[Analysis of the topic of the college electronic competition] - 2019 National Competition G "Wireless transceiver system for dual-channel voice simultaneous interpretation"

gmchen

[Analysis of the topic of the college electronic competition] - 2019 National Competition G "Wireless transceiver system for dual-channel voice simultaneous interpretation" [Copy link]

 

1. Mission

Design and manufacture a dual-channel voice transmission wireless transceiver system to achieve the simultaneous transmission of two-way voice signals on one channel. The schematic diagram of the system is shown in the figure.

II. Requirements

1. Basic requirements

( 1 ) Make an FM wireless transceiver system. The carrier frequency of the FM signal is set to 48.5MHz , the absolute value of the relative error is no more than 1 ‰; the peak frequency deviation is no more than 25kHz ; and the antenna length is no more than 0.5m .

( 2 ) Transmit any voice signal A or B through the FM wireless transceiver system , and the bandwidth of the voice signal shall not exceed 3400Hz . The wireless communication distance shall be no less than 2m , and the demodulated output voice signal waveform shall not be significantly distorted.

( 3 ) Transmit two-way voice signals A and B simultaneously through the FM wireless transceiver system . The wireless communication distance is required to be no less than 2m , and the demodulated output of the two-way voice signal waveform has no obvious distortion.

2. Play part

( 1 ) The carrier frequency of the FM signal in the designed transmitting circuit can be adjusted by a voltage signal v _C ( t ) to simulate the carrier frequency drift in wireless communication. The frequency drift amount caused by the unit voltage adjustment of the voltage signal v _C ( t ) carrier frequency is designed by the contestant.

( 2 ) Under the premise of ensuring that the system can correctly perform two-way voice wireless transmission, the carrier frequency of the FM signal is adjusted by the v _C ( t ) signal to produce a drift of no less than 300 kHz , and the adjustment time τ is required to be no more than 5s (seconds).

( 3 ) Under the premise of ensuring that the system can correctly perform two-way voice wireless transmission, the carrier frequency of the FM signal is adjusted through the v _C ( t ) signal and drifted as shown in the figure below. It is required that the carrier frequency drift range Δ f ₀ of the FM signal is as large as possible.

( 4 ) Others.

3. Description

( 1 ) The voice signal input to the system can be generated by a standard signal source; the demodulated voice signal output should have a test interface to facilitate oscilloscope observation.

( 2 ) The manufactured FM transmitting circuit should have a test port at the transmitting antenna end for testing.

( 3 ) The external voltage signal v _C ( t ) for controlling the carrier frequency drift of the FM signal is input externally through a standard signal source. When the external v _C ( t ) signal is zero, the carrier frequency drift of the FM signal is correspondingly zero.

gmchen

Topic analysis and solution design

FM modulation and demodulation is a basic communication circuit, which is taught in high-frequency circuit courses in college, so the overall solution is not a problem. The design difficulties of this problem lie in the following points: 1. FM signal generator, which requires the FM carrier frequency to have a frequency accuracy of 1 ‰ and a carrier frequency drift of more than 300kHz . 2. Two-way voice transmission. 3. Reception under carrier frequency drift.

I have posted a thread on this forum before and discussed this topic. This time I have carefully analyzed the topic. I will analyze these issues one by one, make some specific quantitative analysis, and correct some deficiencies in the original thread.

gmchen

1. FM signal generation

Common FM signals can be generated by the following methods: direct frequency modulation by voltage-controlled oscillator, direct frequency modulation by voltage-controlled crystal oscillator, frequency modulation by phase-locked loop, and frequency modulation signal generated by DDS . The following analyzes the characteristics, applicable scope, and key points of these methods in the application of this topic.

a) Direct frequency modulation by voltage controlled oscillator ( VCO ).

