Verator front end

[Verator]

Need to work together to make sense of the actions in this directionI have a different approach to the issue of the TX generator. It is interesting to find out what impact the second harmonic of a signal response of the target. I see at this point is emphasized. What's so terrible?The use of any oscillating circuits vyzivaet deviation of the phase of the reference signal. MD is all the same measuring instrument, not just a coin to search. A reference phase deviation of the soil itself and from the various noise increases the noise level in the amplifier channel. The appearance of the phase jitterI took the path of a radical solution to the problem and the rejection of the principle of resonance. Recovery technology allows you to work in this direction. Although there are limitations.

The influence of the soil is sufficiently large. Conduct a simple experiment on a real device and real ground you can see the essence of influence.Applying the quartz frequency stabilization, synchronization, or switch to TX or from the generator can be seen completely different behavior of the device. At what Q of the circuit TX = 5 In larger Q worse.
Find the best solution - it's a great question. An external oscillator or excited TX. You've got to look for a compromise. What is used in a simple cheap MD-Commerce In expensive MD used recuperative bridge and half-bridge TX.
So I do not understand your approach to this issue.

[Davor]

The tank locked loop oscillator uses a tank as a phase reference, but otherwise a tank is connected as a filter. It differs from a typical oscillator in a way that it is driven by a symmetrical VCO, and that it's conduction angle is not skewed against the tank 0° when a loop is locked. In the example above, the tank is driven at precisely 180° (50% duty cycle) and in a bridge configuration it is driven + and - 180° (twice 50% duty cycle). For my own pleasure I made a configuration where 90° conduction angles are used (twice 25% duty cycle), with the idea to better accommodate for QSD switching front end.

As tank is in effect a filter, and 0° is a phase reference, it has very little 2nd harmonic, and it is very close to digitally driven Tx, but with high Q. Best of both worlds if I may. Additionally I can run VCO at whatever multiple of tank frequency, and thus extract digitally accurate phases against the Tx phase, and I'm very flexible in Tx coil design. I plan to use a tank resonance in a wide range of LF band.

In Verator a low Q is a means to avoid large phase excursions on Tx coil, and thus phase to amplitude modulation. In my approach there is no amplitude modulation even with very high Q.

Assuming you are involved in Verator development, there are a few details I simply can't fathom why they are done that way.
First are the phase references. Verator uses I/Q phases by means of phase shifters. It would make perfect sense if Tx was not driven digitally. I think digitally extracted phases, by means of flip-flops would provide punctual phase at exact 90° shift and 50% duty cycle, also consuming less power.
Second, the front end is quasi resonant, as in old Whites or Tesoro rigs. This configuration improves input noise figure by up-transforming the input impedance, but it is phase sensitive, and makes sense only in case of a resonant Tx. There are better solution to this front end, such as Cibola has. It would perform much better with a fixed Tx frequency.

Otherwise I like Verator approach to discrimination and ground balance. The schematic is very criptive though.

[Verator]

In this schema fragment Whites XLT. A similar scheme synchronization Egle spectrum. The quality is poor but the principle is important. It consists of HPF. filter that cuts off part of the influence of soil, its range of subsonic frequencies. 2 comparator controlled very simply. It's not in the younger model "6000" appeared only in these. The expensive devices and apparently is, a compromise.

Either the generator depends on the parameters of the circuit TX, its inductance. Inductance varies from the environment, the current phase of TX is shifting. Has an instant shift. PLL will be looking for a stable point for a while. Synchronization MD will also be some time resent.
This is a parasitic effect, which is then amplified in 1000
If there is a coil and a capacitor is in the circuit - there is phase deviation.
This practical results: put a coin on the ground from above. Change the synchronization of the sync detector circuit TX. All right, the device is working properly and recognizes the coin. Change sync detector of quartz, the device is not working properly, the coin is not defined. This means that the circuit TX with Q = 5 (!) Does not retain a stable phase. The amplitude of the voltage in the circuit does not change.
By itself, the amplitude is not important because the synchronization occurs at a transition sine 0V
Reference phase of the primary field TX should be fixed under any external disturbances. The amplitude is not critical to parents, it is the reference phase. It defines the parameters of the detected targets, the size of the vectors XY.
The high Q of the circuit in the oscillator does not yield immediate stability phase. That it's for. TX generator should be considered as a metal detector, not as the generator itself.
In the heat recuperation scheme only applies coil, there is no capacitor, no phase distortion. The quality factor of such an option TX is 20. But it is not oscillating circuit Q is Q generator (device).

