Announcement

Collapse
No announcement yet.

Need some help Virtual GND circuit.

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

  • #16
    It is the common ground however the gate source capacitance effectively couples different noises/leakage to the common input of the amp which is different for N and P fet. Keep trying .. maybe you will crack it :-)


    regards,

    moodz.

    Comment


    • #17
      Or maybe I should just finish the twin coil one first. then come back to this at a later stage.



      Mick

      Comment


      • #18
        Your coil drive approach will work, I've tried it before. It is usually difficult to get the virtual ground nailed down solid (though yours looks pretty good) so you may need to run a separate shield wire to the coil. If you increase the delay and the hand response doesn't decrease then I would question whether it's due to salt.

        Comment


        • #19
          have you considered a full bridge setup with 4 FETs. eg, if you have LV,LG, RV,RG fets with LV/LG on the left hand side and connected to the inductor, and RV,RG on the right connected to the other side of the inductor. (LV/RV connected to Vcc, LG,RG connected to GND),

          you can enable LV and RG to charge the inductor. opening LV while RG is still shorted will cause the flyback. You can then swap to RV and LG to charge and release RV to cause the other polarity flyback.

          This allows the upper devices and lower devices to be different without loosing too much symmetry. you can also have one side that always stays on during the flyback, in which case you loose the symmetry advantage, but can choose low-voltage devices for the two that won't experience high voltages.

          in both cases you would need to be able to sense differential voltages over a wide common mode range relative to ground.

          Comment


          • #20
            Originally posted by Carl-NC View Post
            Your coil drive approach will work, I've tried it before. It is usually difficult to get the virtual ground nailed down solid (though yours looks pretty good) so you may need to run a separate shield wire to the coil. If you increase the delay and the hand response doesn't decrease then I would question whether it's due to salt.

            I think part of the problem is the breadboard I'm using to test these circuits, So I'm building a small board for the virtual ground supply, to keep all the leads as short as possible, then will try again.

            I guess salt could be a problem, I'll find another test subject.

            I'll post up the results once done.


            Cheers
            Mick

            Comment


            • #21
              Originally posted by permute View Post
              have you considered a full bridge setup with 4 FETs. eg, if you have LV,LG, RV,RG fets with LV/LG on the left hand side and connected to the inductor, and RV,RG on the right connected to the other side of the inductor. (LV/RV connected to Vcc, LG,RG connected to GND),

              you can enable LV and RG to charge the inductor. opening LV while RG is still shorted will cause the flyback. You can then swap to RV and LG to charge and release RV to cause the other polarity flyback.

              This allows the upper devices and lower devices to be different without loosing too much symmetry. you can also have one side that always stays on during the flyback, in which case you loose the symmetry advantage, but can choose low-voltage devices for the two that won't experience high voltages.

              in both cases you would need to be able to sense differential voltages over a wide common mode range relative to ground.


              I had considered a full h bridge, but the problem with this is some purchased coils come with the shield connection tied to one side of the coil, and this would be a problem, because the coil in a hbridge is floating above ground potential.

              Comment


              • #22
                ah, that makes sense. at best you'd have to do the second option I listed as then the shield is at least at either +Vcc or Gnd during the flyback. even then, you'd have to make sure to avoid accidental shorts.

                Comment


                • #23
                  Originally posted by mickstv View Post
                  I think part of the problem is the breadboard I'm using to test these circuits, So I'm building a small board for the virtual ground supply, to keep all the leads as short as possible, then will try again.

                  I guess salt could be a problem, I'll find another test subject.

                  I'll post up the results once done.


                  Cheers
                  Mick
                  As a test, you could also try removing the virtual ground circuit and using two batteries instead. At least this would eliminate that as a possibility. However, you're virtual ground does look stable, so it may be something else.

                  Comment


                  • #24
                    HI Mick,

                    Try putting your cct in a metal case/enclosure of some sort that is also grounded to the coil ground..... and make sure your coil plug is grounded also.


                    Cheers Mick

                    Comment


                    • #25
                      Originally posted by Qiaozhi View Post
                      As a test, you could also try removing the virtual ground circuit and using two batteries instead. At least this would eliminate that as a possibility. However, you're virtual ground does look stable, so it may be something else.

                      Hi Qiaozhi, yes I tried that as well but made no difference.


                      Thanks for the idea though.


                      Cheers
                      Mick

                      Comment


                      • #26
                        Originally posted by Mechanic View Post
                        HI Mick,

                        Try putting your cct in a metal case/enclosure of some sort that is also grounded to the coil ground..... and make sure your coil plug is grounded also.


                        Cheers Mick


                        Hi Mick, I'll try that tomorrow and see what happens.

                        If it doesn't work out I'll get back onto the twin coil version and at least get that one finished.


                        Cheers
                        Mick

                        Comment


                        • #27
                          I'm curious Mick, can you still detect your hand if you turn off the coil pulsing completely? Also as another fun and quite possibly meaningless test, what happens if you grab the virtual ground with your other hand, does that prevent detection?

                          Comment


                          • #28
                            Originally posted by Midas View Post
                            I'm curious Mick, can you still detect your hand if you turn off the coil pulsing completely? Also as another fun and quite possibly meaningless test, what happens if you grab the virtual ground with your other hand, does that prevent detection?
                            Once we reach a certain sensitivity level, the detector starts detecting the human body. the human body is a conductor. Every conductor is subject to eddy currents.

                            If your detector detects the hand or body, without the TX running, it is probably capacitive.
                            If it is detecting only with the TX running, it is inductive.

                            This inductive sensing of the human body is a bit of a problem for walk-through security detectors. How could we solve this problem?

                            What is the time constant (TC) of the human body? Do larger bodies have a different TC than smaller bodies?

                            Could we filter out the body response?

                            Tinkerer

                            Comment


                            • #29
                              Thanks for the replies Tinkerer and Midas.

                              The current gain of the unit is 50 for the ne5534 frontend and unity gain for the inverting ne5534, after the 4066 switching the combined output is integrated with a gain of 100. So not a huge amount of gain in circuit.

                              I disconnected both FETS and tested this showed no response to hand movement, so it cant be capacitive.

                              I've tried sample delays up to 30us but still shows the same problem.

                              Also roughly mounted the circuit into a all metal enclosure but made no difference.

                              Looks like I might put this on the backburner and finish the twin coil detector first.



                              Mick

                              Comment


                              • #30
                                I know this is dumb, but are you wearing a ring or a watch? Have you tried it with other people?

                                Comment

                                Working...
                                X