The VCO can be an RC oscillator or an LC oscillator based on a varactor diode. This method is the simplest FM modulator, and it is easy to obtain sufficient frequency deviation and carrier frequency drift under an applied voltage. The problem is that the frequency accuracy and stability of the oscillator are difficult to guarantee. It is difficult to achieve the 1 ‰ frequency accuracy in the question. Even if you pay great attention to circuit design, component selection, installation and debugging, you can only achieve short-term stability . Therefore, this solution is usually only used for demonstrations and toy-level applications.

b) Direct frequency modulation by voltage-controlled crystal oscillator ( VCXO ).

VCXO is a commercial-grade device that changes the oscillation frequency of a quartz crystal oscillator through a varactor diode, so the frequency accuracy and stability are excellent. The main problem with VCXO is that the relative frequency deviation is very small, usually only ± 100ppm~ ± 200ppm . To achieve frequency modulation with VCXO , the maximum frequency deviation of the VCXO should be controlled within its allowable range, and then the frequency deviation can be gradually expanded through the method of frequency doubling and mixing.

The carrier frequency required in this question is 48.5MHz , and the maximum frequency deviation is 25kHz+150kHz ( 25kHz is the maximum frequency deviation of the modulation signal, and 150kHz is the maximum frequency deviation of the quasi-DC drift). The relative frequency deviation is about 3600ppm , which is much larger than the allowable frequency deviation of the VCXO . Therefore, the frequency deviation needs to be increased. The specific steps are as follows:

First, a 50MHz VCXO is frequency modulated to control the maximum frequency deviation to 1kHz+6kHz ( 1kHz is the modulation frequency deviation and 6kHz is the drift frequency deviation). The relative frequency deviation is only about 140ppm , which is within the controllable frequency deviation range of the VCXO . Then, the phase-locked loop is used to multiply the signal by 5 times to obtain an FM signal with a center frequency of 250MHz and a frequency deviation of 5kHz +30kHz . Then, this FM signal is mixed with a 200MHz signal and its lower sideband is taken, so that its center frequency drops to 50MHz , but the frequency deviation remains unchanged. Repeat the 5- times multiplication process to obtain an FM signal with a center frequency of 250MHz and a frequency deviation of 25kHz +150kHz . Then, it is mixed with a 201.5MHz signal to finally obtain the FM signal with a center frequency of 48.5MHz and a frequency deviation of 25kHz+300kHz as required by the question .

As can be seen from the above example, VCXO direct frequency modulation is still very useful for FM signal modulation with relatively small frequency deviation , so it is often used in digital signal modulation (such as FSK ). However, for FM signals with relatively large frequency deviation , the frequency multiplication - mixing process may have to be repeated many times , and the circuit becomes more complicated. For example, in this problem, since the relative frequency deviation is large due to the need to simulate carrier frequency drift, it is not the best solution.

c) DDS directly generates FM signal.

Since the reference frequency of DDS is based on quartz crystal, this method has excellent frequency accuracy and stability, and can also obtain sufficient frequency deviation and realize analog carrier frequency drift. Theoretically, changing the output frequency of DDS is a discontinuous step change process, but since DDS can achieve extremely small frequency step changes, it can actually be considered as a quasi-continuous modulation process, so it can also realize FM modulation.

However, the FM modulation of the DDS chip is achieved by the microprocessor constantly changing the phase increment inside the chip, so it is more suitable for regular frequency change processes, such as signal generators. The modulation signal in this question is voice, and its frequency and amplitude change randomly. When using DDS to generate FM modulation, the microprocessor must first perform ADC sampling on the input voice signal , and then change the phase increment of the DDS according to the sampling result . The actual structure is relatively complex and is not the best solution.

gmchen

d) Phase-locked loop frequency modulation.

The reference frequency of the phase-locked loop frequency modulation is provided by a quartz crystal, so this method has excellent frequency accuracy and stability, and can also provide sufficient frequency deviation, which may be the best solution for this problem. The following is some specific analysis based on the specific requirements of this problem.

