My AGD analog detector work

[pito]

post #33

https://www.geotech1.com/forums/foru...ge3#post418967

small hot rocks, super small junk metal and fairly small gold = What about discrimination ?

[Auto-Mation-Assist]

As far as discrimination goes it is not very useful for detectors that are primarily used for finding gold. Here the general rule is to dig all which can also be seen with individuals using very high end detectors in YouTube videos in the Australian outback. The most used discrimination used by myself and likely others is the changes in amplitude received back after the TX pulse ends. I have no desire to have a detector designed primarily for finding gold do anything more than that.

[Auto-Mation-Assist]

I have made a number of post that look at the problems associated with the clamping diodes used to protect the input stage of PI detectors against the high voltages generated by the inductive kickback of the coils after the TX pulse ends. In order to show the problems that these diodes cause with some simulations and some captures of the simulated scope pictures of the actual simulations.

For these simulations i used a battery source of 14 volts and a TX pulse width of 20us, 300uH coil with a calumniated load of 615.92 ohms. This load consists of a 890 ohm resistor directly across the coil and a 2k series resistor going to the clamping diodes. When using a TX pulse width of 20us the peak decay voltage was 375 volts. I changed the pulse with in my last simulation to 36us which raised the peak decay voltage to 670 volts. This had no visible affects no the problem I want to demonstrate and only caused the decay time to become a bit longer. The diodes that I used in the simulation are rated as utrafast recovery diodes but other diodes give similar results.

The problem that I see is that as the coils decay voltage starts to decrease the diodes stop conducting as expected but the bad thing is that not the 2k resistor is effectively no longer connected across the coil and that this causes decay wave form changes at a critical time and which causes signal voltage levels to increase levels higher than anticipated. The pictures of the simulated scope traces are all at 0.5us per division horizontal and 100mV per division vertical. They show the very bottom of the coils decay slope and is the area where we are typically looking for signals in the micro volt or sub micro volt region.

This first picture is for the TX coil decay with the clamping diodes shorted out. The red trace is the coil voltage and the blue will be the voltage
across the camp diodes. It looks good with no sign of a problem with the coils load resistor value.

Horizontal is 0.5us per division and verical is 100mv per division in all pictures.
I have made a number of post that look at the problems associated with the clamping diodes used to protect the input stage of PI detectors against the high voltages generated by the inductive kickback of the coils after the TX pulse ends. In order to show the problems that these diodes cause with some simulations and some captures of the simulated scope pictures of the actual simulations.

For these simulations i used a battery source of 14 volts and a TX pulse width of 20us, 300uH coil with a calumniated load of 615.92 ohms. This load consists of a 890 ohm resistor directly across the coil and a 2k series resistor going to the clamping diodes. When using a TX pulse width of 20us the peak decay voltage was 375 volts. I changed the pulse with in my last simulation to 36us which raised the peak decay voltage to 670 volts. This had no visible affects no the problem I want to demonstrate and only caused the decay time to become a bit longer. The diodes that I used in the simulation are rated as utrafast recovery diodes but other diodes give similar results.

The problem that I see is that as the coils decay voltage starts to decrease the diodes stop conducting as expected but the bad thing is that not the 2k resistor is effectively no longer connected across the coil and that this causes decay wave form changes at a critical time and which causes signal voltage levels to increase levels higher than anticipated. The pictures of the simulated scope traces are all at 0.5us per division horizontal and 100mV per division vertical. They show the very bottom of the coils decay slope and is the area where we are typically looking for signals in the micro volt or sub micro volt region.

This first picture is for the TX coil decay with the clamping diodes shorted out.

Horizontal scale is 0.5us per division and vertical is 100mv per division in all pictures.



