EEVblog Electronics Community Forum
Electronics => RF, Microwave, Ham Radio => Topic started by: w2aew on June 23, 2017, 12:26:25 am
-
I posted a new video today about IQ modulators. The video presumes prior familiarity with IQ signals and the operation of diode ring mixers. If you're rusty on that, you can view my "prerequisite" videos on those topics:
https://www.youtube.com/watch?v=RHFZUqUM8DY (https://www.youtube.com/watch?v=RHFZUqUM8DY)
Here are the two pre-requisite videos:
https://www.youtube.com/watch?v=h_7d-m1ehoY (https://www.youtube.com/watch?v=h_7d-m1ehoY)
and
https://www.youtube.com/watch?v=junuEwmQVQ8 (https://www.youtube.com/watch?v=junuEwmQVQ8)
-
Thank you! It's a great continuation of the subject...
-
Thank you :)
-
Nice! Thanks Alan!
:clap: :clap: :clap:
-
BTW, can you use the setup shown to modulate say 16QAM (or higher QAMs)? What would be different then, just what you feed through the I and the Q? More levels than just 'high'/'low', I assume? I may be answering my own questions, but various light bulbs are going off here. If this is so, the magic may be in building the I/Q generator capable of doing all these combinations of multiple levels. Fascinating...
-
w2aew:
In the parlance of the youths, you're really killin' it. Nice job!
BTW, can you use the setup shown to modulate say 16QAM (or higher QAMs)? What would be different then, just what you feed through the I and the Q? More levels than just 'high'/'low', I assume? I may be answering my own questions, but various light bulbs are going off here. If this is so, the magic may be in building the I/Q generator capable of doing all these combinations of multiple levels. Fascinating...
I really don't know the answer to your question, but thinking about what specifications are available and what specifications change as you go up to higher and higher QAM configurations, I think this is just a noise issue. The bandwidth (of the mixers and the spliiter/combiner which) doesn't need to change as you go to higher QAM constellations, provided it is enough anyway. The Mini-Circuits I/Q demodulator he uses (ZFMIQ-10) has a bandwidth of only 2 MHz, which you might think is insufficient for, say, QAM64 or higher. But, for a constant bit rate the bandwidth required for higher QAM encodings decreases (https://electronics.stackexchange.com/questions/178696/what-is-the-required-bandwidth-for-qam-modulation)!
The inputs to the I and Q would need to have more levels (as you state). I think the real figure of merit in keeping a higher QAM constellation making it all the way through to the receiver is a combination of the losses at every stage (e.g. conversion loss for the mixers, insertion loss for the splitter/combiner, insertion loss for any cables) and some sort of noise related figure, and/or the sensitivity of the receiver and its capability to demodulate signals which are very small (just above the noise floor of the receiver).
I remember from The Signal Path videos that Shahriar tested various Tektronix RTSAs by attenuating the constellation all the way down to only a few dB above the noise floor of the RTSA, and it still worked. After the attenuation became so severe that the RTSA was having trouble demodulating the constellation diagram, he would reduce the QAM scheme down a step and he could attenuate and successfully demodulate still further.
-
BTW, can you use the setup shown to modulate say 16QAM (or higher QAMs)? What would be different then, just what you feed through the I and the Q? More levels than just 'high'/'low', I assume? I may be answering my own questions, but various light bulbs are going off here. If this is so, the magic may be in building the I/Q generator capable of doing all these combinations of multiple levels. Fascinating...
Yes - as technogeeky stated, higher levels of QAM would simply mean more levels on I and Q. 16QAM would be 4-level (PAM) waveforms, which would give you the 16 combinations. In this case, you'd get 16 combinations of Phase and Amplitude. QPSK uses the same amplitude at each symbol, just a different phase. 8PSK would be 8 different phase states, and so on. There are lot of different variations that are common. OQPSK (offset QPSK) uses the same I and Q as QPSK, except that the Q signal is offset in time by a 1/2 symbol period. This ensures that the I and Q signals don't cross through zero at the same time, which means the RF carrier doesn't go through zero either, which reduces the peak-to-average ratio of the signal, making the job of the RF amplifier design easier.
Higher levels of QAM require better linearity in the modulator, and better SNR of the signal, because the symbols get closer together in the IQ plane, making it more difficult to demodulate. In modulation parlance, higher level QAM signals have less tolerance to EVM (error vector magnitude), which is a measure of a symbol's deviation in location in the constellation compared to it's ideal location.
-
Thank you Alan (and technogeeky)! I've watched all previous I/Q dedicated videos, but this last one made things click together. A few more questions if you don't mind:
- If I have a stream of bits what is the cheapest way (for education/experimentation purposes) to turn them to I and Q for a 8PSK, QAM, etc? Any relatively easy schematics or chips (google returns mostly the RF modulation details explained here already). Or using older generators from eBay like https://www.rohde-schwarz.com/us/product/amiq-productstartpage_63493-7564.html (https://www.rohde-schwarz.com/us/product/amiq-productstartpage_63493-7564.html) (what other generators do this)?
- If I have demodulated the I and Q components of an RF signal what is the cheapest way (for education/experimentation purposes) to recover the bit stream from 8PSK, QAM, etc. modulated signal? Or display the constellation diagrams for that matter? I do have access to Tektronix RSA306 (without the $1K demodulation SignalVu option) and Agilent E4406 (this one seems to be able to demodulate some preset signals only; or probably I don't know how to drive it). Or relatively easy schematics/chips?