My AGD analog detector work

[Auto-Mation-Assist]

Yesterday I was looking at the values on the TX coils damping resistors that need to stay at a constant value. One of these is the resistor value used to provide proper damping when a DD coil is selected and the RX section is completely removed from its connection to the TX coil including all its components that protect it from the high voltage TX coils decay. Since these parts are no longer connected it is desirable to disable the decay control parts that help control the problem created by the RX clamping diodes as I described in prior posts.

The best way to disable the fine decay control circuit is to not let the coils decay voltage reach the fine decay control circuits input by shorting it to ground. Now we can break the connection that goes to the TX coil and insert whatever value resistor is required to properly damp the TX coil with the RX section disconnected. This works fine and if we now remove the short to ground and select the Mono coil we find that the decay correction circuit is no longer working properly. This is due to two reasons. The control circuits input resistance is now to high and also we have increased the level of positive feedback in the circuit caused by the junction of the added resistor to make the DD coil happy.

There are two things to do to fix this.

First the resistor that is causing positive feedback (R7 62K) needs to connect back to the TX coil and always remain connected there.
Second the total value of the added resistor/resistors (R2 and R5) and the original 6.8K resistor (R3) combined need to be close to 6.8K and not 8.371K. Thus R3 needs to be about 5.1K instead of the original 6.8K. In the picture below R2 and R5, the DD coils TX damping resistor addition are about 1.571K total.

Here is the picture that shows what I have planned for making the damping resistors proper for each type of coils with readily available standard value metal film resistors.



It may not always be possible to come up with a value that is within several ohms from actual test values. If that is the case it is all ways possible to solder a parallel resistor across another on the bottom of the circuit board since these are all through hole resistors. Generally I allow for five parallel resistors for main damping, four for the RX pre-amp and two for the DD coil damping on the AGD24.1 version circuit board. The internet has many parallel resistance calculators to aid in determining what possible combination of resistor values can give the desired values.

[Auto-Mation-Assist]

The AGD24.1.B circuit board has not arrived yet. To get ready for these I checked the length or the wires coming from the front control panel and found that the ones for the volume control, Timing and power input was going to be to short and I decided to replace those instead of splicing them. This allowed me to get a picture of the populated front panel circuit board. There usually is not a need to discuss the front control panel but I thought that a picture may help in case someone is considering building one. It does show how the various wires connect. The coax routes to the variable gain amp in the receive front end. The board color looks weird from the light source.

You can also see the metal case and the rail I made to attach it to the wand. It can use the standard GPX arrangement.



The back for the circuit board looks like this:



The little jumper wire on the bottom left was a change on how the standby feature worked. It used to disable pulsing but that effects component temperatures and was found not to be useful. It was instead changed to just mute the audio output. The power switch works fine for disabling pulsing.

Since updating to AGD24 version is a large change in circuit board layout makes it necessary to make a new rear panel. The old panel can still fit but the connector layout is not the best for use with the new circuit boards which take up considerable more space. The main reason for the updated boards it to make it a lot simpler to and some kind of digital controller if desired. There is plenty of space available on the secondary B board to do that.

In order to determine what the best locations are for the connectors on the rear panel I made this drawing to show were the two circuit boards and the main "A" board connectors are located. This would be like looking into the back of di-cast enclosure.



In the original there was room for a 4 inch speaker in the enclosure. A built in speaker is convenient and I have some new speakers on order that are not as deep as the originals and hope to be able to use them. It is possible to mount them from the outside instead of the inside.

[Auto-Mation-Assist]

The lower AGD24.1.B circuit board arrived a few days back and I was able to get all of its parts installed and checked out. It is working properly but I did have an issue with some connector locations mainly because I designed the board before working on the support tray requirements. This will be corrected on the next blank board order. The picture below shows the circuit board installed with its connections to the front panel in place.



