Announcement

Collapse
No announcement yet.

PI limit to early sampling

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

  • #61
    Originally posted by moodz View Post
    With respect I think your original analysis may be incorrect. It can be shown ( with this cct ) that the coil decay is proportional to the inductance and the current in the coil ie being representative of the energy stored in the H field at turn off and relatively insensitive to coil resistance. In fact its only dependant on the resonant frequency of the coil ( ie L and C ) ...
    The OP was asking if an active damping technique had been tried ... this is an effective active damping technique that is insensitive to temp and mechanical coil changes ... ie damping will always track optimally.
    I did not explain myself correctly. Your damping method uses a constant current sink (MOSFET in saturation zone) instead of a resistor. This surely discharges the coil much faster, but there's a catch... at the end of the transient the MOSFET is back into linear zone (resistive) with a LOW resistance. At this point the coil is paralleled to a LOW resistance and therefore its decay becomes longer. The signal is then a composition of the longer coil decay MINUS the target's signal. Good enough to determine whether there's a target or not, but it prevents analysis of the target or ground signals.

    This problem disappears if your method is used on a Tx coil and detection on a separate Rx coil that's induction balanced. Then you get the best of both worlds. I'm actually designing such a detector but I'll use an inductor as a constant current sink rather than a MOSFET because it does the same while being more predictable.

    If you're interested in the (confidential) details you can PM me, perhaps develop a cooperation.

    Comment


    • #62
      Originally posted by Teleno View Post
      I did not explain myself correctly. Your damping method uses a constant current sink (MOSFET in saturation zone) instead of a resistor. This surely discharges the coil much faster, but there's a catch... at the end of the transient the MOSFET is back into linear zone (resistive) with a LOW resistance. At this point the coil is paralleled to a LOW resistance and therefore its decay becomes longer. The signal is then a composition of the longer coil decay MINUS the target's signal. Good enough to determine whether there's a target or not, but it prevents analysis of the target or ground signals.

      This problem disappears if your method is used on a Tx coil and detection on a separate Rx coil that's induction balanced. Then you get the best of both worlds. I'm actually designing such a detector but I'll use an inductor as a constant current sink rather than a MOSFET because it does the same while being more predictable.

      If you're interested in the (confidential) details you can PM me, perhaps develop a cooperation.
      Hi Taleno ... I understand what you are saying however there is a switch / diode in the source lead of the mosfet ... so yes the mosfet is ON during the flyback and subsequent receive however the switch turns OFF immediately after the flyback leaving only the input amplifier connected via the mosfet to the coil. I will send you a PM with my details :-)

      Comment


      • #63
        Originally posted by Phiphi View Post
        Yes but only in a simulation. Here it is :

        [ATTACH]37745[/ATTACH]

        I use a low noise but fast opamp AD797B with a gain of 10 in order to not saturate the amplifier.

        Below are the waveforms at input and output of the amplifier :

        [ATTACH]37746[/ATTACH]
        [ATTACH]37747[/ATTACH]

        The output reach 1mV (100 uV input with 10 gain) at 3.65 us after Txoff command. So it is possible to sample at less than 4 us.

        Next time, I'll present my ideas for the receiver following the preamp.
        I duplicated your simulation using LTSpice, and was intrigued by the dramatic improvement in early sampling. Initially I thought this might be due to the "damping resistor / protection diodes" topology that was altering the effective damping resistance at the start of the decay. So I ran a simulation that swept the damping resistor values to alter the ratio, but keeping the total damping resistance value constant. Strangely, the early sampling time was improved as the smaller of the two resistors was decreased in value. In fact, the effect was particularly dramatic as the value was decreased to 0.1 ohms. This just didn't make sense .....

        Then I kicked myself!

        The answer was blindingly obvious! The two [damping] resistors form a simple potential divider, and hence the voltage being fed into the preamp is being attenuated. Consequently the sample time will [obviously] be reduced. This is exactly the same scenario as if you were using a lower gain preamp.

