Not affiliated with or endorsed by the FCC, NCVEC, ARRL or any VEC. Questions are the NCVEC pools word for word (public domain); explanations are our own. Exam rules and fees can change: check with your VE team.
E7F: Software defined radio fundamentals: digital signal processing (DSP) filtering, modulation, and demodulation; analog-digital conversion; digital filters
Group E7F of the Amateur Extra pool (2024–2028) has 14 active questions. One of them is on every Amateur Extra exam. Each is shown word for word; open it to see the correct answer and our explanation.
E7F01
What is meant by “direct sampling” in software defined radios?
A Software is converted from source code to object code during operation of the receiver
B I and Q signals are generated by digital processing without the use of RF amplification
C Incoming RF is digitized by an analog-to-digital converter without being mixed with a local oscillator signal
D A switching mixer is used to generate I and Q signals directly from the RF input
Show the answer and explanation
Correct answer: C. Incoming RF is digitized by an analog-to-digital converter without being mixed with a local oscillator signal
Why C is right. C. The converter samples fast enough to capture the RF itself, so no analog mixer or local oscillator is needed to bring the signal down first.
Many SDRs first mix RF down to a low frequency and then digitize it. A direct-sampling radio digitizes the RF itself at a high rate, often 100 million samples per second or more. Tuning, filtering and demodulation all happen in software or programmable logic.
Why not D? A switching mixer still mixes with an oscillator, which is what direct sampling avoids.
E7F02
What kind of digital signal processing audio filter is used to remove unwanted noise from a received SSB signal?
A An adaptive filter
B A crystal-lattice filter
C A Hilbert-transform filter
D A phase-inverting filter
Show the answer and explanation
Correct answer: A. An adaptive filter
Why A is right. A. An adaptive filter keeps adjusting its own settings to follow and remove noise that changes over time.
Noise and interfering tones shift from moment to moment. An adaptive filter compares its prediction of the signal with what actually arrives and updates its settings many times a second. Parts of the signal that don't behave like speech get reduced, while the voice comes through.
Why not B? A crystal-lattice filter is an analog quartz filter, not a DSP filter.
E7F03
What type of digital signal processing filter is used to generate an SSB signal?
A An adaptive filter
B A notch filter
C A Hilbert-transform filter
D An elliptical filter
Show the answer and explanation
Correct answer: C. A Hilbert-transform filter
Why C is right. C. A Hilbert-transform filter shifts every audio frequency by 90 degrees, creating the second signal needed to cancel one sideband.
The phasing method makes SSB from two paths: the audio and a copy shifted 90 degrees. Each path is mixed with a carrier, and the two carriers are also 90 degrees apart. Adding the results cancels one sideband and keeps the other. A Hilbert transform does that wideband 90 degree shift in software.
Why not A? An adaptive filter is for reducing noise, not for producing the 90 degree shift.
E7F04
Which method generates an SSB signal using digital signal processing?
A Mixing products are converted to voltages and subtracted by adder circuits
B A frequency synthesizer removes unwanted sidebands
C Varying quartz crystal characteristics are emulated in digital form
D Signals are combined in quadrature phase relationship
Show the answer and explanation
Correct answer: D. Signals are combined in quadrature phase relationship
Why D is right. D. DSP builds SSB from two signals 90 degrees apart, called I and Q, so that one sideband adds and the other cancels.
I stands for in-phase and Q for quadrature, meaning shifted by a quarter cycle. The software forms the audio and its 90 degree copy, mixes each with matching sine and cosine carriers, and adds them. The math leaves a single sideband.
Why not B? A synthesizer makes frequencies; it cannot remove a sideband.
E7F05
How frequently must an analog signal be sampled to be accurately reproduced?
A At least half the rate of the highest frequency component of the signal
B At least twice the rate of the highest frequency component of the signal
C At the same rate as the highest frequency component of the signal
D At four times the rate of the highest frequency component of the signal
Show the answer and explanation
Correct answer: B. At least twice the rate of the highest frequency component of the signal
Why B is right. B. You need at least two samples per cycle of the highest frequency present, or that frequency cannot be told apart from a lower one.
This is called the Nyquist rate. Sampling 3 kHz audio needs at least 6,000 samples per second, and a bit more in practice to leave room for filtering. Sampling too slowly causes aliasing, where high tones show up as false lower tones.
Why not C? One sample per cycle lands at the same point of each wave, so the tone looks like DC.
E7F06
What is the minimum number of bits required to sample a signal with a range of 1 volt at a resolution of 1 millivolt?
A 4 bits
B 6 bits
C 8 bits
D 10 bits
Show the answer and explanation
Correct answer: D. 10 bits
Why D is right. D. You need 1,000 separate levels, and 10 bits give 1,024, the smallest power of 2 that covers them.
Each added bit doubles the number of levels a converter can produce. Nine bits give 512 levels, so each step would be about 2 mV, too coarse. Ten bits give 1,024 steps of slightly under 1 mV each.
2^n >= range / resolution
levels = 1 V / 0.001 V = 1,000
2^9 = 512 (too few); 2^10 = 1,024 (enough)
n = 10
= 10 bits
Why not C? 8 bits give only 256 levels, about 4 mV per step.
E7F07
What function is performed by a Fast Fourier Transform?
