Announcement

Collapse
No announcement yet.

PI signal snapshots

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

  • PI signal snapshots

    Does anyone have any scope signal snapshots from a Minelab GP 3000/3500 detector or perhaps a description of the signal?

    Thanks,
    Paul

  • #2
    Actually, I'd be interested in seeing output signal plots of any good PI? Or perhaps a good description of the current. I presume the coil is mostly inductive and therefore the current rises at a decent slope for a given time period ... then what?

    Thanks,
    Paul

    Comment


    • #3
      Hi Paul,

      I screwed up and posted a pic I took quite some time back displaying the pulse signals of the SD 2100. The pic is in the thread where you asked about pics of nuggets.

      Most people do not take pics of the pulse current because it can change simply by changing the pulse width, the FET, any series resistance, etc. What is usually displayed is the decay signal since that is where the sample is taken.

      You might want to read Carl Moreland's Hammerhead info. this helps exlain the workings of a PI. He also has some pics at the end that might help.

      Reg

      Comment


      • #4
        Reg,

        Thanks for the scope pic! Wow, does ML stand for what I really think. Cool. :-)

        I see two different pulse lengths. Is there a lot of high frequency ac or is the scope simply out of focus? In other words, the scope line is very thick. If it's noise then perhaps a very small cap would help a little. I was trying to see if there were any hidden changes in between the low and sudden high spike. You wouldn't happen to have a faster scan pic would you? I sure would like to see a blow up of just one pulse cycle. It's hard to tell but it seems they are simply charging up the coil and then releasing it as quickly as possible, which of course causes the magnetic field to collapse, hence a huge voltage spike. Is that what you see at faster scan rate or is something else happening in between?

        Many thanks,
        Paul

        Comment


        • #5
          Hi Paul,

          The "blurry" current signal is more of a problem of the brightness level saturating the camera. I took those pics a long time ago when a SD was available. Unfortunately, that detector is no longer around.

          Minelab simply pulses the coil using a long pulse and shorter pulse. The do change the pulse duration and the pulse voltage. I am not sure why since changing the pulse duration into a low resistance coil will accomplish the same goal.

          One thing unique about the ML is the decay curve is dramatically different between the long and short pulse. That isn't obvious on the pic previously provided. When looking at the decay curve, the longer pulse has a very long decay curve and the short pulse has a fast decay curve.

          Unfortunately, when trying to display the different curves on a scope, the two are superimposed on each other. I have attached a pic showing this. One just has to remember that there are two different decay curves, one for the short current pulse and one for the longer pulse.

          Reg
          Attached Files

          Comment


          • #6
            Interesting. The bottom line represents ground and you're triggering on the start of the voltage spike. So then what's happening is the coil is charged and then the exact next step is reverse in voltage / voltage-spike. Is this true? I just wanted to be certain that the voltage does not momentarily *pause* at zero just prior to the spike. It's hard to tell, but it seems to voltage simply switches polarity without any pause at zero voltage.

            You don't happen to know if the GP 3000/3500 signals look the same do you?

            Thanks for the pics,
            Paul

            Comment


            • #7
              Hi Paul,

              Sorry, but I can't say just what differences there area between the signals on the SD and the GP's. My best guess is the signals are fundamentally the same, but the applied voltages are a little different. I think that ML is creating the applied voltages to assure the voltages remain constant. Fluctuations result in instability which then shows up as noise or "warbles".

              Now, as for the signal going from the current on pulse to the high spike, no, the signal doesn't stop in between.

              What causes the high spike is the fact the current has been turned off in the coil. Since current can't stop in a coil instantaneously, the spike is generated in an attempt to maintain current flow. The spike then decays at a rate mainly dependent upon the damping resistance. Any oscillation that may occur is the result primarily because of the inducance of the coil and all associated capacitances and resistances. This will include the resistance of the coil, the damping and input resistances, the coil capacitances, FET capacitances and all other associated capacitances. I am sure I have forgotten a few minor things, but the basics are there.

              Reg

              Comment


              • #8
                Hi Reg,
                This is a nice scope picture you posted. I am having a hard time understanding what I see on the picture. Maybe you can help.

