Announcement

Collapse
No announcement yet.

WIDEBAND (BROADBAND) TECHNOLOGY

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • mikebg
    replied
    GROUND ELIMINATING WIDEBAND DETECTOR

    The solution for ground elimination in time domain is described in other thred:
    General tech discussions on all types of metal detectors: VLF, 2-box, BFO, off-resonance, PLL, etc. Questions, ideas, and anything else that moves you.


    Here is described solution in Frequency domain.
    To identify frequency characteristic of target, four coils are used in the search head. Here is an example:

    General tech discussions on all types of metal detectors: VLF, 2-box, BFO, off-resonance, PLL, etc. Questions, ideas, and anything else that moves you.

    1. TX coil is used for excitation. It creates magnetic field with suitable frequency spectrum.
    2. AIR reference coil. It is placed with large coupling with TX coil to obtain AIR signal.
    3. GND reference coil. It is placed in position for induction balance with TX coil to obtain GND signal.
    4. RX coil is used to receive TGT signal. At induction balance with TX coil, the RX coil receives suppressed AIR signal. The GND signal is suppressed by subtraction (both identical coils are connected as TWIN LOOP receive system).



    TARGET identification

    The target is represented as block 5. It has Laplace transfer function H(s) between input and output. The name of transfer function in frequency domain is FREQUENCY CHARACTERISTIC (frequency response). If we know the input signal X(s) and output signal Y(s), we can identify the target by calculating its transfer function H(s). For calculation of H(s) is used the mathematical operation DIVISION, because Y(s) is product obtained by MULTIPLICATION of X(s) with H(s). Note that in Time domain we should calculate DECONVOLUTION.


    AIR signal
    This is the excitating signal that induces eddy currents in conductive volume. It is produced in RX coil by mutual inductance, and by self-inductance in TX coil.
    The shape of AIR signal is like di/dt (derivative of TX coil current in time domain). A turn wire placed near to TX coil can generate enough large AIR signal used for refence. The derivative of TX current appears in frequency domain as multiplication by frequency. The induced voltage has phase lead 90 deg for all frequencies.
    To calculate frequency characteristic of ground, we need AIR signal.

    TGT signal
    Since the target is buried, the energy passes through the soil twice and induces TGT signal in RX coil. Note that TGT signal contains distorted information for frequency characteristic of target H(s). Excitating signal X(s) of target differs from AIR signal because it is output of block 4. It is obtained by multiplication of excitating AIR signal in input of block 4 with its frequency characteristic H4(s).
    The output signal from the target Y(s) is distorted. Going to RX coil, the Y(s) undergoes multiplication with the frequency characteristic H6(s) of block 6. To identify target, we should calculate H5(s) by removing the distortion of TGT signal.

    GND signal
    We need GND signal to calculate frequency characteristic of target H(s). Therefore, we use GND reference coil.
    It is assumed that the excitation signal X(s) of target has the same form of frequency spectrum as the GND signal. It is assumed also that spectral characteristics H4(s), H6(s) and H7(s) are similar in shape.

    Computational algorithm

    1. By division of GND signal with the AIR signal is calculated the frequency characteristic of soil H6(s).
    2. By division of TGT signal with the frequency characteristic H6(s) is calculated the target output spectrum Y(s).
    3. By division of output signal Y(s) with the GND signal is calculated the frequency characteristic of target H(s).
    4. The problem is solved. Follows comparison of calculated spectral form H(s) with known forms of different targets in memory.

    The block diagram in frequency domain shows how to design the excitating function i(t) in order to obtain a better SNR and better energy efficiency:

    The form of TX current should contain frequency components in this part of frequency spectrum where frequency characteristics H5(s) of different targets differ significant.

    Sometimes this is possible at only one TX frequency. That means the problem can be solved sometimes with CW (narrow band) metal detector at very high energy efficiency and very low noise.
    Attached Files

    Leave a comment:


  • mikebg
    replied
    STEP-DOWN RESPONSE

    WIDEBAND AMPLIFIER part 2
    The method illustrated above,
    General tech discussions on all types of metal detectors: VLF, 2-box, BFO, off-resonance, PLL, etc. Questions, ideas, and anything else that moves you.

    makes Ь freqency sweep (discrete or continuous) to measure frequency characteristic of an amplifier or attenuator.

