Lab 1 — Tone → RF → IQ → Analysis¶

This showcase is the first complete engineering loop of the course. A simple tone is intentionally used because it makes every mistake visible: wrong sample rate, gain overload, frequency offset, clipping, image components and incorrect IQ interpretation.

Goal¶

Build a minimal but real SDR experiment:

reference tone → FPGA/RF transmitter → independent receiver → IQ file → offline analysis → engineering conclusion

Measurement scenario¶

flowchart LR
    REF["Reference model<br/>MATLAB / Simulink"]
    PARAM["Experiment setup<br/>Fs / Fc / gain / tone"]
    TX["Zynq + AD9363<br/>tone generation / TX path"]
    LINK["RF link<br/>coax + attenuator or antenna"]
    RTL["RTL-SDR<br/>independent receiver"]
    HDSDR["HDSDR<br/>spectrum / waterfall / recording"]
    IQ["IQ file<br/>WAV / RAW / CI16"]
    ANA["Offline analysis<br/>FFT / SNR / spur check"]
    REPORT["Lab report<br/>plots / settings / conclusion"]

    REF --> PARAM --> TX --> LINK --> RTL --> HDSDR --> IQ --> ANA --> REPORT
    ANA -. retune gain .-> PARAM
    ANA -. correct model .-> REF

Hardware setup¶

Block	Role	Notes
Zynq-7020 + AD9363	RF source	Generates or plays the tone through the SDR TX path
Coax / attenuator / antenna	RF channel	Coax + attenuation is preferred for repeatable early experiments
RTL-SDR	Independent receiver	Confirms that the signal is really present outside the board
HDSDR	Visual inspection and recording	Spectrum, waterfall, gain adjustment, IQ recording
MATLAB / Python / GNU Radio	Offline verification	Reproducible FFT and quantitative checks

Do not start with maximum gain

The first lab should teach measurement discipline. Start with conservative TX and RX gain, then increase only after checking overload and clipping.

Signal definition¶

Recommended initial parameters:

Parameter	Example	Why
Tone offset	50 kHz	Clearly visible away from DC
Receiver sample rate	2.4 MS/s	Common stable RTL-SDR setting
Capture duration	2–10 s	Enough for FFT averaging and repeatability
RF path	coax + attenuator	Safer and more repeatable than over-the-air
File format	WAV or raw IQ	Easy to replay and analyze

The exact values are less important than documenting them. Every plot must be traceable to the settings that produced it.

Expected observations¶

flowchart TB
    GOOD["Clean result"]
    PEAK["Single dominant FFT peak"]
    FLOOR["Stable noise floor"]
    NOCLIP["No clipping / overload"]
    OFFSET["Frequency offset documented"]

    BAD["Bad result"]
    CLIP["Flat-topped time waveform"]
    SPUR["Unexpected spurs"]
    DC["Large DC component"]
    DRIFT["Frequency drift / unstable LO"]

    GOOD --> PEAK --> FLOOR --> NOCLIP --> OFFSET
    BAD --> CLIP
    BAD --> SPUR
    BAD --> DC
    BAD --> DRIFT

Analysis checklist¶

Check	Pass condition	Engineering action if failed
Tone frequency	Peak appears at expected offset	Verify sample rate, LO, mixer sign and FFT axis
Noise floor	Stable and below the tone	Adjust gain and bandwidth
Clipping	No saturated IQ samples	Reduce TX/RX gain or add attenuation
Spurs	No unexpected strong images	Check DDS, mixer, filters and RF gain settings
Repeatability	Similar result after rerun	Save configuration and capture metadata

Demo figure¶

Lab 1 FFT

This generated FFT is the documentation target style: readable labels, visible peak, clean grid and enough context for a lab report.

Minimum lab report¶

A completed Lab 1 report should include:

Hardware photo or connection diagram.
TX configuration: carrier, sample rate, tone offset, gain.
RX configuration: center frequency, sample rate, gain, bandwidth.
IQ recording format and duration.
FFT plot with frequency axis and units.
Engineering conclusion: correct / overloaded / wrong frequency / needs retuning.

Why this lab matters¶

A tone looks simple, but it validates the entire experimental discipline:

configuration → physical signal → independent capture → reproducible analysis

After this loop works, the same method scales to AM/FM, BPSK/QPSK, synchronization, EVM and BER experiments.