Cobra search coil

[mikebg]

Cobra search coil for KT315.
Already one year has passed since KT315 announced for the first time that it requires Search coil for King Cobra. Since he again mentioned that today, we tried to understand what that particular coil and looked for pictures to get data based on which we can design and make a coil. We found two photographs in the web from various sources to make sure that the perceived irregularities in one of them are not random. These irregularities are:
1. Used clevis is too narrow (less distance between the ears). This poses a risk of clevis breakage and it certainly bents at waving.
2. Metal bolt is used to tighten the clevis. This creates a parasitic signal in RX winding, which prevents the proper induction balance.
3. The notation "15Hz" on both coils clearly refers to the operating frequency of 15 000Hz. This frequency is too high for use on the salty beach because it creates too strong GND signal. (To be continued)

[kt315]

thank you mikebg for open need for me topic.

there were Gold Mountain King Cobra, Cobra II and GMT-1650. first one works on 15 kHz. you may have seen my detaily report on XLT coil but with nothing of data we get only the nothing. seems Cobra coil is rare as rare are all of these and for me - be out of conus sale and this craigslist - to find original coil is hardly doubly. I have Hot Head coil on the detector, huge coil, but it is not original.

happy hunting

[mikebg]

KT315, you makes advertisement of Bulgarian beach!

Since mentioned above irregularities should necessarily be avoided by following models, we have sought information about the
search coils in the newer models of Gold Mountain Technologies.
Meanwile, on one forum we found a modification made by
Paul (Ca). He moved the trimm pot GND balance in control panel. However his picture of the search coil shows that not
only the bolt is made of metal, but also the large wing nut to it. (to be continued)

[kt315]

from my photogallery...

[mikebg]

The new model search head is "Cobra Max"

The name of the next model search head is "Cobra Max". From the attached picture we can extract the size of the middle diameters to design TX and RX windings. Note that as we have predicted, the operating frequency in new models metal detectors should be decreased, but this is not made enough because the frequency 12500Hz is again too high for 250mm TX coil.
Note that in Cobra Max, the distance between ears (thickness of clevis) is increased. Except that, the bolt and its wing nut are made of plastic.
Then we found the circuit diagram of King Cobra, because we have to determine the equivalent capacity of the TX tank and RX tank. Their values are needed to calculate the inductance of the TX and RX windings inside sensor head.
However, the circuit diagram disappointed us with the incompetent design of the TX and RX front end. We assume that the designer of the circuit shown to the right is Jack Gifford (founder of the company Tesoro) because it uses the same schematics in several models of this company.
The drawbacks of the shown TX circuit are:
1. The designer has not done a mathematical analysis of the effectiveness of the TX coil. If an analysis was made, would have found that at given (limited) weight of TX coil, the TX tank should have large equivalent capacitance, for example 1uF, to obtain a great number of Ampere-turns pro kilogram wire. In the shown circuit the equivalent tank capacitance is about 20nF only, ie we should increase it 50 times!
2. After the analysis would have found that the circuit of Colpitts is not suitable for effective TX oscillator in metal detectors, because if we use 1uF tank capacitance, we need an additional capacitor with approximately 10 times more capacity, ie 10uF. This capacitor is not desirable because of its large dimension. It must be used an oscillator circuit that does not require two capacitors in series.
3. The 2.2k resistor in emitter path decreases dramatically energy efficiency. For better efficiency, the resistance of emitter resistor should be reduced about 50 times!
4. For the most accurate and fastest stabilization of TX current amplitude is required P-I-D controller. The used by Jack Gifford P (proportional) controller can not achieve enough accuration because of too little gain of DC amplifier that can be achieved with only a JFET.
5. Earth and salt water act as a core for TX coil changing its parameters. The bad thing is that when the core is moving relative to TX coil, TX current receives unwanted amplitude and angle modulation and the AIR signal appears modulated as the GND signal. Modulation is greatest when the TX tank is part of oscillator circuit. The used controller can suppress amplitude modulation, but for best results and impedance matching, TX tank should be load of an efficient amplifier driven by a separate oscillator.
6. TX coil should have a tap for impedance matching of load. Capacitive tap is not suitable because of the high value of capacitances as described in point 2.
Now let we see how are these 6 problems solved in TX designs of other companies? (To be continued)

[mikebg]

There are solutions

The mentioned TX problems attributable to King Cobra circuit are solved by other companies as follows:
1. Large capacitance of TX tank.
White's uses several capacitors in parallel to form equivalent capacitance about 1uF inside search head. Fisher forms capacitance 0.44uF with two capacitors 0.22uF.

