Precise Resonant Oscillator

Nice 👍. I'm wondering where the positive feedback is in your circuit. I know that a Colpitts oscillator and opamp oscillators take a fraction of the output voltage and feed it back to the input in phase. Does this work in the same way?

Thanks! Yeah it definitely works the same way, (through positive feedback)…. The output is the blue trace, which feeds directly into a voltage follower which is driving a series LC resonant filter, which is the input to the circuit. (The LC filter right where the grounded resistor is)… The resistor is necessary because it provides an easy path to discharge the capacitor through the neighboring inductor that shares a common node. In-between that series LC filter is the input of the common base amplifier, basically when the emitter goes low, the transistor turns on which pulls the collector voltage low too (closer to the same voltage level as it’s emitter). And visa-versa when the emitter is pulled high. The common base amp is a non-inverting amp, and suffers less from the Miller effect (the Vbe parasitic capacitance, which tends to short out the base emitter junction at high frequencies).

Thank you for the detailed information (considering that my knowledge of transistors and oscillators is VERY basic, your description still made a lot of sense!) I always wondered what the Miller effect was, too. I think I'll build this circuit as soon as I get a hold of some inductors.

Oh awesome! Really glad to hear that it made a lot of since! Also, I forgot to mention what the large capacitor tied to the base of that one transistor does, so I’ll do that now just in case you were curious about it too…. Basically the large 10K resistor charges that capacitor (ideally before the oscillations would start) and provides a flat (ideally ripple free) voltage level that is just enough to bias the transistor itself. Because the Capacitor gets charged through the collector of that transistor, it is the transistor itself that determines the biasing voltage (which will end up being as low as it can go without turning itself off, which would be roughly somewhere around 0.65 - 0.85 volts. This will end up being just the right voltage level (automatically managed by the transistor, through that collector feedback) to put it right in between full saturation and cutt-off…. A.k.A The active region (which makes amplification possible, rather than it just being on or off, it oscillates kinda between the two!). Lastly, the 10k resistor is set that high enough that it’s still kinda isolated from the voltage at the base, yet still low enough that it provides the transistor with enough current to operate as intended in this particular circuit design. There’s actually one more reason why you can’t go much, if any, higher with this resistor though, and it has to do with the Miller effect. The Vbe junction has a certain amount of capacitance right, probably in the single-digit picofarad region, we’ll what happens if you try to simulate that as say just a capacitor across a diode (grounded on one side; same end as the cathode of the diode), being biased with an AC source through a resistor? We’ll, pick a value, say 50kOhms, and scope the voltage between the resistor and cap/diode, and scope your AC frequency source, and start at a low frequency, say 1kHz, and watch what happens as you increase the frequency!!! Then try lowering that resistance value to something like 10k, reparation the frequency sweep, and then do the same thing with only a 100 ohm value, and the results will show you how it affects the maximum operating frequency of that diode, it’s crazy for sure, and it’s the exact same reason why circuit design at moderate and especially really and crazy high frequencies is so incredibly increasing difficult. You can imagine now, how do they do it with all the RF technology we have nowadays? It’s mind blowing, but the answer lies somewhere between a combination of very specialized transistors, and adapting much more creative circuit design strategies the higher you go in frequency. It’s wild no doubt!!!!

Oh also, would love to hear how it goes for you when or if you build that circuit, let me know if you do!

Swell 👍. I'll definitely let you know how it goes. Will this circuit work on a breadboard or should I use a perforated circuit board? I know that breadboards have a lot of stray capacitance which impact circuits in the high frequency range.

Ah you can use a breadboard no problem. I use breadboards for everything honestly. (Although I’d really like to try to learn how to design and order custom PCBs soon) but I’ve even built a 220MHz oscillator and a tiny tesla coil on a breadboard before, although it wasn’t easy or perfect, it did work.

