Announcement

Collapse
No announcement yet.

4 quadrant VLF discrimination

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

  • #16
    Originally posted by Sergey_P View Post
    5 zones => http://md4u.ru/download/file.php?id=4238
    DD and concentric coil does not compensate for the soil
    This concentric coil (http://md4u.ru/viewtopic.php?f=77&t=3866&start=0 ) – compensates
    reception area of target = zone TX * zone RX ?
    There are some experiential comparisons between a concentric and a DD that give an advantage for a DD over a concentric over a difficult soil. My view is that a sharp response change of a DD is responsible for all the benefits, and mostly because of the motion compensated Rx-es that tend to wipe out blunt responses.

    I am not entirely clear on the concentric coils calculations yet, because I'm aware of the magnetic field peaking near the Tx coil. The most of the induction calculations assume a magnetic dipole and uniform magnetic fields, while in reality field concentrates near wires. Guess the calculations for the balance etc work fine in practice if you keep a bit distance against the Tx coil. I like the proposed concentric coil because it maintains the Tx coil integrity, Q factor, etc.

    Reception area is a vector product of Rx and Tx. In case you use differential coils there are zones perfectly illuminated by both Rx and Tx, yet you have a sharp minimum ... due to the vector product.

    ...

    Now about the 4 quadrant...
    Let's observe a non-Fe channel of a regular 2 quadrant Rx.
    We have a GEB channel that has a zero response for ground vector and a positive response for all metals - considering a positive coil phase.
    We have also a Disc channel which has a response ~perpendicular to the GEB. It will have response positive for some metals and negative for others.
    They both go to the comparators to be tested against a threshold value, and these form an analog AND function. If both Geb and Disc values are above the threshold, a Geb value is passed through, and detector beeps.

    It is important to notice here that it is not necessary that Disc channel passes the threshold to obtain a correct response, but instead it must be positive. In a blurry area between a zero and a threshold value, it is the GEB that must overcome a threshold for a clear response.

    To obtain a 4 quadrant Rx we need an analog XOR function, and in practice it is a balanced mixer or a 4 quadrant multiplier. There is a slight problem with both - with Disc and Geb being of the same order of magnitude, the output would be squared, unless... Disc is made binary.
    So when Disc is positive the output of such XOR would be equal Geb, and when Disc is negative the output would be -Geb.

    Easy way to make this is to supply Disc to a comparator and compare its value with zero. The comparator binary output is supplied to a switch which choses between a Geb and inverted Geb. Such signal is further compared with a threshold value, and passed to the tone circuitry. And we have a solution for discrimination of the opposite quadrants for a non-Fe channel in a 4 quadrant solution. It may even be done as a hardware add-on for the existing rigs.

    To get a regular 2 quadrant Rx from this 4 quadrant solution, only the inverted Geb signal is level shifted below zero, and it will work exactly as expected. A Disc polarity switch will operate between a non-inverted Geb value and some negative value. If both Geb and Disc values are positive, a Geb value is passed through when it overcomes a threshold, and detector beeps. Such operation will be needed when using regular coils.

    I'll make a model of this soon, but this concept is open for debate.

    And yes, I already pimped my IGSL Disc channels gains so that they behave more binary-like, and it results with better sensitivity of borderline materials such as Al foil.

    Comment


    • #17
      you have 2 quadrant RX and 2 quadrant CX… no 4 quadrant (pic)
      alternative to the method of phases for GB -> http://www.md4u.ru/viewtopic.php?f=7...4cafed5af084e0 , http://www.md4u.ru/viewtopic.php?f=14&t=5702 , http://www.md4u.ru/viewtopic.php?f=14&t=5918 , http://www.md4u.ru/viewtopic.php?f=14&t=4383 ...
      Attached Files

      Comment


      • #18
        Don't be hypnotised by my simplification to a Cu channel. The example above considers a two channel MD with separate Cu and Fe channels, just like IGSL, and the example was about only one channel.

        BTW, you again just pasted links without emphasis on what to seek there. Took me an hour to sift through. Please put a word, say, "Verator3 metal detector" etc.

        I think extracting GEB by means of phase shifting a switcher drive is much better over a wide dynamic range than partial addition/subtraction in gain blocks. It is less prone to opamp offsets, compression etc. However Verator is a nice idea, and has several points extremely well thought of ... I have some other favorites. No offense. I prefer proportional audio.

        ...

        So here goes the visual representation of the example above, yet with addition of what happens in Fe channel as well. Input signal is considered to originate from a differential coil...
        Attached Files

        Comment


        • #19
          the first link - the theory, the next - a practice
          extracting GEB by means of phase - child's play.
          GEB, GB - the G signal does not compensate … and the received signal with the phase of G - resets. (target + G - gives an error)…
          I'm sorry, good luck…

          Comment


          • #20
            Actually target + here means a positive phase of a differentiated response after the demodulating switchers. I'm talking about the demodulated signal in a frequency range ~10Hz.

            So this is not about target + and ground -. It all happens at earlier stages. Ground signal is either canceled or not, but in both cases I assume it doesn't produce response in a demodulated and motion compensated differentiated GEB channel due to the demodulating switcher being driven by a 90° shifted signal.

            This topic is about making sense of a signal that may have phase reversal due to the use of a differential coil.

            Comment


            • #21
              A small update...