A Converting analog signals to digital form
B Converting digital signals to analog form
C Converting signals from the time domain to the frequency domain
D Converting signals from the frequency domain to the time domain
Show the answer and explanation
Correct answer: C. Converting signals from the time domain to the frequency domain
Why C is right. C. An FFT takes a block of samples recorded over time and calculates how much of each frequency they contain.
Any signal can be built from sine waves of different frequencies. The Fast Fourier Transform is an efficient way to compute those parts from sampled data. SDRs use it to draw spectrum and waterfall displays. Going the other way, from frequency back to time, is the inverse FFT.
Why not D? Frequency domain to time domain is the inverse FFT.
E7F08
What is the function of decimation?
A Converting data to binary-coded decimal form
B Reducing the effective sample rate by removing samples
C Attenuating the signal
D Removing unnecessary significant digits
Show the answer and explanation
Correct answer: B. Reducing the effective sample rate by removing samples
Why B is right. B. Decimation keeps only every Nth sample, lowering the sample rate so later processing has less data to handle.
Once a narrow signal is tuned in, millions of samples per second are no longer needed. Dropping 9 of every 10 samples cuts the rate by 10. A low-pass filter must run first so the remaining samples still describe the signal correctly.
Why not A? Binary-coded decimal is a number format and has nothing to do with sample rate.
E7F09
Why is an anti-aliasing filter required in a decimator?
A It removes high-frequency signal components that would otherwise be reproduced as lower frequency components
B It peaks the response of the decimator, improving bandwidth
C It removes low-frequency signal components to eliminate the need for DC restoration
D It notches out the sampling frequency to avoid sampling errors
Show the answer and explanation
Correct answer: A. It removes high-frequency signal components that would otherwise be reproduced as lower frequency components
Why A is right. A. A lower sample rate lowers the highest frequency that can be represented, so anything above it must be removed or it folds down as false low tones.
This folding is called aliasing. If the new rate is 10,000 samples per second, everything above 5 kHz must go first. The anti-aliasing filter is a low-pass filter applied before samples are thrown away.
Why not D? The filter removes content above half the new rate, not the sampling frequency itself.
E7F10
What aspect of receiver analog-to-digital conversion determines the maximum receive bandwidth of a direct-sampling software defined radio (SDR)?
A Sample rate
B Sample width in bits
C Integral non-linearity
D Differential non-linearity
Show the answer and explanation
Correct answer: A. Sample rate
Why A is right. A. A direct-sampling SDR can capture a band at most half its sample rate wide, so the sample rate sets the maximum bandwidth.
A converter sampling 120 million times a second can represent signals up to 60 MHz. Anything wider would alias. Sample width, the number of bits, affects dynamic range and noise floor, not bandwidth.
Why not B? More bits improve dynamic range, not bandwidth.
E7F11
What sets the minimum detectable signal level for a direct-sampling software defined receiver in the absence of atmospheric or thermal noise?
A Sample clock phase noise
B Reference voltage level and sample width in bits
C Data storage transfer rate
D Missing codes and jitter
Show the answer and explanation
Correct answer: B. Reference voltage level and sample width in bits
Why B is right. B. The smallest step a converter can resolve is its reference voltage divided by 2 to the number of bits, so those two set the weakest detectable signal.
The reference voltage sets the converter's full-scale range, and the bits divide that range into steps. A signal smaller than one step barely changes the output. A lower reference or more bits makes each step smaller, so weaker signals register.
Why not A? Clock phase noise adds noise to strong signals but does not set the size of the smallest step.
E7F12
Which of the following is generally true of Finite Impulse Response (FIR) filters?
A FIR filters can delay all frequency components of the signal by the same amount
B FIR filters are easier to implement for a given set of passband rolloff requirements
C FIR filters can respond faster to impulses
D All these choices are correct
Show the answer and explanation
Correct answer: A. FIR filters can delay all frequency components of the signal by the same amount
Why A is right. A. An FIR filter with symmetric coefficients delays every frequency by the same time, so waveform shapes are kept.
Uneven delay across frequencies smears pulses and distorts data signals. FIR, finite impulse response, filters build each output from a fixed set of past inputs with no feedback, which makes equal delay easy to achieve.
Why not B? Filters that use feedback, called IIR, usually meet a roll-off target with fewer calculations, so B is not generally true.
E7F13
What is the function of taps in a digital signal processing filter?
A To reduce excess signal pressure levels
B Provide access for debugging software
C Select the point at which baseband signals are generated
D Provide incremental signal delays for filter algorithms
Show the answer and explanation
Correct answer: D. Provide incremental signal delays for filter algorithms
Why D is right. D. Each tap holds a past sample, one delay step older than the last, which the filter multiplies by a weight and adds into the output.
Picture a row of boxes holding recent samples; each new sample shifts the row along. The filter multiplies every box by its own coefficient and adds the results to make one output. The delays and weights together shape the frequency response.
Why not C? Taps are internal delay points, not where baseband signals are produced.
E7F14
Which of the following would allow a digital signal processing filter to create a sharper filter response?
A Higher data rate
B More taps
C Lower Q
D Double-precision math routines
Show the answer and explanation
Correct answer: B. More taps
Why B is right. B. More taps let the filter use more past samples, giving finer control over its response and a steeper cutoff.
A digital filter's response is set by its taps and their weights. With few taps, edges are gradual. More taps allow sharper sides and deeper rejection, at the cost of more computing and more delay.
Why not D? Higher-precision math reduces rounding error but does not by itself steepen the cutoff.