Induction Balance Stuff - Single/Multi Frequency Response, GB, Disc, Measurements, Ideas, Fun, etc.

[Sergey_P]

Hi Aziz
Sound BlasterX G6 -> https://reference-audio-analyzer.pro....php#gsc.tab=0
OK
What is high Q-factor used for? Is it preferable to focus directly on the signal from the target, rather than on changes in coil parameters caused by ground effects?

[Aziz]

Hi all,

I have found a ferrite ring core (Do=26 mm, Di=14.5 mm, H=20 mm) to try the step up signal transformer.
I don't know the core material anymore. But the AL is approx. 12000 nH/N². Could be a bad core material with high losses. I don't know.

I have to build a nice tool to wind the secondary windings on the ring core. It is difficult to wind up to 100 or more turns on the ring core. The secondary windings will be partitioned into several sections to lower the parasitic coil capacitance.
I will try to keep well below the resonant frequency (load capacitance C2 + inductance of the secondary coil winding of the signal transformer forms an LC resonant circuit).

At resonant frequency, there would be much much more compensation voltage generated. Ring core losses at resonant frequency could cause heavy temperature drifts too.
Let's try the save region first and look at the temperature drift. Could be a knock-out issue.

Aziz

[Aziz]

Look,

there is a good solution for the AC current limiter on start-up procedure without triggering the output current protection circuit.
Found here:

https://www.linkedin.com/posts/linear-integrated-systems_linearsystems-fetdesigns-jfet-activity-7156544323050729473-CbPC

Nice.

[Aziz]

BTW,

the Sound BlasterX G6 has an output current protection circuit built-in. That's the reason, why I have a high inductance primary coil for the signal transformer. On initial condition, the high current through the primary coil could trigger the current protection circuit. And the sound card switches totally off.

But there is a software work around. Start with 0 V output voltage and increase it steadily to the desired output voltage level during 5 - 10 seconds maybe. This should avoid the current protection triggering. And we could reduce the primary coil inductance at the end.
A series resistor R3 (10 Ohm) helps too. But it has also a damping function to the signal transformer transfer characteristic.
Aziz

[Aziz]

here is the LTspice file for your convenience, playing, tweaking, adapting & arranging ..

[Aziz]

Hi Sergey,

I was referring to my basic & simple solution. Your configuration differs too much from my.

Below is the schematics I am referring to:

Even the source coupling capactor C1 (1n) is quite high, it is degrading the Q down to approx. 30. Still enough to work with. I can push up to almost +/- 500 mA to the TX coil (with only 17 mA source current).
The compensation mixer (C2 + C3) would attenuate the target response 50% and with input impedance R1 (1Meg) and base current of J1 would attenuate it slightly further.
So the j-fet amplifier amplifies the difference signal and does the high input -> low output impedance conversion (Lin impedance approx. 10 kOhm). So the target signal isn't further attenuated.
The sound card Sound BlasterX G6 is operated in the high-gain mode (delivering up to +/-6 V peak voltage).
The signal transformer is 1:10 in this example.

This shows only the basic concept. Fine tuning must be done yet.

Aziz

[Sergey_P]

Hi Aziz
Prototype->
Power supply: 5 V
Half-bridge driver (minimal noise level,+/-2.5V ), series resonant circuit
Coil: diameter: 20 cm; 415 turns; resistance 190 Ω
At a resonant frequency of 8 kHz: Quality factor = 20, peak voltage = 63 V, field = 7 Amp-turns, current = 17 mA, power drawn from source = 26 mW
Signal-to-noise ratio: ~ 150 dB (limited by the Nyquist limit within a 40 Hz bandwidth)
Is the compensating signal identical?

[Aziz]

Hi Sergey,

Hi Aziz
Even the introduction of residual imbalance compensation in DD (-54...60 dB) leads to an increase in noise and a decrease in sensitivity.
<The high-Q TX should not be degraded>
... and the reference signal.
what SNR? (spectrum analyzer, THD... ) desired noise -120dB in the passband

we don't need such a perfect imbalance compensation. Just enough, that the amplified/buffered signal is in the input voltage level (max 1 V peak).
The reference signal should have the same SNR as the stimulation TX signal (approx 120 dB I think).

If we have 30 Vp TX voltage and 30 Vp reference voltage, we would have 50 % target signal loss in the mixer.
If we have 30 Vp TX voltage and 60 Vp reference voltage, we would have approx. 33 % target signal loss in the mixer.
If we have 10 Vp TX voltage and 60 Vp reference voltage, we would have approx. 14 % target signal loss in the mixer.

