Announcement

Collapse
No announcement yet.

field test unit no 001 "model T"

Collapse
This topic is closed.
X
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • OK, beside UF and whatever issues you pursue, there might be some honest people that expect some higher meaning to this discussion. Well, there is none. You don't have to go for Fourier to learn about what a DC blocking capacitor does - Bode plots are just as revealing. Unlike FFT that loses phase information, a Bode plot teaches us that a phase is affected one decade above and below the -3dB cut-off frequency. Now imagine a 10Hz cut-off. For every signal that is above 100Hz (one decade up) there is no amplitude or phase shift to affect anything.

    It is safe to assume that such coupling would improve Rx processing by removing EF prior to sampling. It would not affect the preamp offset though. The offsets can be tackled with a later sample, but also with some clever servo to fix the offset problem at the preamp.

    Beside the DC blocking, series capacitors are useful in lowering the effective coil Z over a large frequency range, which in turn improves the input noise. The final frontier is limited by a coil wire thickness (surprise!).

    Comment


    • Originally posted by Davor View Post
      OK, beside UF and whatever issues you pursue, there might be some honest people that expect some higher meaning to this discussion. Well, there is none. You don't have to go for Fourier to learn about what a DC blocking capacitor does - Bode plots are just as revealing. Unlike FFT that loses phase information, a Bode plot teaches us that a phase is affected one decade above and below the -3dB cut-off frequency. Now imagine a 10Hz cut-off. For every signal that is above 100Hz (one decade up) there is no amplitude or phase shift to affect anything.

      It is safe to assume that such coupling would improve Rx processing by removing EF prior to sampling. It would not affect the preamp offset though. The offsets can be tackled with a later sample, but also with some clever servo to fix the offset problem at the preamp.

      Beside the DC blocking, series capacitors are useful in lowering the effective coil Z over a large frequency range, which in turn improves the input noise. The final frontier is limited by a coil wire thickness (surprise!).
      very true ... but with an ideal capacitor and an infinite input impedance there will be no ( practical ) corner frequency because the impedance of the capacitor is irrelevant in comparison to the infinite input impedance of say an amplifier. So if you know what frequencies you want to pass across a capacitive coupling its simply a matter of making sure that the capacitor impedance << input impedance at the frequencies of interest ( thats a rule of thumb for people who dont make or calculate exact filters ).

      For UF ... signals dont enter and leave capacitors ... they are just like resistors ( frequency sensitive ones ) for AC signals with ***WOW*** "reactance" ***WOW***.





      .

      Comment


      • Ok guys,

        don't pick on UFox anymore. I think, he has learned something. I hope, that PJ did learn something too.

        Yup, I've learned either.
        Aziz

        Comment


        • Originally posted by Davor View Post

          It is safe to assume that such coupling would improve Rx processing by removing EF prior to sampling. It would not affect the preamp offset though. The offsets can be tackled with a later sample, but also with some clever servo to fix the offset problem at the preamp.

          Beside the DC blocking, series capacitors are useful in lowering the effective coil Z over a large frequency range, which in turn improves the input noise. The final frontier is limited by a coil wire thickness (surprise!).
          Inductive or capacitive reactance does not generate noise. In the case of a coil, the thermal noise contribution is only that of the dc resistance. At the present state of the art, noise sources in the preamp will far outway the noise of PI coils which vary between 0.5 and 5 ohms dc resistance.

          The dual sampling and subtraction method is very good at removing EF signals, and personally I do not see any reason to change it. Properly designed with optimal sample spacing, depending what type of targets you are looking for, it is effective at minimising power line frequencies too. Again, properly designed, there are no response holes in the basic two sample method. If greater attenuation is required you can, with a feedback circuit, make the front end have a bandpass response that rolls off at the LF end just short of the TX pulse rate. This can also double as an offset servo.

          Eric.

          Comment


          • Originally posted by Ferric Toes View Post
            Inductive or capacitive reactance does not generate noise. In the case of a coil, the thermal noise contribution is only that of the dc resistance. At the present state of the art, noise sources in the preamp will far outway the noise of PI coils which vary between 0.5 and 5 ohms dc resistance.
            Yes, I checked for the typical values of PI inductances and you are right. There is no way with nowadays preamps to ever come close to such low source resistance. The effect I was referring to is noticeable with several mH coils in IB systems where in effect some impedance transformation happens.

            But then again, you just identified a next optimisation path - PI Rx inductance is far below the optimal for noise performance of nowadays preamps. Rx inductance may be observed as an auto-transformer having a primary inductance of a Tx coil. So the achieved Rx voltage will be increased by a square root of the Rx/Tx inductance ratio.

