Triangular Wave Technology

[Mechanic]

Hi Guys,

Brainwave
So we see that the turn off di/dt slope cancels the eddy currents from the tx on time. So what if we pulsed the coil in such a way that the turn off di/dt canceled all of the eddy currents in a larger target so after the turn off di/dt current slope there were no(close to none) eddy currents left in the target? Ground response would still be there. Then this ground response could be subtracted from the ground/target response from a normal pi pulse...... This would not work so well for small targets, but I'm sure something could be adapted and arranged to work for the smaller targets

cheers Mick

[Mechanic]

Here's a picture of the tx current/voltage waveform that could possibly do the job to cancel eddy currents in large targets and keep the ground response equal after both of the "tx periods"

Cheers Mick

[mikebg]

Mick, why are you trying to answer these questions with TIME DOMAIN? Everything in this domain is shrouded in the mists of the mathematical operation CONVOLUTION.
The analysis should be done in FREQUENCY DOMAIN, where all signals are processed by multiplication and division. Here's the visual answer to your question with normalized locus. Frequency domain shows what frequency spectrum you need to identify small targets. The explanation of diagrams is given in other thread:
http://www.geotech1.com/forums/showt...144#post165144

[Dave J.]

The most basic metal target is a ring, which is an LR circuit.

X = 6.3fL, where f is frequency and L is the inductance of the target in question.

R = a constant for the target in question.

Phase angle of the target in degrees = arctan X/R using the convention that "salt water" is 0 degrees and high conductors approach 90 degrees.

First y'all, do ordinary high school math. Then, I suggest listening to mikebg, he seems to have more time and energy for public education than I do. Plus, I suppose all this is covered in the Bible of Beep (also known as "The Gospel According to Carl & George".)

--Dave J.

[moodz]

Mick, why are you trying to answer these questions with TIME DOMAIN? Everything in this domain is shrouded in the mists of the mathematical operation CONVOLUTION.
The analysis should be done in FREQUENCY DOMAIN, where all signals are processed by multiplication and division. Here's the visual answer to your question with normalized locus. Frequency domain shows what frequency spectrum you need to identify small targets. The explanation of diagrams is given in other thread:
http://www.geotech1.com/forums/showt...144#post165144

Mick is not so far off track .... Frequency domain analysis is not so straightforward for wide band pi signals ... There are at least three distinct states in time .. Tx, fly back and Rx ... It is a mistake to assume that the three states are part of the same Waveform for analysis by freq domain. A priori we don't know the relevant phase and amplitude of the discrete fdm components in the pi pulse with sufficient accuracy to work out if a small target is present in the presence of ground, EMI and earth field as these latter influences have large magnitudes and phase in relation to weak target influences. I can say this with some confidence because the technique I use is a time domain method that properly respects the signal states and isolates the target influence correctly from the others by removing them.

[mikebg]

Moodz, IMNHO you are also as Mick far off track.
Why you use term "wide band PI signals"? Simply remove the word PI and stay with two legs in frequency domain or remove "wide band" and stay in the mist of time domain. There is no PI in frequency domain - there are only two type metal detectors:
http://www.geotech1.com/forums/showt...505#post164505

At a competent designed search head, when analyzed in FD, there is no TX state, no fly back state, no RX state. There are frequency spectrums only. To understand fly back, I will explain how operates an incompetent designed search head with separate TX and RX windings. They are not in induction balanced position.
I showed in other thread that fly back is simply the AIR signal received by self-inductance. Self-inductance search head (MONOCOIL) generates the worst RX signal because of large AIR&GND signal in RX input. It decreases modulation index of TGT signal and limits possible gain of RX preamp. Please read how patent US4506225 operates in all domains. The invention uses excitation with waveform showed above in post #90. See how looks the fly back voltage for this TX current:
http://www.geotech1.com/forums/showt...505#post163505

I will explain again the fly back signal with more diagrams in that thread:
http://www.geotech1.com/forums/showt...-relationships

[Davor]

I showed in other thread that fly back is simply the AIR signal received by self-inductance.

Precisely. The funniest part is that it also has it's own tau, and that very tau can be manipulated. It is also true that the resistive part in Tx coil influences the targets' tau, thus spoiling discrimination.

[moodz]

Moodz, IMNHO you are also as Mick far off track.
Why you use term "wide band PI signals"? Simply remove the word PI and stay with two legs in frequency domain or remove "wide band" and stay in the mist of time domain. There is no PI in frequency domain - there are only two type metal detectors:
http://www.geotech1.com/forums/showt...505#post164505

At a competent designed search head, when analyzed in FD, there is no TX state, no fly back state, no RX state. There are frequency spectrums only. To understand fly back, I will explain how operates an incompetent designed search head with separate TX and RX windings. They are not in induction balanced position.
I showed in other thread that fly back is simply the AIR signal received by self-inductance. Self-inductance search head (MONOCOIL) generates the worst RX signal because of large AIR&GND signal in RX input. It decreases modulation index of TGT signal and limits possible gain of RX preamp. Please read how patent US4506225 operates in all domains. The invention uses excitation with waveform showed above in post #90. See how looks the fly back voltage for this TX current:
http://www.geotech1.com/forums/showt...505#post163505

I will explain again the fly back signal with more diagrams in that thread:
http://www.geotech1.com/forums/showt...-relationships

IMHO You are missing the system model analysis of the problem and assuming everything can be solved with a singular technique. .. Ie frequency domain analysis ...my background in radio and signal processing does not disagree with your use of fd analysis however this is not a linear system and it does have states just as a mixer in a radio is non linear and has states also. it is impossible to analyse a radio system using Freq domain analysis only ... You must also analyse the effect of non linear and state full systems. Unless your system consists of ideal linear components only ....you cannot possibly produce an accurate analysis using fd as you have not factored in the non linear factors and states. Actually that is what fd lacks ... The ability to analyse states whereas you can have states in td.
metal detectors are systems ... Not a detector head by itself.