Now lets remove the short across the decay clamping diodes and see the result of doing that.
Now you can see the result of loosing the load on the coil as the 2k series resistor feeding the clamp diodes is effectively no longer there.
We now have about a 115mv negative going peak to deal with. The red and blue curves are nearly on top of each other. Red being the coil
voltage while blue is across the clamp diodes. You can also see that the decay time is faster now as compared to my first picture on the
horizontal scale which indicates less loading being applied to the coil.



So far I have not shown the effect of adding the FET input first stage to the clamping diode. The below shows what happens when this opamp
load is added and primarily shows that some capacitance is added which causes separation of the two traces and thus a bit of added delay
between signal routing from coil to opamp.



My last picture just show that increasing the pulse width to 36us has almost no effect on the trace even though the coil peak decay voltage
reached 670 volts. It just show a slight additional delay in the horizontal scale.



There is likely away to eliminate most of this problem via a differential circuit that transfers the normal load provided by the 2k series resister
and transferring that to a separate load for the coil at the proper time perhaps via a analog gate. It depends on how critical it is to overcome
the 115mv issue which I demonstrated. 115mv is a substantial unwanted voltage error at a critical point and gets averaged out later via offset
control functions in my detector. This happens in any PI detector that uses clamping diodes in this manner and small value series resistor
values could cause larger problems than shown here due to the overall ratios of resistor values used to provide proper loading of the coil.

[Auto-Mation-Assist]

Since I'm still working on further potential upgrades to my detector I wanted to dive into potential fixes coil resistive load changes that the typical diode clamps cause but the solution for this is not going to be easy or simple. The main problem to deal with is the high voltages involve and the fact that the diodes are usually placed from a ground point to a more negative or positive point depending on the circuit design. I decided to see what would happen with simple approach by setting the diodes above and below ground potential in order to be symmetrical from the o volt ground reference. I used offset levels of 300mv, 400mv and 500mv using the test circuit below.



XSC1 is the scope display and its settings are 0.5us per division for horizontal and 100mv for the vertical like before.

The picture below is for the plus and minus 300mv clamping offset voltage with the blue line being the output that would normally go to
the first stage input. The red is the coils decay voltage after the TX pulse ends. It appears to be at about the -90mv level. You can see that
that the blue trace is raised which would show a better coil load but then the red one is no longer close to it. Thus what is really happing
is that the received signal is being loaded down and due to the diodes starting to clamp without actually doing anything to give the coil
a good resistive load at that point. The next two pictures with 400mv and 500mv show the RX waveform getting more like expected but
also being loaded down more and more. This indicates that the resistive load correction is not happening at the correct place but that
in can indeed be corrected if done properly. Which is directly in parallel with the coil when the clamping diodes stop conducting at the
tail end of the recoil curve.



The 400mv plus and minus curves. Blue curve getting better but loosing a lot of RX signal since there is a lot of spacing between the
two curves below the center zero crossing point.



The blue curve looks fixed at the plus and minus 500mv clamping level but is actually totally useless since much of the actual desired
level has been lost as shown between the spacing of the curves.



I knew that the above would not work but charting it helps to understand the problem better. I hope to be able to find a usable and stable
method to fix the problem of the clamping diodes going to a open circuit at the tail end of the recoil voltage slope and thus not loosing
part of the resistive coil load that is the equal to the resistance in series with the clamping diodes. Eliminating the 115mv dip in the
RX signal curve cause by this problem will help optimizing detector performance and reduce potential early overload conditions.

[Auto-Mation-Assist]

I have worked out a proposed solution to clamping diodes loosing conductivity during the final stages of a coils decay curve. This issue occurs when the voltage across the clamping diodes starts to drop below 1 volt and the diodes slowly stop conducting as the decay voltage decreases to zero. When this occurs the series resistor that connects it to the coils basically becomes a open circuit which varies by degree to circuit design. This then changes to load on the coil which is a problem since it changes the decay waveform.

