# Low Pass Phase Shift Oscillator

I have come across this circuit configuration of an oscillator on a older metal detector.

Searching around I have not been able to find any examples of a similar circuit in the wild. The output is attached to an obsolete amplifier that feeds a search coil. As drawn above the circuit generates a faint signal, but it is being lost through the amp. It looks like there is some issue with either the choice or sizing of components I am using to replicate this. Has anyone worked with a similar configuration before?

• Looks like a very conventional phase shift oscillator to me. Presumably the output buffer is tapped from the stage with the output voltage the designer wanted; he could have chosen another for higher or lower voltage. Is there a question here? – user16324 Nov 3 '17 at 16:05
• It's just an ordinary phase-shift oscillator; the only odd thing is that the output buffer is attached to a point 2/3 the way through the phase-shift network -- for minimum distortion? Also, for some reason, Q2 is configured as an emitter-follower, so there's no voltage gain there. It's probably meant to immediately drive a stage with high voltage gain. – Dave Tweed Nov 3 '17 at 16:07
• @DaveTweed The signal has the lowest distortion at that point in the network. (Nevermind, I see you already guessed it - that was a sneaky edit!) MrPeanutButter: I just simulated the thing in LTSpice using BC547C transistors and it oscillates happily at pretty much exactly 1kHz with a peak-to-peak amplitude of about 100mV. Sounds fine to me, so your problem is most likely with the amp. – Jonathan S. Nov 3 '17 at 16:13
• Thanks Guys, The PCB this was lifted from is pretty beat up, So my first assumption was that I was reading something on it incorrectly. – Mr.PeanutButter Nov 3 '17 at 17:17

Here's the same circuit redrawn a bit to help emphasize the phase-shift oscillator format more clearly:

simulate this circuit – Schematic created using CircuitLab

I've isolated the battery supply over to one side, since it is more of an ancillary bit than a central part of the circuit. It helps to move that away, leaving the rest to examine alone. Now the three stages are brilliantly clear, too.

I've also shown the output more clearly, perhaps, as an emitter follower that taps off at some point in the phase shift chain (midway between the collector and base of $Q_1$.) If I wanted just a little more output swing at the emitter I'd probably "improve" this design with a very simple change:

simulate this circuit

Using some of the very low impedance output of $Q_1$ to help drive some positive feedback into that particular node that is also loaded by $Q_2$'s base.

My own preference is to make an oscillator using high-pass (which is actually always really a bandpass due to parasitics) approach.

simulate this circuit

The above circuit will also be close to $1\:\textrm{kHz}$. Output swing will likely be about $6\:\textrm{V}_\textrm{PP}$ with a $12\:\textrm{V}$ battery and about $4\:\textrm{V}_\textrm{PP}$ with a $9\:\textrm{V}$ battery. It's output is much lower impedance, as well, as well as having a much larger and (less distorted) output.

This circuit keeps $Q_1$ out of saturation (which helps avoid serious sine wave distortion), provides very simple methods for controlling gain, adjusting the set point for average $V_E$ and $V_C$ and it works more reliably (my opinion.) Depending on the $\beta$ of $Q_1$, adjust $R_5$ up or down to the next nearest standard values. Lowering $R_5$'s value is needed for lower $\beta$ transistors (conversely, increasing it is required for higher $\beta$ transistors) but lowering the value also pinches the BJT margins and may move the BJT towards saturated behavior (which means more distortion.)

[Note that if you are using a rather low $\beta$ NPN BJT for $Q_1$, then you may also (in addition to lowering $R_5$ to help properly bias it) need to reduce the values of $R_1$ through $R_3$ and perhaps increase the values of $C_1$ through $C_3$ and perhaps provide a modest decrease in $R_6$. Some or all of (1) more bias current by reducing $R_5$; (2) less loss in the $180^\circ$ phase shifter per this note; and (3) a little more voltage gain by reducing $R_6$ may be required with such BJTs. The converse also would apply to a high $\beta$ NPN BJT for $Q_1$, of course.]

$R_4$ and $R_6$ can also be adjusted independently.

In this circuit, the "natural frequency" is of course found when the high-pass phase shift is $180^\circ$. With three RC stages in series, each set to values such that $f=\frac{1}{2\pi R C}$, the phase shift will be about $128.6^\circ$. It takes a shift in frequency down by a factor of about $2.1$ (where $X_C\approx 2.1 R$) or so to reach the necessary phase shift and damping. So you can roughly compute the expected frequency as $f\approx \frac{1}{4.2\pi R C}$.

• Nice explanation. – rhody Nov 11 '19 at 18:38