Announcement

Collapse
No announcement yet.

FPGA based UNIPI system

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

  • #31
    Originally posted by moodz View Post

    The horizontal time base in for the last pic was 1.6384 milliseconds per value ( but just for the wave form bit ... as we are really looking at a memory buffer here in real time ).
    The update rate of the waveform is 610.35 Hz.
    The numbers are weird in decimal because all the DSP is done in pure binary .. not decimal.
    Hmm however having said that ... those rates above are the "update" rates of the samples .... so think like your CRO with 1024 waveform averaging set on.

    The real time base is still 400 nanoseconds / per sample ... ie 2.5 MSPS.

    Comment


    • #32
      Ok, thanks

      Comment


      • #33
        ...added the code to integrate the waveform across the buffer ... no idea yet if I am supposed to integrate across different sections and add or subtract for ground balance .. so in the mean time just integrating the whole lot and see where we go.

        The integral sum is the first "spike" on the LHS of the waveform as it has wrapped around ... some counter issue maybe.

        Not much to look at here ..but now we have the DC level to feed to the VCO ..( next step ).

        Click image for larger version

Name:	integral.jpg
Views:	202
Size:	545.7 KB
ID:	431455

        Comment


        • #34
          ..here is a video of the integrator in action.

          https://youtu.be/7Kxp_Po1YVQ?si=elCR9sznXg0Dn7UM

          Comment


          • #35
            Good news ...we might have reached the proof of the pudding stage.
            Further coding of the the integrator has vastly improved the detector signal chain and now can show good resistance to noise AND also ferrous / non ferrous discrim indications.
            This test also shows that the UNIPI is a non motion detector.

            No response to magnets at all ... as you would expect with bipolar :-)

            This would seem to indicate that CCPI is an improvement over some other techniques.

            Here is a video of a quick test below ... exciting times !

            moodz.

            https://youtu.be/gnJ_-i5tfsk?si=I_WR8UGLh8POb7pj
            Last edited by moodz; 12-11-2024, 10:13 AM. Reason: added magnet test.

            Comment


            • #36
              Looking good Moodz, interesting experiment for sure, great work.

              cheers.

              Comment


              • #37
                I find the FE response very interesting for a mono-loop. Like you said above once you separate the parts and do the math it will tell the story.

                Thanks

                Comment


                • #38
                  Hi Altra ... This is a dd coil ... CCPI on the Tx. The coil is a large Minelab. I do have a solution for the mono coil but it's not this one. Same Rx but different tx.

                  Comment


                  • #39
                    Ok, great progress in any case!

                    Comment


                    • #40
                      Now added code for a Numerically controlled oscillator . NCO and Pulse Width Modulation PWM output for audio and we only use 1 pin on the FPGA ... yay. The NCO is equivalent to a VCO however this one uses DDS and has about .018 hertz step resolution ( prob overkill )

                      So we have basically a working metal detector from end to end using no extra amplifiers or circuitry for the RX circuit except the FPGA / ADC combo. ( ... ok I did put a small audio amp on the PWM output ).
                      The results are not optimal yet but alot better than I thought they would be. The detector is non motion and the threshold point is rock solid over 12 hours it didnt drift at all since past the RX coil its all data.. theres nothing to add noise or drift.
                      There seems to be a good discrim fucntionality.

                      In fact the results are so good I am consigning the MAGPI work to the bin.

                      Heres a video of the first cut of the end to end signal chain with audio working ... no I wont be doing a field visit.

                      https://youtu.be/U0pcAp-r2bY?si=_SLtSVRCGg3HMvt0

                      moodz
                      Last edited by moodz; 12-12-2024, 10:48 AM. Reason: clarification

                      Comment


                      • #41
                        Results like this from a stored 10 year old project, are very promising Moodz, great work.
                        Have you had any sleep over last few days lol?

                        Comment


                        • #42
                          Originally posted by Mdtoday View Post
                          Results like this from a stored 10 year old project, are very promising Moodz, great work.
                          Have you had any sleep over last few days lol?
                          I am also building a 20 square meter pergola and deck and binging on the Deadwood series DVDs I found in the shed ( inf fact thats when I found the UNIPI board and was going to throw it out .... must be fate ). I cooked two big christmas puddings yesterday ... its the retired life .

                          Comment


                          • #43
                            Originally posted by moodz View Post

                            I am also building a 20 square meter pergola and deck and binging on the Deadwood series DVDs I found in the shed ( inf fact thats when I found the UNIPI board and was going to throw it out .... must be fate ). I cooked two big christmas puddings yesterday ... its the retired life .
                            Haha, so taking it easy then

                            Comment


                            • #44
                              The UNIPI is now better than the MAGPI3.

                              https://www.youtube.com/watch?v=As7Du_a_kCo

                              The final integrator actually achieves a 31 bit resolution value. ( 24 bit ADC + 7 bits of processing gain in the DSP = 31 bits ).

                              The noise is so low I feed bits 0 - 16 of the integrator to the NCO ( VCO ).

                              moodz

                              Comment


                              • #45
                                Breaking new ground ...

                                I set the TX cct of the UNIPI in single frequency mode ( AKA sine wave ) at 19.6 Khz and the results are basically the same but with much lower noise.
                                Additionally the rx peak input signal is much lower at 50 mV vs 600 mV which means there is more headroom for a preamp. The sensitivity is as good as a regular PI now.
                                The coil is a Minelab DD coil ... so its not really for single frequency use. It can detect the 10 cm foil square at around a meter now reliably.
                                I did not change the processing code in the FPGA .. only the CCPI tx mode ( from pulse to sine ).
                                So this is an interesting development.

                                Here is a video of the test ....

                                https://youtu.be/bYvPV9KQYrM?si=SsROecg7kyxG9dFq

                                It would seem that the CCPI is more of a frequency beast than a pulse induction machine.

                                moodz

                                Comment

                                Working...
                                X