My proposed solution is shown in the diagram below. I used the same diagram a s in my prior post and added some parts. The main parts added are below the line on the top right side. These measure the current going through the same type of diodes as the actual decay is in progress and adjust the coil resistive loading automatically via a feed back loop. From all indications this idea will work fine. It is just a concept at this point and I have not verified it by building the circuit or assuring that all parts would operate withing their design limits. Parts selected were just easy to find for running the simulation nothing special. My designs typically use opamps with FET input stages in the front end so this is what this simulation is designed to be used with.

Here is the diagram showing the added parts. I show a 1Mhz signal source on the left and use it to insure that there is no significant signal loss across
resistor RXL4 after the decay ends. I will show this in the last scope picture. Normally the 1M resistor is disconnected.



The below shown the result of the added circuitry. You can compare to the ones shown in prior days.

The horizontal is 500ns/Div and the vertical is 100mV/Div.

In all pictures the red is Coil Decay-Volts and blue output of clamp diodes and signal that would route to front end.



The below picture is the same except the vertical scale is changed to 10mV per division to show more detail of the decay.
The dip below the zero line is about 2mv. The horizontal is 500ns/Div.




The below shows the 1Mhz signal injection after the decay completes. It show near consistent amplitude from left to right and thus
the coil resistive loading of the coil is near optimum even though diode D4 is not conducting.

The horizontal is 1us/Div and the vertical is 100mV/Div.

[pito]

0.015 gram = what is the size of this piece of Alaskan gold. Take Look at this video

https://www.youtube.com/shorts/vUPJmrlSTlg

I think I can get 1 1/4 inches.
---------------------------------------------------
Alaska Gold Detector. = it does mean it is respond only to gold ?

https://www.youtube.com/shorts/txlgNJkJP2g

[Auto-Mation-Assist]

I worked on my proposed fix for improving the coils damping resistance to include compensating for loosing part of the damping load when the clamping diodes stop conduction. I rearranged my concept of how this can be accomplished and now to the point were I build a prototype without worrying that parts will get damaged by the 700 volt recoil voltage of the transmit coil. The reason for this is to help prevent unnecessary changes in the decay wave form at the mV and uV levels and still maintain fast decay speed. The TX coil damping resistance is critical and proper resistance values are typically calculated to an ohm or less when optimum detection sensitivity is desired. Without proper coil damping a lot of sensitivity can be lost in a hurry and can mess up the internal timing in the detector which handles internal signal routing.

The diagram below it my attempt to fix an issue that many detectors may since many also use clamping diodes. These cause a issue at the point when they stop conducting which causes detrimental changes in the decay waveform at the most critical time. Thus it desirable to apply a cure for this. You can think of this problem as a increase in voltage levels that get amplified in later stages, lead to potential early overload and contain no useful information. This is what improper coil damping does. Perfect coil damping is very difficult to achieve.

The prototype I'm working on is shown the the diagram below. It is a reconfigured version of my prior diagrams which show the basic concept that I plan to use.



The below shows the graph of the coil with my added circuitry no connected and with a the RXL1 damping resistor at 679 ohms. It show a negative wave
below the zero point of about 25.25 mV which is cause primarily by the clamping diodes as they drop parts of load off the coil when they stop conducting.
In my detector this 25.25 mV becomes about 631.25 mV at the output of the first stage due to the gain of about 28 dB.

The graph is 500nS/Div horizontal and 20mV/Div vertical. Blue is the signal going to the front end a red is the last parts of the coils decay voltage.



The below graph is the same as the one above but with my prototype circuit applied and the RXL1 resistor value adjusted to 780 ohms
due to the added loading of the added circuitry.

The graph again is 500nS/Div horizontal and 20mV/Div vertical. Blue is the signal going to the front end a red is the last parts of the coils decay voltage.



The below one has a vertical scale of 200 uV and a horizontal scale of 1us per division. Thus the circuitry show in the prototype
simulation shows that it may be possible to correct the 25.25 mV dip and reduce it down to a variation of about 143uv maximum.
In my case this would reduce the output level at the first stage from about 631mv to about 3.5mv.



