Two Birds with One Tone: I/Q Signals and Fourier Transform
93 points
21 hours ago
| 6 comments
| wirelesspi.com
| HN
KeplerBoy
11 hours ago
[-]
The liberal use of AI generated images really cheapens the entire article. Please don't do it. At that point I suspect most of the text is also AI generated.
reply
dsp_man
4 hours ago
[-]
Only two images are AI generated where horses as carriers needed to be shown. Can you please explain why this is a problem? Thank you (Also, all text is written by me :)
reply
yodon
15 hours ago
[-]
Pro tip: If you're writing an article on the significance of something called I/Q, it's cool to somewhere in the first couple pages say something about what I/Q is.
reply
gsf_emergency_2
14 hours ago
[-]
Q=Quadrature, I=In-phase

(As you point out not in the first couple pages, but waaay down)

he "explains" those

https://wirelesspi.com/two-birds-with-one-tone-i-q-signals-a...

Not trying to be charitable like furgot ... The wikipedia page is the first time I've seen authors go pro on the topic

reply
wucke13
14 hours ago
[-]
https://www.pe0sat.vgnet.nl/sdr/iq-data-explained/

This is an excellent introduction to the concept and also to the why complex numbers are used to represent signal samples.

reply
msravi
13 hours ago
[-]
I prefer a more "physical" explanation - you have two carriers: sin(wt) and cos(wt), and you're modulating bits I and Q onto the two carriers and adding them up before transmitting. Now, mathematically, that's the same as representing the two bits as I+jQ and multiplying it with cos(wt)+jsin(wt). Demodulation is simply multiplying that output with the complex conjugate cos(wt)-jsin(wt), which in physical terms translates to mixing with a local oscillator output and low pass filtering.
reply
exe34
11 hours ago
[-]
Why would you want two carriers?
reply
Sesse__
11 hours ago
[-]
Twice as much information.

My go-to for I/Q is: Having two allows you to represent negative frequencies. With a normal, real signal, this is of course impossible (negative frequencies will automatically mirror the positive ones), but if you have a signal centered around e.g. 1 MHz, there's room for above-1MHz and below-1MHz to be meaningfully different. And _that_ allows you to get a complex signal (I/Q), once you pull the center down to 0 Hz for convenience of calculation.

reply
dsp_man
4 hours ago
[-]
Thank you for the suggestion. That's the point. I/Q introduced early gets too complicated. This foundation needs to be built up.
reply
ykonstant
11 hours ago
[-]
Only people with a low I/Q would misunderstand this notation!
reply
furgot
14 hours ago
[-]
Most technical writing is going to assume some familiarity with the discipline. If a reader encounters unfamiliar vocabulary in a technical article, they'd be well advised to look it up.

https://en.wikipedia.org/wiki/In-phase_and_quadrature_compon...

reply
mikewarot
12 hours ago
[-]
Yikes - why even mention the E and B fields? They aren't relevant to the rest of the article.

A few hours playing with Sine and Cosine generators in GNU radio can take you from book knowledge of I/Q complex signals into fully grokking it. You don't even need a radio, just your existing audio I/O.

reply
dsp_man
4 hours ago
[-]
I mentioned E and B fields so that the reader knows why we focus exclusively on sinusoids. Plus, linking the sinusoid to something we see in physics makes it more real.
reply
ygritte
8 hours ago
[-]
> existing audio I/O

I never knew there even is such a thing. Where can I find it?

reply
galangalalgol
7 hours ago
[-]
There is a source block for your mic and or audio in. That was one of the first things I played with to understand sdr. I remember seeing a strong tone on a waterfall plot that I could not hear, thinking it was an articfact, then looking at the frequency and realizing I wouldn't be able to hear it. Turned out it was a crt TV. That kind of dates the story. Fun to be had.
reply
myahio
8 hours ago
[-]
Are there any other high-quality sources for articles about signal processing and its actual application in hardware/software? I've taken signal processing classes at my college and while I have a good grasp of the theory I struggle with actual use case ideas, beyond implementing a simple fir filter on a stm32.
reply
MrBuddyCasino
14 hours ago
[-]
This reads like someone proficient in signal processing is explaining the core concepts to another person who is already proficient in signal processing.
reply
sevensor
8 hours ago
[-]
Exactly right! As somebody who’s spent a great deal of time with the discrete Fourier transform, I thought, “this article reads like it was written specifically for me.” I/Q modulation is new to me though.
reply
esafak
17 hours ago
[-]
Yet another thing from school I've never used in the software world.

By the way, QAM is (still) used in 4G and 5G.

reply
pythonguython
16 hours ago
[-]
Come be a DSP engineer. I take FFTs of IQ data almost every single day
reply
cycomanic
10 hours ago
[-]
Some variation of QAM will always be used in communication. As soon as you deal in with EM-waves, be it physics, engineering or even biomedical stuff you will have to deal with complex numbers, which by extension is dealing with I/Q signals. You probably don't need this for programming a server or a website, but it's indispensable for signal processing.
reply
userbinator
13 hours ago
[-]
Work on low-level software for communications, especially RF, and you will see plenty of this stuff.
reply
cycomanic
10 hours ago
[-]
Not just RF, also optical communications. Really, the only domain left where PAM transmission is used is baseband communication for electronics, and datacom for optics.
reply