Announcement

Collapse
No announcement yet.

Target response tester

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

  • Originally posted by Skippy View Post
    Green : maths for exponential decay:

    V = voltage; V0 = starting voltage; t = time difference; TC = time-constant, Tau.

    V = V0 e -(t / TC)

    => ln (V / V0) = -t / TC

    and: ln (V0 / V) = t / TC

    You do have some dodgy 'scope ADC issues, I had noticed them before. I've forgotten ... your scope data ends up in an Excel table ? You could try the simple filter I posted in post #93.

    Your plots of the 19mm straight wire length and the loops of the same gauge wire appear to show the same initial drop, which I would expect. There doesn't seem to be any evidence of a 'pause' at 22usec rate for any time, before the dominant full loop decay starts.
    Thanks for the formula.

    Tried changing scope scale to get more resolution(smoother trace) with a clad and silver quarter. Tried charting on the same chart but they were on top of each other. Charted on 2 charts, TC lines were added to right chart first then same slope lines were added to left chart. Right chart lot smoother. Negative:minimum input not as low, Positive: smoother and stays out of noise.

    Looking at charts I see I labeled the Y axis scope scale zero's opposite.
    Attached Files

    Comment


    • Originally posted by Skippy View Post
      Green : maths for exponential decay:

      V = voltage; V0 = starting voltage; t = time difference; TC = time-constant, Tau.

      V = V0 e -(t / TC)

      => ln (V / V0) = -t / TC

      and: ln (V0 / V) = t / TC

      You do have some dodgy 'scope ADC issues, I had noticed them before. I've forgotten ... your scope data ends up in an Excel table ? You could try the simple filter I posted in post #93.

      Your plots of the 19mm straight wire length and the loops of the same gauge wire appear to show the same initial drop, which I would expect. There doesn't seem to be any evidence of a 'pause' at 22usec rate for any time, before the dominant full loop decay starts.
      A simple one to try is: Filtered sample value = (0.25 x Previous sample value + 0.50 x Sample value + 0.25 x next sample value)
      Tried filter. Added .04V, one division to original when charting so they wouldn't chart on top of each other.
      Attached Files

      Comment


      • I did a modification to my tester by adding a switch to change the series resistor in the coil circuit. This gives me a choice of 250 or 750mA peak current in the coil. The attached plot uses the higher current and uplifts the amplitude at late times which will enable TC start measurements of 400 and 500uS. I did a plot to compare the differences in the decay of the clad and silver quarters. TX is 1mS long and the top of the trace is the amplifier saturation level. The silver quarter is obviously in saturation for longer, hence the shift at the top. I haven't yet done final TC measurements at these later times.

        Click image for larger version

Name:	Scan_20190211.jpg
Views:	1
Size:	145.4 KB
ID:	353335

        Eric.

        Comment


        • Originally posted by Ferric Toes View Post
          I did a modification to my tester by adding a switch to change the series resistor in the coil circuit. This gives me a choice of 250 or 750mA peak current in the coil. The attached plot uses the higher current and uplifts the amplitude at late times which will enable TC start measurements of 400 and 500uS. I did a plot to compare the differences in the decay of the clad and silver quarters. TX is 1mS long and the top of the trace is the amplifier saturation level. The silver quarter is obviously in saturation for longer, hence the shift at the top. I haven't yet done final TC measurements at these later times.

          [ATTACH]45371[/ATTACH]






          Eric.
          Are your coins located same position on the coil? My clad and silver look a lot closer when I record them. It's like our silver quarters are a lot different. Mine is 1947, what is yours?

          Looking at your graphs reply#204, looks like amplitude might be closer to mine. Have you got a recording with silver and clad on the same chart at the lower Tx current level?
          Attached Files
          Last edited by green; 02-11-2019, 09:45 PM. Reason: added sentence

          Comment


          • Green: It looks like the simple filter is cleaning up the jitter between voltage transitions a bit, most noticeable at the end of the decay. But it's clearly not going to make any difference to the scope's uneven step size - if the scope reads the same value for 10 usecs, an average of 3 samples is not going to have any effect at all. I guess it's useful to have in the 'toolkit' for the noisiest traces, ie. real noise, not repeatable test instrument flaws.

            Comment


            • Originally posted by green View Post
              Are your coins located same position on the coil? My clad and silver look a lot closer when I record them. It's like our silver quarters are a lot different. Mine is 1947, what is yours?