              I became aware of some unique solutions that happen in Verator rig that make the quest for a real 4 quadrant rig even more important. Thanks Sergey. The very special thing that happens in Verator is that instead of a GEB, i.e. an all metal channel, it uses a Disc. channel for extracting audio. With no intentions to go into a deep analysis, this means that Verator is capable of detecting happily in salt water environment (hooray!). It also quiets the reactive response, i.e. ferrite response ... the GEB null, by multiplying the previously mentioned signal with GEB using a two quadrant multiplier. That's just brilliant!

              So, for my 4-quadrant Rx I wish a dedicated audio producing channel that has holes at salt water and at ferrites phases, so that neither produce any tone in my rig. That plus a differential coil would give me a considerable edge over the existing PI rigs, and discrimination too.

              The only thing I'm missing is a proper analogue multiplier. I have some ideas about that too.

              Comment


              • #22
                George Payne patented a 4-quadrant analog multiplication motion discrimination scheme in the late 1970's. I don't remember what he used for the multiplier. In the distant past I've configured LM13700's as analog multipliers, with offset adjustments of course. In principle the limited dynamic range of analog multiplier circuits could be overcome with an AGC system but I haven't seen it done.

                When it comes to analog computation, there are lots of schemes besides the ones in textbooks. Back in the 1970's I designed a circuit which took anemometer reed switch pulses and used that to compute the integral of the root mean cube windspeed, which is equivalent to wind energy. It wasn't even very complex (of course no microprocessor), but unlike most of my designs it was darn clever.

                * * * * * * *

                In my opinion you're trying to solve a problem that doesn't need to be solved. However if you're a hobbyist rather than a commercial engineer, that may not be relevant. I have my day job where what I do is supposed to make practical sense, but I also have hobby projects that intrigue me even though someone else would say they don't make any practical sense.

                My favorite hobby project these days is to see how much medium-fidelity sound I can get out of a compact light weight speaker system that runs for darn near forever on a couple of AA cells. It's been a good education on speaker acoustical design, and what I've learned by actually doing goes against a lot of conventional practice.

                --Dave J.

                Comment


                • #23
                  Originally posted by Dave J. View Post
                  a 4-quadrant analog multiplication
                  http://jre.cplire.ru/win/sep07/2/text.html

                  Comment


                  • #24
                    Hi Sergey, unfortunately, we don't speak Russian.

                    http://www.microsofttranslator.com/b...07/2/text.html

                    (The translation isn't optimal but better than nothing.)
                    Aziz

                    Comment


                    • #25
                      The paper is interesting, and has all the math behind it. It can be implemented by Tayloe-like switches and weighting resistors, I even saw such implementation in one zero-IF Rx. However, for signal extraction I'd prefer a classic lock in amplifier because of complexity difference.

                      What I need here is a real analogue multiplier. I wish to separate phase detection from detection amplitude because of a distinctive advantage: ground balance for both resistive and reactive components - all the time.

                      I'll give an overview of what I want in a few posts to make it easier to follow. First is the "classic" VLF constellation, without the discrimination angles, say an "all metal" response. It is typically used for target response level. In 2-quadrant Rx only the greenish lobe is used:
                      Attached Files

                      Comment


                      • #26
                        If I got it right, and I'm sure Sergey will correct me if I'm wrong, his Verator uses the following tone extraction scheme which uses a Disc signal (the lobes) and punches an additional hole in the left lobe (dashed line) for suppression of the reactive ground signal. IMHO just brilliant! It makes Verator a perfect rig for beach combing.
                        Attached Files

                        Comment


                        • #27
                          So, you can observe the channel that produces a tone as a masking function - it masks ground signal. In classic solutions it masks ferrite-like responses, in Verator it primarily masks high R responses like those of the sea water and additionally it punches a hole in ferrite-like response angle to make it just perfect. Verator does that by means of a VCA (2-quadrant multiplier) produced as a PWM controlled switched capacitor attenuator (phew!).

                          What I want to achieve is a tone response mask as a separate function from phase discrimination vectors. Thus I'd have ground responses at bay regardless of my discrimination settings. It means that my Fe and Cu tones may overlap, but the sea water response will be attenuated anyway. A perfect beach analyser! With 4-quadrant discriminator and a differential coil it will eat the existing PI toys on their own terrain.

                          The mask I'm talking about is a clover-like with 4 deep holes fixed at ferrites and sea water. Discrimination vectors may be set freely and independently (red and green).

                          To obtain that clover-like mask I'll need a real analogue multiplier. It is produced by multiplication of two channels at angle.

                          Of course, patent trolls go to hell!
                          Attached Files

                          Comment


                          • #28
                            Ah, I just noticed that I've put R and X coordinates wrong ... anyway, you got some idea what I wanted to say.

                            Comment


                            • #29
                              Click image for larger version

Name:	fff.GIF
Views:	1
Size:	14.9 KB
ID:	333803

                              Comment


                              • #30
                                In my analysis of Verator rig I concluded that the signal produced by the X path is the one with left and right lobes since it is switched in phase with Tx, and that would produce a nice pair of holes at the sea water vector. So in any case and regardless of the differential coil it would outperform nowadays VLF detectors on the beaches. Even more so with the differential coil. The "No Signal" part of the mask I indicated as white in order to emphasize that I'll use it in a 4-quadrant solution.

                                Otherwise we are on the same track. However complex to follow, a Verator design is a very nice and unique rig.

                                Comment

                                Working...
                                X