Announcement

Collapse
No announcement yet.

IGSL

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

  • Now a reworked comparator stage. I gave it a proper thought and said - heck, it works anyway, and LM358/324 are workhorses that can not fail even working as comparators. Their inputs will not complain. Major problem was with odd capacitors, and it is fixed already.

    I just tinkered a bit with resistor values to hit the Goldilocks zone for a LM358 stage, and I think it is OK now. My goal was finding values tat will work equally good with gain stages as they are (clipping at 0.7V) and my tinkered ones with peak at 6V, and to allow for a little bit of saturation there. All unnecessary components, except for a diode, are gone. For new builds (mine included ) I'd put 47k resistor instead of a diode, which is completely useless there. 339's are comparators with open collectors, which means there is no positive pulse on their output. With diode you can use any comparator with the same pinout, even push-pull ones - that is the purpose of such a diode. Using 339's with a 47k resistor there, or even a piece of wire (short) shall do fine.

    Next will be audio, but odd as it may be - it just might be OK. I'll check it anyway.

    BTW, comparators are notorious for their slow convergence in LTspice. If that happens to you, just go to LTspice control panel and check the alternative solver engine. If you are still stuck you may relax a bit on Gmin and Abstol, and put them to 1e-10. That should work.
    Attached Files

    Comment


    • Originally posted by Aziz View Post
      Hi SB,

      why don't you use the latest sound card detector controller platform for this? You don't need the TGSL or any of the other platforms.

      But you should be able to code with C/C++ and Win32 (SDK).
      Cheers,
      Aziz
      You mean that slick general purpose board you've been working on? Looks nice. If I could just directly use the sound card input that would be easier.

      Problem is, I want to sample from about .1Hz to maybe 2K Hz probably -- I think many sound cards don't go near DC frequency.

      -SB

      Comment


      • Originally posted by golfnut View Post
        Its around U8 I would like to see some changes.

        I like the change in volume you get with target proximity on some dets,, even the surf pi does this a bit.

        With IGSL all targets crossing the Threshold come in the same.


        It would be good to have a log range - rather than binary.

        Plus on Igsl often its impossible to advance the thrshold pot close enough to quiesent level to max sens to small objects.

        A small target doesnt trip sound (if you scope pin 3 you could see a volt of detected signal with no notion of a Hit)

        I wired my threshold pot between plus and minus rails to get the two inputs closer together for more sens to small targets. But yes I still dont get a range of volume ..


        S
        I used to think I wanted volume-proportional detection. But after using actual MDs, I now prefer "binary" volume. I think the reason is this: deep signals are the ones we especially want to not miss. If the volume is lowest for those, it becomes a strain to hunt, listening so carefully; any loud noise outside covers the target sound. With binary signal, you hear best, and weak signals will start to break up, and that is how you know they are weak. I find you can estimate depth more by moving the search head and noticing the pattern of signal. Just my preference, but I'd rather hear loud chirps to indicate a deep target than some soft hum.

        -SB

        Comment


        • Originally posted by Davor View Post
          Now a reworked comparator stage. I gave it a proper thought and said - heck, it works anyway, and LM358/324 are workhorses that can not fail even working as comparators. Their inputs will not complain. Major problem was with odd capacitors, and it is fixed already.

          I just tinkered a bit with resistor values to hit the Goldilocks zone for a LM358 stage, and I think it is OK now. My goal was finding values tat will work equally good with gain stages as they are (clipping at 0.7V) and my tinkered ones with peak at 6V, and to allow for a little bit of saturation there. All unnecessary components, except for a diode, are gone. For new builds (mine included ) I'd put 47k resistor instead of a diode, which is completely useless there. 339's are comparators with open collectors, which means there is no positive pulse on their output. With diode you can use any comparator with the same pinout, even push-pull ones - that is the purpose of such a diode. Using 339's with a 47k resistor there, or even a piece of wire (short) shall do fine.

          Next will be audio, but odd as it may be - it just might be OK. I'll check it anyway.

