Balanced VLF Tx

[Davor]

I did it again

Lo and behold, a PLL driven balanced oscillator that is locked to the tank zero phase at resonance. It uses a XOR phase detector to reduce noise. There is no point using a bang-bang charge pump phase detector because it is noisier and the phase is always within + and - 90° regardless of a frequency. It works perfectly with very low Q tanks as well.

The XOR detector locks to a quadrature signal which is provided by the flip-flops, and from there I also have perfect sources of quadrature drivers for Rx, Tayloe ... just about everything.

Just a small mind stretch ... with additional divider this contraption could provide a clock for a micro and make it tick synchronously with the rest of the rig.

I did not try it yet, but it seem to me as a perfect material for various off-resonance and loaded coil solutions as well.

I'm so proud...



The LTspice kit and caboodle coming in the next post.

[Davor]

And here are the LTspice files. Despite 4046 being a very complex device, this hierarchical model works beautifully and very fast. I think This one will be my future toy for various rigs. I made this simulation a bit too simple for the sake of speed and clarity. The switch is in reality a comparator, and the tank will have to be supplied with some kind of buffers, but the way you see it here simply proves the point.

@Sergey, this one works nicely with very low Q. You may try screwing it up completely, and the only result is a lower amplitude

[Aziz]

I'm so proud...

I'm proud of you too. Good job. As always.

Aziz

[WM6]

Davor, congratulation on new development.
Can you give us some more applicable solution?
It works very interesting in my LT-Spice simulator, but I have no idea how to use this in some real "off resonance" design.
Thanks in advance.

[Davor]

I'll come to that soon enough. The whole point is that this "oscillator" will continue working past the limits of every other creation ever built. It locks to the phase, while the amplitude is of minor importance here.

There are several approaches to turn this thing off-resonance, but the simplest is to put, say, 100k resistors in series with the comparator inputs (switch inputs on this schematic) and, say, 100p in parallel with input . It should still lock happily, but the amplitude of oscillation should reduce and become sensitive to just about any metallic object in it's vicinity. On the plus side, you should also have some level of discrimination by means of monitoring the phase lock control signal ... it is directly related to the resonant frequency of the tank, so ferrous objects should act opposite to non-ferrous.

Guess this will be a hart of my pinpointer.

[Davor]

My latest invention, a single ended PLL driven oscillator created mainly as a replacement upgrade for IGSL, IDX, SMW etc. detectors. While it is NOT balanced, it will still provide a great deal of signal purity that is ensured by amplitude stabilisation circuitry, and symmetrical drive, thanks to PLL. It has significantly reduced 2nd harmonic over all free running, and I dare say crystal controlled single ended designs. Here is the picture, and sims will follow soon

[golfnut]

OK

For IDX, IGSL.. it is not improtant to have a 50 50 duty cycle for the wanted signal.

As its a single nded det, 45 OFF 55 On wont give any different result.

If it was a balanced det (full wave demod) keeping a good duty cycle is more relevant as both conducting halves need to graft together on the zero crossings.

S

[Davor]

On the contrary, good behaviour happens only if duty cycle of these devices is precisely 50:50. Trouble with a varying duty cycle in SPST switches is that the difference in duty cycle transfers directly as floating of the reference. Because there is always some residual unbalanced Tx signal on input, this floating is interpreted by the rest of Rx as a false target because loading of the coil changes with proximity to ground surface.

So how about exact 50:50 duty cycle now that you know this?

Even better than just 50:50 would be a balanced mode configuration of any or both coils, so that the capacitively coupled unbalanced signal disappears, together with the 2nd harmonic that promotes IM and also floating of the reference.

Unfortunately most of the nowadays commercial coils are not built to support balanced mode operation, they are mostly single-ended, so there you have it - a 50:50 duty cycle Tx with high purity signal. In my vocabulary it reads upgrade.

[Verator]

Something theme is not supported for a long time. Loss of interest?

[Davor]

No. Eventually at the end of every venue you hit the wall of some kind. This one beautifully served it's purpose so far. The fruits of this development will be used in my ongoing LF project, and most probably some others. You are free to use any of these concepts and improve them if you like.

I made some improvements over the tank locked loop concept presented here for my own pleasure, but basically it originates from the design presented here, a kind of prior art. I'll post some of it eventually. My focus is somewhere else at this moment. It's spring. There's no snow. Ancient coins are bored to death already. They so badly want to come out.

[Aziz]

Oh yeah!,

this interesting thread led to the invention of the TEM V2.0 and Class-E-TEM transmitter.
And much more.
You see the benefit of collaboration?

Aziz

[Davor]

I'm considering adding a mechanism to push the oscillator from AB -> C bias once it starts oscillating. The benefits of this approach are twofold: lower consumption, and greater loaded Q of the tank. Simulations show about 10dB lower harmonics, most probably due to the higher loaded Q, and lower than half supply current.

[Davor]

I thought of something like this ... all harmonics below -50dBc, not bad at all.

[Davor]

One of my occasional musings ... a way to use a single ended oscillator to drive a coil in a balanced way.

Attached is a network that works fine in a simulator. Trouble with these networks is that they are supposed to provide a proper bias and simultaneously enable using a search coil with no bias current, while not influencing the resonant frequency by much. This solution can be seen as a tank in parallel with the left two ports of a Pi network comprising a search coil and two capacitors. At resonance this Pi configuration forces a coil into a balanced mode operation.

The resultant frequency response of this network reveals two peaks, of which the left one is lower than the right one. At the left peak frequency, both sides of a search coil are operating at the same phase, and a coil current is low. This would be mostly unfortunate outcome, but in a practical oscillator the right peak sustains oscillation and everything is nice and dandy (encircled by pink ellipse).

The proper peak oscillates only with the inductance of L1 up to the inductance of a search coil. Beyond that the left peak becomes dominant, and balanced operation is lost. Resonant tank with low inductance results in some significant currents in a tank L1C1 and the choke L1 must be designed well within it's saturation limits.

Motivation to obtain a nice choke is in twice the amplitude in a search coil (6dB gain), and also near to balanced operation. This eliminates using coils with one end connected to a shield, but otherwise any other coil could do.

[Davor]

Here is the example of an oscillator employing the above mentioned network, with only difference that L1 is roughly the same inductance as a search coil, and the capacitors C1 and C3 are joined into a single capacitor.

This circuit also presents AB->C class conversion by means of V- voltage that appears with oscillation (a clever reuse eh!) so once the oscillation starts the conduction angle is reduced to just a little bit below 50%, resulting in more economical, yet cleaner oscillation.

As mentioned earlier, this oscillator provides twice the amplitude of a simple single ended oscillator, and coil ends potentials are opposite against the zero, hence balanced... well almost - even harmonics are still here.