This board has the CMOS logic, VCO, Demod Signal handler, Audio amp and an indicator driver which comes in handy. The VCO is on the bottom left and the CMOS logic is on the Bottom right. The demoded signal handler for the signals from the main circuit board is on the top left. The auto amp and indicator driver is on the top right. The audio amp is on the top center. The board is 4.1 inches wide and 5.9 inches high. I mostly designed this board to make it possible to add some digital processing, likely just for the VCO, CMOS logic and add a scope like display along with wireless audio interface. For now it will basically be an all analog machine which has proven to work very well for me.

The DEMOD signal handler/VCO driver on the board above gives a 25 Hz increase in VCO frequency for a input voltage of +10mv and 2.5 Hz frequency shift for a input level of +1mV. A 2.5 Hz frequency shift is easy to hear while sweeping the detector at a rate of 1/2 to 1 second across a very small target. The VCO idle frequency that I use is 400Hz which is suitable for most small loudspeakers. The VCO outputs a sine wave.

The boards current requirements are, 20ma from the -12V source (usually about -13.5 volts), 47ma from the +10 volt source and 39ma from the -10v source.

I expect the main circuit boards, the AGD24.1.A boards arrive some time next week. It will mount on the back side of the board shown above and interface with it using PC/104 type connectors. Two 6 pin and one 10 pin which allowed for a good board layout on both circuit boards.

[moodz]

Some of your posted images look like two small squares ... ( I think its a problem with this bulletin board software ... Carl ? )

see below. ( unless its happening to me also )

[IBGold]

Some of your posted images look like two small squares ... ( I think its a problem with this bulletin board software ... Carl ? )

see below. ( unless its happening to me also )

Might be an Aussie thing same with me
Regards, Ian.

[Qiaozhi]

The images have not been uploaded properly to the forum. They're probably linked to some files on the local computer so are not visible to anyone else.
The proper way to upload attachments is to use the button at the bottom of the page, conveniently labelled "Upload Attachments".

[Auto-Mation-Assist]

I think you are in part correct, but the pictures were properly uploaded.

I think that the problem is with the forum software. When I log on to make a post and it takes sometime to make it, I get logged out for some reason and receive messages that state a server error. My computer indicates that I'm still logged on. In order to finish my post I have to logout, log back in and then restore my post from the auto saved forum feature. When I look at what is restored it is correct and the pictures are also there. But it actually appears to loose the links to the pictures that were uploaded somehow.

Thus the problem source is being logged out during a post and trusting that the prior auto saved data and links to uploaded pictures are correct. The text is correct, but the picture links are likely not since it appears to save the source location of the pictures and not the uploaded location. My posts all look good to me and have the pictures which would seem to verify that the wrong location is being saved by the forum software for the pictures.
I

[Qiaozhi]

I think you are in part correct, but the pictures were properly uploaded.

I think that the problem is with the forum software. When I log on to make a post and it takes sometime to make it, I get logged out for some reason and receive messages that state a server error.
I

Did you click on the "Remember Me" button when you logged in?
If you don't do that I believe you get automatically logged out after a certain time.

[Auto-Mation-Assist]

Did you click on the "Remember Me" button when you logged in?
If you don't do that I believe you get automatically logged out after a certain time.

No I have not done that. I will try that.

[Auto-Mation-Assist]

I wanted to post something about multiple TX pulses in a row of variable frequency and why I think that this is not usefull for a PI detector unless it is being done to overcome a specific problem caused by choice of RX signal processing.



We here a lot about multi frequency metal detectors and the benefits of that technology is primarily limited to its use VLF detectors which use tuned circuits. The application of multi frequency in PI detectors is not really very useful in the transmit side of PI detectors. Using variable width TX pulse in a PI detector turns into nothing more than a TX power level control. We all know that voltage leads current when charging the TX coils magnetic field. Thus if we have longer and shorter pulse we are changing the amount of stored energy that is in the TX coil. This energy is released into the ground when the charging current suddenly stops at the end of the TX pulse. This is a simple way of looking at this but useful just the same.


