Announcement

Collapse
No announcement yet.

MPP Ground Balance

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

  • MPP Ground Balance

    Ok so I hooked my MPP up to my chipkit UNO and have been playing with GB. As its in the workshop I am unable to test properly on the ground but would like some feedback on my methodology.

    Timings:

    TX pulse width 58us

    Main Sample Delay 20us
    Main Sample Width 20us

    GB Sample Delay adjustable from 27us to 43 us after end of main sample
    GB Sample Width 10 us

    EF Sample Delay 150us after GB sample
    EF Sample width 10us

    The main Sample is via Q3 and the EF and GB samples are via Q4 so the result is (Main - GB) - EF
    I am also going to look at making the EF width variable as well

    At the moment I am detecting a 5mm sq aluminium (from can) 27us EF delay = 50mm and 43us EF delay = 65mm using small fast coil.

    Feedback please

  • #2
    Errr... no. GB to be effective must resemble pause/pulse relationship of the main sample, and EF must be their difference in duration. Your scheme may work if you change gains of inverting/non-inverting channels. Also, you may wish to process audio after absolute value circuit, or otherwise you'll have response to short tc targets within a coil perimeter, and for large tc targets outside.

    Considering your main pause/sample is 20:20us, and you wish to tune by GB sample position, and it starts at average 35us after main sample - which is a good choice, its duration should be 20+20+35=75us, providing you with average 75:75us pause/pulse relationship. The EF duration should be 75-20=55us, and position as far as practical bordering with the onset of the next Tx pulse.

    Optionally as for Tx pulse duration, I'd suggest resistor-limited duration over entire main+GB samples (not EF though), 20+20+35+75=150us, which would yield constant current performance, but also greater battery drain. You may consider this if you intend to go for extreme OZ soils because it will considerably improve GB, and to some degree also target responses.

    Edit: you may wish to keep the original MPP timings for easy terrains, as you'll have better depth that way. Every GB scheme reduces sensitivity.

    Comment


    • #3
      Thanks for that info. Its pretty easy to change the timings as its all just code. I am using a 14.8v lipo battery so I can run a longer TX pulse with a current limit resistor.
      I want to reduce the main sample delay down to 5us min and have it adjustable up to about 100us. I can then use code to calculate the various delay/pulse values as per your info.
      Is there an advantage for a smaller/larger main sample?

      Comment


      • #4
        Originally posted by Davor View Post
        The EF duration should be 75-20=55us, and position as far as practical bordering with the onset of the next Tx pulse.
        Is the 20us you used in the EF calculation, the main sample delay or the main sample width?

        Comment


        • #5
          In PI you must find a balance between a few things, so having option to adjust timings is a plus. If you sample to early, you gain some sensitivity to very small targets and short tc-s, but you lose depth as the coil is still discharging. If you sample too late, you already lost all the small stuff, and again you lose on depth.

          Main sample longer - more energy is integrated and you get better S/N. You get the same by shorter sample and faster repetition rate. These are in balance.

          Ground, as in viscous effects, decays much slower, and in log-log scale, and you may always apply some sort of ratios scheme that cancels it. Therefore, you wish to include sample timing variability, so you need to keep the ratios roughly the same over the same terrain.

          With GB you introduce a hole in response of certain tc, and the easiest way of removing a hole is to simply toggle between two base durations upon which you set the ratios of other samples. With this you may even have some crude discrimination, but of a sort that does not differentiate iron.

          Comment


          • #6
            Originally posted by CAS View Post
            Is the 20us you used in the EF calculation, the main sample delay or the main sample width?
            It is the main sample width. With EF you are balancing the main sample and GB sample, so that any offset gets cancelled.

            Comment


            • #7
              Davor, thanks very much for all your help. It is really appreciated.

              Comment


              • #8
                Got all the timings sorted but was getting strange clicks in audio. Seems that using higher battery voltage and longer tx pulses is causing breakdown of the IRF740. Eventually killed the FET. When I get to work tomorrow I will change it for a higher voltage one and perhaps drop supply voltage back down to 12v.

                Comment


                • #9
                  Alternatively you can add a snubber circuit.

                  Comment


                  • #10
                    Thats an option but my idea with using the MPP was only as a test platform and not to have to add too much to the circuit. Once I get all this sorted and I am happy with it, maybe then I will modify the design.

                    Comment


                    • #11
                      Sure, that's sensible.
                      You could alternatively put a larger resistor to further limit the current to, say, 1A and go with plain old IRF7xx.
                      I tend to analyse what mods are to keep and put them right on. My problem is that I can afford only a small portion of my time and effort to actual soldering, but I often steal 10 to 15 minutes of quality time with LTspice. MP+ is also a learning platform for me.

                      Comment


                      • #12
                        Ok well I replaced it with a FPQ1N60C - higher voltage but higher on resistance offset with lower capacitance. All works fine. I may even try a higher series resister later and see what effect that has.

                        Timings:
                        TX pulse width 150us

                        Main Sample Delay 10us
                        Main Sample Width 58us

                        GB Sample Delay 33us
                        GB Sample Width 101 us

                        EF Sample Delay 502us after GB sample
                        EF Sample width 41us

                        The TX width and Main Sample delay are adjustable and the ratios vary accordingly.

                        With the settings above I detected my 5mm sq of al (cut from drink can) at 10cm. A $1 AUD coin at 22cm all using a 6" coil. Still air tests and in workshop with lots of fluros and electrical devices operating.

                        Comment


                        • #13
                          Ratios are off. For GB you need main sample width to main sample delay ratio to be in average equal to GB sample width to the sum of (main sample delay + main sample width + GB sample delay)

                          By varying any of the delays you may tune GB to perfection and keep the same EF arithmetic.

                          Comment


                          • #14
                            Ah ok I see where I went wrong. I will have to make the main sample width a lot smaller - maybe 10us because if I use my current settings the GB sample width will be huge.

                            Comment


                            • #15
                              It all seems to be working now with the correct ratio's. I have an adjustable TX width and adjustable Main Sample Delay. Even can have various main sample widths without causing timing issues. A width of about 25us seems to be good overall.

                              All the ratios are now correct as per Davor's info. There is a very slight decrease in depth but overall I am happy with the results.

                              I have waved a mass of ferrite cores over the coil with no effect whatsoever. I will grab a couple of red house bricks from home tomorrow and check that.

                              Would there be any advantage in an adjustable GB delay? I was thinking of trying a method of one TX period having x GB delay, then the next TX period having y GB delay, next period back to x delay and so on. Maybe this alternating GB delay will lessen the 'hole' issue?
                              Last edited by CAS; 03-06-2015, 05:25 AM. Reason: added more info

                              Comment

                              Working...
                              X