              Looking at your graphs reply#204, looks like amplitude might be closer to mine. Have you got a recording with silver and clad on the same chart at the lower Tx current level?
              The coins are mounted in a close fitting plastic pot that positions them accurately in the middle of the 45mm diameter x 50mm long solenoid coil. The field lines are then perpendicular to the surface over the the whole area. My silver quarter is 1964 and it is the only one I have. Attached is a plot at the lower current level with the same graph scaling as the previous one at the higher current.
              According to Google, the silver content is the same. '1932 – 1964 Silver Quarter is 90% silver'.

              Click image for larger version

Name:	Scan_20190212.jpg
Views:	1
Size:	142.8 KB
ID:	353338

              Eric.

              Comment


              • Originally posted by Ferric Toes View Post
                The coins are mounted in a close fitting plastic pot that positions them accurately in the middle of the 45mm diameter x 50mm long solenoid coil. The field lines are then perpendicular to the surface over the the whole area. My silver quarter is 1964 and it is the only one I have. Attached is a plot at the lower current level with the same graph scaling as the previous one at the higher current.
                According to Google, the silver content is the same. '1932 – 1964 Silver Quarter is 90% silver'.

                [ATTACH]45374[/ATTACH]

                Eric.
                Thanks
                Wasn't thinking, your graphs reply#213(higher Tx current) didn't come out of saturation near same time. Mine wouldn't at 250usec either. Maybe my figure eight coil or coin position is causing our silver coins to chart different. I'll try some things. The 1947 quarter is the only silver I have.

                Comment


                • Tried some things. Was thinking straight line decay after one to two time constant's. All decay's did straight line decay. Starting amplitude seems to effect when straight line decay starts. Don't know why or if it's real. Changing scope scale, constant current or constant rate Tx doesn't seem to effect when straight line decay starts. Starting amplitude is effected by target position, maybe position effected decay. Still get faster silver quarter decay than Eric. Mine shows some wear, not sure why decay is different yet. This is an exercise in measurement. Other than trying to understand why or what happens, not sure it's important to detecting. Maybe need to get a set of targets, 2 or 3 of each that chart the same for me. Keep one set and send some to England. Skippy mentioned rings while back. Maybe rings and plates? The question is, can we or are we still learning something? Building the tester with log out has been interesting. Learned a lot, still have some questions.

                  Constant current Tx(.5A peak) does seem to straight line decay a little sooner than constant rate Tx(1A peak) at the same target location. Constant rate, higher amplitude though so is it position, amplitude or something else?
                  Attached Files
                  Last edited by green; 02-12-2019, 05:49 PM. Reason: added sentence

                  Comment


                  • I had a look at another method of measuring the TC of an object with my Mk2 unit and Picoscope. That was to set up the decay waveform and measure the TX width at which the waveform started to change, indicating that the TX was passing the optimum point of width for that object. The easiest way was to set the width high i.e. > 1mS; put a time cursor on a steep part of the waveform, and then reduce the TX width until the waveform just started to move to the left of the cursor. The scope being triggered by the TXoff edge. My instrument has the benefit of a 10 turn pot as a fine adjustment of the TX width. The movement is quite small to start with as we are at the 5 Tau point (99%) but I was using 100uS per large division on the X axis and could resolve 1uS of movement. Taking this to be the 5 Tau point, I then switched to the TXon edge to measure the width of the TX pulse. It is then just a matter of dividing by 5 to get the TC. The TX widths I got were 650uS for the clad and 850uS for the silver, This gives TC's of 130uS for clad and 170uS for silver quarters. There is not much change for a 4 Tau TX width as we are still at 98%, but below that, the decay waveform gets noticeably shorter.

                    Carl's comment in the thread 'pi discriminator of metals' spurred me to look at this.

                    Eric.

                    Comment


                    • Originally posted by Ferric Toes View Post
                      I had a look at another method of measuring the TC of an object with my Mk2 unit and Picoscope. That was to set up the decay waveform and measure the TX width at which the waveform started to change, indicating that the TX was passing the optimum point of width for that object. The easiest way was to set the width high i.e. > 1mS; put a time cursor on a steep part of the waveform, and then reduce the TX width until the waveform just started to move to the left of the cursor. The scope being triggered by the TXoff edge. My instrument has the benefit of a 10 turn pot as a fine adjustment of the TX width. The movement is quite small to start with as we are at the 5 Tau point (99%) but I was using 100uS per large division on the X axis and could resolve 1uS of movement. Taking this to be the 5 Tau point, I then switched to the TXon edge to measure the width of the TX pulse. It is then just a matter of dividing by 5 to get the TC. The TX widths I got were 650uS for the clad and 850uS for the silver, This gives TC's of 130uS for clad and 170uS for silver quarters. There is not much change for a 4 Tau TX width as we are still at 98%, but below that, the decay waveform gets noticeably shorter.