          BTW, comparators are notorious for their slow convergence in LTspice. If that happens to you, just go to LTspice control panel and check the alternative solver engine. If you are still stuck you may relax a bit on Gmin and Abstol, and put them to 1e-10. That should work.
          Davor:

          That LM358 stage is not just a comparator as I see it.

          It actually does two things:

          1) It provides some proportional audio due to the feedforward signal through the 100K resistor from your GEBout.

          2) It is a "chatter" filter. The original RC network you removed filters out short pulses from the comparator because they don't have time to rise (integrate) above the reference threshold at the LM358 inverting input. (The LM358 inverting input reference is usually near ground, and the comparator pulses start rising from the negative rail.)

          I have tried running the circuit without the RC network and could not get a pleasant action. But I feel we can make a definite improvement here with some other designs.

          The main downside to the "chatter filter" is that it strips off a constant time duration from every target pulse that gets through it -- which we don't want, it makes target signals skinnier, we want fat ones.

          So there is probably a better way to provide a little more "chatter" filtering without whittling down the target pulse.

          Regards,

          -SB

          Comment


          • LM358 is not supposed to work as a comparator, however, the circuitry around the inverting input is designed to provide minimum 11x gain (plus some 1.5V level shifting to trigger darlington audio stage). Since the preceding stage supplies diode limited ~0.7V at any moderate to high target response, timed 11 (at minimum gain) ... works pretty close to a comparator. My "comparator" criterion kicks in the moment there is a voltage between inverting and non-inverting input, and there you have it - it saturates with all but the faint target responses.

            I'm a bit confused about the chatter filter action here because of two reasons:

            1) At this stage GEB and DISC are in sinc all the time and frequency limited to under 100Hz - they are either in phase or in counterphase, but always in sinc. Comparator here provides only the analog AND action.

            2) If there is some chatter going on because a DISC channel barely passed a comparator triggering criterion, it must also be very low in amplitude, hence, with LM358 gain set properly (to the lowest gain in original setup) it would be just a low peep and nothing else.

            Every other chatter must be of some other origin and it would be worth further examination. The first gain stage introduces some asymmetry viz ground, but it does not happen at low signal levels

            Comment


            • Originally posted by simonbaker View Post
              I now prefer "binary" volume.
              No problem, with the "treshold" trimmer set to 0 gain rises to open loop and LM358 becomes a real comparator, and there you have it. Even my reworked setup has that feature.

              Comment


              • Originally posted by simonbaker View Post
                I used to think I wanted volume-proportional detection. But after using actual MDs, I now prefer "binary" volume. I think the reason is this: deep signals are the ones we especially want to not miss. If the volume is lowest for those, it becomes a strain to hunt, listening so carefully; any loud noise outside covers the target sound. With binary signal, you hear best, and weak signals will start to break up, and that is how you know they are weak. I find you can estimate depth more by moving the search head and noticing the pattern of signal. Just my preference, but I'd rather hear loud chirps to indicate a deep target than some soft hum.

                -SB
                Both ways is better!
                That's why i tried monstable in the past. Sometimes binary is better and sometimes "linear" is better. Depends. Both ways are having advantages and drawbacks. Usually we don't pay much attention on audio at some design, we tend to be more focused on earlier stages, but as time goes by i do realize that audio is probably one of the most important parts in detectors. Can pull all the quality at some detector behavior. All this i am talking from a users point of view. Sometimes spice simulators can not substitute real life.

                Comment


                • Originally posted by simonbaker View Post
                  I used to think I wanted volume-proportional detection. But after using actual MDs, I now prefer "binary" volume. I think the reason is this: deep signals are the ones we especially want to not miss. If the volume is lowest for those, it becomes a strain to hunt, listening so carefully; any loud noise outside covers the target sound. With binary signal, you hear best, and weak signals will start to break up, and that is how you know they are weak. I find you can estimate depth more by moving the search head and noticing the pattern of signal. Just my preference, but I'd rather hear loud chirps to indicate a deep target than some soft hum.

