Now I know that this question might sound silly, that is because apparently I am missing something here. Basically my question is, if ASK works as on/off keying meaning that provided we have an oscillator with fc carrier frequency and binary input to produce the fc signal (using oscillator) when there is 1 in input we just using that one frequency - the carrier frequency! So why we need a range of frequencies (which as I understand it is what bandwidth basically means)?
-
2\$\begingroup\$ Your signal is not one signal frequency. It is not a pure sine wave, it is being switched on and off. Such a signal has a theoretically infinite bandwidth. \$\endgroup\$– Eugene Sh.Commented Jan 10, 2020 at 17:52
-
1\$\begingroup\$ Indeed, and likewise, a sine wave signal that has only energy at single frequency, it must be infinitely long. Any amplitude change means it has bandwidth. \$\endgroup\$– JustmeCommented Jan 10, 2020 at 18:42
4 Answers
The rise time of Carrier Amplitude spreads BW just as the rise time of the signal determines the -3dB BW = 0.35/ t for t= 10 to 90% rise time.
So the BW of the signal translates to both sides of the carrier. (unless single sideband carrier is suppressed by design to conserve BW.)
You mustn't forget two things:
- The convolution property of the Fourier transform (i.e. multiplication in time domain is equivalent to convolution in frequency domain)
- the fact that you're multiplying your carrier with your data signal
So, your signal is
$$s(t) = c(t) \cdot d(t)$$
where \$c\$ is the carrier (typically, a cosine or a complex sinusoid), and \$d(t)\$ is your amplitude that changes over time.
Now, we don't know much about \$d(t)\$, but assume it's just a rectangular function (so, the thing is "on" for some finite time).
That means that your data signal has a spectrum with a bandwidth – the Fourier transform of a rectangle is a (scaled) sinc function, and that has a bandwidth.
With the carrier multiplication (the carrier is just one or two dirac impulses in frequency domain), you just shift the spectrum of the data signal from being centered around 0 Hz to being centered around the carrier frequency. The bandwidth stays the same.
TL;DR: Your data carrying, i.e. the carrier-modulating, signal has bandwidth. So does the RF.
What would happen to the output of a very, very small bandwidth band pass filter when a pure carrier of the matching frequency is suddenly applied at its input. Would the output of the band pass filter instantly spring into action and produce a signal at the carrier frequency or, would you see a very, very sluggish carrier build up in amplitude over several seconds or minutes until finally you saw the full carrier amplitude?
You may wish to consider the Shannon Limit from information theory, which states that you need a combination of bandwidth and signal-to-noise ratio to successfully transmit information over a channel. The better SNR you have, the less bandwidth you need, but the latter never actually goes to zero.
The bandwidth of an ASK signal can be very low, allowing many signals to share a limited piece of radio spectrum, but it is closely related to the rate at which the signal is keyed. If you impose a bandwidth limit of 1Hz, for example, then you can expect the signal to take about a full second to respond adequately to each key-on and key-off event. Clearly for a crisp 20WPM transmission, you need significantly better than that.