                      Carl's comment in the thread 'pi discriminator of metals' spurred me to look at this.

                      Eric.
                      Interesting. Thought I would give it a try. Don't have enough resolution with my scope. Charted clad quarter decay at 150, 300, 450, 600, 750, 900 and 1050us Tx. Looked like voltage was about 63% of volts at 1050us with Tx=150us. Made a chart for starting at 2 or 3V with long Tx time(at least 6x target tau), wondering if it might be correct. Could you try with your tester? Start with 1ms Tx. Adjust cursor at 3V(time at 3V), then adjust Tx time for 1.8964V(1 tau)at same time and record Tx time.
                      Attached Files

                      Comment


                      • Charted some copper rings. Charted to see the effect of doubling diameter and increasing resistance eight times. 20 and 40mm diameter with AWG 19 and AWG 10. AWG 19 eight times the resistance AWG 10.
                        Attached Files

                        Comment


                        • Charted the HyperPhysics chart with 10 equal minor divisions/decade to match the charts, log out with the Target Response Tester. Can compare calculated change in detection distance for 1minor change with different size coils at different detection distances.

                          forgot to label X axis, target distance mm
                          Attached Files

                          Comment


                          • I've done a few more refinements to my Mk2 test setup and another set of measurements on clad and silver quarters. I could not fully comply with the suggestions in post 220 but made tests on both coins starting at 200, 400, and 600uS after Toff with TX width of 1mS. I expanded the Picoscope traces as much as possible and took readings from the Y cursors at the three different measurement start points. I then took 36.8% of that reading in mV to find the level for the second cursor. The time difference for the TC was read from the two X axis cursors.

                            Starting with the clad at 200uS start TC = 131.8uS. At 400uS, TC = 133.4uS. At 600uS, TC = 147uS.
                            For the silver quarter at 200uS, start TC = 166.1uS, At 400uS, TC = 172.3uS. At 600uS, TC = 183uS.

                            Virtually the same results occurred with both 250 and 750mA TX levels (constant current).

                            It looks as though the single exponential occurs quite late in the decay, probably starting around 600uS (a bit less for the clad), but becomes more difficult to measure accurately as the signal decays into the noise level. The TX amplitude could be increased to lift these lower signal levels, but even at 750mA the sensor coil is warming up and I wouldn't like to take it above 1A.

                            The other method of measuring TC was to take the whole decay to -99% and divide by five. This really neglects diffusion effects at early and mid times and gives an average figure over the whole decay. This may be more appropriate for wire rings using thin wire.

                            This has been an interesting exercise and useful for me in helping to refine my test instrumentation for magnetic soil measurement. The Mk2 instrument I am working on, hopefully can do both this and conductivity measurements on conductive objects as well. A recently acquired book entitled 'Soil Magnetism - Applications in Pedology, Environmental Science and Pedology' by Neli Jordanova shows how important this subject is not only for metal and mine detection, but in other areas of scientific interest.
                            My posts on the subject of target responses to coins, nuggets etc will now taper off for two reasons. One is that I am shortly moving house and workshop to a new location and there is a lot of packing and dismantling to do over the next couple of months. The other is that I must eventually get back to magnetic soil testing so that I can demonstrate an improved PI intrument in this area. (Magnetic Viscosity Meter Mk2).

                            Health has also been an issue for me for almost a year but last week was given the 'all clear' to carry on normal life. Having been confined to largely stay indoors, practical metal detecting was not an option. Doing a bit of development and experimental work helped greatly in preventing boredom during that period, but in this coming year I want to get out more and as the new location is close to the beach, I will no doubt be again spending some time detecting.

                            I shall certainly follow what Green, Skippy and others come up with on this subject and make short posts, if appropriate. However, I must start packing up my test gear this week and will not be able to do further measurements.

                            Eric.

                            Comment


                            • Hi Eric

                              Good luck on your move. Thanks for the testing and suggestions. Looking forward to more after your move.

                              Comment


                              • Yes, best wishes for your house move, Eric.

                                I may as well add that I'm going to be very busy for the next month due to a sad family-related event. So don't expect too much from me, though I'll try.

                                Comment

                                Working...
                                X