But we want a high TX voltage (=high TX current), to increase the SNR too. So we need a high reference voltage. As much as possible. So we can operate the TX at its max. voltage.
Aziz

PS:
The target signal isn't really lost. It is only attenuated. The amplifier could compensate the overall mixer attenuation.

[Sergey_P]

Hi Aziz
Even the introduction of residual imbalance compensation in DD (-54...60 dB) leads to an increase in noise and a decrease in sensitivity.
<The high-Q TX should not be degraded>
... and the reference signal.
what SNR? (spectrum analyzer, THD... ) desired noise -120dB in the passband

[Aziz]

Hi Sergey,

It can be made as much sensitive as a DD coil configuration now.

No
Noise(mono coil) = Noise(TX)+Noise(Vcomp), / Noise(TX)<->Noise(Vcomp) - are not correlated /
Noise(DD) = Noise(TX)-Noise(TXcomp), / Noise(TX)<->Noise(TXcomp) - correlated /
mono coil VLF - It works, but very poorly

I haven't had time yet.
To increase the sensitivity of the mono coil VLF, I have to wind a good low loss signal transformer yet ( < 50 mW, it's still too much, < 20 mW would be better).

The idea is using a simple high impedance voltage mixer, buffering / amplifying the difference (j-fet input) and then processing the residual target response. Effectively subtracting the high TX voltage by the high reference voltage. The high TX voltage (let's say up to 60 Vpeak) is being capacitive mixed with a high voltage reference signal (let's say up to 60 Vpeak or even more if it is possible). The high voltage reference signal is being generated on the second line output channel and transformed up with the good signal transformer.

The mixer is causing some target signal loss however (Ohm's law). The higher the reference voltage compared to the TX voltage, the lower the TX coupling mixer capacitor and hence the lower the target signal loss.

I have a reference signal up to 6 Vpeak on the second line output channel. And this will be transformed up as much as possible. The more, the better.
But I don't know, what is really possible. The signal transformer is quite critical. I would be really happy to get 60 Vpeak out of 6 Vpeak voltage.

The high-Q TX should not be degraded. So the signal path is a high impedance path till the j-fet buffer/amplifier.

The AI gives me good tips to wind the signal transformer however.
Aziz

[Sergey_P]

It can be made as much sensitive as a DD coil configuration now.

No
Noise(mono coil) = Noise(TX)+Noise(Vcomp), / Noise(TX)<->Noise(Vcomp) - are not correlated /
Noise(DD) = Noise(TX)-Noise(TXcomp), / Noise(TX)<->Noise(TXcomp) - correlated /
mono coil VLF - It works, but very poorly

[Aziz]

Hi all,

instead of the projection method for ground balance, there is a more elegant and efficient way to do it.
In the I/Q vector space, you have the unity ground vector G, obtained by ground balance procedure. |G| = 1 (unity vector)
Your signal S is also a vector in the I/Q space.

The vector cross product P of the vectors G and S is:
P = G x S (vector operation, P is also a vector)
A = |P| is the surface area (scalar) of the vectors G and S (forming a parallelogram). |P| is the lenght of the vector P.


Reactive response will show always towards the direction of ground vector G. And hence, A is becoming zero.
Anything different than zero, A is your target signal.
A threshold for A is for manual GB adjustment.

Any objections?

[Aziz]

Hi all,

I haven't built any new coils yet. Minelab's Z-Set coils made me busy.
The recently bought tinned litz coil cable is good and I have desided to buy more other cables: 1 mm² and 0.75 mm²

For the IB coil stuff, I want to design a new easy to build induction balanced coil. For this purpose, I need lot's of coil cables.
Maybe a Class-E transmitter too.
I hadn't enough time to make any progress on the High-Q mono coil VLF yet.
Cheers

[Aziz]

Hi friends,

my 100 m long 1.5 mm² tinned litz wire came yesterday. This is a nice coil cable. 2.6 mm diameter total thickness. I can make the final flat spiral (pan cake) search coils soon.
But I will focus to the mono coil VLF design first. This gives us more detection depth potential.

What I can say is, that we really need active circuit design (high impedance buffer/amplifier). Active nulling or a difference amplifier is mandatory to get maximum out of the mono coil design.
First of all, we have to look, where the target signal gets lost or attenuated. And how we can increase the SNR.

I don't have much time right now however..
Cheers

[Aziz]

Hi all,

I don't have much time at the moment.

There is free lunch and SNR improvement available.
The mono coil VLF should work fine without any amplifier with active nulling. It can be made as much sensitive as a DD coil configuration now.

More next week..