# Single supply opamp colpitts oscillator

I'm designing a Colpitts oscillator with a single supply rail-to-rail opamp powered with 3.3 V. The positive input of the opamp is set to VDD/2 to offset the negative part of the oscillations so everything stays between 0V and 3.3V.

I have to make a few design considerations and I'm wondering which one is best and why.

What would be better and why, to attach the capacitors to the virtual ground or the real ground? Both circuits work (simulation) but I wonder which is best, I have a small preference to the second circuit because then the virtual ground opamp doesn't need to deliver much current.

Virtual GND = VDD/2

Real and Virtual GND combined

If your virtual ground is stable, it makes no difference. Your two capacitors behave the same, regardless of the DC bias they're subject to.

Since you say it isn't, go for the second circuit. In fact, since practically no current should flow into an opamp input, you can omit the voltage buffering second opamp alltogether, too. And: 100 Ω is far too small a resistance – at the femto- to microamp currents that might flow into an Opamp input, a 10 kΩ - 10 kΩ voltage divider should work just as fine; replace the gigantic 10 µF cap with a smaller one when you increase the resistors, as that will otherwise take "eternity" to charge.

Generally, don't forget this is just a simulation with what I presume are idealized component models – in real world, a 100 µH coil has some DC resistance (simply because it's not made from superconductor), so read that from the coil's datasheet (measure if you're just using an inductor you've got lying around without datasheet) and adjust your series 30 Ω accordingly.

Same goes for your opamp: it has some finite bandwidth, i.e. high changes in phase at high frequencies. I'm too lazy to calculate your Colpitt's frequency, but most opamps should be save at these frequencies; still, it's rather uncommon to build oscillators from opamps, where simple transistors would do (and probably has nicer phase behaviour at higher frequencies).

• Thank you for your answer. I thought about a using an op-amp as it is a recenter component so "it must be more stable and capable". But please tell me if I'm wrong or why you would prefer to use a transistor. It will be used as a inductance meter for an inductive loop detector (vehicle detection) so it has to be able to oscillate properly with a wide variety of inductors. Also when not using an op-amp there is still a choice to be made between a mosfet and a transistor. Could you give me a few arguments on why you would consider a transistor or maybe mosfet? Commented Jan 3, 2019 at 14:21
• as said, phase margin is the main aspect here. If you want to have a stable oscillation for measurement purposes, you'd use no Colpitts Oscillator at all, but a crystal-based one, typically; for measurement purposes, you'd, in fact, usually use a PLL chip with an integrated voltage-controlled oscillator (VCO) that you discipline with said crystal; that way you can get a range of reliable, clean, yet adjustable oscillations and don't depend on the properties of some random capacitors and even worse inductors over a range of temperature. Commented Jan 3, 2019 at 16:04
• I understand what you mean, but the way it is measured is using a oscillator built around a inductor. In which the inductor is the coil placed under the ground where the vehicles will pass. As the inductance changes, so will the frequency and this is easily measured with a microcontroller. Commented Jan 4, 2019 at 8:27
• Ah, nice! What are the frequencies such systems operate at? Commented Jan 4, 2019 at 11:46
• Frequency is very depended of the used coil but mostly between 30-200kHz. Also the oscillator should be able to be changed in frequency range by adding/removing a capacitor. And thus be able to work with different capacitor values. This is used for multiple loops near each other so they won't interfere each other. Commented Jan 4, 2019 at 12:12