So if the TX coil is charged for a shorter period of time it will have less stored energy to charge the ground or any objects in the ground. Less stored TX coil energy meas that the detection distance will be decreased and thus its best to settle on a single TX pulse width that is suitable for the inductance of the TX coil. During the TX coils charging cycle the slope versus time is actually quite large and things in the ground start charging but not very rapidly. Rapid charging of any metal object will primarily occur when the TX pulse ends which causes most of the coils stored energy to unleash into the ground. At that time metal objects start bouncing back the signals to us and happiness sets in. In the detector the coils damping resistor helps with getting the energy of the coil into the ground by the high current passing through it. You can compare what is going on to a cars ignition system that generates the spark. Different use of basically the same thing and with no need to make it overly complex.


Those of use who have test equipment know that the recoil voltage of the coil will vary greatly depending on how wide TX pulse is, with width determining frequency. A shorter pulse produces less TX coil stored energy than a longer pulse. So why have longer and shorter pulses? Very likely because some issues had to be overcome, and the method chosen was suitable to solve that problem.


Changing pulse TX pulse width does nothing more than change the TX power level via the amount of the coils stored energy. The start of return signal detection always starts after the TX coil has dumped its stored energy into the ground, and then very shortly after that we start looking for return signals from metal objects. At this point in time we can generate channelized signals that are based on the time differences to see the signal level difference between channels or make comparisons between them based on other potentially useful parameters.


My point is that only a single TX pulse is required on the TX side, and that frequency-time-based related sampling can all be done on the RX side of a detector by channelizing the RX signal with the use of analog gating and routing each separated signal to its associated sampling system. Personally I have found not found the need for more that two channels, one fast and the other slower for gold detectors which in general are not the same as detectors used for finding relics.


In the past I have measured the TX pulse width of one commercially available detector and seen one long pulse followed by a number of shorter TX pulses all of the same width. In thinking about why this was required for their design I suspect that it is due to the lack of speed in the RX signal processing chain and that thus newer RX return signals needed to be generated since the old ones had died out already and thus multiple pulses were required to refresh the RX signal. Sending multiple pulses in a row wastes a lot of time so it really not a good thing to do before starting the looped routine over again. To even make this possible it is required to to TX pulse width or the applied voltage supplied to the TX coil must be quite small in values, other wise decay time will be to long. Again causing potential loss of detection sensitivity due to the TX coil now having lowered stored energy.


It is possible that a detector with just a single TX pulse at a properly chosen repeat rate, and with a good RX system to build a detector with great detection sensitivity. It does not need to have digital signal processing to do that. Know matter how the circuitry is implement you can put as many channels as you wish in say the first 50us window after the TX pulse ends to play with the RX signal and make all types of sounds or display the results.

Personally I think that two channels are likely sufficient for most use, one for fast signals and the other for the slower decaying signals. I do not see any need to go past a single TX pulse per loop rate since the RX signal chain can and should be designed to be fast enough to do whatever channelization of the return signal is desired to be done.

[Auto-Mation-Assist]

I think that this was one of the pictures that ended up missing.




And this one about how things will fit in its enclosure viewed from the rear panel end.

[lucifer]

I wanted to post something about multiple TX pulses in a row of variable frequency and why I think that this is not usefull for a PI detector unless it is being done to overcome a specific problem caused by choice of RX signal processing.



We here a lot about multi frequency metal detectors and the benefits of that technology is primarily limited to its use VLF detectors which use tuned circuits. The application of multi frequency in PI detectors is not really very useful in the transmit side of PI detectors. Using variable width TX pulse in a PI detector turns into nothing more than a TX power level control. We all know that voltage leads current when charging the TX coils magnetic field. Thus if we have longer and shorter pulse we are changing the amount of stored energy that is in the TX coil. This energy is released into the ground when the charging current suddenly stops at the end of the TX pulse. This is a simple way of looking at this but useful just the same.


So if the TX coil is charged for a shorter period of time it will have less stored energy to charge the ground or any objects in the ground. Less stored TX coil energy meas that the detection distance will be decreased and thus its best to settle on a single TX pulse width that is suitable for the inductance of the TX coil. During the TX coils charging cycle the slope versus time is actually quite large and things in the ground start charging but not very rapidly. Rapid charging of any metal object will primarily occur when the TX pulse ends which causes most of the coils stored energy to unleash into the ground. At that time metal objects start bouncing back the signals to us and happiness sets in. In the detector the coils damping resistor helps with getting the energy of the coil into the ground by the high current passing through it. You can compare what is going on to a cars ignition system that generates the spark. Different use of basically the same thing and with no need to make it overly complex.


