X vs R

[Tepco]

Please note that a transformer is not needed because a common mode of LM339 inputs includes a negative supply rail. I didn't test it though.

Transformer is badly needed because LM339 will “reverse” if input is forced below negative supply rail with AC signal present at R108.

[Davor]

Texas Instruments data sheets for both LM339 and LM393 says "The low input voltage state must not be less than −0.3V (or 0.3V below the
magnitude of the negative power supply, if used)." which makes perfect sense because of the PNP input transistors in their inputs. This 1ohm current sensing resistor will hardly going to exceed the -0.3V limit.

[Tepco]

Rated .3V limit will not be exceeded across 1R resistance, circulating current in LC tank is smaller. This limit is however, maximum allowable parameter, not related to reversal behavior. Some popular op amps (324, 35 and comparators (393,339) are rated for common mode range down to negative supply rail, but going even slightly below will cause reversal. Interestingly, many other popular op amps or comparators, not specified for negative rail common mode are more resistant to reversal behavior. Some, usually newer generation of single supply devices are specified, allmost to input damage level (current forced into input pin is responsible for reversal, not voltage).

[Davor]

OK, makes sense, when Vcb exceeds that 0.3V limit there is a current that rises because the input transistor loses it's marbles and a base starts acting as a diode. Guess it can be cured by adding some resistor in series with the inputs. Because there is not much excursion of voltage from one input against the other, I guess it can work without reversal anyway. I'll try that some day soon. I'm still re-decorating my apartment

Would you recommend some comparator that you are happy with?

[mikebg]

WHERE IS THE CARRIER WAVE?

WHERE IS THE CARRIER WAVE?

To identify target, we should calculate its transfer function H(s) illustrated in the functional diagram attached below. In frequency domain, the calculation seems easy:
H(s)=Y(s) / X(s)
(like to calculate gain of an amplifier - simply divide its output to input).
However in time domain the calculation of target characteristic is dificult. The name of transfer function is "impulse response" and it is calculated by a math operation called "deconvolution". [ If you know somebody who is participant in MadLabs Inc.(c)(tm)(r), he can explain this to you ]
The problem is how to design phase reference in a narrow band metal detector. To make carrier recovery we should know where the carrier wave is?
When search head is lifted far from ground and there is no target, we receive AIR signal and it is the carrier wave.
When tere is target, the TGT signal appears as modulation with side band of AIR signal. This is true if we make AIR TEST of a metal detector.
When search head is lowered to ground and there is no target, we receive AIR&GND signal. The Bulgarian designers of REKS think that AIR&GND signal is the carrier wave and they use it as phase reference. However this not true. Functional diagram illustrates that transfer functions of blocks 4 and 6 are involved in TGT signal (We should use MaDlab to solve the problem :-) .

[Davor]

WHERE IS THE CARRIER WAVE?

Some ~10km away at these frequencies. In terms of radio we are speaking of near field. Actually very near field because our antennas have effective height of <<<λ. Also the mirroring effect does not function in the conditions of electrically soft surface. There is no reflection, hence there is no hard surface.

Otherwise the diagram is OK.

An electromagnetic wave initiated by lightning - which is a phenomenon of kilometre dimensions just right for the job, can interact with a plasma in ionosphere that consists of heavy particles that remain of the meteorite impacts, and create an "plasma wave" by exciting the plasma particles with their energy. When the heavy particles are excited, the ions, we get a ion wave that is electrostatic in nature (i.e. cyclotron movement) or electromagnetic nature (Alfven or Magnetosonic waves). These waves pick up energy in ionosphere by means of accumulated energy in ions and travel much slower than light, the entire Earth ionosphere is their "resonant box" and that's why we have so much EM noise in VLF.

[moodz]

Conductive GND signal

We can use NETWORK ANALYZER to explore the conductive GND signal.

Why fiddle with theoretical analyzer when you can measure with real thing ....

http://www.analog.com/static/importe...ets/AD5933.pdf

Smart guys and gals will see that this chip can make one Kick-***** X-R type detector ... it has almost everthing you need with just a few more cheap bits added on.

For those who want head start without soldering. ( but you will need to put some code together and coils / amps etc )
http://www.digilentinc.com/Products/...m?Prod=PMOD-IA

My multichannel design based on this chip exceeds or gives PI detectors run for money.


You can thank Aus corporate detector company for me not sharing design at this point in time. However I know some are smart enough to come up with own designs.

Regards ...

[Dave J.]

This thread proves that there is no limit to how something relatively simple can be made impossibly complicated.

When I'm training new engineers I draw three phase circles on the whiteboard-- the magnetic loss angle circle, the circuit voltage and current circle, and the target response vs. frequency circle (which looks more like a Smith chart). The magnetic loss angle circle I tell them to commit it to memory so they can see it in their sleep: it's the foundation of VLF metal detector engineering. The other two circles they have to remember what they are and how they work so they can be re-created as needed.

It's all USA (!) high school level math and really basic LCR electricity. Good thing I never got a college education, this thread would have had me absolutely bewildered!

--Dave J.

[mikebg]

TWO FREQUENCY METAL DETECTOR

TWO FREQUENCY METAL DETECTOR

Let TX creates magnetic field with two frequencies A and B. Attached figure illustrates where points A and B appear in complex plane. The low frequency point B forms a signal with small amplitude OA. The high frequency B forms large
amplitude OB.