    We can use something like PI technology for faster measurement of frequency characteristic. The method illustrated below is known as "step response". A function generator 7 delivers signal as step-down function. In output of the device under test 2 (inverting amplifier) is connected a processor 8 to compute two functions of frequency response or to plot a locus of transfer function in complex plane.

    The algorithm of measuring process is:
    1. Measure or collect data for the step response.
    2. Calculate the impulse response. This is computing of derivation because the impulse response is derivative of step response.
    3. Transform impulse response in frequency response. The Fourier transform generates two frequency functions: Magnitude and Phase or Real and Imaginery components. The computer can display both functions as a locus in complex plane.
    Note that for measurement in time domain no need of pulse train. We need only a step signal for excitation. It operates as a flash-light to take a picture of the measured object.
    Instead step-down function, we can use step-up function for excitation.

    APPLICATION for WIDEBAND METAL DETECTOR
    We can use the illustrated method in a metal detector to measure targets frequency characteristics. However there are two differences at metal detecting application:

    1. When the search head moves, we need a train of step-down pulses (as series of flash-lights to take a video). This creates problem because the response is not to a true step down signal. It contains response to the previous pulses in the train.

    2. RX coil (mutual inductance or self-inductance) operates as a derivator. The voltage induced in TX and RX coil is derivative of targets response. That means no need of step #2 in algorithm to calculate the impulse response.
    Attached Files

    Leave a comment:


  • Aziz
    replied
    Originally posted by Aziz View Post
    This happens with targets having TC=100µs. Same configuration and colors as above.

    [ATTACH]23524[/ATTACH]

    Can you see, why the standard PI sucks?

    BTW, with TC=1000µs, you can't see much response in a PI configuration. Almost zilch.

    I'm going to detect the decent sized monster gold nuggets....
    ALL MINE!!! ALL MINE!!!! ALL MINE!!!!!!!
    *LOL*

    (F$%uuu"$%k", not anymore. You know my secret invention now... )

    Cheers,
    Aziz
    Did anyone see the real value of the idea?
    And the genius of the TEM2.0 and Class-E TEM transmitter?
    The idea is released for the benefit of all. Free!!! Absolutely greed free!!! Not patented!!! Not patent trolled!!! Not monopolized!!!

    Aziz

    Leave a comment:


  • mikebg
    replied
    HOW TO MEASURE FREQUENCY RESPONSE

    WIDEBAND AMPLIFIER part 1

    Let us imagine that we have a metal detector that works very well. Worth is for us to copy something from it. However,
    the manufacturer foresees that the machine will fall into the hands of people like us, so the preamp is sealed in black
    plastic. We do not do anything else but to measure how this preamp amplifies different frequencies .
    How we would do this if we think in the Time domain?

    We must pass to input sinusoidal signal and change frequency in discrete steps. For this purpose we need a
    sine generator 1. To output of preamp 2 we should connect devices 3 and 4 that measure amplitude and phase, as illustrated below with the left diagram. If there are none such devices, we should use an oscilloscope. The method is slow and imprecise because we use discrete frequency values.

    To the right is shown the block diagram of a faster method. It takes sawtootth signal from the oscilloscope 6 to control
    frequency of a sweep function generator 5. Despite the improved performance, the method is not accurate.

    Remains to see what will offer as fastest method the Frequency domain.
    Attached Files

    Leave a comment:


  • mikebg
    replied
    SINC WAVEFORM

    THE SINC WAVEFORM

    See how it seems in colors!
    General tech discussions on all types of metal detectors: VLF, 2-box, BFO, off-resonance, PLL, etc. Questions, ideas, and anything else that moves you.