                I assume this is a picture of the voltage applied to the coil before waiting to see the target's return signal. From what I can see, it looks like 2 applied voltage signals superinplsed over each other, with the short pulse stopping before the coil has reached a stable voltage level.
                It appears this short pulse decays exponentially immediately after reaching a peak voltage. The second trace appears to be the long pulse which reaches a peak, then oscillates slightly before stablizing around 6.5 volts. The long pulse finally decays exponentially in a curve that looks similar to the short pulse. The only difference I see is the long pulse has a 3 usec stable pulse width at 6.5 volts before decaying.
                Am I reading it right?
                Attached Files

                Comment


                • #9
                  Hmm, now I'm confused because I thought that was the collapsing magnetic field; i.e, just after the coil pulse.

                  Paul

                  Comment


                  • #10
                    Hi J,

                    On the SD 2100, I took all pics just using an external coil and a loading resistor. Now, when you look at the pic I posted on the other thread, you will see the pulse train of this detector. That pulse train indicated two different pulses are happening, a short pulse and a much longer pulse.

                    Now,when trying to look at just what happens when the pulse spike occurs, one can't readily separate the two on the scope and what is displayed is the short pulse decay time and the long pulse decay time with one superimposed on the other.

                    As for the decay time, I usually refer the time from when the FET is turned off, which is basically represented when the voltage suddenly rises. On your pic, this is about the time of the first vertical yellow line. So, the actual decay time would be from the first yellow vertical line to when the particular decay curve levels off. This would be the time needed before a sample could be taken.

                    The actual pulse applied signal is not present on this particular pic. If you go back and look at the pic I attached under the thread of electrical resistance of gold nuggets, you can see the actual coil signals of the SD as seen using a coil and a loading resistor spaced a finite distance from the ML coil. On this pic you will see the current pulse or current on time as the small vertical down step before each high spike. I have labeled the LP coil current time, where LP stands for long pulse.

                    Time wise, the LP coil current is on about 250 usec or so. The short pulse time is about 50 usec

                    Hope this helps.

                    Reg

                    Comment


                    • #11
                      Hi Paul,

                      You are correct, the two signals superimposed are in fact the signals that are generated by the collapsing field. One of the reasons ML has such different decay times between the long pulse and the short one is the fact that ML tries to recover some of the energy used in the long pulse. So, it tries to divert some current into a capacitor where it can be used again. This results in a much longer decay time.

                      Reg

                      Comment


                      • #12
                        Reg,

                        How many micro seconds delay after the pulse is a good time to detect the nugget? I see 10 divisions on your scope-- standard. The short pulse peaks after 1 micro second and is nearly decayed 2 micro seconds after peak. You think 2 micro seconds after peak pulse is good time to sample?

                        Also, I'm curious about the characteristics of your receiving coil. I mean, one micro second is fast for a typical coil. Do you think the rounded ness of the voltage spikes are partially or perhaps mostly due to your receiving method by any chance. To detect such short pulse I'd probably use a very small coil, like 1" diameter max. That way you could record a more precise signal. Just wondering.

                        Paul

                        Comment


                        • #13
                          Hi Paul,

                          How fast you sample will determine the size of the gold you can detect. A very fast sampling time of 10 usec will work for most purposes but it will still be beat by a good VLF gold machine. A PI sampling at 10 usec will detect gold down to 1 to 2 grains or so, depending upon the gold itself. It is nice to be able to sample sooner but few people are able to accomplish it. Special techniques are required to make much headway less than the 10 usec mark.

                          It has been estimated the SD has a sample time of about 15 usec. The GP series is probably a little quicker, maybe 13 usec or so. Both of these two detectors will ignore extremely small gold or certain nuggets with very strange surface characteristics. That is to be expected since they are designed mainly to find large gold that is buried deep. The longer pulse time of 250 usec and the heavy pulse current indicate that.

                          Such machines work great in Australia where there is a lot of large gold nuggets. They work here great also, but do miss some of the extremely small gold or speciality gold simply because of the longer pulse length and heavier currents. They will find very small gold even down to 1 grain or so under certain circumstances, but very irregular surface or porous gold becomes much more of a problem, especially on the small stuff. A lower limit of about 3 grains is more common, although, as mentioned before, certain nuggets even down to 1 grain can be detected by the ML's under the right conditions.

