Announcement

Collapse
No announcement yet.

LM394 on my Minelab PCB

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

  • Hi there Mick,
    I have tested the machine in two different setups:
    First, when I connect the coil and turn it off, that is BUK455 FETS off, I have about 4 volts peak2peak noise on output.
    Second, when connect input to preamp to +Batt, and BUK455 FETs off, no difference coil connected or removed, noise on output comes within 400mV P2P. Turnning the coil on in this case (I have used wire for connection between coil receive and input of front end so I can connect input of front end to +Bat while the coil is in service) increases the noise a bit.
    I measure output on base of BC857 before 100k res, pin 6 of TL061 in Audio CCT.
    In both cases, removing 4066 analogue switch of Ground Balance reduces the noise and putting it in place increases the noise amplitude and turning the GB pots has just transient effects and nearly no steady state effect on output.
    Frequency of noise is random but in order of 1~2 Hz.
    Mick, if yellow wire in transmit CCT is receive wire so what are the red and balck ones?
    Another question, should the ground plane be connected to +BAT?

    Comment


    • Hi Ghaj,

      The red wire connects to the tx cct through the byv28-200 and also the damping resistors. The black is coil gnd.

      The ground plane is also connected to batt+.

      Did you do the test where you short out the output of the second last opamp in the front end? This result will be interesting...

      cheers Mick

      Comment


      • Hi there Mick,
        Before any thing, I modified my LM394 front end and replaced the LM394+NE5534 amplifier with an AD797 chip. So, now my front end is exactly as SD2000 just using AD797 instead of LM394+NE5534. Still I can,t see any apparent differences in performance.
        I did the test you said. I connected input to second NE5534, that is before 3k3 res, to GND and measured output of dc blocking stage. When Ground Balance 4066 is in place, output swings between -15mv to +20mv around +BAT. When GB 4066 is removed it comes to -10mv~+10mv.
        I can't judge these voltages since I don't have access to an original SD2000.
        Can the problem be because of splitting whole board into separate boards? specially separation between front end receive and transmit CCT?
        Should the ground plane be considered for all boards? That is logic, power, transmit, ...?
        By the way, as I have got from your previous post, are both red and yellow wires connected to the same location?
        Thanks.

        Comment


        • Hi Ghaj,

          Ok well +and-10mv is much better! With an original one, it is about +-7.5mv(15mv peak to peak) and the noise is random very low frequency, .1hz to 2hz.

          So that tells me that your problem is in the preamp or the front end fets. Try shorting the + input to your ad797 to gnd(also have the input to the fets shorted to gnd) and see what the output noise is doing. What are you using for fets? Are you sure they are connected correctly? You can pm me if you don't want to give out specifics...

          Yes for a mono coil the yellow wire is connected to the red wire, but if you use a DD coil, the yellow wire is then hooked up to the RX coil(this is why a DD coil needs to have the RX winding separately damped)

          Now yes the ground plane should be considered for just about all areas, although I think the tx area is not so important except around the tlc271 . At very minimum for the logic cct you need to have a ground trace for your bypass caps to terminate to. The front end and signal processing definitely need a good ground plane, as well as a good connection to the coil gnd. Make sure the preamp ground(connection to the - input trough the 8ohm resistor) is nice and close to the coil ground ground also. Make sure there is also a connection from your analog front end ground back to your power supply ground also as there will be currents that will want to source their way back to the power supply caps. Mount the 4093 and associated parts on your front end board, not on the tx board.

          Cheers Mick

          Comment


          • Hi Mick,
            I did what you said. I shorted + input to ad797 and input of FETs to GND. The noise at the output of the dc blocking stage went up to -20mv~+20mv for ch2 and -100mv~+100mv for ch1. Putting 4066 for ground balance makes things worse twice.
            Could you explain connection from analog front end ground back to power supply more? Does is it matter which point on analog ground of front end and power supply?

            Comment


            • Hi Mick,
              what are does trimmers doing there? their R is very low compared to surrounding resistors, for example compared to 10k resistors. Are they really 200R and 500R trimmers? does changing them have any effect on performance? For example changing 500R trimmer does not change gain of first Ground Balance low pass filter very much as long as 5k9 and 10k resistors are in the path.

              Another question. Are 62k and 220k resistors below ground balance pots, in ZED schematic, in right position? Shouldn't they be replaced by each other? I mean 62k in position of 220k and vice versa? I say this because duration of ground samples are very larger than ch1 and ch2 samples. So I guess gain of GB channel must be at least one third of ch1 and ch2 channels (I mean 100k/220k for GB compared to 100k/56k for ch1 and ch2).
              Tnx.