            Comment


            • Originally posted by Davor View Post
              But then again, you just identified a next optimisation path - PI Rx inductance is far below the optimal for noise performance of nowadays preamps. Rx inductance may be observed as an auto-transformer having a primary inductance of a Tx coil. So the achieved Rx voltage will be increased by a square root of the Rx/Tx inductance ratio.
              Correct. Too much reliance in the past on the simple mono coil, because of its simplicity, where the constraints are for the benefit of the TX.

              Eric.

              Comment


              • Hey we aren't finished.

                Anyone want to try the subtraction of integrals and wanna see the transfer function? This is more complex and difficult to do it as you have to calculate the differences of integrals this time.

                Who is going to bite the bullet this time?
                Aziz

                Comment


                • Originally posted by Ferric Toes View Post
                  Correct. Too much reliance in the past on the simple mono coil, because of its simplicity, where the constraints are for the benefit of the TX.

                  Eric.
                  I am going to start a new thread with this, so as not to highjack this thread.
                  Lets look at some alternatives to the simple mono coil configuration.
                  Separate TX-RX of same or different amount of turns and inductance.
                  IB configurations, concentric and not. Different amount of turns or different inductance.
                  Differential coils
                  etc.

                  Tinkerer

                  Comment


                  • I will........Aziz....what you mean is aaaaaadddd the intergrals

                    Comment


                    • Originally posted by Aziz View Post
                      Look at this transfer characteristics (red):
                      http://www.musicanddesign.com/images/DW_eq4a.GIF
                      (not to scale with our table)
                      Yes, that's what I would expect to see.

                      Comment


                      • Originally posted by Carl-NC View Post
                        Yes, that's what I would expect to see.
                        D'accord.
                        But very difficult to make it visible in a simple excel table. You know, the disbelieving guys here wouldn't believe me, that it is a high-pass filter in the interesting (low) frequency region.

                        I'm now making an appeal to the guys to show the integrated window subtraction method, which usually occurs in a PI technology. This hasn't been proved (or showed) yet.

                        But I'm not very keen on to bite the bullet this time. (Ok, I admit, that it's beyond my scope. Honestly, no kidding .) Who wants be a hero? Sido?
                        (Or PJ? UFox? Robby_H? Rafferty? Lord Fatcat? Anyone else? )
                        (BTW, one can make use of the math symmetry to make the proof simple. Just focus to the worst case condition in the low frequency region.)
                        Good luck. I believe in you guys.

                        Aziz

                        Comment


                        • Well this is what I have. Not by me but by John Alldred, employed by myself in the 1970's. He had a Masters Degree in physics and was one of those rare persons who could work out all the maths and theory and then build something that worked. Sadly he died 10 years ago at age 61 of lung cancer.

                          Aziz, it would be interesting if you could refine this and bring it up to date. Also to include extra samples for GB.

                          There is data on coils but this will be posted later in Advanced Coils for PI thread.

                          Eric.

                          Click image for larger version

Name:	PI Theory003.jpg
Views:	1
Size:	433.2 KB
ID:	334126
                          Attached Files

                          Comment


                          • Nice, thank you.
                            There are, however a few points that may be emphasized... vo can in fact be measured directly in case of a separate Rx coil placed in induction balance, and EF can in fact be cancelled by both bipolar pulsing and HPF.
                            Furthermore, I'd say keeping sampling periods equal may stand for easier math, but is certainly not good for S/N at later samples.
                            You may consider a period t-to to be a bandwidth limiting one, so in fact you are subtracting amplitude responses of ever so lower cut-off low pass filters, thus enhancing the high frequency content e.g. short tau. With 1/(t-to) being the boss, I can't see equal time spacing to be optimal. I also fail to comprehend why is a late sample taken after every pulse. Some kind of iambic pulsing would provide more punch per time, while keeping the goodies of the late sampling.

                            Comment


                            • We need to do what geophysics has done in TDEM for years. Bipolar pulses and logarithmically spaced sequential time gates. This gives good resolution at early time for fast decays and at late time maximises the signal. Samples the whole space between TX pulses and cancels EF too, as you say.

                              Eric.

                              Comment


                              • Hi Eric,

                                thank you for your published interesting papers. I enjoyed reading them.

                                Aziz, it would be interesting if you could refine this and bring it up to date. Also to include extra samples for GB.
                                We aren't finished yet and I'm waiting for a math freak to solve the proof.
                                The GB is a whole different topic and a PhD work worth. One have to flip over between the time domain (TD) and the frequency domain (FD) to describe and understand it.

                                Aziz

                                Comment

                                Working...