2. Different from Colpitts oscillator circuit, able to operate with large TX capacitance.
White's uses in Classic and Coinmaster 3900 an oscillator circuit with negative resistance as shown in the attached circuit diagram. Fisher also uses similar TX circuit in FRL in 1225. In model FRL 1210, the company uses TX coil with center tap, connected in Hartley oscillator circuit. The tap increases efficiency of TX coil because of better load resistance matching.

3. Low ohm resistor in emitter of TX transistor. This is done in the shown circuits.

[mikebg]

"Dismodulating" controller

4. P-I-D controller for AM "dismodulation" of TX.
It is not known a company to use theoretical required P-I-D controller just for amplitude stabilisation, but Garrett uses P-I controller as shown in the attached circuit digram. Note that an adittional capacitor C4 is connected in parallel to TX coil. It increases triple the equivalent tank capacitance. The circuit is specific case of Colpitts oscilator where capacitive tap is used to reduce the influence of transistor parameters on tank circuit and also the capacitance of feedback capacitor C6. The (R) EMI group wanted to design TX "dismodation" by a quick acting P-I-D controller, but after measuring the P-I controller of Garrett, it appeared that it operates enough fast at modulating frequency 2Hz, so this task was dropped.
The attached in right circuit diagram shows TX circuit with P (proportional) controller used in Fisher FRL 440 to "dismodulate" the TX amplitude. However the gain of this DC amplifier is very small in comparison with DC gain of IC integrator used in Garrett circuit.

[mikebg]

Driven TX

5. TX tank as load of an efficient (class C) amplifier.
White's uses such circuit in TM808, Coinmaster Pro, Eagle etc.
Fisher also uses an amplifier as end stage in TX circuit of FRL 1265 as shown. Note that it has lower resistance in emitter 33 ohm instead 330 ohm in White's. However, without amplitude control these circuits can not suppress amplitude modulation and AIR signal acts as an additional GND signal with the same modulating frequency but having different phase. The shown circuit of Spectrum operates even worse in low power mode. The shown circuit of Fisher operates with reduced efficiency of TX tank because is added an 1 ohm resistor to produce reference voltage for synchronous demodulation.

[WM6]

..........., it appeared that it operates enough fast at modulating frequency 2Hz, so this task was dropped.

You mean

..... it appeared that it operates not enough fast at modulating frequency 2Hz, so this task was dropped. - ? -

[mikebg]

You mean

..... it appeared that it operates not enough fast at modulating frequency 2Hz, so this task was dropped. - ? -

At fastest swing of search coil, the modulating frequency of TX (caused by change of height) not exceeds 2Hz, that's why Garrett circuit operates well, despite the controller has no derivative function. But let I continue with the last drawback:

6. Tap in TX coil for impedance matching of load.
All "push-pull" circuits have better TX efficiency because of improved load matching. They use TX coil with center tap. Note that each "push-pull" circuit can operate nonsxmmetric using only one transistor as Hartley oscillator. The second transistor can not increase the power. but reduces even harmonics. The Hartley circuit shown in posting # 6 also has better load matching because it also uses a tap. The load of a TX used for metal detecting is the resistance of TX coil. However for metal detectors is not important the power dissipated in this resistance (it is radiated as heat, not as EM wave), but ampere-turns of TX coil. The method with tap of TX coil is so efficient that a German trader of metal detectors Mustafa Uzman
www.uzman-import.de
started to improve machines sold by him and even file a patent pending for this. Note that in circuit for patent pending, there is an excess capacitor (pos. 6). The USPTO refused to issue patent for this.
The left circuit is my idea for powerful TX intended to operate as a box of two box machine. It is powered by two laptop batteries and amplitude is stabilized by PI controller. I will test this circuit when I have time.