    Read how in 2001 Carl Moreland reinvented the (sine x)/x or SINC INDUCTION under the name "sine(x)/x looking waveform":


    The following threads show how several famous professional designers of metal detectors are tumbling in the CONVOLUTION mist of Time Domain because not use Frequency Domain as powerful tool for analysis and design of wide band metal detectors.

    However they use the terminolgy of Frequency Domain to explain processes in Time Domain.


    Leave a comment:


  • Aziz
    replied
    This happens with targets having TC=100µs. Same configuration and colors as above.

    Click image for larger version

Name:	PI,TEM,TEM2.0-Response, TC=100us.gif
Views:	1
Size:	18.9 KB
ID:	335159

    Can you see, why the standard PI sucks?

    BTW, with TC=1000µs, you can't see much response in a PI configuration. Almost zilch.

    I'm going to detect the decent sized monster gold nuggets....
    ALL MINE!!! ALL MINE!!!! ALL MINE!!!!!!!
    *LOL*

    (F$%uuu"$%k", not anymore. You know my secret invention now... )

    Cheers,
    Aziz

    Leave a comment:


  • Aziz
    replied
    Originally posted by tony_av View Post
    Yes, it is. But it is nonlinear in high frequency region. We all need DSP to cancel ground.
    That's the reason, why we should use DSP coding in the future. Or the laptop/netbook/tablet PC platform.
    Aziz

    Leave a comment:


  • Aziz
    replied
    Ok, I have made a simple target response comparison for you to convince you further. (Neglect the ground effects at this stage).
    If you integrate the response signal over time (to get the total response energy, we take the absolute value of the response signal and integrate it over one TX period) and compare the surface area between the response signal and the horizontal time-axis, you can see the difference of the signal integrity of the different architectures.

    The standard PI sucks.
    The TEM is better.
    The TEM 2.0 is much better.
    (The class-E TEM is similar to TEM 2.0)

    Click image for larger version

Name:	PI,TEM,TEM2.0-Response.gif
Views:	1
Size:	17.4 KB
ID:	335158

    Enough motivated?

    Aziz

    Leave a comment:


  • tony_av
    replied
    We have more ground response too. But this is dominant in the low frequency region.

    Yes, it is. But it is nonlinear in high frequency region. We all need DSP to cancel ground.

    Leave a comment:


  • Aziz
    replied
    Originally posted by Tinkerer View Post
    Aziz, I find all your ideas very interesting, even if many times I understand nothing like in the case of the TEM V2.

    I did play with the simulations and got the wave forms you show, but I don't understand how this is good for large targets.

    With a sine wave, or any continuous wave current, we stimulate the target and generate eddy currents on each slope, when the rate of change is high.
    With every inversion of current, we kill the existing eddy currents and then create new ones.

    How is that good for large targets?

    Tinkerer
    Hi Tinkerer,

    it is good for all targets.

    We get more spectral response energy if we compare it to a standard PI or the standard TEM TX.

    Best to use the power efficient mode (0.5, 1.5, 2.5, .. radiant sine waves during TX-on period) and we have additional half sine wave response, which produces a wide band response. The full radiant sine waves produce a resonse for one frequency (resonance frequency). More sine waves, more response. And we have the half cosine/sine current wave form during the TX-off period, which also produces a wide band response.

    Sampling all the response signals will increase the response signal integrity. So don't leave anything. Collect as much as you can sample (integrate).

    One important thing to know:
    We have more ground response too. But this is dominant in the low frequency region.

    And if you want to use the boost-mode, just change your TX frequency.

    The sampling and usage of the new transmitters is quite easy stuff on my laptop/netbook/tablet PC platform as I'm using the continiuous wave sampling (always sampling the RX signal).

    Cheers,
    Aziz

    Leave a comment:


  • Tinkerer
    replied
    Originally posted by Aziz View Post
    Yeah, that's the reason, why we have now the TEM V2.0 or the class-E TEM transmitter. The quasi linear current ramp up period is replaced with sine waves, which will stimulate the large targets much much better (low frequency).

    Anybody had a look at the new inventions?
    Aziz
    Aziz, I find all your ideas very interesting, even if many times I understand nothing like in the case of the TEM V2.

