Announcement

Collapse
No announcement yet.

What is the best frequency response for the preamp in a PI ?

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

  • What is the best frequency response for the preamp in a PI ?

    LOW PASS HI PASS BAND PASS FLAT ( also why opamp noises specs are not what you think they are for PI s )

    So you are designing a new Pulse Induction metal detector ... what should the preamp frequency response be and what factors are important in choosing an opamp ?

    (... and here is a clue. That 0.8 nV opamp you have been drooling over in the catalogue may not be optimum at all ... noise figures for opamps are driven by marketing not engineering )



  • #2
    Hi moodz,

    I think, the best frequency response for the preamp in a PI is Low pass at appr. 10MHz. This will allow very fast response. But most important for the firs amplifier, in my opinion, is noise figure of opamp. This will allow managing of signals under 1uV. For PI both voltage noise and current noise have to be low. If the waving speed is 1m/s, these noises have to be measured at appr. 2Hz frequency. Opamps with good parameters are:
    1. OPA1611AID - best
    2. ADA4895-1ARZ
    3.LM6171P
    4. NE5534 - very popular

    Comment


    • #3
      I've only used low pass, never tried band pass. For gold designs I usually set the BW to be 10x the smallest target I want to detect. That is, if I want to detect a 1us nugget then I set the feedback tau to 0.1us. For a walk-through I tend to use a much slower response, 10s of us. This sounds like it would not be able to pick any low conductors but they tend to get integrated (smeared) in the slower response so you can still see them, though maybe not quite as well. A beach design would be somewhere in between.

      On opamps I pay more attention to overvoltage recovery which is a primary limiting factor in the sample delay. Noise is secondary; besides, it makes no sense to pick a 1nv/rtHz opamp when it's preceded by a 1k clamp resistor that has 4nv/rtHz. On some aggressive designs I've even increased the clamp resistor to several kΩ to improve clamp recovery. Noise at the audio doesn't seem any worse. Overvoltage recovery is rarely spec'd in the data sheet so you have to try each opamp.

      Comment


      • #4
        Copy the ML design, of all the opamps for a single stage the AD797 is the best choice and I have tested nearly all offerings.

        Comment


        • #5
          The AD797 is one of my favorites as well, so is the LME49990. For basic circuits I often use a MAX410.

          Comment


          • #6
            Good points .. but ... the PI signal stimulating the target caused eddy currents to flow during the collapse of the TX current waveform and the target response has no signal information below the TX primary modulation frequency ?

            So if we say have a TX frequency of 1kHz then we are looking at demodulating target signals at 1 Khz upwards.

            But the typical opamp noise performance looks likes this ... so why would we amplify anything below the primary TX frequency ?

            Click image for larger version

Name:	image.png
Views:	311
Size:	36.2 KB
ID:	421671

            Comment


            • #7
              Originally posted by Carl-NC View Post
              The AD797 is one of my favorites as well, so is the LME49990. For basic circuits I often use a MAX410.
              I think they discontinued the LME49990 ... I have a couple of hundred on the shelf but will have to find something else. doh.

              Comment


              • #8
                Originally posted by Carl-NC View Post
                The AD797 is one of my favorites as well, so is the LME49990. For basic circuits I often use a MAX410.
                I have 10 LME49990 saving them for a special project, I have replaced the AD797 devices in the old ML detectors and I did obtain lower distortion and less peaking, also better at 0-10 hz noise.

                Pity it’s no longer available.

                Comment


                • #9
                  Originally posted by moodz View Post
                  But the typical opamp noise performance looks likes this ... so why would we amplify anything below the primary TX frequency ?​
                  True. However, all of my designs lately have been bipolar pulsing which cancels 1/f noise.

                  Comment


                  • #10
                    Click image for larger version  Name:	image.png Views:	0 Size:	43.6 KB ID:	421676Click image for larger version  Name:	image.png Views:	0 Size:	45.2 KB ID:	421677

                    Aanalogue Devices design note 140 discusses noise performance of opamps at different frequencies ... in this case 1 Khz and 10 Khz
                    of course its slanted at their products but shows the noise performance of opamps vs Source Resistance at different frequencies.​

                    Comment


                    • #11
                      I have 10 LME49990 saving them for a special project, I have replaced the AD797

                      has SLEW RATE 22V/us, while obsolete LF357 has 50. we say about PI or VLF?

                      Comment


                      • #12
                        Hello folks,

                        a question... I've been looking at the datasheets of all the opamps suggested here. Is it really worth paying about 20 euros for an AD797, for example, compared to about 1 euro for an NE5434? In my opinion, the performance under real conditions is not much better to make it worthwhile. Now the Analog Devices OpAmps are really very good. But from a commercial point of view they are also completely overpriced. A money-making machine for Analog Devices. For the price of an AD797 you can almost buy the components for about 2..3 complete hobby detector kits the size of a Hammer Head PI detector project. Is it really worth it to have an AD797 in the circuit?

                        What is your opinion?

                        Comment


                        • #13
                          Limitation of metal detector sensitivity/range
                          1. interference
                          2. noise
                          I use MCP6022, gain 100
                          1. amplifier output noise 2mv,
                          2. 25mV interference

                          What is the limiting factor in your case?

                          Noise can be easily filtered using an LP filter.
                          interference are more difficult to limit,

                          Instead, AD797 consider an active antenna design​/Discrete components

                          Click image for larger version  Name:	image.png Views:	0 Size:	7.7 KB ID:	421683Click image for larger version  Name:	image.png Views:	0 Size:	33.1 KB ID:	421684

                          Comment


                          • #14
                            Hi Carl,
                            It is very interesting for me what is the value of the change of the recovery time of 1N4148 with changing of the value of the clamp resistor from 1K to 5K?

                            Comment


                            • #15
                              Hi GeoMax,
                              All depence of your final aim. If the parameters of HH PI are enough for you -no problem.
                              But if you want to reach top level of some prameter - you have to pay big money.
                              When I designed pinpointer for very small nugets (0.5grain), I trayed many OPAMPS with different slew rates and different noise figures. The best result give OPA1611 for 6.5USD. This is the reality!

                              Comment

                              Working...
                              X