Balanced VLF Tx

[Davor]

I'll not go too deep into the WHY a balanced Tx. Whoever thinks it is not needed, well, most probably does not need it at this point.
I think I do.
My primary objection to non-balanced Tx-es is that they generate excessive 2nd harmonic, which in turn spoils timing at the switchers, and that enables various problems entering the Rx on the back door.
Unbalanced Tx at the coil also spoils the coil induction balance by capacitive coupling (it rhymes )

Anyway, I see 3 possible approaches, two low tech and a high tech:
- a digital source with perfect 1:1 timing supplied to a resonant tank by two counter phase bridge sources
- a balanced oscillator ... there is a vast number of papers generated about the cross-coupled oscillators and their superior phase noise
- a digital source with phase sense at the tank to enable frequency steering of the digital source

I think I have a nice contender for the second solution, a balanced oscillator.
Now about the cross-coupled oscillators. They generally require a center tapped coil (but not in my solution), and they resemble a two transistor multivibrator. Trouble with these is that you must supply DC and it is usually achieved via center tap. I gave it a second thought and concluded that the center tap is not required if the cross-coupled drivers are configured in H-bridge. It fixes the DC problem. Additionally, if the capacitor is done "center tapped" I reap the fruits of the balanced design WITHOUT A CENTER TAP. It additionally cancels a 2nd harmonic.

Stability of this design relies on the stability of the rails voltage. Power consumption is very little, and the best of all - it works happily with large resistance coils as well.

IMHO any serious VLF design should have a balanced Tx source.

Here it goes:

[Davor]

I tried some sims with L at 300uH and caps at 2u with corrections to R3 and R8 to 33ohm, and it works of course. Coil current is ~1App, no 2nd harmonic and 3rd harmonic lower than -50dBc. Nice.

[Sergey_P]

Qtx = ???

[Carl-NC]

Interesting... almost exactly the same ckt I'm using on a new design. Sorry, no further comment.

[Davor]

That's understandable, patents, IP and stuff.
Cross-coupled oscillators are a logical step in VLF metal detectors, and they are extremely well documented in other walks of the art, mainly VHF to microwave VCOs. They are mostly built using mosfets, and their superior phase noise is very well documented. In fact there are at least two models explaining their good behaviour. Configuring them in a H-bridge was kind of obvious. True, I did not find them in this configuration yet, but most probably someone else already did that. Even if this is new it still falls under the "obvious" category.

Let me guess, your configuration employs mosfets

Qtx = ???

Try the above attached LTspice sim and see for yourself. It is extremely flexible. The coil that I put in the schematic has a realistic resistance for the coil of that size, in fact I used measured values of my own coil. Using a 4 times lower inductance means half the turns, and a double wire cross section, e.g. a quarter resistance, so 300uH is equally feasible. If coil Q is a problem, just lower the R3 and R8 to increase overall gain and you get a nice rail to rail oscillation again.

Below is a solution with all lumped elements with the coil resistance in series with the coil, and without the initial DC point solution (as if you applied voltage at 0s, and it was 0 before that). It draws about 12mA for ~1App output. Try the zipped sim

[Sergey_P]

It draws about 12mA for ~1App output

Qtx ~ 30
very bad!
Qtx < 7

[Davor]

Bad what? Please elaborate.

[Aziz]

Hey Davor,

good find. I like it.

This TX cct is well destined to the sound card processing application as well. You have to use the TX reference on one input channel (using a voltage divider) and the (amplified) RX signal on the other input channel using the stereo input channel of the sound card.
Even if there is any phase noise of the TX cct, the lock-in amp demodulation from the TX reference is very easy and very very stable.
The sound card output channels can be used to beep the detection signal out. So you would need only one sound card for this application.

Cheers,
Aziz

[Sergey_P]

high Q - good for PP
for IB - low Q (the phase and frequency stability in the presence of soil)

[Davor]

I think your reasoning is perfect for external oscillator driving a tank, such as with Verator rig, where ground Δμ results in Rx phase change.
Considering Tx and Rx coils sharing the same environment, whatever Δμ affects Tx coil will certainly affect Rx coil as well, so their mutual phase change will compensate in case of a Tx being a part of a Tx oscillator tank, thus following the Δμ change. With oscillations limited by the rails voltage, even amplitude will not change. As for phase noise, again a higher Q is beneficial. I think this kind of oscillator is a clear cut choice for VLF MD.

[Sergey_P]

this kind of oscillator is a clear cut choice for

…PP and for children's toys «VLF MD»

[Davor]

OK, I'll give this a second thought.

[Carl-NC]

Yes, I've done it with both bipolar & Mosfets. Actually, I'm using bipolar right now. It is nicely fuel-efficient & linear.

In a previous life I designed a similar circuit on a GSM receiver chip to run at 2.4GHz. It used Mosfets with tuning varactors. I recall achieving -106dBm/Hz phase noise but don't recall the offset, 1kHz I think.

[Sergey_P]

ideal Tx (Rx) for IB MD - is 90 degrees phase
Resonance - is not optimal. Compromise: Qtx <7 and Frx> Ftx

[golfnut]

Thats an elegant circuit Davor - its good if you can keep the Tx coil <2ohms.


I have a electronic speed controller from model aircraft motor - there is some similarity here in the drive bridge.

You can buy them for £5 - you get the mosfets, interface bipolars, pcb, a small processor you could reblow and use the AtoDs on it and make a simple £5 detector.

S