Announcement

Collapse
No announcement yet.

ground vs target

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

  • #16
    In this case the Target residual on the lower input (geb) will reduce the differential gain.

    Or Sens to a target.


    So, why have a filter on GEB ch which is centered to pass 9Hz Target residuals - which reduce Sens?
    Attached Files

    Comment


    • #17
      Hi golfnut

      I've just looked at the IGSL schematic and I can't see an instrumentation amplifier configuration anywhere. U7 a,b c,d is shown as an LM339, which is a quad comparator IC with open collector outputs. The outputs are linked in pairs such that both outputs have to go "high" (actually open circuit) for the combined output to be pulled high, and an audio output to be produced. A similar arrangement is used in the TGSL.

      Or maybe I'm looking at the wrong version of the schematic? Which version are you using?

      Gwil

      Comment


      • #18
        look for a part of IGSL that mirrors this.. Its just a buffered Diff amp.
        Attached Files

        Comment


        • #19
          Do you mean this?

          Sorry, but this isn't an instrumentation amplifier - for one thing it has no differential stage.

          It still looks to me like a pair of comparators with a common output.

          Gwil
          Attached Files

          Comment


          • #20
            Originally posted by Gwil View Post
            Do you mean this?

            Sorry, but this isn't an instrumentation amplifier - for one thing it has no differential stage.

            It still looks to me like a pair of comparators with a common output.

            Gwil
            It's two open-collector comparators in a wire-ANDed configuration. In other words, both outputs must go high before a signal is passed to the next stage.

            Comment


            • #21
              Thanks. Red face .... I got that wrong. Epic fail .


              I feel as if Im getting there but it is a rocky road.



              Disc - Big Target + with some gnd
              Geb - Small gnd + small Target


              Both signals have to 'trip' predefined level set by pot.

              So with the attached if the lower ch is only 10mv it doesnt 'trip'

              Is it fair to say the Target amplitude on GEB will be the smaller of the two - between Disc and Geb?

              S
              Attached Files

              Comment


              • #22
                So in the absence of a target, we tune up the geb to min signal.


                The gnd sig on GEB ch is very small.
                The gnd sig on Disc is relatively big.


                Set Threshold (SENS on IGSL) to just above the level of the GEB ch gnd signal.


                But it does not false like this - even with the disc ch gnd sig over the Threshold because BOTH of the gnd signals would have to be over the Threshold as they are logic AND'ed together.



                huff, Im sure rest of world knew this from the off. It may of helped me to buy a cheap machine and get used to this stuff rather than jump in on IGSL.

                I had made a wrong assumption early on that we exploit the difference between disc and GEB channels to get high sens.


                Rather, it is both of the channels have to creep over the threshold level to trip a valid target.


                The fact that the GEB ch gets less Target signal thru isnt a big limitation, as, if the GEB is set correctly then the gnd sig is very low and the Threshold can be very low too.


                A small Target signal coming thru GEB only has to be a tiny bit above the level of the Threshold, the same signal on the Disc ch will be bigger, so it will satisfy the other half of the 'AND' requirement - and boom, get your shovel out.

                Im going to a darkened room

                S

                Comment


                • #23
                  Hi golfnut

                  I don't know this for sure but I suspect the target signal coming out of the GEB channel could be fairly large, because as Davor said in post #14 the phase of most target signals is well away from the ground signal so they aren't attennuated very much by the GEB phase sensitive detector.

                  Maybe...

                  Gwil

                  Comment


                  • #24
                    Sure, and using the same argument there will be a fair bit of ground signal in disc ch too.

                    Its clear how this all works now. Finally.



                    It says to me that the sens is governed by the ability to get the ground signal to almost zero - this enables the SENS threshold to be very low - so you can trigger on small targets fractionally above the tuned out/balanced gnd sig level.


                    If you have a software based det you could dynamically set the threshold level automatically - as close as possible to the GEB gnd signal.

                    Simply by monitoring the ground sig level on the GEB channel and adding a small guardband factor to the measurement value - this becomes the current Threshold value almost in real time.

                    S

                    Comment


                    • #25
                      Originally posted by golfnut View Post
                      The gnd sig on GEB ch is very small.
                      The gnd sig on Disc is relatively big.
                      It is the other way around. Think of all metal detection - it comprises only a GEB channel.

                      Originally posted by golfnut View Post
                      Sure, and using the same argument there will be a fair bit of ground signal in disc ch too.
                      ...or not. That would depend on your discrimination criteria. If you align discrimination to give 0V for foil, it will scream in GEB channel, but will give nothing in disc.

                      The way discrimination is set in IGSL you are able to overlap Cu and Fe channels in a way that you accept tin foil (and small gold) as well. Without overlap such targets would be missed entirely.

                      Comment


                      • #26
                        Despite I'm bad educator, I will try to explain why we need BPF (band pass filter) to suppress GND signal and noise.

                        Attached image shows pulses of GND signal (demodulated signal from ground at a coil sweep) and TGT signal (demodualated target signal at the same sweep). For convenience, the pulses are shown separated.
                        In this example, the sweep duration is 1s, the GND signal has duration
                        T/2=0.8s and the TGT signal has duration T/2=0.2s because the target is small relatively to coil diameter and coil moves fast.

                        To design the bandpass filter, we should transform pulses in frequency spectrums.
                        The spectrum of GND signal starts from zero because there is large DC component and contains fundamental frequency Fgnd=2/0.8=2.5Hz and a harmonic 5Hz.
                        The spectrum of TGT signal also has DC component, but it is small and not important. The fundamental frequency is Ftgt=2/0.2=10Hz and there is harmonic 20Hz. In practice we have no spectral information for target above the harmonic 20Hz. That means, to suppress noise, the upper frequency of our BPF should be 20Hz. To suppress the GND signal, the lower frequency of BPF should be
                        somewhere between 5Hz (the harmonic of GND signal) and 10Hz (fundamental frequency of TGT signal).

                        However what happens at convential pinpointing (when the search head moves slowly over target)? The TGT signal will have much larger duration, that means lower fundamental frequency and the BPF will suppress it. Follows that we need other method for pinpointing - the search head should sweep quickly scaning in small sters over target.
                        Attached Files

                        Comment

                        Working...
                        X