The figure below is the basic structure of the phase-locked loop FM modulation circuit. In this circuit, the reference frequency of the phase-locked loop comes from the quartz crystal, which has extremely high stability and can be considered to be fixed. However, the frequency of the FM signal output by the VCO is constantly changing, so most of the time, the two input frequencies of the phase detector are different. In other words, the phase-locked loop in the phase-locked FM circuit works in an unlocked state most of the time. When the phase - frequency detector ( PFD ) is used for phase detection, the PFDthe FM signal frequency is close to the carrier frequency , and outputs alternating positive and negative pulses, while the PFD works in the frequency detection state for most of the rest of the time, and its output is high or low, so the output of the PFD is approximately a square wave with the same frequency as the modulation signal frequency. The purpose of the loop filter is to take out the average value of this approximate square wave as the center control voltage of the VCOthe FM signal is always synchronized with the reference frequency. In this sense, it seems more appropriate to call the phase-locked loop a "frequency-locked loop".

gmchen

The following figure shows the relationship between the phase detector PFD output and the loop filter LF output. In one cycle of the modulation signal, the frequency of the FM signal changes periodically around the carrier frequency, and the output of the PFD is an approximate square wave. Since the RC time constant of the loop filter is usually much larger than the modulation signal cycle, the LF output voltage is approximately a triangle wave, and its average value V _H /2 corresponds to the carrier frequency of the FM signal, and the LF output voltage fluctuation corresponds to the fluctuation of the carrier frequency in the FM signal.

After approximating the LF output voltage to a triangular wave, its maximum deviation from the average voltage can be written as

This value corresponds to the maximum value of the carrier frequency fluctuation in the output FM signal (note that it is the fluctuation of the carrier, not the maximum frequency deviation caused by modulation). Since V _H /2 corresponds to the carrier frequency, the maximum relative error of the carrier frequency fluctuation is

The above formula can be used as a basic design formula for phase-locked loop FM circuit.

Back to the topic, _{the relative error of the carrier frequency} is required to be no more than 1 ‰. Substituting it into the above formula, it requires ₍ R1 + R2 ) C > 1000T / 4 . The topic stipulates that the modulation signal is a voice signal. Usually, the frequency range of the voice signal is 300Hz~3400Hz , so the maximum period of the modulation signal is T =3.3ms . Substituting this value, one of the conditions for the design of the loop filter in this circuit is ( R1 + R2 ) C >0.83s _, which can be taken as 1s during design . This is a large time constant. In order to ensure that its value _will not be reduced due to the influence of load resistance, it is usually necessary to connect a buffer with high input impedance at the output of the loop filter, such as a follower composed of an op amp.

Another basic formula in this phase-locked loop is the damping factor

Where Kd is the _{gain of the} PFD , Ko is the gain of the VCO , _Kf is the gain of the loop filter (including the subsequent voltage buffer and the weight coefficient of the adder), and N is the frequency division coefficient of the feedback loop. Considering the stability of the phase-locked loop, ζ is usually required to be between 0.7 and 1 .

The design of the loop filter can be completed based on the above two basic formulas.

gmchen

There is also a requirement for simulating carrier frequency drift in wireless communication in the question . The indicators are summarized as follows: a voltage signal v _C ( t ) is input externally through a standard signal source to make the carrier frequency change (drift), with a maximum change of ±150kHz . When the external v _C ( t ) signal is zero, the carrier frequency drift of the FM signal is correspondingly zero. The following analyzes the circuit that realizes this requirement and its points for attention.

The drift voltage and the audio signal voltage are superimposed as the modulation voltage of the VCO , _{and the drift of the output frequency can be achieved. The circuit structure is shown in the figure below. Among them,} R3 ~ R5 constitute a resistor adder, and the resistance values of the three resistors determine the weights of the three signals. The modulation signal is an AC signal, and the use of capacitor coupling can ensure that it will not affect the average voltage of the input _VCO , that is, it will not affect the carrier frequency. The drift voltage is a DC signal, so DC coupling must be used.

