While studying 'Qaudrature carrier multiplexing', I came across these statements in my text material:

"Precisely varying the phase of a high-frequency carrier sine wave in a hardware circuit according to an input message signal is difficult. A hardware signal modulator that manipulates the amplitude and phase of a carrier sine wave is expensive and difficult to design and build, and, as it turns out, not as flexible as a circuit that uses I and Q waveforms."

Thus practical hardware design concerns make I and Q data the better choice. This way we can avoid manipulating the phase of an RF carrier directly.

Why manipulating the phase directly is not a good choice?

  • \$\begingroup\$ What part of the two sentences that you quote do you fail to understand? \$\endgroup\$ – Dilip Sarwate May 17 '15 at 16:07
  • \$\begingroup\$ @DilipSarwate:Sir, I am unable to understand that why/how it is difficult to precisely vary the phase of a high-frequency carrier sine wave in a hardware circuit according to an input message signal. Is this the same reason for which we prefer non-coherent detection over coherent detection? \$\endgroup\$ – a.s. May 17 '15 at 16:49
  • \$\begingroup\$ The why is a question for electronics.SE or engineering.SE, not dsp.SE. \$\endgroup\$ – Dilip Sarwate May 17 '15 at 17:08
  • \$\begingroup\$ How do I migrate my question to electronics.SE? \$\endgroup\$ – a.s. May 18 '15 at 1:20
  • \$\begingroup\$ Click on the flag link and you can send a request to the moderators to migrate the question to electronics.SE \$\endgroup\$ – Dilip Sarwate May 18 '15 at 3:31

I think an alternative way of rephrasing the statement would be "The easiest way to vary the phase of a high-frequency waveform is often to generate I and Q waveforms and modulate those". While it may be possible to adjust a reference oscillator's phase by momentarily changing its frequency, adjusting the frequency by just the right amount is apt to be very difficult. If one wants to maintain a consistent frequency despite the phase modulation (so that after the phase is e.g. modulated forward 90 degrees and then back 90 or--even harder--after it's been modulated forward 90 degrees four times, its phase will be the same as if it had never been modulated at all) one will almost certainly have to do something to the signal after it comes out of a reference oscillator which is unaffected by the modulation.

Phase-shifting a signal would thus require adding a variable amount of delay to it. While it might in theory be possible to use a multi-tap delay line structure with some form of interpolation between the taps, there's really only a need for one delay tap; things work out most conveniently if it's 90 degrees separated from the reference. If one has a pair of four-quadrant multipliers, any phase may be generated by scaling the reference and 90-degrees-offset wave by phase-dependent amounts and summing the results.

If one needed to have a variable time delay on a waveform which was not sinusoidal, jinxing the reference oscillator or using a multi-tap delay line would be reasonable ways of accomplishing that, whereas the I/Q approach is only really usable with sine waves. Indeed, the Atari 2600 Video Computer System (1977) had a color circuit which would generate 15 different chroma (3.57945MHz) phases using delay lines, and I believe a number of other 1970s/1980s computers did so as well. The 2600 started with a roughly-sinusoidal 3.579545Mhz sine wave, but having a chain of 30 inverters with a combined delay of about 8.7ns each was cheaper than the analog circuitry that would have been required to amplify, scale, and sum sine waves.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.