    I did play with the simulations and got the wave forms you show, but I don't understand how this is good for large targets.

    With a sine wave, or any continuous wave current, we stimulate the target and generate eddy currents on each slope, when the rate of change is high.
    With every inversion of current, we kill the existing eddy currents and then create new ones.

    How is that good for large targets?

    Tinkerer

    Leave a comment:


  • Aziz
    replied
    Originally posted by mikebg View Post
    I have a look at the future reinventions.
    Soon somebody will reinvent the SINC function as most suitable for wideband metal detectors.
    Soon MINELAB will nake a patent pending for it.
    Yup, I'll have to look from time to time to see the bad guy lodging a patent application.

    But I have a lot of tar and feather for this guy.

    I don't think, that ALLMINE is going to try it.
    (Although, you never know..., their profession is to "adapt and arrange..")

    Anyway. I like to share it with you (see my MadLabs conventions below).

    Aziz
    Last edited by Aziz; 02-22-2013, 07:04 PM. Reason: adapted and arranged ...

    Leave a comment:


  • mikebg
    replied
    Originally posted by Aziz View Post
    Yeah, that's the reason, why we have now the TEM V2.0 or the class-E TEM transmitter. The quasi linear current ramp up period is replaced with sine waves, which will stimulate the large targets much much better (low frequency). Anybody had a look at the new inventions? Aziz
    I have a look at the future reinventions.
    Soon somebody will reinvent the SINC function as most suitable for wideband metal detectors.
    Soon MINELAB will nake a patent pending for it.

    Leave a comment:


  • Aziz
    replied
    Yeah, that's the reason, why we have now the TEM V2.0 or the class-E TEM transmitter. The quasi linear current ramp up period is replaced with sine waves, which will stimulate the large targets much much better (low frequency).

    Anybody had a look at the new inventions?
    Aziz

    Leave a comment:


  • mikebg
    replied
    Originally posted by Tinkerer View Post
    Professor,
    if you are saying that the linear ramp of the triangular TX wave form does not create a signal you are wrong. The words linear ramp imply a linear rate of change in the current. Therefore there there are eddy currents generated.
    Further more, you state your self, that there is a linear change of TX current, therefore a rate of change, therefore there are eddy currents generated.

    We, the students, ask you to please revise your statements.

    Tinkerer
    Thinker,
    You have no reason to call me "Professor", because I'm just a former ham with primitive knowledge for design TXs, RXs and antennas. However I have reasons to call you "Thinker" because you think what I'm saying:

    1. You think that ramp functon is part of triangular TX wave. I do not think so. I showed the frequency spectrum of triangular wave in other thread:

    General tech discussions on all types of metal detectors: VLF, 2-box, BFO, off-resonance, PLL, etc. Questions, ideas, and anything else that moves you.


    General tech discussions on all types of metal detectors: VLF, 2-box, BFO, off-resonance, PLL, etc. Questions, ideas, and anything else that moves you.


    The spectrum of sawtooth wave also is shown here - in this thread, but it also is not ramp function:

    General tech discussions on all types of metal detectors: VLF, 2-box, BFO, off-resonance, PLL, etc. Questions, ideas, and anything else that moves you.


    2. You think that ramp function does not create eddy current. I do not think so. Even connected a battery in eddy current loop to illustrate that eddy current created by source "e" is DC.
    General tech discussions on all types of metal detectors: VLF, 2-box, BFO, off-resonance, PLL, etc. Questions, ideas, and anything else that moves you.


    Participants of this forum are not my students because they (and you) are much more educated than me. Radio clubs in Bulgaria do not issue diplomas. They are trained hams to make primitive analysis in order not make mistakes at design.

    Thinker, please think whether your draft contains something stupid before to post it in a worlds forum! Stupid is to think that triangular wave and sawtooth wave contains ramp function. The ramp function is a mathematical abstract for analysis because starts from infinity and goes to infinity. However analisis shows that d2i/dt2 - second derivative of TX current creates signal.
    Professor Mike

    Leave a comment:

Working...
X