Since the average voltage output by the PFD is V _H /2, which corresponds to the carrier frequency of the FM signal of 48.5MHz , and according to the requirements of the question, the external drift voltage varies positively and negatively. When it is 0, the carrier frequency of the FM signal is 48.5MHz , so the zero potential of the external drift voltage should be shifted to V _H /2 . In the above circuit, a superimposed bias voltage is used to represent this voltage shift. In the debugging of the actual circuit, this bias voltage can be fine-tuned so that when the external drift voltage is 0 , the phase-locked loop enters the locked state normally, and the control voltage of the VCO is close to V _H /2 .

gmchen

The state where the applied drift voltage is not equal to 0 is discussed below .

Assuming that a larger drift voltage is input, the output frequency (carrier frequency) of the VCO will change in a step under this step input. Due to the negative feedback of the phase-locked loop, the PFD will output a level (VH or 0) that is inversely proportional to the drift voltage . After _passing through the loop filter, this level will output a voltage that changes exponentially. This voltage will partially offset the input drift voltage, reducing the actual carrier frequency drift. Therefore, after the initial step change, the output frequency of the VCO will gradually approach the initial frequency as the output of the loop filter changes. However, since the final output of the loop filter is limited by the power supply voltage, _it will approach its limit after 3τ ~5τ . Therefore, as long as the weights of the two input resistors ( R3 and R5 ) are _{reasonably arranged} , it can be ensured that the VCO finally reaches the predetermined output frequency. The following figure is a schematic diagram of the above frequency change .

There is a requirement in the question: "Use the v _C ( t ) signal to adjust the carrier frequency of the FM signal to produce a drift of no less than 300 kHz , and the adjustment time τ is required to be no more than 5 seconds." This description is a bit vague, but according to the literal understanding, it refers to the drift voltage of the above input step change. According to the above analysis, if the time constant of the loop filter τ = ( R ₁ + R ₂ ) C = 1s , then after 5s , that is, 5 τ , its output is close to the limit, and the circuit is stable, which can meet the requirements of the question.

From the above discussion, we can also see that under the action of a large drift voltage, the loop filter is finally in an output saturation state (the output is always maintained at a high level or always maintained at a low level). In fact, the phase-locked loop is already in an open-loop state, and the entire circuit degenerates into a VCO direct frequency modulation circuit.

However, within a small range of input drift voltage, due to the negative feedback of the phase-locked loop, the output voltage of LF cancels out the input drift voltage, the phase-locked loop remains in a locked state, and the center frequency of its output remains unchanged (frequency pulling phenomenon).

The third item of the question requires that the output carrier frequency changes in a triangular wave. If the input offset voltage signal is a triangular wave, due to the frequency pulling effect of the above-mentioned phase-locked loop, the output of the phase-locked loop will have a period of frequency-invariant locking process near the center frequency of 48.5MHz . When the external offset reaches the point where the phase-locked loop cannot lock, the output frequency will jump to a certain frequency and then begin to change. Its frequency - time curve is somewhat similar to the crossover distortion of a class B push-pull amplifier. However, since the question does not have this requirement, it seems that there is no need to improve it.

gmchen

2. Problems with two-way simultaneous interpretation

Since the spectra of the two-way voice overlap, they cannot be simply added together using an adder, as this would make the two signals indistinguishable. The correct approach is to first move the spectrum of one of the signals so that the spectra of the two signals no longer overlap, and then add the two signals together.

The specific method can refer to FM stereo broadcasting: AM modulate a subcarrier frequency with a certain audio signal B (FM stereo broadcasting uses AM modulation with suppressed carrier frequency , and ordinary AM modulation can be used here for the convenience of later demodulation) to form a signal B ' , and then superimpose it with another audio signal A to form a composite audio signal. The above spectrum relationship is as follows:

In order to facilitate filtering during demodulation, a certain frequency interval should be kept between signal B ' and signal A. It is known that the bandwidth of signals A and B is 3400Hz , so the subcarrier frequency should be greater than twice 3400Hz . For example, the subcarrier frequency can be 9kHz . At this time, the highest frequency of the composite signal is 9000+3400=12.4kHz .