I will plan to build the circuit and test it in one of my detectors in the near future and make any changes required.

[lucifer]

Looks like a hard fight for tiny returns. Wondering how that performs in a real circuit.
My oversimplified view on the subject: I would pick the damping resistor so that the coil is critically damped. During the flyback the damping resistance is lowered somewhat as the clamp series resistor is in parallel with the damping resistor. That's why picking a higher value series resistor helps (at the expense of some more thermal noise).
After flyback has collapsed to 0.7V or so and the clamp is no longer active, the coil should be again critically damped. However the input signal going to the preamp is affected by the diodes junction capacitance. That's why I would experiment with fast diodes with less junction capacitance.

[Auto-Mation-Assist]

Lucifer I agree with everything you mention. I have used many types of diodes and for the most part all have shown that high frequency losses even at 1Mhz are very low even with 6k series resistor. My simulations usually give 0.03 dB loss and real circuits have never shown a issue with excessive looses due to diode capacitance. I think that the problem with the diodes changing the decay waveform to be more important and not that hard to fix as my proposal will show. However I can only show the results for my own overall design.

Today I build my proposed circuit and applied to one of my detectors and perfectly happy with the results. We all know that a good normal decay waveform slope contains signs or ringing, overshoot or undershoot of any kind and that loading the coil down with to much damping can get bad enough that small gold has no chance of ever being found. Thus wave form is important and so is the settling time. Some people want discrimination and others like me don't care about that very much. But if it is desired to start measuring phase angles instead of just averaging signal levels then wave form becomes more important and anything we can do to improve that is benificial. Lets look at my first picture.

This picture is the wave form that I always get and the S shape that goes negative is strictly caused by the clamping diodes no longer conducting. The error in amplitude between the two peaks is about 350mv and is significant.



The above picture shows that the decay waveform is not correct due to the clamping diodes. The slope going to the right is correct and if you follow it up from right to left is should go up and exit the scope picture towards the top left and not with a downward slope going down and then back up. The other timing wave is my receive gate for the time I look at this signal, a bit under 2us in duration. Thus I find that I mostly depend on a signal that is not at all correct.

Now lets look at the below picture of the waveform after it has been corrected with the first prototype circuit after installing it in one of my detectors for testing.
Now the decay slope finally looks like it would in nature with just some very minor deviations here and there. Now my receive window for the fast channel 1
has something nice to work with. It is now much more usable for measuring correct phase angles during this early <2us sampling period.




Here is the last picture which shows more of the receive signal after the DC offset control has been automatically adjusted to set the correct zero level for this signal.
Zero reference volts are at the center line and the DC offset amount applied is controlled by the first early sampling period to set its zero reference point.
The second sampling period ends after right after the first sampling period ends. Sampling totally ends at the dotted line at 11.20 us and the ground signal is a negative voltage at this point. This can be considered the ground signal and is now a negative voltage. Later it is dc offset and its zero reference level is automatically set. Both the early time period and 2nd ground time periods are independently adjustable in the AGD detector.



Here is a picture of the schematic for prototype 1. It is really simple. In order to apply this the damping resistor in a detector will need to have its have increased. For me this was a change from 546.6 ohms toe 647.40 ohms. This is because the circuit to be added has parts to the damping resistor built into it so that it can have some control over damping when the RX diode clamp stops conducting. I did not have a ESD diode to try so I used back to back 5.1 volt zeners to help protect the opamp output a bit more. Nor did I have 1W 62k metal film resistor so I used 3ea 21k ones in series for this first prototype. It is easy to adjust the circuits action via the 200k trimmer. To determine a correct coil new damping resistor value make a scope picture of the original first and use it as reference for adjusting later when the new circuit added. .I installed the circuit but did not supply the opamp in the circuit with power so it would not affect setting the now damping resistor value. Set it just like your reference picture. When done apply power to the opamps and adjust the trimmer that controls the circuits gain.

