RELIC HAWK

[mikebg]

This is continuation of post #965. Attached is the SPICE file of posted image.

4. Suitable amplitude detector.
To maintain amplitude of TX constant, the controller needs information for amplitude of oscillation, ie it needs an amplitude detector. When the circuit has symmetry (push-pull), to suppress more the second harmonic, the amplitude detector should operate as full wave rectifier (see the circuit in post #955). How this is designed?
Relic Hawk:
There is no amplitude detector. The information for amplitude is obtained because the P-I controller integrates the negative half-wave. An information error will occur when the negative half-wave is distorted. This is an incompetent design. The ripple of controller output decreased when I formed an amplitude detector by connecting a capacitor after rectifier diode (see C3 in post #937, C4 in post #941).
Fisher 440 and project VLF770:
There is an amplitude detector formed by transistor U1c, C6 and R11. However there is no second diode connected to other lead of TX coil as illustrated in the post # 961.

5. P-I-D controller.
The P-I-D type control can suppress amplitude modulation most quickly and most accurately.
A proportional controller can*t maintain the amplitude accurately. For more accuracy it should have very high gain.
The P-I type controll operates accurately, but not so quickly as the P-I-D controll. However the P-I-controller operates enough well for metal detecting because no need of fastest setting.
If you need more knowledge for P-I-D control, please visit
http://www.ecircuitcenter.com/Circuits/pid1/pid1.htm
How the controller is designed?
In Relic Hawk:
SPICE analysis shows that P-I controller used in Relic Hawk operates enough fast, ie it has enough small setting time.
Fisher 440 and project VLF770:
The P-controller has low gain. It can't suppress enough the amplitude modulation of TX.

6. Suitable reference voltage for amplitude comparator.
I avoided generation of reference voltage by using single supply opamp for P-I controller. In this case the TX coil is connected to its negative supply rail (see posts # and #).
When TX coil is connected to positive supply rail, I use the Vf of a LED as voltage reference. Depending on color, a LED has Vf in range from 1.7V to 3V.
Relic Hawk:
The TX circuit is complicated with using several reference voltages.
Fisher 440 and project VLF770:
As reference are used Vf of two emitter junctions. The circuit needs significant modification to adjust duty cycle of pumping pulses.

CONCLUSION:
We can make significant improvements in both TX circuits because they are not competent designed.
Remains to revise and redesign the RX circuit of Relic Hawk.

[Davor]

There is another way of providing symmetry to a single ended design: capacitors center tap. In effect it is a pi configuration, and it may get a bit asymmetric when powered single ended.
I played a bit with this idea, and it can yield incredibly simple solutions, such as this one:

[mikebg]

Hi mikebg
Do you make a RH with standart generator? Which type a coil you use? How long distance you can detect 1 Euro in air

Plboy, your question should be "How to design the most sensitive metal detector?"
I know the answer because my hobby is design of TXs, RXs and antennas for QRP amateur radio.
I posted in other threads how seems the most suitable search head for all kind metal detectors. In this thread, I posted the design of most suitable TX for VLF metal detectors. My posts relates to revision, imrovements and redesign of a project, but this information leads step by step to design of most sensitive metal detector.

Follows continuation of post #963 REVISION AND REDESIGN OF AN ARBITRARY TX
but note that it contains information HOW TO DESIGN THE BEST TX FOR VLF METAL DETECTORS. See above the best block diagram.

1. Efficient transmit antenna.
The efficiency of TX coil is measured in Ampere-turns pro kilogram wire, because the weight of search head is limited. To increase coil efficiency, the designer should increase supply voltage for TX end stage and decrease resistance of TX coil. At given weight limit, the resistance can be decreased when capacitance of LC tank circuit is increased. How this is done in TX circuits of Relic Hawk and VLF770?

Relic Hawk:
The TX coil is competent designed because coil resistance is enough low (2 ohm), tank capacitance is enough large(1uF) and TX end stage is powered by whole battery voltage.

Fisher 440 and project VLF770:
This block is incompetent designed. The tank capacitance is too small. That means the weight of TX coil is increased. In project VLF770 only 9V is used to supply TX end stage, despite there are 18V battery rails available.

2. Antenna matching.
Most suitable method for matching of TX coil is to use center tap. When TX coil is driven between a lead and center tap, the TX can be designed for almost four times more power relative to circuit without tap. However the battery drain also increases. The center tap makes symmetry which can suppress distortion. To suppress more the second harmonic, we need two pumping transistors connected in push-pull circuit. Note that the additional transistor can not increase the power. It reduces distortion only. How this is designed?
Relic Hawk:
Antenna matching with tap is not done.
Fisher 440 and project VLF770:
This is done. A center tapped coil is connected in push-pull circuit.