AIR and lossless ferrite signal
Points A and B lie on straight line 1. Both amplitudes OA and OB have phase angle + 90 deg relative to phase of TX
current (Re axis).

LOSSY ferrite signal
Points A and B lie on straight line 2. Both amplitudes OA and OB have phase lead almost but less than + 90 deg
relative to phase of TX current.

Signals from iron meteorites, hot rocks, bricks, terracota, dry clay and dry magnetic soils lie in 1st quadrant on straight
lines 1 or 2. If a signal, at two frequencies has the same phase lead angle, that means the object is magnetic and has
no conductivity.

CONDUCTIVE substance
Points A and B lie on straight line 3. At the frequency A, the amplitude OA has small phase lag relative to TX current. At
high frequency, the phase lag of amplitude OB is increased, but it is less than - 90 deg. Signals from nonferrous metal
objects and conductive nonmagnetic soil lie on line 3.

COMBINATION of conductivity and permeability
Each point on line 4 is vector sum of points from line 2 and line 3. At low frequency, the amplitude OA has phase lead
less than 90 deg. At high frequency, the amplitude OB forms phase lag less than - 90 deg. Note that there is a
resonance frequency which point lies on Re axis.
Signals from magnetic ferrous metals and magnetic conductive soils lie on line 4.
Both lines 3 and 4 have points with cutoff frequency at which the phase lag is - 45 deg.

[Dave J.]

When illustrating theory, I plot the locus of conductive targets as that of a simple LR circuit target. R constant, reactance proportional to frequency. With such a target, max amplitude of the resistive component occurs when the magnetic loss angle is -135 degrees (electrical loss angle 45 degrees), at which phase the vector length is 0.707 of the amplitude at infinite frequency (and the magnetic loss angle is -180 degrees).

Real targets aren't simple LR circuits, but to explain how and why, the LR model is the starting point.

--Dave J.

[mikebg]

When illustrating theory, I plot the locus of conductive targets as that of a simple LR circuit target. R constant, reactance proportional to frequency. With such a target, max amplitude of the resistive component occurs when the magnetic loss angle is -135 degrees (electrical loss angle 45 degrees), at which phase the vector length is 0.707 of the amplitude at infinite frequency (and the magnetic loss angle is -180 degrees).

Real targets aren't simple LR circuits, but to explain how and why, the LR model is the starting point.

Dave, please make a drawing!
Do you mean as is illustrated in attached publications? Their authors mean that AIR signal induced in TX coil is the Re axis.

[Dave J.]

I don't have a drawing, but the one you posted a couple posts back was fairly good. However you didn't specify scaling nor did you state what sort of nonferrous target the red curve was intended to represent.

If the intent of the red curve was to model an LR circuit it should have been a semicircle in the lower right quadrant, the diameter of the semicircle equalling the R component of LR. Then you drop the whole chart into a circle of unit radius. The end result looks a bit like a Smith chart.

I often embellish the chart with vectors corresponding to the signals of specific targets at specific frequencies.

Many metal detector patents have phase diagrams and in most cases they're reasonably correct. However you'll run into a metal detector industry convention where, when phases are plotted as magnetic loss angle, zero degrees is vertical (the ferrite axis), and the loss vector rotates clockwise. Although this is not how trig is taught in school, in the world of beeps this abuse of what professors taught us is very useful because it's how you make magnetic and electrical circuit phase diagrams plot on the same axes and quadrants so you don't get screwed up switching back and forth between the domain of magnetic fields and the domain of circuit design. What is real in magnetics is imaginary in circuit analysis, and vice versa.

[edit] I'd suppose that the C & Q Beeper Bible includes phase diagrams, but I don't have a copy handy right now to check it out.

--Dave J.

[Qiaozhi]

I'd suppose that the C & Q Beeper Bible includes phase diagrams, but I don't have a copy handy right now to check it out.

C & Q Beeper Bible - I like that. ..... Two skeptics producing a bible.

Although we refer to the Gifford Patent (U.S. 4,486,713) it is not explicitly shown in the book. So here it is below ->




However, in this particular diagram, the ferrite axis is vertical (as Dave mentioned) but the phase vector rotates anti-clockwise. I have also seen similar diagrams with the ferrite axis horizontal (pointing left) with the phase angle increasing clockwise. This second method equates more readily with what you see on a detector meter or LCD, where ferrous (iron) targets are on the left, and non-ferrous targets increase to the right.

[Dave J.]

That's the phase diagram I tell engineers to commit to memory so they can see it even in their sleep, it's the foundation of VLF metal detector engineering. Difference being that here in El Paso we rotate it clockwise.

--Dave J.

[mikebg]

I don't have a drawing, but the one you posted a couple posts back was fairly good. However you didn't specify scaling nor did you state what sort of nonferrous target the red curve was intended to represent.

Locus 3 in post '69 relates to a target which is not anular. The locus is scaled in normalized form.

FOR BEGINNERS:
Anular means something like ring, donut, bracelet, shorted turn of wire.
If anyone wants to know what means "normalized form", can use the following URLs:
http://www.uni-magdeburg.de/iwfzfp/EC/en/EC_fxEddy.htm
http://www.geotech1.com/forums/showt...1570#post71570
http://www.geotech1.com/forums/showt...047#post104047
http://www.geotech1.com/forums/showt...230#post104230