Announcement

Collapse
No announcement yet.

PCB Coil?

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

  • #31
    bugwhiskers,

    Can you share with us, the self resonant frequency of both the single-side coil and the dual-side coil connection?

    Typically, when you double the number of turns, you make the inductance 4 times larger. Your single-sided 250uH coil should work for making a pretty good gold seeking coil at low delays. Shielding is the only challenge.

    The best way to even attempt a double sided coil is to put the traces on the second side directly in line with the space between the tracks on the first side to minimize capacitance. When the tracks on both sides align, you have one giant capacitor.

    The real problem with making a PCB coil is attempting to shield it. Any gain in making a high self resonant PCB coil may be offset by the capacitance added by the shield. That is why a simple bundle wound coil, made with Teflon insulated wire (AWG 30, 10.5" ID, 19 to 21 turns, short coax about 3 ft.) and two layers of polyethylene (PE) spiral wrap spacer, makes a pretty good coil that can fit a variety of coil housings and be easily shielded.

    bbsailor

    Comment


    • #32
      Precision makes a difference.

      Precision makes a difference.

      Bbsailor,

      There is no doubt that it is easier to make a good bundle coil by using Teflon coated cable. However, after winding many coils, I found that there are always differences between the quality of the coils. Tie the coil a little bit tighter, and the capacitance goes up. Leave a bit more slack, and the coil is just perfect.
      This happens when I do it myself. Good days, bad days, good coils bad coils.
      Of course this does not become apparent when you have 15 or 30 uS first delay, but when the delay has to be between 5 and 10uS the difference shows up.
      Now, consider somebody else making the coils. A person that does not understand the importance of precision.
      I wonder sometimes if that could be the cause that people complain that some ML detectors are not as good as others of the same series.

      This I why I find the PCB coils interesting. Very precise. Always the same. Designed to be manufactured in quantity and quality.

      As for the shielding, I follow the same line of thinking. One of these days I will post a detailed description of the shielding that I now use. It seems to work quite well.
      Tinkerer

      Comment


      • #33
        Originally posted by bbsailor View Post
        bugwhiskers,

        Can you share with us, the self resonant frequency of both the single-side coil and the dual-side coil connection?

        bbsailor

        Will get back to you about the resonant frequency today or tomorrow. As for the second side, well, I ground it away with a sander after seeing its capacitance.

        What you propose about putting the opposite side traces in the gaps between the first side traces is correct but alas to get good inductance the traces have to be as close as possible...ye olde catch 22 at work again !

        The best possible scenario is to get 1oz boards made by professionals who can get the track spacing down to a few thousandths of an inch and then get the tracks silver plated up to 3oz or 4oz.

        Comment


        • #34
          Teflon PCB

          Bugwhiskers,
          Getting the track spacing closer will increase the capacitance. Increasing the amount of turns increases the inductance and capacitance. But somewhere there, there is a sweet spot, where the best compromise is reached.
          To find that sweet spot it needs a precise repeatable design and precise measuring instruments. From there on, results can be predicted by calculation.

          Some excellent Pi designs use coils with tens of Ohms of resistance. They seem to be just as effective as other designs that use brute power.
          It would be interesting to know what the dielectric number of the PCB material is. I believe there are Teflon boards, probably very expensive. Then there is the problem of dielectric on one side and air on the other side of the trace.
          Interesting subject.
          Tinkerer

          Comment


          • #35
            PCB coil resonance

            The resonant frequency of the coil described a few posts back is in excess of my 100kHz Frequency generator. I will have to add a capacitor to bring it within range.

            I concur with you about coils with "supposed" parasitic impedance Tinkerer as I was given a high quality "Lietz" wire coil by a friend and its performance is pathetic when used on my PIMD. All it seems to do is guzzle current.

            For a long while I have thought that a PIMD coil if run at its natural resonant frequency or harmonic thereof then its performance would be enhanced.

            A few weeks back I was tinkering with settings and the whole detector changed drastically. It was going off at my test targets 3 or 4 times further away than normal and even putting my hand within a foot of the coil sent it off.... I never did figure out what caused it and being an old fart I can't remember what I changed just prior to the event !

