FM modulation Block with Sawtooth Signal Source, I/Q Vs Time shows issues with Signal Generated.

[smileyrekiere]

What would you like to know?

O.k. have a simple vector generated off a sawtooth wave and using analog.frequency_modulator_fc, where radians/sample = amplitude sensitivity.
Question: How to clean up IQ values for higher amplitude sensitivity (radians/sample) for analog.frequency_modulator_fc?

At a high level modulating a signal (5 - 20 Mhz) on a carrier frequency in the 5 Ghz range with a pulse width from 50 - 100 micro seconds the IQ values and modulated signal is not correct.

At lower frequencies (-2.5Mhz - 2.5 Mhz to -7.5 Mhz - 7.5 Mhz) the signal looks as expected and the I/Q Vs Time looks close to what is expected. Between -7.5 Mhz - 7.5 Mhz and -10 Mhz - 10 Mhz , the IQ values look incorrect and there is an anomaly on the edges of the modulated signal.

Lower modulated frequency on carrier frequency

Higher modulated frequency on carrier frequency (I/Q values issue)

[smileyrekiere]

Could the hackrf hardware version affect the signal modulated signal?

Host at 2023 and hackrf_info at the following hackrf_info version: git-13765055 libhackrf version: git-13765055 (0.8) Found HackRF Index: 0 Serial number: 000000000000000057b068dc22276763 Board ID Number: 2 (HackRF One) Firmware Version: 2023.01.1 (API:1.07) Part ID Number: 0xa000cb3c 0x00644754 Hardware Revision: older than r6
Hardware supported by installed firmware: HackRF One

From the revisions see there is changes with the clock distribution HackRF One r9¶ MAX2837 was replaced by MAX2839. Si5351C was replaced by Si5351A with additional clock distribution. A series diode was added to the antenna port power supply. Manufacturing year: 2023

https://hackrf.readthedocs.io/en/latest/list_of_hardware_revisions.html

[smileyrekiere]

What is the correct settings for the osmosdr_sink_0.set_bandwidth(X,0) some of the readings say to default to X to zero. When doing so see the unusual I/Q values. When I set the set_bandwidth( tx_sample_rate / 2, 0) see the signal is cleaned up a fair amount. Other discussions mention .8 of the tx_sample_rate (looking into other osmosdr_sink settings) two trials below

[martinling]

What sample rate are you using?

[smileyrekiere]

20 Mhz (wanting a chirp from -10 Mhz to 10 Mhz max to low of -2.5 Mhz to 2.5 Mhz)

That is the tx sample rate.

20 samples per microsecond. Pulse width is 100 us , with 2000 samples

[smileyrekiere]

Closing out. The I/Q issue is mitigated by having a precise equation gnu radio uses radians/sample = amplitude sensitivity. Manually testing various values based on frequency. The empirical values were then graphed and the resulting equation : frequency_modulation (pi /tx_sample_rate) this valure was then used as an input to the gnu radio analog.frequency_modulator_fc . Empirical data for cleanest results at each frequency 5 Mhz - 20 Mhz

[smileyrekiere]

Closing as issue resolved