Those of use who have test equipment know that the recoil voltage of the coil will vary greatly depending on how wide TX pulse is, with width determining frequency. A shorter pulse produces less TX coil stored energy than a longer pulse. So why have longer and shorter pulses? Very likely because some issues had to be overcome, and the method chosen was suitable to solve that problem.


Changing pulse TX pulse width does nothing more than change the TX power level via the amount of the coils stored energy. The start of return signal detection always starts after the TX coil has dumped its stored energy into the ground, and then very shortly after that we start looking for return signals from metal objects. At this point in time we can generate channelized signals that are based on the time differences to see the signal level difference between channels or make comparisons between them based on other potentially useful parameters.


My point is that only a single TX pulse is required on the TX side, and that frequency-time-based related sampling can all be done on the RX side of a detector by channelizing the RX signal with the use of analog gating and routing each separated signal to its associated sampling system. Personally I have found not found the need for more that two channels, one fast and the other slower for gold detectors which in general are not the same as detectors used for finding relics.


In the past I have measured the TX pulse width of one commercially available detector and seen one long pulse followed by a number of shorter TX pulses all of the same width. In thinking about why this was required for their design I suspect that it is due to the lack of speed in the RX signal processing chain and that thus newer RX return signals needed to be generated since the old ones had died out already and thus multiple pulses were required to refresh the RX signal. Sending multiple pulses in a row wastes a lot of time so it really not a good thing to do before starting the looped routine over again. To even make this possible it is required to to TX pulse width or the applied voltage supplied to the TX coil must be quite small in values, other wise decay time will be to long. Again causing potential loss of detection sensitivity due to the TX coil now having lowered stored energy.


It is possible that a detector with just a single TX pulse at a properly chosen repeat rate, and with a good RX system to build a detector with great detection sensitivity. It does not need to have digital signal processing to do that. Know matter how the circuitry is implement you can put as many channels as you wish in say the first 50us window after the TX pulse ends to play with the RX signal and make all types of sounds or display the results.

Personally I think that two channels are likely sufficient for most use, one for fast signals and the other for the slower decaying signals. I do not see any need to go past a single TX pulse per loop rate since the RX signal chain can and should be designed to be fast enough to do whatever channelization of the return signal is desired to be done.

Hi,
I think the multiperiod TX is used to help eliminate ground response. Suppose you have a single TX pulse of 100us duration followed by 5 TX pulses of 20us duration each. The RX responses from the larger pulse and the 5 shorter pulses are processed separately. The responses are proportional to dB/dt so after integration over the the same time constant the RX signal would be about the same for the single longer pulse and the 5 shorter pulses. That's so because for the same amount of time 5X more responses are integrated from the shorter pulses.
The ground response does not depend on the TX pulse width and should be the same for the longer and shorter pulses. The target responses however will be affected by the different pulse lengths. I suppose the method can be somewhat extended further to achieve a level of discrimination.
That's in theory, in practice things might be different.. I haven't tried it.

[Detectorist#1]

Hi lucifer,

Very good explanation of working of Minelab SD2000 metal detector from year 2000. But now we try to make PI detectors for small nuggets of for long distances in hard grounds.

[Detectorist#1]

Hi lucifer,
It is easy for ivconic - many popular detectors have to be used for coins in 15-20 cm depth. But this searching of coins in our country is strictly not allowed! We have to find other variants to practice our hobby.

[Auto-Mation-Assist]

Hi lucifer,

Very good explanation of working of Minelab SD2000 metal detector from year 2000. But now we try to make PI detectors for small nuggets of for long distances in hard grounds.

I think that the technology may go back a good number of years but it still in use in 2024. Today the analog section may feed some some digital processing but that is not at all a requirement to make a very good detector.