            Comment


            • #36
              PIMD

              Would you care giving us some more info about your PIMD? Without any information it is not possible to recomend a coil. It is very important that the coil matches the first sample delay.
              Tinkerer

              Comment


              • #37
                My PIMD

                I use an IRF9640 MOSFET, 200V P channel pulsed with 19.5 volts generated along with +5 & -5 volts for op amps and an Atmel Micro by a SMPS fed by 4 AA NiMH batteries.
                Three samples are taken from the LM318 (gain 1k)front end and fed via an 4053 to 3 integrators (gain 1k).

                The width of the second sample is adjusted in relation to the first sample so that ferrous targets generate an identical voltage.

                The output from the first 2 samples is then fed to a Differential OP amp (gain 10). All samples are fed to the micros AtoD including the Diff op amp output. Averaging is done in software on the Diff op amp output by getting the last 64 samples, averaging them then dropping off the oldest. The resulting "number" is then fed to the micros PWM and on to a piezo speaker.

                This is a poor mans PIMD.

                A rich mans PIMD has a DSP running at 300MHz and taking 100's of 24 bit samples of the decay curve then running Fourier over them !

                Comment


                • #38
                  Digital PIMD

                  Very impressive! I take it you design it for nugget hunting. What is your pulse length? What is the delay to the first sample? From the size of your test nuggets and batteries, my guess is that you use short pulses and low repetition rate.
                  A 1k amplification on the front end needs a fast coil so that you can sample early, otherwise these small nuggets don’t show up.
                  The choice of the cable is critical for a short delay.
                  To shield the PCB coil I suggest some material like Styrofoam or other that contains a lot of air, as spacer. The PCB is very thin, so you can add a spacer of 5 to 10 mm on either side without making the coil heavy or clumsy.
                  For the actual shielding I use a compound similar as what is used to make carbon film resistors. Send me an email and I will return with the formula.
                  Some time ago, Reg or bbsailor (sorry, my memory is real bad) posted a link to a calculator for calculating inductance and capacitance of cables. I found this very useful since it gives an ides about the spacing between the wired of a coil. I found that, up to a certain limit, it is best to increase the spacing between the wires to the maximum that the overall dimension of the coil allow. It will need more turns of wire and somewhere there is a reasonable compromise between capacitance and inductance.
                  Tinkerer

                  Comment


                  • #39
                    50% open?

                    Originally posted by Carl-NC View Post
                    I expect taking the spiral all the way to the center will hurt performance. You want at least 50% of the diameter open.

                    "What about multi layer PBC's?? No one thought of that yet?"

                    I laid out a 2-layer PCB coil years ago, but never made it.

                    - Carl
                    Carl,

                    could you please elucidate on that statement? Is there a formula that helps predict these results? Does it have to do with the shape of the Tx or RX field?
                    Tinkerer

                    Comment


                    • #40
                      PCB Coil

                      Tinkerer,

                      Years ago I use to design microstrip antennas for GPS [we only had 5 birds then], at the time, this [GPS PCB microstrip] was a relatively new area, however there were some observable guidelines to insure success. It's be a while since I did any design work so I hunted around to see what was available on the net for the latest design rules.

                      My search lead me to some interesting articles and design "cookbooks" that are used for PCB antennas. The best two sites were not the type of sources that I expected, these are Microchip and Texas Instruments. In the case of the Texas Instruments literature, the designs could also be used for two-box detectors and GPR [Ground Penetration Radar]. The Microchip offered the best review and design of all types of antenna's, including the single coil that we use on the Hammerhead and of course a spiral PCB design.

                      Microchip, application node AN710, see pages 11 and 16. The coils start at page 9.

                      http://www.microchip.com/stellent/id...pnote=en011776

                      Texas Instruments, RFID section, "HF Antenna Cookbook" and "HF Antenna Design Notes".