[mikebg]

Search head as "Cobra Max"?

There is something interesting for KT315 in the above mentioned site of Mustafa Uzman.
The attached picture shows a metal detector that it offer only for 79 euro. He insisted that he can increase with his "invention" the depth of detecting with about 20% using modification in TX circuit. Note that the search coil of MD318 seems exactly as the coil "Cobra Max" shown in posting # 5. We should search WEB for parameters of metal detector MD318 and similar to it (and information for their China producer).

[WM6]

Excellent analysis.

Thank you mikebg.

Do you plan to build "mini" two box device with small distance between boxes or as usual?

[mikebg]

WM6,
I hate the term "two box" because it is misleading. For me, none of loop configurations shown below is "TWO BOX" because they are induction balanced. "Two box" means something very bad for balance - variable mutual inductance between TX and RX coils because they are not fixed.
However you gave an interesting idea to use small "almost orthogonal loop configuration" for deep coinshooting. An Italian company (CEIA) uses such small loop configuration with variable distance between loops for mine detecting (see the figure in the middle). Induction balance is adjustable in the field by operator as in Gemini of Fisherq but it can operate as two box (separate boxes).

[mikebg]

RX coil and RF amplifier

Ideal designed RX coil and RF amplifier

The term "ideal designed" is used when the designer is not restricted at practical realisation of his project. In our case let we asume that he is not restricted by weight of RX winding. For ideal design we can calculate RX coil with a not realisable value of coil resistance.
Let's calculate the ideal RX coil resistance.
Given is that the frequency of received from target signal is f=15kHz for King Cobra.
As we kmow, the bandwidth of received signal is no more than BW=12Hz and the BW is independent from TX frequency. It follows that we need a low noise narrow band RFA (Radio Frequency Amplifier). I'm using the radio term RFA because it amplifyes the frequency radiated by transmitter.
King Cobra has an incompetent designed RFA. Unfortunately this is valid not only for this model, but for most MD designs. I named the RFA circuit "TESORO" because it is widely used by this company. For comparison, below is given RFA used by White's, operating at different carrier frequency f=6.59kHz, but the necessary bandwidth is also BW=12Hz max.
Let's see the difference:
NOISE RESISTORS:
a) Thermal noise is increased at Tesoro by adding two noisy resistors R1 and R3. They act as a 10.2kohm noisy resistor.
b) In RFA of Whites there is only one noise generating resistor R4 and its resistance is below 1 kohm. We know that 1 kohm resistor generates noise density 4nV/rtHz (as an IC NE5534 or MC3307.
IC NOISE:
a) LM308 used in RFA Tesoro has noise density 32nV/rtHz. For military variant LM108 is given parameter "Noise Popcorn" 40uVpk at BW=10Hz - 5kHz and at source impedance Rs=100k.
Other given parameter is "Noise Broadband" 15uVrms at BW=10Hz - 5kHz and at Rs=0.
b) MC33078 used in RFA Whites has voltage noise density 4.5nV/rtHz and current 0.5pA/rtHz.
BANDWIDTH:
a) RFA Tesoro amplifies all frequencies. The high pass filter C2-R5 is connected after output pin 6. That means at pin 6 we have amplified not only AIR an GND signal, but all kind interferences including mains frequency and its harmonics and all noises including flicker and popcorn. This limits the allowed gain of RFA because it should not distort the processed signal. Flicker and popcorn noise are very important because are used primitive synchronous demodulators. They can not suppress these noises.
b) RFA Whites suppressts frequencies below 4100Hz with C1 and R4. Additional suppression makes timeconstant C8 and equivalent of resistors R9-R11.

[Mrand]

The discussion is interesting and going in the very good direction. Set up however from the detector CobraKing. I send the part "of the true" schema KingCobra.
The oscillator is very strange. The input amplifier is also atypical. The all this detector is atypical, but works sensationally.
Mrand