              Comment


              • Hi Ghaj,

                The trimmers are there to match the + samples to the - samples, so when the integrator does the subtraction the result is 0 when there is no target or ground present. They are to cancel static magnetic fields as the coil is swept through the air as this will cause an offset which needs to be cancelled. See this topic: http://australianelectronicgoldprosp...rs-sd20002100/

                Now most of your noise problem is from the front end fets, as when you have eliminated them the noise has dropped from 4v peak to peak down to 20mv and 100mv with the ground channel on. There could be some error in your layout causing the extra noise from the ground channel. The setup of the 220k is correct.

                Cheers Mick

                Comment


                • just wondering has anyone gone to the trouble of starting a schematic of the 2100 ?

                  Comment


                  • The biggest problem of all the P.I designs is the resistive loss in the front end design... Minelab with the N and P fet and GPX N and N are resistive across the junction and degrade the signal to noise ratio, there are others that use a resistor and diode arrangement that is far worse due to high resistive losses, added thermal noise, breakdown noise and non linear junction noise components. The LM394 can surpass the AD797 if it is used as a matched current amplifier. The so called ZVT is a curly one. Zero Voltage = zero current = no power = no signal. What they are attempting to say is "getting rid of ,or reducing the back emf voltage spike" to sample closer to transmitter turn off. This in itself is not the answer due to lighting up ground minerals like a spot light.... I suggested this 6 years ago on voltage annihilation to reduce the back emf transient... its not rocket science , it has been done for over 40 years. Use Low R base spreading resistance bipolar transistors ... directly connected to a low source impedance receiver coil... that is the ducks nuts for the best front end design that is known at this present time. Fets or resistors in the input path is crap, a cop out on state of the art.

                    Comment


                    • Originally posted by detectormods View Post
                      Zero Voltage = zero current = no power = no signal. What they are attempting to say is "getting rid of ,or reducing the back emf voltage spike" to sample closer to transmitter turn off. This in itself is not the answer due to lighting up ground minerals like a spot light.... I suggested this 6 years ago on voltage annihilation to reduce the back emf transient... its not rocket science , it has been done for over 40 years. Use Low R base spreading resistance bipolar transistors ... directly connected to a low source impedance receiver coil... that is the ducks nuts for the best front end design that is known at this present time. Fets or resistors in the input path is crap, a cop out on state of the art.
                      I suspect it's simply a labeling shift from the DVT method to Constant Current, where in the coil Voltage is asymptotically Zero during maximum current pulse.
                      Last edited by Kev; 12-16-2014, 11:00 PM. Reason: clarity

                      Comment


                      • hello detectormods
                        How does the the modded gp3500/gpx4500 compare to the ZVT technology what are the benefits ZVT compared to what we already have ????

                        thanks in advance

                        Comment


                        • I am sure that a representative from Minelab will be explaining it to everyone soon enough.

                          Comment


                          • Originally posted by mario View Post
                            Hi,

                            That schematic is spot on!!!

                            I tried getting capacitor sizes to enter them on the photo, but I can't really measure them directly unless I physically remove them, which I won't do. Unfortunately, that white gunk, which I can not remove with alcohol or scrape off (I wasn't able to find that bug repelant at my store) is all over the capacitors, so I have no way of knowing their sizes.

                            Also, I tried obtaining values for the diodes, and that's kind of hard to do as well...

                            --------------

                            I noticed that IC 7 feeds IC 8, not IC 5.
                            ------------

                            Should we go on and Id the remaining chips? do you have any ideas what they might be (I don't mind continuing )?

                            What other chips could we replace to obtain an increase in performance?


                            Any ways, based on your work (thanks btw), here's what we have so far:


                            High res photo is here

                            it such a shame all these pictures are lost to the forum most of the pictures in this thread are long gone would someone like to re post them

                            Comment


                            • Originally posted by Ronk View Post
                              it such a shame all these pictures are lost to the forum most of the pictures in this thread are long gone would someone like to re post them
                              This is one of the problems of posting attachments on external sites, as these sites are often transitory, and the images are only there for a limited time.
                              It is always preferable to use the attachment facility on Geotech, so that they remain in the database.

                              Comment


                              • hi guys i am trying to find out the circuit for the pulse generation for the Front end FETs sd 2100 early version uses a microchip

                                Comment

                                Working...