The transmitter uses the above composite audio signal to perform FM modulation on the main carrier (the modulation circuit has been introduced earlier) to obtain a modulated wave for dual-channel simultaneous transmission.

gmchen

The receiving end performs a series of processing such as down-conversion and intermediate amplification on the received FM signal (detailed analysis will be given later), and then performs FM demodulation (frequency discrimination) to obtain a composite audio signal. Then a low-pass filter can be used to directly obtain signal A , and a band-pass filter can be used to obtain signal B ' modulated by the subcarrier . Then this signal is AM demodulated (detected) to obtain signal B. The figure below is the demodulation circuit structure for this problem.

The discriminator can be a phase-locked discriminator, a slope discriminator, etc. Among them, the slope discriminator composed of an LC resonant circuit is the easiest to make and debug. The detector is the simplest and most convenient diode large signal detection circuit.

The figure below shows _an LC resonant circuit slope frequency detector. The core of _the circuit is the resonant circuit composed of L , C1 and C2 . The circuit has two resonant points: parallel resonant frequency

Series resonant frequency

The two resonant frequencies are located symmetrically on both sides of the intermediate frequency.

gmchen

When the frequency of the input FM modulated signal is higher than the intermediate frequency, the amplitude of vp increases because it tends to the parallel resonance frequency _{; when} the frequency of the modulated signal is lower than the intermediate frequency, it tends to the series resonance frequency and _{the amplitude of vs} increases . The DC components of vp and vs are extracted through diode detection, and the subtractor composed of the last operational amplifier subtracts and filters out the intermediate frequency components, which becomes the demodulated output of the FM signal.

The relationship curve between the output voltage and the input signal frequency of this demodulator is the inverse superposition of the resonance curves of the LC loop at two resonant frequencies (parallel minus series), and its shape is called an S curve. The middle part with better linearity is the effective frequency discrimination range.

The Q value of the inductor affects the slope of the S curve, that is, the frequency discrimination gain of the frequency discrimination circuit. The resistor R4 affects the ratio of the two resonant peaks, that is, the symmetry of the S curve. During debugging, this resistor should be fine-tuned to obtain good frequency discrimination characteristics.

gmchen

3. Reception problems under carrier frequency drift

The difficulty lies in the processing at the receiving end, which involves the selectivity of the receiver and the sensitivity of the frequency discrimination circuit.

The receiving end usually adopts a superheterodyne receiving circuit. The selectivity of this circuit mainly depends on the frequency selection characteristics of the intermediate frequency amplifier. On the premise of meeting the spectrum requirements of the modulated signal, the better the frequency selection characteristics of the intermediate frequency amplifier, the lower the noise of the receiver.

In addition, FM demodulation often uses circuits such as slope discriminators or phase-locked discriminators. These circuits have one thing in common, that is, the narrower the operating frequency range of the demodulator, the higher the demodulation sensitivity.

The local oscillator frequency of a conventional superheterodyne circuit is fixed. When the frequency of the input signal drifts, the frequency of the signal sent to the intermediate frequency amplifier will also drift. Considering the possible drift of the input frequency, the bandwidth of the intermediate frequency amplifier's frequency selection network and the operating frequency range of the discriminator are usually designed to be slightly larger than the signal bandwidth. This question stipulates that the maximum frequency deviation of FM modulation is less than 25kHz , so in general, the bandwidth of the intermediate frequency amplifier and the operating frequency range of the discriminator can be received normally as long as they are slightly larger than 50kHz .

However, in this problem, the frequency drift range is very large. According to the requirements of the problem, the carrier frequency drift is greater than 300kHz ( ± 150kHz ). If a fixed local oscillator frequency is used, the intermediate frequency amplifier must have a passband bandwidth greater than 300kHz , and the operating frequency range of the discriminator must also be greater than 300kHz . However, this is in great contradiction with the normal 50kHz bandwidth mentioned above. Such a large intermediate frequency bandwidth and discriminator frequency range will result in a very poor signal-to-noise ratio of the output signal.

gmchen

The key to solving this problem is to try to ensure that the intermediate frequency remains unchanged. If the local oscillator frequency can be changed with the drift of the input signal frequency, then the entire intermediate frequency amplifier and discriminator can be designed according to the conventional superheterodyne circuit. This circuit is usually called an automatic frequency control ( AFC ) circuit. The specific circuit structure is shown in the figure below.

