![]()

Today, integrated circuit for FM/AM radio are readily available,
why would anyone want to build an FM only radio using transistors?
Well, like most of the projects I posted on my website, I started the
design of this FM radio as a project so I can learn a bit of RF
design. It also serves as a flexible circuit that can be modified to
receive signal from other frequency bands. This circuit is also
slightly departed from all the other project I have done so far, as
this project is a pure analog project and lack any digital
components.
The capability of this FM radio is not that impressive. It is a simple receiver that can receive commercial FM band, with a bit of tweaking, it can also receive TV band signal. The output is mono only, however, the bandwidth should be high enough to glue the circuit on to stereo decoding circuit. Though it would be quite difficult to design a FM stereo decoding with out transistor.
There are also some self imposed limit to this design. My goal is to design the circuit with as little magnetic components as possible. This is partially due to my lack of access to good magnetic component and inexperience in working with them. Also, most people who would like to replicate the circuit would find getting the right magnetic component with the correct magnetic core more difficult than to find the right transistor for the job.
The design of the radio is a simple heterodyne design with IF at 10.7MHz. A simple pre-filter that select the FM broadcast band low gain HF(High Frequency) amplifier made up the front stage of the radio. The post filter was omitted in the first cut of the design, the result is excessive noise on the lower end of the reception band.
The output of the HF amplifier is feed to the mixer. The LO(Local Oscillator) generates a signal that is 10.7MHz higher than the channel intended to be received. The mixer multiply the local oscillator's signal with the incoming HF signal to generate the IF signal
The IF stage is the main gain stage. The objective is to provide around 40dB of gain at 10.7MHz. To simplify the IF stage, a ceramic filter is used for frequency selectivity instead of IF can transformer. Thus, alignment of the IF stage is unnecessary.
The output of the IF stage if feed to Quadrature Decoder which demodulate the FM signal back to the audio signal. A 10.7MHz ceramic resonator is used to generate the 90 degree phase shifted signal required for quadrature decoder.
The front stage of the radio consists of a fix frequency, low Q bandpass filter with passband centered around 100MHz. This filter is fed into a amplifier with gain around 10dB. The inductor L1 consists of 6 turns of 22 gage magnetic wire formed on a pen with 5mm diameter.
Tuning of this filter is not important as the selectivity of the filter is very low. The objective is to remove the grossly out of band signal such as 60Hz hum, AM broadcast signal and TV signal.
The design of the local oscillator is pretty conventional. The low end of the LO has been tuned to be around 90MHz, this should be sufficient to cover the low end of the FM broadcast spectrum. The tuning is archived though L2 and the combination of C4, C5 and D2. D2 is the main tuning diode. By using the tuning diode, I can skip the task of trying to find the proper variable capacitor, which is always in short supply. The same tuning voltage can also be used to supply the post-filter stage, which I skipped in this design.
Q2 is the main oscillator transistor, while Q3 is used to lower the impedance of the output. In traditional radio design, can transformer usually fills in this role, but in this design, you will find liberal use of current follower to replace the role of the transformer.
The L2 used in this design is made with 5 turns of 22 gage magnetic wire formed on the core of a pen that is around 3mm in diameter.
The mixer design is a simple switched current mixer. The LO signal is used to chop up the input signal from the HF stage. The down side of this mixer is excessive harmonic and leakage of LO signal and HF signal. However, since the all these signal are way higher than the IF frequency of 10.7MHz, the simplicity of the design wins the argument.
The bias current "AMP BIAS" and "MIX BIAS" is shared across the entire circuit. The "AMP BIAS" is supplied by a potential divider trimmed by a small trim pot. The "MIX BIAS" is half of the supply voltage produced though the simple application of a resistor divider.
The first part of the IF stage is a 10.7MHz filer rip out of a FM radio. The filter is a mechanical filter with Q so high that no LC filter can reach. The bandwidth of the filter is around 300kHz.
Number of website has good analysis regarding the application of the ceramic IF filter. Such as this one
The gain stage of the IF stage is 3 identical transistor amplifier connected in series. Each stage of the amplifier has gain around 13dB. There is no AGC circuit included in this design. I have not figure out a good way to reduce the gain of the IF stage without introducing distortion. Usually, a transformer is used at each stage of the IF amplifier, allowing gain to be adjusted by changing the collector current. However, with solid state design, this is not possible without introducing clipping. This is one area for improvement for the future.
The detector stage is copied from one of the classic FM detector design. The IF is fed to a tuned filter consists of a single resonating crystal Y2. The crystal shifts the phase of the signal depending on the deviation from the center frequency of the crystal. The IF signal and the phase shifted signal is mixed together in a switched current mixer. The output of the mixer is filtered using a simple RC filter to remove high frequency component, leaving the audio frequency as the output.
One of the issue with the detector design is the limitation of the switched current mixer. The mixer do not have good dynamic range, excessive input signal can quickly cause harmonic distortion. This can be a problem as the IF stage do not have AGC support. Digital detection might be the easier way out. Passing the IF though a comparator such as TL3016 than sampled though the use of a FPGA should be a fun project that extends the possibility of this project, however, this wouldn't be a transistor FM radio if FPGA is used...
There you go, this is my simple FM radio design, a bit more complicated than the three transistor AM radio that was popular on the net. I hope you find the circuit useful.
Page created with VIM
Page
last updated: May 1st, 2007
Email: rihuang ([at]) gmail
(*dot*) com
This page is best viewed with Mozilla
Firefox
My Mozilla Ad blocking also blocks Micromedia Flash
Ads: userContent.css