3. Efficient pumping circuit (end stage of TX).
An efficient end stage of TX should operate as amplifier class C with duty cicle about 1/3 (ie 33%) or conducting angle about 120 deg . Voltage efficiency of end stage depends on saturation voltage of pumping circuit. Bipolar transistor have enough low saturation voltage. However efficiency decreases when a resistor is connected in emitter path and/or when when two transistors are connected in series. See in post #957 how efficient can operate a simple pumping stage with MOSFET. Let we see how this is designed in both circuits:
Relic Hawk:
The TX end stage is incompetent designed because the cascode amplifier has high saturation voltage. SPICE shows that the conducting angle is incorrect designed. That means low efficiency.
Fisher 440 and project VLF770:
The end stage also has low efficiency because pumping current pulses flow through two transistors and a resistor connected in series. SPICE shows bad duty cycle (more than 50%). It is difficult to adjust width of pumping pulses in this circuit. There is distortion of pumping pulse because there is no symmetry in amplitude detector. The distortion will disappear when we connect second rectifier diode as shown in post #961.
(To be continued)

[plboy]

Hi mikebg
Do you make a RH with standart generator? Which type a coil you use? How long distance you can detect 1 Euro in air

[mikebg]

REVISION AND REDESIGN OF AN ARBITRARY TX

generator for more then 60Vpp

Hi Plboy,
I agree that this is very powerful generator because of impedance matching. However the efficiency of your circuit is reduced by resistors connected in emitters.

The following text is continuation of my post #961.

REVISION AND REDESIGN OF AN ARBITRARY TX

Every design of an electronic device should start with study what is the best block diagram. Then the designer should make analysis of parameters that need to have each block and next step is to use the most suitable circuit for each block. I guess that professional designers of Relic Hawk have done this.

When an amateur designer starts to revise and redesign the circuit diagram of an arbitrary metal detector, his first step should be to understand how the block diagram seems. The next step is to compare its block diagram with the best block diagram and to find the differences. The third step is to understand if the suitable circuit for each block is used.

Here is the best block diagram for a CW TX (Continous Wave or non-modulated transmitter) suitable for metal detectors. It should contain the following blocks:
(To be continued)

[plboy]

generator for more then 60Vpp

[mikebg]

I really wanted to stop with revision and redesign of TX in Relic Hawk and wish to start the analysis of its RX. I thought that procedure for TX is understandable. Were created TX circuits, which will work well not only in the Relic Hawk. The redesigned TX circuits are suitable for all CW metal detectors.

However an incompetent designed TX circuit, published in this thread (post #959) without comment, indicates that participants or did not understand the procedure used for revision and redesign of TX, or reluctant to apply it to an arbitrary CW TX. So let's again make revision and redesign of two TX circuits.

Here is the original of TX circuit used in VLF770.
I see an useless resistor 470 ohm, but what are shortcomings of this TX?

[Davor]

...Your expression "Well stabilised and low noise voltage supply" means waste of battery energy to heat the regulator because its output voltage differs from voltage of battery rails.

(a late reply) Yes, but not too much. You may lose a few milliamps here, also a few there, but at the very end - Tx oscillator is not the most power hungry device in your design. QRP is a bit different, and you don't have much going on besides Tx while transmitting - it is not a full duplex.
I try to follow the KISS principle in my designs, and my thoughts about the IGSL oscillator were that there is nothing wrong with it. Amplitude is stabilised by virtue of 7808 voltage stabilisation, and it is very much in line with the KISS principle. It is true that you lose over 50% power when stabilising 12V to 8V and using such reduced voltage to power the oscillator, but it produces well stabilised oscillations, hence power well spent.
We can turn this topsy-turvy:
1) Oscillators DO perform well when supplied with well stabilised power supplies.
2) Amplitude stabilisers DO reduce amplitude obtained by the oscillator, so in effect there is no difference between amplitude stabilised oscillator and an oscillator with stabilised power supply, yet the latter is simpler.
3) There are ways other than 78xx stabilisers that can provide isolation and noise reduction with smaller voltage drop, and not involving super-sexy amplitude stabilisers that are also power-hungry (and using them does not spare much juice in a process... if any)
4) We need power sparing stabilisation that does not involve additional "spend to save" features that make no sense, but look very intimidating and complex - suggesting a well design that it most probably is not.

[o.mag]

hi
hi
in the same style but the oldest of the oscillator vlf 770
olivier

[ivconic]

Usually i am collecting ideas during the Summer and start to work on those when Winter comes.
So i will write down all your suggestions for sure.
I am interested in such oscillator and most probably i will make it.
Thank's for sharing!

[mikebg]

Ivica, for Davor I can't agree with you, but incompetence of Minelab designers is because they have not knowledge of radio amateurs.
The powerful TX showed in post #955 is suitable for operation in a two box system. At conductive trace of a metal
pipe, it can operate with 200 ohm load between electrodes delivering 16W heat into ground :-).
The conductive method is described in Gemini-3 operating manual.
For a conventional metal detector, no need of such TX power and so expensive batteries. Below is showed SPICE
analysis with CircuitMaker of push-pull TX which uses a simple centre tapped TX coil and 12V lead acid battery. Note
that unlike original TX in Relic Hawk, here the pumping transistors operate at almost battery rail to rail amplitude.
To make SPICE analysis of this variant using LTspice, you can open the compressed file attached in post #955, delete
two sections (windings) of TX coil and change other parameters and battery voltage.
With this TX circuit we can end the redesign of TX section in Relic Hawk and start redesign of its RX section. The RX is
also incompetent designed.

