanna see the full range(24-1800)?
Is this possible to get the raw signal from ADC and manually apply FFT to see the full band spectrum?

How can I monitor the spectrum from 24 to 1800MHz simultaneously?

Thanks in Advance

Brightnoise
Posts: 20
Joined: Fri Jan 19, 2018 10:37 pm
Location: Netherlands, Nuenen
Contact:

Re: Raw digital output from usb dongle

Post by Brightnoise » Thu Feb 15, 2018 2:18 pm

The complete spectrum on the screen is also my wish. The AD converter bandwidth is 14.4 MHz (half sample frequency). The filters in the baseband chip however limit the maximum BW to slightly above 2 MHz wide (baseband I&Q signal in frequency range zero plus and minus 1 MHz).

Using the dongle as a spectrum analyzer requires the front-end chip to tune over the full frequency range. The IF BW setting limits the step size (do not jump over a signal; don't miss it). There is a "spectrum analyzer function" available, but I can't get it working. See Youtube and search for "SDR spectrum analyzer").

Scan speed calculation: for RBW = 100 kHz the frequency steps may not exceed 33 kHz. From 24 MHz to 1.024 GHz there are now 30k steps. The settling time in the 100 kHz filter is 20 microseconds (accuracy-related). The full sweep has a theoretical limit of minimum 600 ms, but the PLL-settling is probably dominant.
So roughly: one sweep will cost between 1 and 30 seconds......

I should like to have this application anyway.
Let me know when you find a working spectrum analyzer!

Brightnoise
Posts: 20
Joined: Fri Jan 19, 2018 10:37 pm
Location: Netherlands, Nuenen
Contact:

Re: Raw digital output from usb dongle

Post by Brightnoise » Thu Feb 15, 2018 3:11 pm

Working now! RTL-SDR dongle panorama needs these files (see figure below) in one and the same subdirectory; that is all.

Download from: https://sourceforge.net/projects/guiforrtlpower/
and follow the instructions. 64 and 32 bits versions are available.

Program start: RTLPAN.EXE
And as expected: scan of 1 GHz with 100 kHz RBW is around 30 seconds.
Math and assumptions seem correct.

The frequency tuning is using the 2 MHz wide span (visible as a repetitive high-low noise floor variation, since the frequency response of the low-pass filters in the RTL-chip are used at the edge) AND the PLL-tuning; of course. The frequency step size is 0,66 times the bandwidth; that is a compromise between speed and accuracy.

The software crashes when during a scan more than one button is pressed.

Improvement options: update display during the scan. Now you have to wait for a refresh after the full scan. Also then you can see where the actual frequency window is, and that the program is up and running. And add fixed max-min dB values for range. Make more steps with the PLL and use the 2 MHz window for a smaller part.
Anyway: fantastic software!

Direct sampling (zero to 14 MHz) does not give the expected results.

See figure below for the necessary files.


*****************************************************************************************************
Added on February 21: On a DELL E6500 with WIN10 and 64 bit operating system I can't get it working!
On two other computers (Lenovo T530 and HP 2560 it is running OK).
*****************************************************************************************************
Attachments
Zowerkthet2.jpg
Zowerkthet2.jpg (24.63 KiB) Viewed 6955 times
Last edited by Brightnoise on Wed Feb 21, 2018 8:55 pm, edited 1 time in total.

Hesam
Posts: 2
Joined: Wed Feb 14, 2018 7:02 pm

Re: Raw digital output from usb dongle

Post by Hesam » Thu Feb 15, 2018 5:58 pm

Hey dude,
Thank you so much
It seems to be perfect

hotpaw2
Posts: 126
Joined: Sat Jan 14, 2017 11:07 pm
Contact:

Re: Raw digital output from usb dongle

Post by hotpaw2 » Thu Feb 15, 2018 9:55 pm

How does one determine how much PLL settling time is needed after each frequency jump? I assume the IQ data is mostly garbage until after the PLL + internal filter settling time.