[Davor]

I understand your point of view, except that it omits the fact that the Rx coil is also sensing the ground, and it's inductance is also changing as Tx does.
The Rx voltage is related to Tx voltage, only phase shifted and at value related to instantaneous balance mismatch. The eddy current response is superimposed to this, but ferrous response that relates to target permeability only shifts amplitude of balance mismatch. This one is invariant of the resonant / non-resonant approach. This means that amplitude stability DOES matter as amplitude fluctuation can be observed as a dynamic vector superimposed to cyclo-stationary carrier, picked by Rx as a false response. Furthermore, as Tx phase / frequency shifts (they are basically the same thing) the Rx tightly follows by means of phase shifters, so in such system any phase deviation of Tx is invariant as the Rx demodulators follow.

The tank locked loop can deviate only a little in phase, even if harshly banged around by drastically changing the Tx coil inductance. Remember, a Tx coil is a filter, and it does not oscillate on it's own, so a phase deviation is ALWAYS within + and - 90°, but practically much less. In locked conditions it is fast following any changes, so the phase deviation is negligible. There is no way to make this device lapse a phase beyond 90°, as there is in fact only one oscillator, a VCO.

You can imagine a tank locked loop as a crystal oscillator driven Tx, except that the crystal is magically following the Tx tank resonance.

[Verator]

These questions have already been raised many time ago. But were ignored.

If we talk about the RX. and here it is impossible to apply the resonance! What's the point then? The receiving coil should have a linear phase. In general, the sensor must be linear phase.For a device with a phase notch (Tesoro, etc.) is irrelevant phase and amplitude stability. There it all is forgiven. But in the devices vector type Whites. phase stability is crucial. Not namely amplitude phase. Because the amplitude of the changes, both vectors XY. subtracting the same proportional change in the channel, these G. Phase deviation will change the ratio of the vectors and the inability to subtract the interference of the soil or in the channel G.You and I have a different understanding of the issue. For you to sample the instrument conditional Tesoro, for me, a different type, vector.Phase methods clippings / discrimination suffer interference. But there is worse than the accuracy of the detection, the soil is not deductible, it is ignored.In the vector method, the soil is subtracted. But there is a phase noise of ground which is much larger than the noise. He is very big! Especially when changing the sensor type. The approach to the issue should be integrated. And the reference phase TX fixed. Then, the receiving system must provide a linear signal reception target / soil. But if you go on the principle of resonance, it will not work. RX coil can not track the phase as well as the TX. They are completely different and are in different positions relative to the ground and the object. They have a different impedance.In systems such as sensors Whites Blue max. receiver line, no resonance.
Going back to the PLL would like to remind that in order to stabilize the loop tracking takes time. The loop can not operate in real time. She is always late. This is what raises questions about the influence of soil on the signal. Do not need it for the sake of noise and to compensate for the soil. This is the most important.
Any perturbation TX PLL will track it also is interference, as Sync all derives from the same PLL. The fact of the phase change of position is unacceptable. It can not move at all!

[Verator]

I agree with your conclusions and methods. This is really a normal subject. I will check it as much as possible.You did not try to bridge PLL?

[Davor]

The Elektor circuit is a fine catch, I was not aware of that one. There are a few problems with Elektor Tx that I can see, primary in sense of off resonance phase sample. But in essence it is very similar to my tank locked loop. Funny thing is that I also made some experiments with simulating this principle in off resonance, and it has potential for a good pinpointer.

...in order to stabilize the loop tracking takes time. The loop can not operate in real time. She is always late...

It takes very little time. You must have realised already that Verator, and all other rigs with crystal controlled Tx are not tracking anything, let alone a ground influenced phase. To avoid conversion of phase to amplitude all such rigs, including Verator, use low Q, but as tank locked loop follows the phase I can use high Q and minimise problem even further.

Regarding phase noise, it is essential that a loop filter is faster than the channel filters and that removes such noise from detection. High Q of the tank also minimises phase noise away from resonance. The best part - it is not that important if there is no phase to amplitude conversion.

IMHO the benefit of driving a Tx coil with a symmetrical exciter is in absence of 2nd harmonic, and it by far exceeds the phase related problems.

Please see Tesoro Cibola Rx front end for inspiration to further reduce phase deviation in future Verator. It can be improved by a few transistors to reduce input noise, but it shows nicely that a series resonance at Q<0.5 provides a much better solution.

[Davor]

You did not try to bridge PLL?

