Group Actions
Engagement
  • Author Author: strb
  • Date Created: Date Created
  • Views 1472 views
  • Likes 11 likes
  • Comments 6 comments
Related
Recommended

Oscillator design

strb

I was surfing the net looking for inspiration to build something new to me when I stumbled across Building a Voltage-Controlled Oscillator (VCO) by  .

His blog post inspired me quite a lot, but also showed me the amount of work and test needed to build something like this and since I have absolutely no experience in this field, I took a step back and decided to build a simpler oscillator.

Going back to university lessons, I brought back the one where we talked about oscillators and used it as base to design a Colpitts oscillator. Added a couple resistors for DC biasing my bjt and started designing it. Since it's just for learning, I don't have any stringent requirement. I'm shooting for an oscillation frequency around 10MHz and an output waveform that could resemble a sine.

After a couple iterations on paper I was happy enough with the result. After a quick verification with LTspice using a 2N2222 I was ready and impatient to build my circuit.

imageimage

Powered up the circuit and... nothing happened. The output sadly wasn't showing any oscillation.

Back to the paper (and the simulator), I realized I made some rookie mistakes:

  • First of all, 2N2222 used for the first simulation is nowhere similar to the D882 I'm using in the actual circuit. Using a similar bjt the simulation is correct and does not oscillate
  • Feedback capacitors value is too small: I miscalculated loading effect on feedback network thus obtaining the wrong feedback ratio
  • I did not include in my analysis the attenuation due to resistor network to bias my bjt instead of an ideal current/voltage source

After these corrections, I obtained this scheme:

This circuit no longer has the resonant frequency I initially hoped, instead it should resonate around 4MHz, which for me is "high enough". C3 in the scheme could be easily removed, but it was there from the previous revision and I did not remove it.

image

image

It works! Looking at the waveform in the time domain I was quite happy with the result: 500mVpp with an oscillation frequency just under 4MHz and no major distrortion.

FFT unfortunately tells me another story, with a calculated THD just over 4%. I guess it's fine, but I was hoping for a slightly better result.

Just for fun I decided to try measure the oscillation frequency using a DIY near field probe I made a couple years ago.

image

With the H field probe I was fortunate enough to be able to pick up a very faint signal, just enough to display first and second harmonic with the FFT.

image

image

image

Overall it was great fun and I've learned (once again! :) ) that it's better to not rush things (like I've done in the first revision of this circuit) just because I was impatient to see if the circuit works.

  • in reply to shabaz

    Thank you for your suggestions! And you're right, second harmonic is about 26dB down.

    I looked at BFR92P and indeed it seems very very nice, I will probably buy some of those in my next order. BF199 looks nice too but I usually prefer SOT23 compared to TO92 package Slight smile.

  • Nice work! 

    That's actually good results, if you've got your second harmonic down by what looks about 26 dB. It's hard to optimise, it just takes experimentation, swapping out components etc, because LTSpice is only partially useful for oscillator circuits. Component books with loads of pF capacitors helps a lot, just sitting there experimenting all day! The 2N2222 will struggle to reach 10 MHz oscillation, due to limited fT value. The old BF199 NPN transistor is good for experimenting since it is TO92 package, but unfortunately it's hard to find now (it is probably available from ebay etc but they are probably not genuine which doesn't help!). Anyway since you're comfortable with surface-mount, the SOT-23 BFR92P is really nice, well worth getting a few of those for experimentation, and Farnell has loads of them in stock currently although I bet they often run out of those. You'll easily get that to function for 10 MHz operation.

  • in reply to colporteur

    I finished my electronic degree in 2019 and since then I work as electronic design engineer (but nothing that I would strictly call analog design). I think that is safe to say that I'm still a newbie Slight smile.

    I agree, Shabaz project is great and well explained

  • What is your background? What was your educational discipline? What are you doing now? My career after 17 years heavy in electronic support and design took me away to computer systems administration for 20 years. Tens years ago I started reviving the skills and knowledge. I would never rate my knowledge and skill level with a Shabaz. You picked a good project, he is still around to support it.