Announcement

Collapse
No announcement yet.

Goldscan IV, 4 position switch question?

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

  • #16
    Hi Reg,

    Thanks for the drawings. I was thinking along the lines of your second drawing except the opposite sides of the diodes would be connected to the switches. So basically, instead of the gain for that path being zero(by shorting with the switch), that path is essentially open circuited with a switch instead.

    Thanks again,
    Brian

    Comment


    • #17
      Reg,

      I've been playing around with my homebrew GS4, and I'm having difficulty getting the Low/High conductive modes working. It turns out, that on my board, the Gold channel always overpowers the Iron channel regardless of how the ground balance is adjusted.
      In other words, I never get a negative target signal at the summing(GND) pot no matter what my target is...even a large piece of high conductive metal. The ground balancing appears to work as I can cancel out a large piece of metal as I move it towards the coil. I could probably fix this by adding some gain in the iron channel or increasing the value of the 22K resistor from the Gold channel to the GND pot, but I'm wondering if you have experienced this, or know of a schematic error in this area?

      Thanks,
      Brian

      Comment


      • #18
        Hi Brian,

        To be honest, I haven't taken a microscope to the schematic so I can't say there are not other errors. However, this design uses two identical channels so they should have equal gain in the various stages up to the GB pot area. So, if you keep that in mind you can't miss.

        Now, the easiest metal to get a negative response on is either silver or copper. So, you might try them first.

        Next, what value do you have for a coil inductance? If this is way off, you can get strange results. I personally use the most common inductance of 300 uh. Also, if you have the delay less than 15 usec, it takes more iron signal to ground balance.

        If you have the adjustments off on the various timing networks, the same problem can occur. You can set the pulse width at 100 usec and it will work fine. The sample width can be 10 usec to 15 usec. Play with the iron channel delay. If it is out too far, you can get the results you indicated. An actual delay of 10 usec to 15 usec is sufficient between the two channels. So, you might try 10 usec for the gold channel and 30 to 35 usec for the iron channel. BTW, I usually call the iron channel the GB channel.

        So, there are lots of signals to check that could cause what you are experiencing.

        Now, make sure to look at the output of U13b to see if it ever goes negative.

        BTW, I changed the quad opamps to very low offset IC's so the need of zeroing the amps becomes basically no problem. The LT1014 or OPA4277 work well but can be pricey. I stumbled upon a good deal on the OPA's on Ebay. If the standard IC is used, the offset balance adjustments used to zero the amps become very critical and can cause all kinds of problems if not correct.

        Reg

        Comment


        • #19
          Hi Reg,

          Thanks, I'll check the timings and my component values to see if I messed something up. I also made sure the pulse delay isn't too short which would make the gold channel too 'hot'. I'm using TDI coils for testing, and I have tried a solid silver spoon for testing, as well as a copper clad PCB material. With these targets, the Gold channel still wins.

          I'll let you know what I find after some more testing,

          Thanks again,
          Brian

          Comment


          • #20
            I would use a thicker piece of copper. Use a penny if you are in the US. If not, a simple copper coupling for pipe will work.

            See if a nail about 3" long changes tones.

            Reg

            Comment


            • #21
              Reg,

              It looks like my timings might be messed up. From the falling edge of the TX pulse(250uS wide), I get the following timings(approximately):

              Q5: 300uS
              Q6: 75uS
              Q7: 350uS
              Q8: 30uS

              Sample widths: 20uS

              The secondary delay on the GB channel looks long to me, but maybe that's ok?

              -Brian

              Comment


              • #22
                Hi Brian,

                Your initial delay of 30 usec is way too long making it difficult to even ground balance, let alone differentiate metals by conductivity.

                Adjust this delay down to between 10 and 15 usec and try again.

                As for the 250 usec pulse, this will work but using a 100 usec pulse will do almost as good depth wise but use much less current. Also, the shorter pulse time will make it easier to obtain a working 10 to 15 usec initial delay.

                Reg

                Comment


                • #23
                  Reg,

                  My measurements were taken from the falling edge of the TX pulse, not the flyback signal. The decaying portion of the flyback signal looks OK in relation to first sample. I'm sampling at about 15uS minimum.

                  The suggestion for a smaller pulse width is a something I'll try as well.

                  Do my secondary delays look Ok? Also, I seem to have roughly 50uS between the two channels. Maybe I should make it 35uS as you suggested.

                  Thanks,
                  Brian

                  Comment


                  • #24
                    "My measurements were taken from the falling edge of the TX pulse, not the flyback signal. The decaying portion of the flyback signal looks OK in relation to first sample. I'm sampling at about 15uS minimum."

                    I don't understand the statement above at all. The falling edge of the TX pulse should be when the flyback signal starts. The 10 usec to 15 usec measurement should begin when the flyback starts.

                    Reg

                    Comment


                    • #25
                      I made the mistake of measuring from the fall of the TX pulse to the fall of the first sample pulse. So, if I subtract the sample width of 15uS, I get a minimum sample delay of 15uS. I'm hoping to improve upon this with a two stage preamp.

                      thanks,
                      Brian

                      Comment


                      • #26
                        Just realized that the minimum capacitor value stated in the spec for the HCF4538 that I used in the timing circuit, is 4nF. Oops.

                        Comment


                        • #27
                          legendary pi detector goldskan 4
                          Attached Files

                          Comment


                          • #28
                            One other thing I should mention is, when I refer to positive or negative responses, this can't mean the tone will change to a positive or negative signal, but, rather which of the two diodes will conduct on U13B.

                            Because of the fact the two signals from the diodes will ultimately result in a positive tone, then shutting off one of the two diodes will still have the same audio response but will simply cause the detector to ignore either high or low conductor objects.

                            In other words, the positive or negative response I referred to before has to do with which diode will conduct and what the polarity of the output of U13B will be.

                            Maybe, this will make more sense now.

                            Reg

                            Comment


                            • #29
                              D15 and D14 are not oa 85 they were originally 1n4148 signal is positive polarity.
                              Last edited by Orbit; 04-24-2011, 02:44 PM. Reason: error in the answer

                              Comment


                              • #30
                                Hi Orbit!
                                Can you make more pic's on Goldscan pcb?
                                10x!

                                Comment

                                Working...
                                X