        My simulation contained a standard PI setup, alongside a modified copy with Phiphi's configuration. Next I changed the standard configuration to lower the gain to match the attenuation produced by the resistor divider network, and "lo and behold" the results were virtually identical.

        So I'm sorry, but there's no free lunch here. Pity ... as it looked promising.

        mickstv's approach is different. I'll look at that later .....

        Regarding the 47uH choke you placed in the receive path:
        As I said in an earlier post, this has the effect of reducing the overall coil inductance, which will (by default) decrease the sampling time. However, I can see that this could be useful if (as in your case) you're using a commercial coil, and you want to make it sample faster.

        Comment


        • #64
          Here's some more screen shots, scale is 10us per div. First pic is flyback, the second is at the input of the amp and the third pic is the output of the amp.

          Can detect 6x6mm side wall coke can, between 11 and 14cm.







          Attached Files

          Comment


          • #65
            mickstv - I don't doubt that your design is working as described. It is fundamentally different to what phiphi was suggesting.

            Comment


            • #66
              Originally posted by mickstv View Post
              Here's some more screen shots, scale is 10us per div. First pic is flyback, the second is at the input of the amp and the third pic is the output of the amp.

              Can detect 6x6mm side wall coke can, between 11 and 14cm.







              Tried the 6x6mm can with my circuit(detect at 125mm, GEB off). I have about 10x your amplifier gain. Integrating between 6 and 16.2usec after turnoff(total integrator gain about 100). I think the average signal is about 1/6 the signal at 6usec. Wondering how you are sampling amplifier out. With an A-D(how many bits?). With an integrator(sample times and total gain?) Calculated your amplifier gain at 32(10k/68*100/460). I see reply #54 shows a 5usec sample(missed it before)
              Last edited by green; 11-13-2016, 03:05 PM. Reason: added sentences

              Comment


              • #67
                Originally posted by Qiaozhi View Post
                mickstv - I don't doubt that your design is working as described. It is fundamentally different to what phiphi was suggesting.
                I just thought I would post some more cro shot's of it running.


                Originally posted by green View Post
                Tried the 6x6mm can with my circuit(detect at 125mm, GEB off). I have about 10x your amplifier gain. Integrating between 6 and 16.2usec after turnoff(total integrator gain about 100). I think the average signal is about 1/6 the signal at 6usec. Wondering how you are sampling amplifier out. With an A-D(how many bits?). With an integrator(sample times and total gain?) Calculated your amplifier gain at 32(10k/68*100/460). I see reply #54 shows a 5usec sample(missed it before)
                It runs 3 samples S1 (5us), S2 (130us) and S3 (125us). I use a single ended integrator with 4.3k input resistor and a 1m resistor with 22nf cap in parallel. The output of the integrator goes into a LPF with a gain of 2.2x.

                Comment


                • #68
                  What size of coil is used for the detection of the 6X6mm target @ 11 to 14cm? Is the test in air?

                  Thanks

                  Comment


                  • #69
                    Originally posted by Qiaozhi View Post
                    The answer was blindingly obvious! The two [damping] resistors form a simple potential divider, and hence the voltage being fed into the preamp is being attenuated. Consequently the sample time will [obviously] be reduced. This is exactly the same scenario as if you were using a lower gain preamp.
                    Yes, you are right, the damping resistors and the diode form a potential divider but not a linear one. When the voltage is high, the current in the diode is high and it's equivalent resistance is very low so there isn't a significant current in the 22 ohms resistor but at low voltages, the current in the diode is very low and all the current runs in the 22 ohms resistor and we have a potential divider 22/( 22 + 738 )
                    That's the same in the classical solution for high voltages but for low voltages, the equivalent resistance of the diode is very much higher than the 1k resistor and there isn't any attenuation.
                    However, I think my solution remains to be considered. It seems possible to optimize the ratio between the two damping resistors to decrease the decay time without attenuate too much. For example, with 675 ohms and 75 ohms, the attenuation is a 5 factor and with an amplifier with a 5 gain, the equivalent circuit has an early sampling at about 4.5 us to reach 100uV at amplifier output.

