Announcement

Collapse
No announcement yet.

Targets frequency response

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

  • Hi Carl,

    I think that such a simplified (adapted to the wishes rather than reality) approach together with the idealized and simplified mikes equation is dropped as soon as we make a small intervention in the structure of visualized object such as this:



    Maybe equation can be usable if we observe object inside inductor (like in inductive heater in last Aziz post here) but such demo have only in very general simplification something to do with coin detecting in earth.

    In case of coin in soil there are not only two main variables (as in given equation) that are enough to make the virtual proof. What is enough for mathematical idealised needs is not enough for reality. What is enough for doctorate in physics is not enough for working detector.

    So I do not see how the pictured adjustment was more convincing, regarding state of the art, than the previous original, except that it is more convenient for some from reality isolated calculations.

    Comment


    • Some additional view:

      The equation for calculation how eddy current density changes in a semi-infinite conductor is:



      Where:

      d = Standard Depth of Penetration (mm)
      p = 3.14
      f = Test Frequency (Hz)
      m = Magnetic Permeability (H/mm)
      s = Electrical Conductivity (% IACS)
      Form of field of penetration:



      A little different observation than pictured theory I think.

      Good lecture about how to eddy current calculations and experiment place together one can find here:

      http://www.cwscholz.net/projects/proj3/

      Comment


      • The problem is that both the original diagrams and the new ones are correct, depending on your viewpoint.

        When the TX electromagnetic field intersects a metallic object there are lots and lots of small eddy currents generated in the target. Because these circulating currents intersect there is inevitably some cancellation that occurs. If you plot the current density in the target the result is a series of concentric rings. This gives the illusion that the current flows in a series of discrete rings (like an onion) around the target. This is not the case. Also, the current density does not decrease, from the surface to the interior, in a linear manner. In fact, this is shown in the graphs that WM6 presented in his last post.

        I still think the original diagram is basically correct.

        I'll let you know the results of my experiment later this week. Unfortunately, there's too many other things getting in the way at the moment.

        Comment


        • Results of initial tests

          Today I searched through my coin collection and found 6 identical British 2p coins from 1994. The first thing that surprised me was how hard they were to drill and/or file. Perhaps this should have been my first clue. In fact, they were so hard to drill that I only used 4 of the coins, as follows:
          1. Original (undrilled) coin
          2. 5mm hole in center
          3. 12mm hole in center
          4. 20mm hole in center

          Now for the unexpected result ... there was no measurable difference in detection distance between these 4 coins.

          I used a Fisher 1266-XB with a 10.5" spider coil on minimum sensitivity, and all 4 coins were detected at 24mm. After this test I had a suspicion about the composition of these coins, so I tried a Garrett GTAx-1000 and found it was unable to clearly ID coin #1.

          Therefore I did some research and discovered the following. The 2p coin was initially minted from bronze in 1971, but after 1992 it was minted in copper-plated steel. No wonder it was so hard to drill out the holes!
          The diameter of each coin is 25.9mm with a thickness of 1.85mm.

          Conclusion - removing the center of these coins has no effect on detection distance.

          I will next try to search out some pre-1992 2p coins to repeat the test, or acquire some aluminium discs (for easier drilling).
          Last edited by Qiaozhi; 12-30-2009, 07:30 PM.

          Comment


          • Here is what I originally used to test coin vs ring response. I like the idea of making variable-sized holes to see where the best response is. Another idea is to machine a series of aluminum rings that will snuggly nest inside one another.
            Attached Files

            Comment


            • Test continued

              This time I used 4x British 2p coins dated 1971. These coins are solid bronze and were modified as follows:
              1. Original coin.
              2. 3mm hole in center.
              3. 17mm hole in center.
              4. 17mm hole in center, but with a 3mm gap in the rim.

              Again I used the Fisher for some subjective testing. The result was that #1 and #2 were detected at 24cm, #3 was detected at 25cm, and #4 was detected at 18cm. The results for coins 1, 2 and 3 were not so dramatic, so I then performed a more objective test.

              In this case the 4 coins were each placed at a distance of 5cm from the center of a metal detector coil, and the DC voltage at the output of the synchronous demodulator in the GEB channel was measured. Coins #1 and #2 each gave a reading of 23.3mV, #3 gave 24mV, and #4 was 6.7mV.

              Perhaps removing more material from the center of coin #3 will improve the response (must try this later). However, by far the greatest difference was coin #4. It appears that breaking the ring causes the target response to fall quite distinctly.

              Comment


              • Originally posted by Qiaozhi View Post
                This time I used 4x British 2p coins dated 1971. These coins are solid bronze and were modified as follows:
                1. Original coin.
                2. 3mm hole in center.
                3. 17mm hole in center.
                4. 17mm hole in center, but with a 3mm gap in the rim.

                Again I used the Fisher for some subjective testing. The result was that #1 and #2 were detected at 24cm, #3 was detected at 25cm, and #4 was detected at 18cm. The results for coins 1, 2 and 3 were not so dramatic, so I then performed a more objective test.

                In this case the 4 coins were each placed at a distance of 5cm from the center of a metal detector coil, and the DC voltage at the output of the synchronous demodulator in the GEB channel was measured. Coins #1 and #2 each gave a reading of 23.3mV, #3 gave 24mV, and #4 was 6.7mV.