                  -SB
                  Hi Simon,

                  My Garrett At Pro has both binary and proportional audio and I use the proportional mode almost exclusively. It is a matter of personal choice but I think the proportional mode is the best argument for using a good set of head phones so you can hear the nuance of what the detector is telling you.

                  That said, I did add the threshold control to my TGSL so by juggling the settings on sensitivity and threshold, I can make the TGSL behave either way. Today I had it set up for more of a binary signal and most of what I found was at five to six inches deep.

                  I agree that you can tell about how deep the target is by how it behaves when moving the coil over it.

                  Jerry

                  Comment


                  • I was wondering why this particular control was left on the PCB as a trimmer, and not as a real potentiometer. It has a function of a real volume control, and IMHO it should be put on the front panel, and preferably as a stereo potentiometer. It really works as a volume control. It is too bad that the rest of the circuitry is dimensioned for minimum gain of sky-high, but if it is replaced by a 100kΩ a minimum gain drops to 2 and it can really behave.

                    Actually the way level shifting is done here works perfectly. It produces a ~1.5V shift regardless of "disc treshold" control. In my reworked setup I kept 10kΩ trimmers because if someone already has the kit, he'd most probably find 10kΩ there. It is perfectly OK to keep 100k, 330k resistors as they are, but use a trimmer, or even better a potentiometer for a real volume control at 100kΩ linear. If you crank it up -you get a binary response. So everyone is happy.

                    I'm waiting on my Musketeer IGSL PCB. It is almost itching. I'll surely put a stereo pot there, most probably at 100kΩ and the resistors at 100k/330k values as they are. I'll surely rework the gain stages as per my previous observations. Most probably I'll reduce a 220kΩ near "Disc sens" to obtain more control. And I plan to use headphones with it.

                    Comment


                    • Originally posted by Davor View Post
                      LM358 is not supposed to work as a comparator, however, the circuitry around the inverting input is designed to provide minimum 11x gain (plus some 1.5V level shifting to trigger darlington audio stage). Since the preceding stage supplies diode limited ~0.7V at any moderate to high target response, timed 11 (at minimum gain) ... works pretty close to a comparator. My "comparator" criterion kicks in the moment there is a voltage between inverting and non-inverting input, and there you have it - it saturates with all but the faint target responses.
                      The preceding GB channel stage (LM308 ) does not have the diode in the original TGS, so it should pass through a proportional volume that can go up to positive rail I think. That would allow a proportional volume once the signal rises above the inverting input level of the LM358 and before the output saturates. I'm not sure if the original TGS has 11x gain there.

                      I think the proportional audio was calibrated mainly for intermediate to deep depth signals -- very loud signals are clipped by the first LM358 after the SD possibly as you mention and also by saturation of the LM358 in that final stage, depending on gain.


                      I'm a bit confused about the chatter filter action here because of two reasons:

                      1) At this stage GEB and DISC are in sinc all the time and frequency limited to under 100Hz - they are either in phase or in counterphase, but always in sinc. Comparator here provides only the analog AND action.

                      2) If there is some chatter going on because a DISC channel barely passed a comparator triggering criterion, it must also be very low in amplitude, hence, with LM358 gain set properly (to the lowest gain in original setup) it would be just a low peep and nothing else.

                      Every other chatter must be of some other origin and it would be worth further examination. The first gain stage introduces some asymmetry viz ground, but it does not happen at low signal levels
                      Yes, but noise in the 20Hz to 100 Hz range can make plenty of chatter compared to your target signal of say 1/4 to 1 second wide. I agree that with proportional audio, noise just triggering chatter would probably be low in volume -- but that is where the deep targets are also, in that noise, which is why some people may prefer binary audio.

                      But I do feel the "chatter filter" looks like an afterthought that probably could be much improved.

                      -SB

                      Comment


                      • Originally posted by Jerry View Post
                        Hi Simon,

                        My Garrett At Pro has both binary and proportional audio and I use the proportional mode almost exclusively. It is a matter of personal choice but I think the proportional mode is the best argument for using a good set of head phones so you can hear the nuance of what the detector is telling you.

