4

I`ve got a capacitive vibration probe whose base capacitance may vary from 4 to 10 pF due to manufacturing/installation tolerances, whereas vibrations it detects can only alter its capacitance in sub-pF range.

To translate capacitance into voltage I`m using an LC oscillator, since the probe is so constructed that one of its sides must be grounded and also the circuit must be insensitive to EMI. As a consequence, RC oscillators and capacitance to digital converters cannot be used.

The problem is, I want to get as little phase noise as possible with a lossy probe. I`ve got decent results with Vackář and Lampkin topologies but I want to go further. I`ve found an active Q-enhancement technique for inductors (see here) but I need to enhance the Q of the capacitor, not the inductor.

I`ve also found out that coupled resonators can reduce phase noise (see here) but I seem to be unable to find any examples of practical oscillators with coupled LC resonators except for the microstrip ones. I`ve made an attempt to design an oscillator with coupled LC resonators using this technique but it turned out to require relatively high value capacitors while I need to keep Δf/ΔC ratio as high as possible, so that didn`t work either.

Please help.

hidefromkgb
  • 307
  • 1
  • 13

2 Answers2

5

XY problem, you have what is effectively a capacitor microphone element, with one side grounded and you wish to build an RF Mic.

Were it me, I would not make it part of the oscillator instead using a crystal osc, and putting the transducer in a bridge circuit. Use a couple of fixed inductors and a piston trimmer or such (shunt it with a variable capacitance diode so you can servo the operating point to compensate temperature drift), then do phase detection on the bridge output voltage.

A couple of inductors, a piston trimmer, a varicap diode, a crystal osc, and some sort of phase detector (maybe a diode ring mixer?) followed by a lowpass filter and some opamp bullshit, job done.

Dan Mills
  • 17,266
  • 1
  • 20
  • 38
  • I don\`t quite understand which kind of bridge you mean. You\`re suggesting a quadrature detector with my probe in its phase shifting circuit, right? AFAIK the noise of the quadrature detector is reciprocal to the Q of the RLC network. Does that approach still give less noise than an LC oscillator? – hidefromkgb May 09 '18 at 21:45
  • Okay, I\`ve come up with a new method based on what you propose. Having added an AFC circuit I\`m now able to use a high-Q RLC network. – hidefromkgb Jul 08 '18 at 11:59
3

The HP3033 synthesizer used a VCO something like this

schematic

simulate this circuit – Schematic created using CircuitLab

Notice how the two VDD bypass capacitors have their RETURN nodes tied into the resonator RETURN, for quiet operation. Also, C5 is part of the resonant path, thus its AC current is affecting phasenoise.

You might install a Schottky diodes across R2, if you want a fixed Vout amplitude of 0.3v PP.

You should make C3 quite small, so only a portion of the resonant energy is dissipated in that base resistor.

analogsystemsrf
  • 33,703
  • 2
  • 18
  • 46
  • Can you please explain why this circuit is supposed to have less phase noise than the other LC oscillators? I simulated it and Q1 runs into saturation every oscillation cycle; doesn\`t that increase the phase noise? Or am I not fully understanding the circuit and thus doing something wrong? – hidefromkgb May 09 '18 at 21:51
  • 1
    the voltage swing across the collector resistor of Q1 will be 15 volts * 100/1000 or 1.5 volts, at the most, if the differential-pair has enough base drive to fully steer left-then-right. And reduce C3 to 100pF or 1nF. – analogsystemsrf Aug 24 '19 at 03:55
  • I\`ve finally come to a conclusion that the answer I originally accepted is unsuitable for me since the sensitivity gets modulated by probe loss, so I opted for the circuit proposed in this answer. Never seen such oscillator topologies before, so I started searching and stumbled upon the HP8662 bulletin where this osc was described. I modified the circuit a bit because I need higher output and the oscillation was stable despite all my attempts at destabilizing it with added loss. Thank you for your guidance. https://imgur.com/a/qjMCsZY – hidefromkgb Aug 31 '19 at 20:41
  • I'm delighted you can use this circuit. Its been in my bag of tricks for decades, but I've never needed it. Based on R1 and R3, can you predict the steady-state tank current and voltages? – analogsystemsrf Sep 04 '19 at 08:57