Project schematic on 01-17-2024 - I renumbered the parts after putting on a note. The total of R8 and R5 were set to 132.35K - Not R2 and R3.

[Auto-Mation-Assist]

I have started to make a circuit board layout to add this into my TX board later this year. While doing that I decided to also see what would like happen if the extra set of diodes were not added and instead a buffer was added to pick up the normal clamping diode signal without introducing to much receive signal loss.

My simulation shows that this should also work but the resistor values had to be decreased significantly and the circuit has increased delay and loading of the RX signal as shown the the graphs trace separations but it does provide a alternative way to overcome the clamp diode issue. The simulations also seem to tell me that this way is not as stable, meaning harder to adjust. I do not plan to build this circuit since I'm happy with the prototype that I have already constructed and tested.

Here is the circuit diagram for the simulation of the alternate way that I used along with the graph it produced. The graph is 500us/Div horizontal and 20mV/Div vertical. Also beware that the opamps have FET inputs as in the prior diagrams.



And the simulation result. 500us/Div horizontal and 20mV/Div vertical.

[Auto-Mation-Assist]

I made a prototype board of the decay waveform corrector to make testing a bit easier. Since the circuit is working properly I worked on changing the layout of my TX AGD23.3.B circuit board to a upgraded AGD23.4.B circuit board. I was able to get all the required added parts to fit by moving a lot of parts around, deleting one mounting screw hole and placing the six added large metal film resistor on the bottom layer of the board. This also required converting the TX board into a four layer board. This change will make it compatible with the AGD23.4 RX board. Besides adding the decay waveform corrector circuit there is also a change for connector J2 pin 9 which now gets its decay complete pulse which start the receive sequence from the output of the MC14013BDG pin 13 instead of the BU4S584G2-TR pin 4. This sets the start decay pulse to change state to the start on the TX pulse and eliminates a short delay. Also changed is the diode D2 to a AIDK10S65C5ATMA1.

The decay waveform correction circuit, which becomes active when the normal clamping diodes in the receive side stop conducting during the last part of the coils decay slope replaces resistor R7, a 7.5 K 1W metal film resistor. The combination of R6, R8, R9, R10 and R11 provides the normal coil load of 685 ohms. The waveform corrector is wired in parallel with the 685 ohm load. I have attached three page schematic of the new AGD23.4.B TX circuit board. I will likely send the board layout to my board house tomorrow.



The above picture shows my latest decay waveform at the output of the second variable gain stage. The first analog gate is just before this stage. The lower waveform of about 1.8us is the fast signal while to rest would be considered the ground signal. The center horizontal line is the zero volt reference level. Internal processing has centered the receive waveform to about the plus and minus 4.2 volt level at this point. The effect of this is increased dynamic range of a signal that is normally only positive. The gain stages operate on plus and minus 10V and thus the 4.2v levels are well within the maximum of +/- 9.8 volt range. Moving further up the receive waveform curve is not required and is a area that goes up very steep very rapidly.

This brings up another way I have been thinking about for handling that and If the decay curve is more or less known and can be simulated in a CAD or math program.

Math can be used to make a reference curve which can then be subtracted from the receive curve which would the turn it into a nearly straight line. Continuous digital sampling and subtracting the reference curve values during the receive gate period should make it easy to look for the signal level bumps at any point in the line after and generate alerts as desired. It is easy to generate a digital reference curve with thousands of points for such subtraction or addition as required. Only the numerical values of the reference curve are required to do this. Once you have a flat line representation of the curve determining time related phase should be simple.


Anyway below are some pictures of my experimental work for my decay waveform correction work.
This shows the four connections the proto board to the TX board



The test board before I started working on changing the actual TX board to add the required parts.
I used some isolation milling to make the board.



The AGD23.4.B TX board schematic is attached below as a pdf file.

