Hi everyone,
I’m currently studying frequency demodulation circuits and have some questions. I understand that these circuits are used to extract information from frequency-modulated (FM) signals, but I’m struggling to grasp how they work in detail.
Specifically, I’m curious about how common designs like phase-locked loops (PLL) or Foster-Seeley discriminators achieve demodulation. How do they convert frequency variations into voltage changes, and what factors affect their accuracy?
If anyone could share explanations, examples, or helpful resources,
I’d greatly appreciate it!
I’m currently studying frequency demodulation circuits and have some questions. I understand that these circuits are used to extract information from frequency-modulated (FM) signals, but I’m struggling to grasp how they work in detail.
Specifically, I’m curious about how common designs like phase-locked loops (PLL) or Foster-Seeley discriminators achieve demodulation. How do they convert frequency variations into voltage changes, and what factors affect their accuracy?
If anyone could share explanations, examples, or helpful resources,
I’d greatly appreciate it!
I've worked with quite a few PLL's over the years, starting with a design project in college to make an AM receiver that used a PLL to synthesize the local oscillator frequencies. That was in 1973.
At their heart PLL FM demodulators are pretty simple. A PLL is a type of servo control loop that uses a voltage controlled oscillator (VCO) and a phase detector, whose (low-pass filtered) output voltage is the control voltage to the VCO. The phase detector's output is proportional to the phase/frequency difference between an external input signal and the VCO. It is used in a feedback loop to force a constant-phase difference between the VCO and external signal. So if the external signal is frequency modulated, the VCO control voltage varies to follow the modulation. So its output is the demodulated version of the FM signal.
Discriminators use an older approach. They use a filter circuit whose center frequency is (typically) 10.7 MHz. The FM-modulated signal is fed into the tuned circuit, designed so the amplitude of the output signal is proportional to the input frequency. That is rectified and filtered, and, again, the output of the filter represents the demodulated signal.
The discriminator approach requires a signal with a fixed amplitude going to the filter, so variations in signal strength at the antenna are greatly reduced by using an amplifier chain that has enough gain to go into limiting -- so signal strength variations have no impact on the output of the discriminator. Only the frequency.
Have you ever experienced a situation where you were listening to one FM station on your car radio, but when driving by another, stronger station, it broke into your station? That's because the PLL based discriminator locked onto the stronger signal and demodulated that one. The Foster-Seeley discriminator can't lock onto an adjacent, stronger, station so it's not susceptible to that effect. I have a (very) old Radio Shack multi-band radio receiver that uses the older-style discriminator, and out of curiosity I took it with me on a test drive, along with my newer car radio. The old RS receiver didn't have any problems with multi-path or adjacent-channel breakthrough, compared to my newer "fancy" car radio. Sometimes technology takes us backwards, not forwards 🙂
At their heart PLL FM demodulators are pretty simple. A PLL is a type of servo control loop that uses a voltage controlled oscillator (VCO) and a phase detector, whose (low-pass filtered) output voltage is the control voltage to the VCO. The phase detector's output is proportional to the phase/frequency difference between an external input signal and the VCO. It is used in a feedback loop to force a constant-phase difference between the VCO and external signal. So if the external signal is frequency modulated, the VCO control voltage varies to follow the modulation. So its output is the demodulated version of the FM signal.
Discriminators use an older approach. They use a filter circuit whose center frequency is (typically) 10.7 MHz. The FM-modulated signal is fed into the tuned circuit, designed so the amplitude of the output signal is proportional to the input frequency. That is rectified and filtered, and, again, the output of the filter represents the demodulated signal.
The discriminator approach requires a signal with a fixed amplitude going to the filter, so variations in signal strength at the antenna are greatly reduced by using an amplifier chain that has enough gain to go into limiting -- so signal strength variations have no impact on the output of the discriminator. Only the frequency.
Have you ever experienced a situation where you were listening to one FM station on your car radio, but when driving by another, stronger station, it broke into your station? That's because the PLL based discriminator locked onto the stronger signal and demodulated that one. The Foster-Seeley discriminator can't lock onto an adjacent, stronger, station so it's not susceptible to that effect. I have a (very) old Radio Shack multi-band radio receiver that uses the older-style discriminator, and out of curiosity I took it with me on a test drive, along with my newer car radio. The old RS receiver didn't have any problems with multi-path or adjacent-channel breakthrough, compared to my newer "fancy" car radio. Sometimes technology takes us backwards, not forwards 🙂
The coil based FM detector is a phase shifting network that has 90 degrees phase at 10.7 MHz. The limited input signal is multiplied by this phase shifted signal - result zero.
Now if the input signal is FM modulated lower, the phase is less than 90 and there is a positive result. Modulate higher and the phase is more than 90 and the result is negative.
The product voltage is almost proportional to the frequency deviation. The dual coil detector is an attempt to make the phase shift more linear over the channel bandwidth.
Now if the input signal is FM modulated lower, the phase is less than 90 and there is a positive result. Modulate higher and the phase is more than 90 and the result is negative.
The product voltage is almost proportional to the frequency deviation. The dual coil detector is an attempt to make the phase shift more linear over the channel bandwidth.
another detector is the pulse count detector. at every zero crossing a pulse is counted or integrated. the higher the frequency the more pulses. A relative insensitive method but very linear. It is used in some high-end FM receivers of panasonic, where the frequency deviation of the 10.7 IF signal is tripled by tripling the 10.7 to 32.1Mhz, then subtracting 31.3Mhz until an 800khz wide bandwidth signal results to be pulse counted.
In the past I designed higher order phase shift networks for the FM multiplying detector that most FM chips use. Otherwise it is a tradeoff between linearity and sensitiviy. the filter design book of Sverev is very helpful in creating higher order filters.
The FM demodulator in the Studer A726 uses a length of coaxial cable to convert frequency into phase difference, and then passes it through the usual multiplying phase detector/hard-clipping double-balanced mixer.
Modern FM radios often have a complex-valued (I/Q) IF signal that gets digitized and digitally filtered (channel-selective filtering). The result is then put into a CORDIC to calculate the momentary phase. Take the derivative to time and you have the momentary frequency.
Modern FM radios often have a complex-valued (I/Q) IF signal that gets digitized and digitally filtered (channel-selective filtering). The result is then put into a CORDIC to calculate the momentary phase. Take the derivative to time and you have the momentary frequency.
there are even higher order phase linear coil circuits possible, the filter synthesis handbook of anatoli sverev can be useful . the other alternative for the linear phase shifter is a delay line, like a roll of coaxial cable.
I have built experimental delay line detectors for a very wide deviation FM video link over fibre optic cable. They are not useful for FM radio detection as the output voltage would be tiny.
In practice, no matter how linear you make your detector beyond your typical dual coil detector, other distortion due to filter group delay is more important.
In practice, no matter how linear you make your detector beyond your typical dual coil detector, other distortion due to filter group delay is more important.