                Perhaps removing more material from the center of coin #3 will improve the response (must try this later). However, by far the greatest difference was coin #4. It appears that breaking the ring causes the target response to fall quite distinctly.
                Broken or open rings are notorious for their weak signals.
                One interesting test to add to the series, would be a full coin but cut to the center.
                Tinkerer

                Comment


                • Originally posted by Tinkerer View Post
                  Broken or open rings are notorious for their weak signals.
                  One interesting test to add to the series, would be a full coin but cut to the center.
                  Tinkerer
                  Thanks Tinkerer ... I'll try that.

                  And maybe a full coin with a cut that does not go all the way to the center ... perhaps only 5mm.

                  Comment


                  • Or this!!!!

                    http://www.youtube.com/watch?v=H9b3lePwN4I

                    Comment


                    • Hi Ivconic,

                      I have a large sheet of magnetic field paper that I could use.
                      Here's a website that shows something similar -> http://www.coolest-gadgets.com/20071...c-field-paper/

                      Comment


                      • These methods work well for DC magnetic fields, probably not for AC fields.

                        Comment


                        • What to experiment?

                          What information is needed on one amateur designer, to design metal detector possibly surpassing existing models?
                          Can we acquire this information through strange experimens as drilling of coins?
                          Two years ago my goal was to design a metal detector for meteorites location. Some of these "heavenly stone" are purchased at prices higher than gold. I began to look for relevant information and because I could not find it in the WEB, I turned to this forum with a question of interest to me "frequency response". Note that the question is in frequency domain. He could be asked in another way in frequency domain: "Does published somewhere cutoff frequencies of different targets?"
                          If I had to formulate both questions in time domain, they would be respectively:
                          1. "Does published somewhere impulse characteristics of different targets?" And
                          2. "Does published somewhere time constants of different targets?"
                          This information is necessary to know how targets differ in "color". My "meteorite locator" would receive signal not only from stones stuck in the dust, but dust and if there are any other targets. To identify targets by "color" the TX must illuminate them with white light. Using wide band TX and RX has disadvantaged for two reasons. The first reason is the huge power consumption. Wide band of TX can be achieved only by dumping process which means waste of energy in resistances. This implies a large and heavy batteries.
                          The second reason is that the wide band RX provides an increased level of thermal noise and interference (note the dimension "sqrtHz").
                          What information about targets can be obtained at SI (Sine Induction)? This is like lighting objects with colored light? You can experiment with lighting by green light to a green and a white object. They will look identical green in color because their spectral response is the same in the green region of the spectrum. Then illuminate both items with a red light. The white object will appear red, but the green - black.
                          This means that you can discriminate targets with SI (narrow band) metal detector, if appropriately chosen the operating point in frequency spectrum. The narrow band machine has three advantages:
                          1. Economical consumption of battery.
                          2. Low noises and low interferences.
                          3. At SI can be increased modulation index of the useful signal by automatically compensating signal at the input of RX preamp. If so, we can increase the gain of RF amplifier in RX, which will provide greater depth of detection.
                          When I asked my colleagues on the hobby (the REMI group) how to make a compromise between two conflicting requirements for wide band and narrow band, they simply answered: "Using a key SI / PI (Sine Induction - Pulse Induction)". Then they gave me 3 old circuit diagrams (over 100 years!). These are ancient AF metal detectors made of telephone parts.
                          I think we need to measure spectral characteristics of different targets and soils. Computer sound card is adequate to do the job. Wolfgang Buescher created such a program for amateur radio. Search WEB for DL4YHF's Spectrum Lab (Audio Signal Analyser). Maybe Aziz will offer a program designed specifically for measuring the spectral characteristics of targets and soils.

                          Comment


                          • Mike,

                            I agree, the discussion has shifted from the original topic, although a good understanding of eddy response is not a bad thing to know.

                            What needs to be done is a thorough investigation of target phase response vs frequency. This is easier said than done. You either need a excellent phase-linear wideband detector, or you need to characterize its phase response and compensate target results. The latter would require a target with a predetermined phase response.

                            I would suggest a DDS-based TX circuit for easy programming and a good clean sinusoid. Then a traditional I/Q receiver driving an ADC+micro, where the micro computes a phase angle and drives a simple display. The hard part is designing the coil, and ensuring the whole thing has a good inherent phase response. But it can be done.

                            The analogy with light doesn't hold for detectors. Changing frequency just shifts target phase responses. This would be like something appearing to be blue under a red light, but orange under a green light.

                            - Carl

                            Comment


                            • Latest test results:
                              1. Original coin
                              2. 5mm long slot from circumference towards center.
                              3. Slot extends from circumference all the way to the center.

                              There was not detectable difference in the response to these targets.

                              Regarding the possibility of using magnetic field paper to view the actual field ... of course, that was a stupid idea. As Carl said, the field is a.c. not d.c.
                              In addition, the magnetic field paper acts as a target to the metal detector.

                              If you imagine the huge number of small eddy currents being generated in the target, and the manner in which they superimpose each other, then it is clear why the slotted coins have the same response as the original coin.

                              Comment

                              Working...
                              X