UHF Colpitts Oscillator in Class C mode problem using Harmonic multiple

Question

I want to increase the frequency of the crystal oscillator oscillating at 25MHz frequency. For this, I added an LC tank circuit to the collector in parallel with each other. I experimented to see if the frequency would change.

But, even though I tried many times for capacitor and coil values, the frequency never changed. The amplitude changed, but the frequency remained 25MHz. Why did the frequency not change? Where am I going wrong? Here is the circuit I designed.

I replaced the coil in the collector with a coil with 3 or 5 turns of winding. But, still the frequency did not change. I played with the capacitor values, and the frequency did not change. When I removed the capacitor from the tank circuit, it oscillated again at 25MHz without any change in the signal. The presence or absence of the capacitor did not change the result. Below is an article about this method.

http://www.koreascience.kr/article/JAKO201509057414006.pdf

In the article in the link, the crystal oscillator increased its frequency to 2 or 3 times. How did he achieve this? I built a similar circuit. But, I could not increase the frequency. In the circuit in the article, using an inductance at the emitter, your circuit. I added a coil to the emitter, but still the result did not change.

Maybe someone can't open the article. I am attaching a picture from the article.

Although I soldered the circuit on the PCB, I could not increase the frequency.

---

@Tony Stewart EE75 @glen_geek

I have done trial and error many times. I changed the eelman values of many circuits. I have been dealing with this for 1 month. No result.I read whatever I found on the internet to see if something was missing. I even have a lot of information about the history of frequency.The fact that the real experiments do not match with the ones in the books makes me very tired. It is very easy to say "look, this experiment gave this result" in simulation. It makes me sad that I can't set up the circuit and get results from the experiments.

Here is the actual circuit I built, the circuit diagram and the oscilloscope image.

/****************************************/

edit:17/02/2022 I built the circuit on a perforated plate and soldered it. In the tank circuit, I used 1uH coil and 10pF capacitor. In this case, the oscillation frequency of the LC tank circuit is 50Mhz. But 50 Mhz oscillation did not occur. The output is still 25 Mhz. Therefore, soldering the circuit did not work either.

Answer 1

Why did the frequency not change. Where am I going wrong?

The tuned circuit added in the collector is resonant at 27.7 MHz. This is due to: -

$$f = \dfrac{1}{2\pi\sqrt{LC}}$$

And, the basic crystal oscillator running frequency is 25 MHz. The whole idea is that your tuned circuit in the collector is meant to pick up on harmonics of the basic crystal frequency but, 27.7 MHz is not a harmonic of 25 MHz.

If you look at the circuit in the linked article, they are tuning to 4x the crystal frequency: -

In other words they are tuning into the 4th harmonic of the basic crystal frequency and, as I said earlier 27.7 MHz is not a harmonic of 25 MHz.

Answer 2

There are 2 critical parameters in harmonically tuned resonators.

• Q or Zo and n*fo.

• The impact on choice selection is the tolerance of the passive components and harmonic content of gain and attenuation in the fundamental oscillator at the desired harmonic.

• If Q is too high from the R/X(f)=Qp for the parallel resonant circuit then the value of LC is critical hit or miss. with 1% to 20% typical range then you have to add or subtract capacitance with a varicap or trimp cap to tune it. This sensitivity can be reduced by shunting the collector current sink with a parallel R. But this also affects the attenuation of distortion from more harmonics. So 100 is a reasonable limit for LC tuning Q unless extreme selection or tuning is used with temperature compensation.

• It is also critical to have short leads when you get down to < 0.1uH as wire is ~10 nH/cm and put an RF decoupling Cap across the supply close to the oscillator.

• Here I simulate it with a voltage source and a series R, where you can see the effects with a mouse-wheel tuning R over its value.

For a simple tuning take your 3.3 uH and /4 [uH] for 2f or /9 [uH] for 3f or /16 [uH] for 4f then use very tight construction and short leads. But each time you increase n x f the Zo reduces and you must shunt the collector with a suitable R for Q<=100 or the inverse of your tolerance error for tuning.

• the choice of transistor also has an effect with GBW or current gain hFE-BW product, but your choice of SS9018 is an excellent one. Worst case loop gain >1 is your goal.

Conclusion

Tolerances and Q (tuned by your R2) are critical tradeoffs, as well as layout improvements.

• Possibly all you had to do was change your load from 10k to 1K or less to pass the signal thru the harmonic filter

I assume you know all the equations for Q, f, and X(f). Refs: https://www.cs.ccu.edu.tw/~cwlin/courses/electronics/notes/CH13.pdf https://www.wikiwand.com/en/RLC_circuit

Other info

• Saturation and cutoff may prevent square waves by limiting gain and false oscillation on 3rd harmonic, but that is prevented by the LC filter.
• Most Xtals purchased >> 20 MHz are actually harmonic frequencies or an overtone crystal, guaranteed by filtering.

Answer 3

It is unclear what you're trying to achieve from the various schematics you posted. Are you trying to pull the crystal frequency or use a harmonic of the crystal frequency. You can only pull a crystal a small amount. Normally, you choose a harmonic of the crystal frequency.

In the top schematic, the center frequency of the RLC circuit is \$ {1 \over {2 \pi \sqrt{LC}}} = 27.7 MHz\$. With the extra capacitance and wiring inductances, the resonant frequency is probably closer to 16MHz.
The lower schematic is in the 100s of MHz.

Some things to watch out for. In the red box, there are two capacitances to watch out for. One is the breadboard and wiring capacitance. On a breadboard, the capacitance between rows can be significant. You also have lead inductance which is significant when you use inductances under 1uH.

At the frequencies you're operating at, get rid of the breadboard and either prototype this on a perf board or dead-bug style (no board is used). Trim the lead lengths to minimize lead inductance.

Add an emitter follower buffer on the output to minimize the scope probe capacitance on the tank's center frequency.

Answer 4

Here is an example of what you search for, but be aware that VHF multipliers are "difficult" to "work".
These are found in the old amateur radio "schematics".
See the lower part of this schematic (x3 used). And this.
I use an "old" transistor in this schematic ...
And A "breadboard" is not an option.
Simulation may be "long", very "long" ... and not "representative" of success.
For some reasons, multiplication factors are generally odd numbers (1,3,5 ...).

Here is an amplifier (first stage) at 25 MHz, the second stage is a 4x multiplier at 100MHz.

A more complete spectrum view analysis shows the waveform at output and a view of the spectrum content. :-) --> filtering needed.

And AC analysis

Answer 5

Q1 is very much in the class A mode due to R4. Class A is renowned for low distortion which means low harmonics compared to the fundamental. Place an experimental resistor from base of Q1 to batt neg to starve Q1 pushing it into more nonlinearity meaning more harmonics which should be easier to find. If Q1 starts class AB and runs class C it would be better.