                    Comment


                    • #70
                      Originally posted by Qiaozhi View Post
                      Regarding the 47uH choke you placed in the receive path:
                      As I said in an earlier post, this has the effect of reducing the overall coil inductance, which will (by default) decrease the sampling time. However, I can see that this could be useful if (as in your case) you're using a commercial coil, and you want to make it sample faster.
                      Yes but with a lower coil inductance, the magnetic field will be lower but, with my solution, the field stays the same so there is an advantage to use it !

                      Comment


                      • #71
                        Originally posted by Phiphi View Post
                        For example, with 675 ohms and 75 ohms, the attenuation is a 5 factor and with an amplifier with a 5 gain, the equivalent circuit has an early sampling at about 4.5 us to reach 100uV at amplifier output.
                        Sorry, it's not a 5 factor but a 10 ones !

                        Comment


                        • #72
                          Originally posted by Phiphi View Post
                          Yes, you are right, the damping resistors and the diode form a potential divider but not a linear one. When the voltage is high, the current in the diode is high and it's equivalent resistance is very low so there isn't a significant current in the 22 ohms resistor but at low voltages, the current in the diode is very low and all the current runs in the 22 ohms resistor and we have a potential divider 22/( 22 + 738 )
                          That's the same in the classical solution for high voltages but for low voltages, the equivalent resistance of the diode is very much higher than the 1k resistor and there isn't any attenuation.
                          However, I think my solution remains to be considered. It seems possible to optimize the ratio between the two damping resistors to decrease the decay time without attenuate too much. For example, with 675 ohms and 75 ohms, the attenuation is a 5 factor and with an amplifier with a 5 gain, the equivalent circuit has an early sampling at about 4.5 us to reach 100uV at amplifier output.
                          If you run a comparison simulation between a standard PI topology and the proposed version, it appears that a 1us or so reduction in sample time can be achieved. This is due to at least two factors:

                          1) More potential coil energy is being expended in the damping resistor during TX-on, which results in less energy being stored in the coil. Hence [by default] you will be able to sample earlier.

                          2) Due to the potential divider network, there is a lower RX signal at the input to the preamp, which means that the preamp spends less time in saturation.

                          The drawbacks are:

                          1) The target eddy currents peak at a higher maximum value, but decay more quickly due to less stimulation.

                          2) Since the RX signal is attenuated prior to the preamp, it needs to be re-amplified. This will introduce additional noise into the RX chain.

                          Personally, I think you need to get this idea out of the simulator and onto the bench; and I highly suspect that the "magic" will then disappear.

                          Comment


                          • #73
                            Originally posted by Phiphi View Post
                            Sorry, it's not a 5 factor but a 10 ones !
                            Yes - I saw that, but forgot to comment in my last post.

                            Comment


                            • #74
                              I have done virtually no simulation, just basic calculations and bench experiments. After all, I started in PI when personal computers were just a dream. We had to book time on the laboratory's computer which filled a room. Mostly we didn't bother.

                              What I would like to see simulated is the effect of running the TX at a lower current say 0.5A rather than 5A but with 10x the repetition rate with damping adjusted accordingly.

                              In both cases amplifier gain 500x wideband dual type; sampled into an integrator whose TC gives a useable quick response when sweeping at 1m/S. Integrator TC's adjusted for the higher rep rate to give the same response.

                              The coil inductance to be the same i.e. 300uH. TX supply volts the same i.e. 12V, but coil series resistance increased to bring the peak current down to 0.5A.

                              How do the two configurations compare for early sampling, noise reduction and target signal at integrator output?

                              Anyone prepared to give it a go?

                              Eric.

                              Comment


                              • #75
                                Originally posted by Ferric Toes View Post
                                I have done virtually no simulation, just basic calculations and bench experiments. After all, I started in PI when personal computers were just a dream. We had to book time on the laboratory's computer which filled a room. Mostly we didn't bother.
                                Eric.

                                Comment

                                Working...
                                X