                      RFID Documentation Site
                      http://www.ti.com/rfid/shtml/doc-center-technotes.shtml

                      Cookbook:
                      http://www.ti.com/rfid/docs/manuals/...naCookbook.pdf

                      Design Notes:
                      http://www.ti.com/rfid/docs/manuals/...esignNotes.pdf

                      Search for "spiral", "helix", "pcb trace antenna" or "microstrip" to get some more information on Google.

                      One thing to keep in mind, well perhaps two :-), we are dealing with audio frequencies which can result in some relatively large wave lengths as far as calculations are concerned. Many or most of the RF designs are for 1/4 or full wave-length and expect the length to be reasonable [at least for RF] somewhere in the "inches or mm". Well audio is not really RF is it? or is it? Typically the RF stuff will get you in the ball park, and then it's experimental time.

                      Not sure who mentioned it in the forum here, but they were correct, keep the PCB traces wide and put some space between them. Also as Carl pointed out, keep the center free, there are other factors here, such as velocity factor. Gary [Chemelec] may have some good idea's on PCB coil design, he has built some of the PCB coils. Not sure if he did some heavy R&D with a calculator using some design formulas or just got a ballpark idea and then did the building.

                      Hopefully this is useful. Now did I really answer your question??? Hmm not sure.

                      Andy

                      Comment


                      • #41
                        Originally posted by Tinkerer View Post
                        What is your pulse length? What is the delay to the first sample?

                        Tinkerer
                        The figures change depending on the coil but for a fast coil:

                        Pulse on time: 37.2uS (Flat top 1.3uS)
                        Delay to first sample: 7.12uS

                        Many people specify the sample delay as starting from the start of the droop in the decay curve... I actually think that it should be specified from Pulse Off which is the end of the Tx pulse.

                        The delay figure I have quoted is from Pulse off.

                        By using a 200V MOSFET the back emf is killed off quicker by the protection diode than the 400V varieties.

                        I also start sampling just before (0.5uS) the decay curve starts to droop. This assists in detecting very small nuggets.

                        Damping resistor is 470 ohms 1/4 watt metal film (low capacitance of coil = doesn't even get warm)

                        You are correct, I have no interest in steel relics this machine is to be used for gold only with total rejection of Ferrous materials.

                        Comment


                        • #42
                          Thanks Andy, for the information. Lots of stuff to read. Hope I dont blow a few circuits in my brain with overload.
                          Tinkerer

                          Comment


                          • #43
                            I like your precise numbers.
                            Not sure what you mean with flat top at the TX. Do you mean more than 5 TC of the coil, when the current has reached its full potential? I don’t see how you do that with 1 Ohm and 250 uH.
                            OK for the delay figure, I do the same.
                            I don’t understand what you mean “the decay curve starts to droop”

                            Here some figures I took today to see how the decay time of a small nugget and the gain interact. The numbers are not very precise, only as good as my equipment permits. The test object is a 0.6 gram nugget. Output of the coil amplifier. Coil shielded and connected to board by a 2 conductor shielded cable of about 6 ft.

                            Gain 2,2k – 14uS to first delay – 100mV
                            Ik - 12uS - 75 mV
                            470 - 11uS - 50mV
                            220 - 9uS - 50mV
                            100 - 8uS - 15mV
                            57 - 6.5uS - 10mV
                            Tinkerer

                            Comment


                            • #44
                              A metal detector coil is not an antenna, it is a transformer winding. There is no classical electromagnetic wave propagation going on; it is all near-field magnetic coupling. So none of the classical antenna design methods will work for detector coils, including microstrip.

                              When you design a printed spiral coil, I think you want to leave the center open. Taking the coil all the way to the center means that the magnetic flux from the outer windings will cut through the inner windings and degrade efficiency. Actually, this is true for each turn that's added after the first one, but each turn also adds inductance. At some point, I suspect the degradation outpaces the improvement, but I don't know where that happens.

                              - Carl

                              Comment


                              • #45
                                Leaving a void in the middle of PCB coils

                                Interesting post Carl.

                                I have noticed from experimentation that with PCB coils the field extends out as a sphere centred on the middle of the coil.

                                I have not been able to demonstrate yet what effect running the traces right thru to the centre has.

                                Comment

                                Working...
                                X