[Qiaozhi]

IMHO You are missing the system model analysis of the problem and assuming everything can be solved with a singular technique. .. Ie frequency domain analysis ...my background in radio and signal processing does not disagree with your use of fd analysis however this is not a linear system and it does have states just as a mixer in a radio is non linear and has states also. it is impossible to analyse a radio system using Freq domain analysis only ... You must also analyse the effect of non linear and state full systems. Unless your system consists of ideal linear components only ....you cannot possibly produce an accurate analysis using fd as you have not factored in the non linear factors and states. Actually that is what fd lacks ... The ability to analyse states whereas you can have states in td.
metal detectors are systems ... Not a detector head by itself.

These are good points to consider. One thing some people fail to understand when performing SPICE analyses, is that frequency domain analysis is linearized around the operating point, which is why it's also called a "small signal ac analysis". In simple terms, this means that all the nonlinear characteristics are lost, because the models are linearized and are only valid for small changes in signal. If you want to do large signal ac analysis, then you need to use an RF simulator with a harmonic balance algorithm.

[mikebg]

MULTI ANGLE TECHNOLOGY?

People, you have forgotten the title of this thread. Look at it because you started talking about completely different things.
Just log into the role of a ham designer and try to answer the question whether Triangular Wave Technology is efficient for a wideband metal detector. Frequency domain analysis shows, this is not an efficient technology.
It now remains for you to explain me how Time Domain will show what is the best technology. I know that Frequency domain can show this.

[mikebg]

The most basic metal target is a ring, which is an LR circuit.

X = 6.3fL, where f is frequency and L is the inductance of the target in question.

R = a constant for the target in question.

Phase angle of the target in degrees = arctan X/R using the convention that "salt water" is 0 degrees and high conductors approach 90 degrees.

First y'all, do ordinary high school math. Then, I suggest listening to mikebg, he seems to have more time and energy for public education than I do. Plus, I suppose all this is covered in the Bible of Beep (also known as "The Gospel According to Carl & George".)

--Dave J.

Dave,
I can't agree with convention that "salt water" is 0 degrees. The GND signal at salt water has significant phase lag relative to phase of TX current even at extremely low frequencies (below 3kHz). To decrease magnitude and phase of salty GND signal, Eric Foster reduces diameter of TX loop for beach combing. The timeconstant of salt water (conductive halfspace) depends on square of TX coil diameter D:
http://www.geotech1.com/forums/showt...462#post104462

I also can't agree with your term "high conductors". More suitable is to say targets with "large timeconstants" because the timeconstant depends on square of target eddy diameter:
http://www.geotech1.com/forums/showt...172#post104172

The diagram in post #93 is valid for two pieces of copper or two nuggets with 1:5 eddy diameters. Note how large are phase differences despite both targets are made of the same metal.

Most suitable for phase test is a ferrite core because the phase is independent on frequency. It generates signal with phase lead 90 deg for all frequencies.

[Carl-NC]

First, I agree that FD analysis is difficult at best on some systems, and impossible to do in Spice on others. Try doing a Spice noise analysis on a PLL, or a Bode plot on a Gilbert cell; you will fail, whether you do it in FD or TD.

Second, I disagree that the triangle wave method is "inefficient" without some kind of statement as to what is meant by "efficient." Is it spectral efficiency? Power supply efficiency? Circuit hardware efficiency? Signal processing efficiency? Target response efficiency? Take a look at the top $1000+ metal detectors currently being made. Which detector has the reputation for hitting the deepest silver, and what technology does it use?

Third, I think Dave uses -90° for ferrite, 0° for salt, and +90° for superconductor. I tend to use 0°/90°/180° respectively, some people use -180/-90/0; same difference. His term "high conductor" is commonly used in detector talk to mean "AC conductance," not "DC conductance." A US silver dollar is a "high conductor," a US 3-cent silver coin is not, despite having the same, err, "conductance." I don't make up the terminology, I just live with it.

Finally, I'll also disagree with the "incompetent-designed" such-and-such. There are 1000 different ways to design metal detectors, and they usually involve trade-offs. I've yet to see a single "best way," either for an overall metal detector or for a component like a search coil. Again, what are you designing for? A $100 kid's detector? A meteorite hunter? A 10mm crevassing probe that will detect fly-sh!t nuggets? A salt-surf machine that will detect a 2mm serpentine chain? A detector that can "see through" nails? A security walk-through? Can someone name a single metal detector design that will meet all of these goals?

However, I will say that I enjoy these discussions immensely! Thanks Mike!

[moodz]

Can someone name a single metal detector design that will meet all of these goals?

Human eyeball with shovel and fine mesh sieve

eg for security detector ... you dong em on the head with the shovel then empty their pockets into the sieve.

[Davor]

... you dong em on the head with the shovel ...

...at 90°

[moodz]

.. 180° also works very well ( ie from behind )