Brightnoise
Posts: 20
Joined: Fri Jan 19, 2018 10:37 pm
Location: Netherlands, Nuenen
Contact:

Re: Raw digital output from usb dongle

Post by Brightnoise » Fri Feb 16, 2018 5:32 pm

PLL settling time in good designs is ten times the period duration of the reference frequency. (Frequency error smaller than 1 % of initial frequency step, within one octave).

In the RTL front-end chip the VCO is running in the GHz range, with a max/min frequency ratio for that VCO of preferably slightly over two, and the wanted mixing oscillator signal is obtained at the output of a divider chain. This divider allows the generation of complex LO-signal AND the generation of LO-signal with low (odd) harmonic content by using waveform synthesis. The even harmonics are by nature suppressed (50 % duty cycle).

The loop filter for the PLL is on-chip; the reference frequency is probably 28.8 MHz or 14.4 MHz; settling in one microsecond is thus probable.

The calculation of the PLL-division number and the calculation of the LO-divider numbers and wave synthesis factors (---> these factors are probably already fixed-values in the hardware) takes some time.
(In early days there was a separate microprocessor, controlling a separate PLL).

Brightnoise
Posts: 20
Joined: Fri Jan 19, 2018 10:37 pm
Location: Netherlands, Nuenen
Contact:

Re: Raw digital output from usb dongle

Post by Brightnoise » Mon Feb 19, 2018 12:13 am

The question was:
How does one determine how much PLL settling time is needed after each frequency jump? I assume the IQ data is mostly garbage until after the PLL + internal filter settling time.

My answer was detailed because I am curious how fast a well-designed spectrum analyser with the RTL-Dongle can be.

To come back to the impulse response of the 100 kHz wide RBW-filter (that is 50 kHz low-pass in I and Q branch): see figure attached. Step at t = 100 microseconds. Filter is Bessel, 5th order.

At the cost of accuracy the speed can be increased to 15 microseconds, or to be defined.
Attachments
Response-example-50kHz-LPF.jpg
Response-example-50kHz-LPF.jpg (26.78 KiB) Viewed 6340 times
Last edited by Brightnoise on Mon Feb 19, 2018 1:50 pm, edited 2 times in total.

Brightnoise
Posts: 20
Joined: Fri Jan 19, 2018 10:37 pm
Location: Netherlands, Nuenen
Contact:

Re: Raw digital output from usb dongle

Post by Brightnoise » Mon Feb 19, 2018 12:29 am

And then the speed of the PLL. The PLL-filter is not integrated in this chip. The filter in the standard application is pole at 1800 Ohm and 330 pF, zero at 1800 Ohm and 6800 pF. The frequencies belonging to these values are given in the plot below. The unity-gain for the PLL is 100 kHz and the settling (frequency step change) is faster than 100 microseconds (I will probably come back on this with better figures). With this filter not the mentioned 1 microsecond settling......
Attachments
PLL-bodePlot-standard-application.jpg
PLL-bodePlot-standard-application.jpg (50.53 KiB) Viewed 6337 times
Last edited by Brightnoise on Mon Feb 19, 2018 1:49 pm, edited 1 time in total.

Brightnoise
Posts: 20
Joined: Fri Jan 19, 2018 10:37 pm
Location: Netherlands, Nuenen
Contact:

Re: Raw digital output from usb dongle

Post by Brightnoise » Mon Feb 19, 2018 12:49 am

The PLL settling with the proposed component values from the Rafael application:
see figure attached.

This is the response to a frequency step (t_step at 10 microseconds). When the initial step was 1 MHz then the final error can be obtained as ratio from the default factor "one".

For a frequency sweep in a spectrum analyser the steps are small. So the error is acceptable.

I should like to program the analyzer software, but I have no clue about EXE and DLL files. Can somebody help me, or can I help somebody that can program such things?
Attachments
PLL-settling-standardcomponents.jpg
PLL-settling-standardcomponents.jpg (20.1 KiB) Viewed 6328 times

Post Reply