                        That said, I did add the threshold control to my TGSL so by juggling the settings on sensitivity and threshold, I can make the TGSL behave either way. Today I had it set up for more of a binary signal and most of what I found was at five to six inches deep.

                        I agree that you can tell about how deep the target is by how it behaves when moving the coil over it.

                        Jerry
                        Yes, with great headphones it probably is nice -- I tend to use an open air type so I can hear what's going on around me -- to the detriment of some detection I admit.

                        -SB

                        Comment


                        • Originally posted by Davor View Post
                          I was wondering why this particular control was left on the PCB as a trimmer, and not as a real potentiometer. It has a function of a real volume control, and IMHO it should be put on the front panel, and preferably as a stereo potentiometer. It really works as a volume control. It is too bad that the rest of the circuitry is dimensioned for minimum gain of sky-high, but if it is replaced by a 100kΩ a minimum gain drops to 2 and it can really behave.

                          Actually the way level shifting is done here works perfectly. It produces a ~1.5V shift regardless of "disc treshold" control. In my reworked setup I kept 10kΩ trimmers because if someone already has the kit, he'd most probably find 10kΩ there. It is perfectly OK to keep 100k, 330k resistors as they are, but use a trimmer, or even better a potentiometer for a real volume control at 100kΩ linear. If you crank it up -you get a binary response. So everyone is happy.

                          I'm waiting on my Musketeer IGSL PCB. It is almost itching. I'll surely put a stereo pot there, most probably at 100kΩ and the resistors at 100k/330k values as they are. I'll surely rework the gain stages as per my previous observations. Most probably I'll reduce a 220kΩ near "Disc sens" to obtain more control. And I plan to use headphones with it.
                          Again, I think the original use of the pot was a chatter filter too, since the voltage level determines how short a pulse from the comparators is rejected -- assuming you have the capacitors in there. Unless you're talking about a different pot.

                          -SB

                          Comment


                          • Originally posted by simonbaker View Post
                            You mean that slick general purpose board you've been working on? Looks nice. If I could just directly use the sound card input that would be easier.

                            Problem is, I want to sample from about .1Hz to maybe 2K Hz probably -- I think many sound cards don't go near DC frequency.

                            -SB
                            Hi SB,

                            you don't need the near DC frequency range. Direct sampling at the pre-amp output and direct digital demodulation. That's it.
                            Cheers,
                            Aziz

                            Comment


                            • Originally posted by simonbaker View Post
                              Again, I think the original use of the pot was a chatter filter too, since the voltage level determines how short a pulse from the comparators is rejected -- assuming you have the capacitors in there. Unless you're talking about a different pot.

                              -SB
                              That is the exact pot I was thinking about, however, in a 0Ω position a capacitor parallel to a 100kΩ resistor does exactly nothing. Delay is associated primarily with a capacitor next to a LM358 non-inverting input. The remaining capacitor next to a 1MΩ does exactly nothing. The very 1MΩ does exactly nothing as well. You may associate delay function with anti-chatter, but only in case of a binary response, but you may also omit it and anti-chatter with disc sens pot, and gain some real control over it.

                              I think it is a waste not to use this pot for volume control. For the existing rigs I'd suggest extracting the trimmers and wiring a linear stereo potentiometer of 100kΩ instead. At position 0Ω you'll have a pure binary response anyway, so everyone is happy.

                              Comment


                              • Originally posted by Aziz View Post
                                Hi SB,

                                you don't need the near DC frequency range. Direct sampling at the pre-amp output and direct digital demodulation. That's it.
                                Cheers,
                                Aziz
                                Yes, but unless you supply the oscillator signal as well you lose discrimination (no phase information). It can work well with a stereo input. The high side of this approach is completely avoiding 1/f noise. If processing is done properly, say using a decimating digital filter to gain S/N, you'd be able to obtain very sharp responses for very low signals.

                                Baseband is troublesome because all soundcards are high pass cut at~20Hz, and you inevitably lose DC information.

                                It is all in signal autocorrelation and crosscorrelation properties.

                                Comment

                                Working...
                                X