This circuit uses a voltage-controlled oscillator ( VCO ) to generate a local oscillator signal, the center frequency of which is one intermediate frequency higher than the nominal value of the FM signal, that is, 48.5MHz+10.7MHz=59.2MHz . The DC component in the discriminator output signal is used to control the oscillation frequency of the VCO .

When the carrier frequency of the input signal is higher than the nominal value, the intermediate frequency will decrease because the intermediate frequency is equal to the local oscillator frequency minus the input frequency. Assuming that the characteristic of the phase detector is the positive S curve shown in the previous figure (note that the S curve of the phase detector can be positive or negative, depending on the input connection of the subtractor), the average voltage output by the phase detector will decrease accordingly. This voltage is added to the control end of the VCO after inverting amplification , so that the oscillation frequency of the VCO increases, thereby maintaining the intermediate frequency basically unchanged. The same result can be obtained when the carrier frequency of the input signal decreases.

Pay attention to the positive and negative directions of the S curve. If the discriminator used is an inverted S curve, the amplifier can be changed to a common-mode amplifier.

gmchen

The discriminator output of the AFC circuit contains both the quasi-DC voltage indicating the deviation of the intermediate frequency and the demodulated voice signal. The function of the filter - amplifier in the feedback loop is to extract the quasi-DC signal and attenuate the voice signal as much as possible (if the voice signal is fed back, its output will be reduced, which is the so-called anti-modulation phenomenon). Therefore, the cutoff frequency of this filter circuit should be much lower than the lowest frequency of the audio signal, 300Hz . Considering that the question requires the drift signal to change with a triangular wave with a period of 2s , the cutoff frequency of this filter should be higher than 0.5Hz . After comprehensive consideration, it can be selected between 1 and 10Hz . For example, if 3Hz is selected , it will have a 40dB attenuation for the 300Hz audio signal , which can be considered sufficient.

The above AFC circuit is called a zero -order zero-difference control system in the automatic control system . Its characteristic is that the feedback control signal is maintained by the output error signal, so after the input signal frequency drifts, the output intermediate frequency still changes. The purpose of automatic control is to minimize the deviation of this output frequency.

Assume that the frequency discriminator gain is Kd ( _V /Hz ), the voltage control gain of the VCO is Ko ( Hz/V ), _and the DC gain of the filter - amplifier in the feedback loop is Ka _. According to the negative feedback theory, when the drift frequency deviation of the input signal is _ΔfS , the intermediate frequency change is

It can be seen that due to the negative feedback effect, the frequency drift of the input signal is greatly reduced. As long as the reduced carrier frequency drift does not affect the amplification of the intermediate frequency amplifier and the frequency discrimination of the discriminator, it can be considered that the receiver can work normally under the condition of carrier frequency drift.

The maximum value of the input carrier frequency drift in this question is _150kHz . Assuming that the effective frequency range of the LC slope discriminator is ±50kHz , and the output within this range is ±1.5V , the frequency gain Kd = 3 × 10-5 ( V/Hz ); assuming that the frequency change of the VCO is ±10MHz when the control voltage is ^2.5V ±1V ^, _the voltage control gain Ko = 1 × 107 ( Hz/V ) ; assuming that the gain of the filter amplifier Kf = 1 , then _KdKfKo = 300. _{Knowing that} the maximum carrier frequency change of the input signal is ±150kHz , the _intermediate frequency change under the action of AFC is only about ±0.5kHz . As long as the passband width of the intermediate frequency amplifier and the effective frequency range of the discriminator have a margin of ±0.5kHz , normal reception can be guaranteed in the case of carrier frequency drift. By appropriately increasing the passband width of the intermediate frequency amplifier and the effective demodulation range of the demodulator, the requirement to increase the input carrier drift frequency can be met.

说说说

good