                          I have a couple of "invisible" nuggets that, if I remember correctly weigh about 8 grains or so and they can't be detected by most PI's including the ML's. These particular nuggets have a very unique surface quality. When my PI was set for a delay of 10 usec, I could just barely detect these nuggets. I had to get the delay down to less than 8 usec to get a decent signal.

                          Now, most of the gold here in the US is way less than 1/4 oz, with most probably being under 1 gram, so real heavy currents or long pulse times are not needed. In fact, I have used pulse widths of about 20 usec just fine and small pulse currents. At such short times, my guess is the currents are between .1 amp and .2 amp peak.

                          On my small PI's, I am sampling at about 7 usec or so. At this rate, I am using a 11" coil or something similar in size. I do have different shape and type coils including rectangular ones I use also. Such short delays are extremely difficult to obtain even on low powered units.

                          My recommendation is anyone who wants to learn about PI's should build the Hammerhead. This is a really good unit to begin with.

                          Now, on a detector such as the Hammerhead, one could easily set the pulse width at 50 usec and detect much of the gold normally found here. This pulse setting will easily detect very large gold as well as small stuff.

                          A DD coil helps on the Hammerhead but reducing the current will help the DD work better also.

                          One could build a ground canceling coil and that really helps with ground signals.

                          One could also limit the coil current some and not lose much depth if any. If the pulse sample time can be shortened, then one might even gain depth.

                          The bottom line is a lot of experimenting can be done with something like the Hammerhead to see just what happens. This detector is really a good one to experiment with to see just what happens and what can be done.

                          Now, for those who have a willingness to experiment, one could build a ground canceling circuit for this machine. This would enhance things further. However, I don't recommend one building something like that until they have mastered much of the rest that has to be learned.

                          I have mentioned the Hammerhead a lot because it is a decent PI with a lot of capability. Carl has done an excellent job of providing information about the building and set up of one. Once built, one can experiment a lot so see just what works best. This is far cheaper and better than taking a much more expensive one and trying to work with it.

                          Reg

                          Comment


                          • #14
                            Reg,

                            Originally posted by Reg
                            How fast you sample will determine the size of the gold you can detect. A very fast sampling time of 10 usec will work for most purposes but it will still be beat by a good VLF gold machine. A PI sampling at 10 usec will detect gold down to 1 to 2 grains or so, depending upon the gold itself. It is nice to be able to sample sooner but few people are able to accomplish it. Special techniques are required to make much headway less than the 10 usec mark.
                            I'm curious why sampling faster than 10 usec is difficult. Is is due to hot-rock signals, or too much noise from the coil, or what? I'm certain it's not an electronic limitation since 10 usec is nothing for modern chips.

                            In reference to VLF's being better than PI's at small gold. Is this in reference only to dual frequency VLF's or perhaps a new VLF technique I'm unaware of?

                            Thanks,
                            Paul

                            Comment


                            • #15
                              Hi Paul,

                              The sampling problem of not being able to sample less than 10 usec has to do with a combination of things, mainly the coil. However, the amplifier used as a preamp also has a large part as does the FET and all associated circuitry. It is also extremely difficult to get a coil to dampen correctly in the time needed to samle less than 10 usec, simply because of the associated capacitances, among other things.

                              The preamp needs to be extremely fast and extremely quiet, otherwise serious sensitivity is lost. Unfortunately, it has been extremely difficult to find an amp that is better than the NE 5534 and it is not the fastest in the world.

                              Now as for your statement regarding VLF's, the gold units are simply detectors operating at higher frequencies that have extremely high sensitivity. Higher frequencies, by nature will detect smaller gold better simply because of how they work.

                              Once again, the size limitation is somewhere around 1 grain to 3 grains. VLF detectors designed for gold hunting, generally have a frequency of close to 20khz or greater. Those in the 50khz and above are the ones that do better on the really small gold, the 1 to 2 grain or less size.

                              The more you study on the subject, the more this will become clear. One thing to look for is eddy current response and how frequencies affect this.

                              Reg

                              Comment

                              Working...
                              X