See the post #196. My initial intention was a bridged balanced configuration, but I realised that the existing rigs could benefit from a single ended configuration. Most of the commercial coils support single ended Tx excitation. Balanced is better.

[Davor]

Anyway, if you wish to go further and develop this kind of Tx exciter - together with phases for Rx, I'll be glad to assist you. I made some progress with these, but I did not see much interest and I kept it for myself until I get to my LF project Tx exciter. There are a few more new details that ensure exact 0° matching, including an integrator in a PLL loop.

[Verator]

I have all these ideas of self check and see for yourself and make conclusions. If you have questions ask for help. No interest in this to the people because few people even do it. Basically, make ready, for whom it was once painted patterns found on the Internet. 90% did not realize that.
I understand the problem of the stabilization phase and the ground effect on the quality of discrimination

While the best solution is a bridge-stage recovery. He gives high Q coil. No "C" only "L". But the problem with this device in the fact that the spectrum is dirty and the problem with increasing frequency, quality factor too falls with increasing frequency. In the oscillatory circuit on the contrary, the frequency increases Q increases. But with the help of the bridge you can get a current in the coil at low frequency 2-3 kHz can not be done in the circuit LC. These principles complement each other.
All the questions I check almost no simulators.
Integrator in the PLL loop I set 25 years ago, only on that occasion.

[Davor]

Yes, I understand. I mentioned the integrator in a loop just because I did not put that design here, but done it for me only.

I'm afraid you'll not be able to neutralise all the phase fluctuations, so discrimination will always suffer a little. It will always be a bit fuzzy on the border. Therefore I adopted a technique with my IGSL that uses overlapping Fe and Cu discrimination tones. So all the fuzzy discrimination happens in the area where both tones are triggered simultaneously, and it works surprisingly well.

I'll keep this approach in my future LF build too.

[Verator]

Here I am saying that there is nothing better than quartz. Resonance Systems TX will not be able to work steadily on the ground. Once the bridge and meander in the coil.IGSL and similar systems with phase tenderloin is not able to improve the quality of detection. No suppression of the soil, it is ignored. Everything is very simple. This is a common resonance system. It has a high phase distortions. The use of an integrator in the loop PLL will increase the accuracy of the retention phase.

[Verator]

Why experience for 2 harmonica? What is it terrible? When excited by a meander of harmonics is much bigger and stronger. ... 1 = 80% of the rest of 3-5-7

[Davor]

2nd harmonic, as produced by regular oscillators causes asymmetry about the zero. When such signal is digitised by means of a comparator, it's duty cycle is different from 50%. Moreover, the 2nd harmonic content changes twice the rate a fundamental frequency changes with loading, in effect producing PWM offset. It's phase also changes twice as fast, and it is related to the oscillator conduction angle. Since Tx and Rx are running on the same phase difference all the time, phase differences in Tx are compensated at Rx switches for fundamental frequency, but 2nd harmonic, being twice as fast, does not.

The easiest way to introduce all these problems to detection audio is by half wave switches. 2nd harmonic is a powerful false signal.

With Tx and Rx running on the same phase difference, some frequency or phase changes will not matter, and that's why devices with free running oscillators work fine.

Easy fix is just replacing the single ended regular oscillator with a balanced one, or a single ended with forced 50% duty cycle.

Verator has no such problems because of precise 50% duty cycle Tx. It also uses full wave switch in Rx.

I still don't understand why Verator doesn't use digital phase shift. Nor do I understand a quasi resonant Rx front end.

[Verator]

The digital shift is tightly bound to the TX. Lack of flexibility of choice relations XY. At low frequencies, 2-6kHz have to shift the phase of the signals for low-conducting objects ... Nickel, shallow gold alloys. The ratio is not 0 \ 90Using the shifted resonance receiver RX feature allows you to get. how to filter LPF. and align the phase. You can even do without resonance but worse sensitivity and a lot of garbage. For example the receiver coil in the sensor Blue 950 max. has 400 turns and is set to 4,6 kHz. TX circuit has Q = 5 (30 turns of aluminum) 6,6 kHz.Now remember this is a vector unit working in statics. Conversion function of the target signals of all nodes device must be linear phase, Resonance is a bad phase. As a result - bad signal vectors, poor balance of soil. In simple devices such IGSL no phase linearity (linearity and it is not necessary). So always say this is the ground he so strongly influenced by the depth decreases. But the influence of soil were concluded in the ground and in the instrument of his abilities.