[ivconic]

Davor, I'm sure that your hobby is not design of QRP amateur radio :-).........

I agree!
There is one of his posts indicating exactly that, somewhere on IGSL thread.
I suspected that long time ago!
Very conversant and fine radio expert - that's seems to tend to be our fellow Davor!
That's why he already founded so many "lacks" in a typical VLF/IB design!
I knew it from a start!


"...Thus, the amplitude of TX
oscillation decreases with depletion of the battery, but this change is slow, so no impact."..."

That's excatly the case here.

"...1. What happens when its loudspeaker beeps.
This makes extremely small change of battery rails voltage, but the synchronous demodulator senses it..."

Now you also gave proper answer for that audio transformer question from the past.
One of it's role is exactly to smooth up the unwanted "pulls" from the speaker (at Musketeer, RH is not having such issue).

"...Therefore we should redesign not only TX section of Relic Hawk. We should redesign supply circuit for its audio section. The battery should see the audio as slowly (below 0.2Hz) changing load...."

Good points. Fist task would be easier than second one.

P.S.
"Uncle" Bozo Metzger is laughing now from heaven, i suppose...!

[mikebg]

IMHO every oscillator that is running near saturation, supplied with well stabilised and low noise voltage supply will perform very well. The only difference comes from the oscillator design, and if it is symmetrical like push-pull - it will do very well.

When you look at it, every amplitude stabilisation circuitry works the same way voltage stabilisation would do, and that's the whole point.

You can always make a PI regulator that will push the oscillator transistor deeper into C class as amplitude rises. Its stability and noise will depend solely upon the power supply.

Davor, I'm sure that your hobby is not design of QRP amateur radio :-). An amateur designer of QRP thinks for maximal TX efficiency and minimal TX modulation. Your expression "Well stabilised and low noise voltage supply" means waste of battery energy to heat the regulator because its output voltage differs from voltage of battery rails. For max efficiency, we need battery rail to rail amplitude across pumping transistor despite battery voltage changes.
The TX of Relic Hawk is powered direct from battery rails, but it can't deliver rail to rail amplitude. More than 3 volts are wasted in an incompetent used cascode amplifier.
Here is attached the SPICE analysis for a powerful and efficient push-pull TX designed according QRP principle. The idea for this I posted in Jan. 2010
http://www.geotech1.com/forums/showp...3&postcount=59
"The idea is TX to be powered without voltage stabilizer for maximal efficiency. The oscillation amplitude is
stabilized by P-I controller. For its reference voltage is used a LED. Thus, the amplitude of TX
oscillation decreases with depletion of the battery, but this change is slow, so no impact."
We can take any conventional metal detector and make two tests for unwanted TX modulation:
1. What happens when its loudspeaker beeps.
This makes extremely small change of battery rails voltage, but the synchronous demodulator senses it.
There are Garrett detectors where for better sensitivity are used separate batteries to supply audio section despite all other supply rails are stabilised with very low efficiency (36V battery rails are stabilised to 15V rails).
The threshold audio hum of a maximal sensitive machine should sound as random noises in frequency band 0.2Hz-16Hz. This band modulates as AM or FM an audio carrier frequency (for example 400Hz).
2. When TX coil changes height (distance to ground) and/or passes over different soil. The modulation spectrum of AIR signal in this case also coincidents with mentioned frequency band attributable for TGT and GND signal.
If we design a machine to detect 1uV target signal, a modulation index of AIR signal in order of 0.0001 will cause false alerts.
Therefore we should redesign not only TX section of Relic Hawk. We should redesign supply circuit for its audio section. The battery should see the audio as slowly (below 0.2Hz) changing load.
Here is an explanation for the attached circuit:
The TX circuit contains an additional schematic (not shown) to avoid deep battery discharge.
U1 is shown as powered with low voltage V2 to set initial operating point of M1 and M2. In reality U1 is powered direct from battery rails via a slow charged circuit to avoid oscillation start with large drain currents of M1 and M2.
Voltage V3 is obtained by a red LED.

[o.mag]

the green curve is the current IC (missing scale)
the yellow curve is the voltage

olivier

Are you sure the test points you are indicating are correct?
It would seem to me that the readings would/should be identical if taken from those points.
Are the test points located far apart on the PCB?
What am I missing here?

[HECTOR]

Are you sure the test points you are indicating are correct?
It would seem to me that the readings would/should be identical if taken from those points.
Are the test points located far apart on the PCB?
What am I missing here?