I noticed the edit to the circuit. I agree the larger capacitor makes more sense. It struck me as odd that it was so small after reading from your comments that it provided a steady bias point. How come then the waveform is more distorted now with the larger capacitor?

P.S. I'm still waiting on my order of inductors to arrive.

P.S.S. If you're interested in making PCBs, I recommend Kicad EDA 6.0. It does have a learning curve, but I caught on pretty fast (better than Eagle EDA in my opinion). It is also free and has a 3d view option.

Yeah I didn’t pay close enough attention to the size of that cap before I posted it at first, Sorry for that confusion….

I think now that the bias point is more like d.c. it is pulling the emitter signal quite a bit higher, and so it’s not able to drop negative much if at all anymore like it would naturally as it resonates if it weren’t susceptible to clipping like it is now. As for reason of the distorted shape of the signal as a whole (like why it doesn’t look more like a perfect sine wave) I think all this is due to minor fluctuations in the feedback; for example, the voltage follower just feeds directly into the filter for feedback without any consistency in the feedback voltage and current waveforms, so some of it cuts corners I guess from the way the transistor itself gets hard biased/reverse-biased. Overall though, If you look at how the entire signal transforms throughout the circuit, it definitely looks much different in shape in other spots, but, due to the filtering side affects it smooths out fairly well by the time it reaches the output

Oh also I didn’t realize you had to order parts, you probably could’ve still made it work with other size L and C you have laying around, it’s fairly interchangeable with this circuit, with maybe a few minor exceptions just depending on what components you have to work with perhaps. On the PCB design, I have Easy EDA downloaded, and I’ve looked through it a little, but I couldn’t find a specific component that I ordered on there, I’d have to make a custom layout or footprint I think, but i had a hard time finding all the tools that I had seen in videos, not sure if I had a different version maybe. I’ll have ti try those other options you mentioned sometime. How do you do multi-layer though I wonder?

You can make this circuit oscillate at many different frequencies with different sizes of LC components with whatever you already have is what I was meaning to say.

That's the thing. I don't have any inductors on hand at all! 😄 It's OK though because I wanted to make an order of components anyway.

I know in Kicad 6.0, to make a multilayer board, you go to the physical stack-up menu and select the number of layers (supports up to 32, I believe. Crazy!).

Haha oh no…. You could’ve hand made one still though, right?! Yeah 32, that’s a little nuts, but super cool that’s an option 😁

My electronic order finally arrived. Yeah, I could have hand wound some inductors, but I wanted to get an assortment kit for future projects. Anyway, I built your circuit and it worked fine (the first time I ever generated a signal in the Mhz region!)

Here are the results: I probed a 3Vpp 7.6Mhz clean sine wave on the output. I used 330pF film capacitors, 1uH inductors and 2n3904 transistors. On my next trial, I switched to 2n2222 transistors. I got a much higher amplitude (12Vpp), but it was distorted to a rounded square wave with a spike at the beginning. Next, I switched capacitors to 100pF ceramic, but this time it didn't oscillate.

Interesting… I’m guessing the 2n222 has much different parasitics. I don’t know why it wouldn’t have worked with 100pF caps…. Did you make sure to replace both capacitors, so that they have the same value? Both sets of caps and inductors need to be tuned to resonate at the same frequency. Only other thing I can think of is possibly adjust the base resistor to a lower value the higher you go in frequency. And breadboard connections can be very finicky a lot of times, being a little loose one second, but not the next when you blink your eyes ;) Otherwise, I’d say check the transistors, you may have blown one of you switched to one that had lower power tolerances. Sometimes the small gauge inductors overheat easily too, so it couldn’t hurt to check those also

Swell tips. I cleaned up the wiring and checked out the 100pF capacitors. They where the same, but still no oscillations. I switched out the transistors. Still nothing. It's like it wanted to work, because I'd see brief bursts of oscillations that quickly faded out. Next, I switched to the larger capacitors (330pF) and it started working! I guess I don't know 😄. Anyhow, it's a fun circuit to play around with.