What is the purpose of the marked band-pass filter?
The output of an AM (Amplitude Modulating) mixer contains a number of frequency components. These components are 1) the modulating signal. This signal is typically audio frequency, and in the diagram is the signal that comes out of the DAC and passes through a filter before reaching the mixer. The second component is the carrier signal, which is provided by the LO (Local Oscillator) in your circuit. The third and fourth components are the lower and upper sidebands. These are new components that are created in the mixer, and are neither the carrier, nor the modulating signal.
The purpose of the filter between the mixer and the (RF) radio frequency amplifier (in your circuit, labeled "PA") is to attenuate (reduce the amplitude of) the modulating signal, so that very little of it appears at the input of the RF amplifier. The filter between the mixer and the RF amplifier should not remove the carrier or either side-band in an amplitude modulated signal in an AM (amplitude modulated) system. There are other types of modulation, such as SSB (single-sideband suppressed carrier) where the carrier is suppressed, and one of the sidebands as well. Finally, there are systems such as phase modulation and frequency modulation that sometimes generate an AM signal as an intermediate stage. However, your diagram does indicate that the output of the mixer is used in any of these ways.
Every mixer produces quite a deal of unwanted frequencies.
- frequencies contained in both input signals
- the above frequencies multiplied by various integer numbers
- all of the above added to or subtracted from each other
- as well as some amount of a honest noise
Some of these may even be stronger than the "payload" signal.
In the general case, we don't want to propagate the unwanted frequencies down the stages.
In your particular case, the unwanted frequencies will:
- be transmitted in frequency bands that the device is not allowed to transmit, interferring with other radio communications. This leads to the second point:
- In the general case, transmitting radio is heavilly regulated by law. A device that emits too much radio power outside of its intended frequency band may not be allowed to operate. And, in the general case, "too much" is in fact very tiny amount.
- require (and waste) additional power from the source powering the final amplifier. The available energy is a limiting factor in a lot of designs.
- the next stages in the circuit (PA in your case) may misbehave in a number of ways when fed with unwanted signals along with the wanted signal - e.g. they may lower their efficiency, generate additional heat, stop working altogether, etc, etc...
Luckily, most of the unwanted ingredients of the mixed signal lie well outside of the useful signal frequency band and can be attenuated to the needed low level just by applying a frequency-dependent filtering.
This is why every mixer in a sane radio frequency diagram is always followed by a band-pass filter.
Before sending the signal to the antena the signal's bandwidth must be limited using a band pass filter in order to avoid interference with adjacent stations.
Most of the RF spectrum is regulated by an agency or a goverment body within a country. Among the regulated attributes of a RF tranmission, there is the channel size. For example, the channel spacing of AM broadcast in North America is 10 KHz; so there is a station bradcasting at 540KHz and another one broadcasting at 550KHz. Therefore, the bandwidth of both stations must be limited to 10KHz (i.e. station 1 from 535 to 545 KHz and station 2 from 545 KHz to 555Khz) to avoid both stations to interfere to each other.
Another example is WiFi that has a 20Mhz (or a multiple of 20 MHz) channel bandwidth. Using channels is possible to have multiple routers in a neighborhood without interference.
This is the basic configuration of a radio transmitter which actually generates the modulation digitally in the computer. The output of the DAC can have for ex. data which is amplitude-phase modulated to a carrier which has low enough frequency to be handled by math (=DSP).
The first bandpass filter (=IF) limits the output of DAC to a certain bandwidth which is narrow enough to fit in the allowed bandwidth to the becoming radio signal, but also wide enough to carry the modulated data so that it can be later extracted.
The mixer shifts the filtered output of the DAC upwards in the frequency. The 2nd bandpass filter kills non-ideal mixing products that practical mixers always make. Only a mixer which does an ideal multiplication is free of non-ideal mixing results (assuming LO outputs a pure sine wave). Fortunately proper mixer design and selection of frequencies make possible to filter out all non-ideal mixing products so well that there's low enough spurious emission level.
The ideal mixing result still contains 2 components - the upper and the lower sideband. The names are taken from DBS-modulation, because essentially this DSB modulates to LO frequency what the 1st filter (=IF) outputs.
If the passband of the 1st filter is far enough from 0Hz, the 2nd filter can in practice be made to remove also the lower sideband. In that case this radio transmitter really outputs what is got from the output of the 1st filter, only every frequency component of its spectrum is moved upwards. The shift for every component is = the LO frequency.
Not asked: In radars the same LO output should be used in the receiver to ensure perfect coherency between the making of the transmitted signal and mixing down & detecting the echoes.