Can you combine ASK, FSK and PSK?
2026-08-16
Can you combine ASK, FSK and PSK?
A reader wrote in about Seeing signals: I and Q from first principles in the September RadCom with a question I did not see coming. The article works through ASK, then FSK, then PSK, and then describes QAM as amplitude and phase together. So, he asked, is there an all-three version, ASK plus FSK plus PSK, and would it carry any advantage?
The answer turns out to say something about what a constellation plot is actually showing you.
Every mode named below is in the Signal Explorer: CW for ASK, RTTY and FT8 for FSK, BPSK through 8PSK for PSK, and 16-QAM and 64-QAM for amplitude and phase together. Turning the noise up on the three PSK panels, then on the two QAM ones, shows the trade-off discussed here rather than asking you to take it on trust.
There is no third axis
Start with what a single symbol has to work with. A sample of IQ data is two numbers, I and Q, which is the same as saying amplitude and phase. That is the whole of it. QAM already uses both, so on that count there is nothing left to add.
Frequency feels as though it ought to be a third thing, but it is not independent of the other two. Frequency is how fast the phase is changing, and the article gives it in exactly those terms: f = dθ/dt. So frequency is a property of how a run of samples moves around the plot, not of where any single point sits. FSK is not a third dimension alongside amplitude and phase. It is a rule about how the phase is allowed to evolve.
That is why the answer cannot simply be yes. There is no free axis to hang a third modulation on.
Two things it could mean instead
Which leaves two readings of the question, and both describe real systems.
The first is to treat frequency as a separate resource rather than a third dimension. Run a constellation on each of many tones at once. That is OFDM, and if you use WiFi or listen to DAB you have used it today. It appears on HF in PACTOR-III. The tones do not all have to carry the same thing: DAB and PACTOR-III put a phase-shift constellation on each one [1][2], while WiFi and DVB-T use QAM [3], which is the version that really does combine all three ideas at once. The advantage is not subtle. Each tone is narrow enough to see a simpler version of the channel, so multipath becomes something you correct tone by tone instead of fighting across the whole passband, and you can put a dense constellation on the tones that are behaving and a sparse one on the tones that are not.
The second is to encode bits in the choice of which tone you transmit on, as well as in that tone’s amplitude and phase, all within one symbol. This also exists, under the name FQAM, and it is a real scheme rather than a curiosity. Some of the bits pick the tone and the rest pick a QAM point on it [4].
The advantage claimed for it is a narrow one. Because only one tone of the set is transmitting at any moment, the interference it creates for neighbouring cells does not look like Gaussian noise, and non-Gaussian interference is easier for a receiver to work around. That is a problem worth solving in a crowded cellular network. It is not a problem any of us have.
Why the second one does not turn up on HF
Here is the part worth the detour.
The great virtue of FSK is that the receiver never has to recover phase. It only has to work out which frequency carried the energy. That is why FSK survives conditions that defeat PSK, and it is part of why so many of the weak-signal modes are frequency modes.
Add PSK on top and you throw that away. Now the receiver has to track phase properly across every hop, and phase coherence is exactly what fading and Doppler destroy. You would pay the extra bandwidth that FSK costs and lose the robustness that made the bandwidth worth paying for. On the trade-off arrow at the end of the article, it falls into a gap rather than onto a useful point.
The categories are ours, not the signal’s
One last thing, which only struck me when I went back to the article to answer the question.
ASK, FSK and PSK are a convenient way to teach modulation, but they are not a real partition of the possibilities. Minimum shift keying can be described equally correctly as continuous-phase FSK or as offset QPSK [5]. Same waveform, two names, depending on which description you pick up first. It is the precise point where the frequency account and the phase account of a signal turn out to be the same account, and it is sitting in WSJT-X as MSK144.
So the honest answer is that you cannot add a third thing, because there were never three separate things to begin with. There is one phasor, moving. ASK, FSK and PSK are three rules about how it is allowed to move, and the shapes in the article are what those rules look like when you plot them.
My thanks to the reader who asked. Questions like that are the best part of writing for a magazine.
References
[1] Radio broadcasting systems; Digital Audio Broadcasting (DAB) to mobile, portable and fixed receivers, ETSI EN 300 401, which specifies the differential QPSK applied to each OFDM subcarrier
[2] Wavecom, Advanced Protocols: PACTOR, http://www.wavecom.ch/content/pdf/advanced_protocol_pactor.pdf
[3] Digital Video Broadcasting (DVB); Framing structure, channel coding and modulation for digital terrestrial television, ETSI EN 300 744 V1.6.2 (2015-10), clause 4.3.5, which specifies the QPSK, 16-QAM and 64-QAM constellations carried on each data carrier
[4] S Hong, M Sagong, C Lim, S Cho, K Cheun and K Yang, ‘Frequency and quadrature-amplitude modulation for downlink cellular OFDMA networks’, IEEE Journal on Selected Areas in Communications, vol 32 no 6, June 2014, pp 1256-1267
[5] S Pasupathy, ‘Minimum shift keying: a spectrally efficient modulation’, IEEE Communications Magazine, vol 17 no 4, July 1979, pp 14-22
Comments
You don't need an account, just a name. Comments are read before they appear, so there may be a short delay before yours shows up.