Radio Design 101 - RF Oscillators (Episode 4) 

You're very welcome. Thanks for the complement 🙂

Thank you!

Awesome. I inherited teaching this course from others in the department back in the 1990s. The combined lecture/lab/build paradigm is definitely the way to go (at least at the upper level classes). Have also created a lecture/lab course like it for antennas and microwaves and I'm working on an "Antenna Briefs" series to support that too (at least the antennas part of it). Thanks again for the feedback on how the videos come across. Very nice to hear 🙂

Awesome, thanks for the feedback. Watching a few times is definitely needed for some of this. Glad it is helpful !

Great to hear! Thanks :-)

Awesome. I'm glad to hear the series does that. Thanks for the excellent analogy with jigsaw puzzles. I hadn't thought of it that way ! I don't know if you've seen it yet, but there's a companion website for this channel. It has all the slides used in the videos as well as some info on "RF circuit prototyping boards", and raw notes from the course from which this series was derived. If you are interested, see: ecefiles.org/ Thanks again for the encouraging words! 73

@@MegawattKS Yes, I've seen it and downloaded some of it into a binder. Just had open heart surgery this week so I'm running a little behind lol but anyway God bless you sir

@@Mark_KE8YCV Wow. I hope you're feeling better and get well soon, and have a happy holiday / Merry Christmas!

You're welcome. Thanks for the comment !

Glad you like them!

Thanks ! Slides for next one are done. Just refining them now...

Thanks!

Awesome. Glad it was helpful!

Thanks! Glad you liked it.

You're welcome. Its nice to hear that it is helpful !

Thanks for the comment. The slides are available at ecefiles.org/ Hope they're helpful.

There are a number of performance issues for both transmitters and receivers relative to phase noise. For high-dimension digital modulations (e.g. 256 QAM like in cable systems), it impacts receiver sensitivity and error rate floor as well as spread the constellation in transmit mode. More generally it can cause "blocking" problems in receivers from nearby strong interferers or cause spectral pollution from transmitters. But to be honest, except for things like high-dimension QAM, it's impacts are often over-rated and its just one of those things that products compete against each other with (like noise figure in a terrestrial receiver which is arguably not very important given the elevated noise floor from terrestrial interference). Sorry - that's a long answer...

Thanks. I think the answer to Q1 might be yes. However, it depends on what you're trying to do and what the FM radio requires (and of course getting to the appropriate circuits). The TinySA has a maximum amplitude of about - 7 dBm when it's in output mode. Mine (and maybe all of them), can generate either fixed-frequency CW or swept signals. The NanoVNA (at least the one I have) can also generate CW outputs, and they are stronger: about +2 dBm below 300 MHz. Neither unit has any ability to add a Marker to the sweep in the traditional way (briefly increasing amplitude at specific frequency). But they're synthesized, so you can always set them to whatever frequency you want to investigate closely and do that manually, perhaps. Sadly, nothing is going to help with stereo tests. That requires a 38 kHz DSB modulation in the audio plus a 19 kHz pilot (also in the audio). The TinySA can create wideband FM (its less than +/ - 75 kHz deviation, but still wide in the technical sense) with audio modulation. But there is no audio input and the internal audio is not pure sinewave and only works up to something like 5 kHz. Hope that helps.

Good observations and questions. Yes - its a homebrew board created for the class use. There is a ground plane on the bottom and yes - it's just 2 layers. While the lines are technically microstrip due to the underlying ground plane, the frequency is so low that it's functionally just a PCB with the ground plane being used to limit parasitics and EMI issues. The wavelength at 100 MHz is 3 meters in free-space and about 1.5 to 2m on the PCB (depending on trace width to thickness), so traces are generally 1/100 wavelength or less on the board and the controlled impedance aspect of microstrip isn't needed.

Interesting question. There are a couple of reasons I can think of (but I confess to not having really considered this before). For example, crystal oscillators can have good stability and good phase noise (because their Q is very high) even at low power levels in the amp the oscillator is based on. Lower C values mean higher impedance, which means generally lower current in active devices. Hence low power operation. It's also possible that the crystal frequency is most "accurate" (relative to what its marked as) at a particular C value, whereas LC oscillators are looking for the highest Q value and C will be varied anyway. On the question of the oscillator amp being driven into class C. Yes - you are absolutely correct. The active device (e.g. transistor) is definitely switching on and off over a cycle and the LC tank circuit takes the pulsing current and smooths it into a nice sinewave :-) How deep it goes into class-C mode depends on the starting loop gain - which is usually at least 2 to guarantee the oscillation starts. It turns out that the signal level with feedback is limited. Once the amplitude builds up and the current waveform becomes pulse-like, the amplitude in the fundamental frequency falls and the loop gain settles to one.

@@MegawattKS Thanks for your comments. It's great to find material on oscillators. Books that I have read in the past have always avoided oscillators and mixers. Your videos are invaluable. In the future maybe you could do something on PLLs? I'm particularly interested in the filter design i.e the choice of pole/zero frequencies in a simple RC filter. I read application notes in the past but they only deal with PLL filters for a capture/settling time perspective. I'd like a practical design methodology that looks at keeping the phase noise low in ham radio applications. An approach based on the well established CD4046 would be ideal. Mike (in Germany)

@@mikewhelan5992 Here is another good book and it seems to cover the things you're interested in. The Design of CMOS Radio-Frequency Integrated Circuits, Second Edition 2nd Edition by Thomas H. Lee www.amazon.com/Design-Radio-Frequency-Integrated-Circuits-Second/dp/0521835399 . The Amazon page also has a "Look Inside" where you can see the table of contents and the very extensive index. The sample text and title are a little misleading. It says "CMOS" in the title, but its really just about radio circuits. And the sample text doesn't go far enough into the book to show how it covers circuit design. It is only the radio pre-history part. The actual technical material is pretty detailed later in the book and covers up through oscillators, plls, phase noise, etc, It's written at a university level, but I like how Lee has a history similar to my own in ham radio, and comes at the subject from an understanding that goes well beyond the math alone. 73 Bill

@@MegawattKS Thanks Bill, Looks like a very good book. It covers all of the topics that interest me. I must pick up a copy. Mike

Yes - that's the basic value of the total C resonating with L1. However, that assumes the output coupling capacitor ( shown but not labeled) is small and/or it is driving into a very high Z load. Otherwise, that capacitor and the load will have an effect on the frequency (or even stop it from oscillating). Assuming that output coupling C is small relative to Ctotal, then the formula you gave should get you close. There are 2nd order effects also from Cbe and Ccb in the transistor - but they're usually negligible if we design as described in the video and the transistor used has an "f_T" value (max operating frequency) sufficiently high, which is the case for something like the class project's MMBR5179 transistor used in an FM radio design.

A better answer is probably to just make Cc small compared with Ctotal. Just keep in mind that the smaller you make it, the smaller the output signal level will be...

Sorry for the late reply. Unfortunately no. The course involved a lot of lab work and was never offered in that format. Now that the NanoVNA and TinySA became available, that could possibly change in the future, but I don't think the school has any current plans along those lines for that course. Thanks for the interest though. I'll pass it on to the current faculty. Again - sorry for the delayed reply.