[Auto-Mation-Assist]

I have been working a number of updates to my AGD detector such as the “decayed waveform corrector” and now “decayed waveform flatten-er” which I will describe in this post. The decay waveform corrector is working very well and new TX boards that have this incorporated should arrive here at about mid-month. Adding it does present a problem since correcting of the waveform to what it should be without the prior flat and depressed points now raises the voltage levels that have to be processed. Thus, it became easier to reach the point where the final receive signal handling op amps start running into a potential saturation area at the very peaks of the receive signal. To increase dynamic range I had mentioned in a prior post to a way to flatten out the RX signal coming from the TX coil during the very last part of the TX coils decay. This post shares the circuit that that I intend to use for doing this and simulations indicate that it will reduce peak RX signal levels by a factor of five and thus add roughly 14 dB of headroom which is very significant and without causing and actual desired receive signal to be impacted to any major degree as shown on my pictures that have simulated 2 Mhz receive signals added.

It does not need a lot of parts to implement the “decayed waveform flatten-er” which effectively flattens the TX coils decay waveform during the period that the waveform is usually sampled or past on to following receive stages. This is also the area of the decay curve where the decay voltage increases by an extremely large amount per microsecond when we move up the curve TX coils decay curve. Having the ability to do this without causing saturating problems in the receive gain stages if you want to find small gold is very important.

The circuitry that I plan to use to do this is rather simple and as I mentioned before can reduce the voltage levels to be handled by a factor of 5 or about 14 dB according to my own simulations. This represents a very large increase in signal handling capability without running into final gain stage saturation issues.

To accomplish this all that is required is a reference waveform that subtract from the received signal at the "proper time" and with no impact on the actual desired receive signal.
We really not interested in the entire coil decay curve and only care about the curve once the receive front end recovers from overload which may occur about .4 to 1us before we start thinking about passing the signal along to following stages. The important thing here is that the receive front end be fast enough to do this so very high speed >100 Mhz op amps are an absolute requirement. If you do not have enough front end speed or bandwidth then you will find that what I use will not work for you. Trying to modify a fast moving signal with a much slower one will not generate any usable results. Time alignment is the all-important thing and thus the reference waveform to be used must be generated at the same time as the TX decay waveform, be reasonably identical in shape and also be totally independent. Only then can one be subtracted from the other and the difference found and past on as a new and flattened received signal.

The diagram shows how I presently plan to generate the required reference waveform using a small separate 300uh inductor and at what approximate voltage level it will operate at. It also shows where I plan to inject its output signal into the front end first stage.

Below is the picture of the ”decayed waveform flatten-er” basic schematic.




I have also attached some waveform curves for normal and with the " flatten-er" added that can be displayed side by side for comparison if desired. Some have vertical lines which mark the +10mV and 0 volt cross points. Blue is the generated reference curve while red is the output of the simulated 1st stage op amp. The simulated signal when used as an injected sine wave of 2Mhz at a 0.1Vpk through a 1M ohm resistor into the simulated TX coil.

[Auto-Mation-Assist]

I built up a quick test circuit board to test the flatten-er I described in my prior post and wired it into one of my analog detectors to verify the circuits operation.
Not a lot of parts required.



I wound a suitable coil to generate the reference waveform and it measured 311uh at 10Khz. I drove the coil with ZVN4424GTA MOSFET and used a NXPSC1060D6J as the series diode. My real first RX gain stage has a gain of 28.08 dB and not the 40 db in my simulation circuit that I shared in my prior post and ended up adjusting the 20K trimmer pot in the flatten-er circuit to 6.292K. I'm pleased with the early results.

I took some pictures of two points in the analog detector. The first being the output of the first gain stage which is ahead of any signal gating. The pictures here show that with the flatten-er in the circuit first stage saturation is never reached during the TX coils decay period.

Pictures for the first stage output are: 'FirstStageNoFlatten.png' and 'FirstStageWithFlatten.png'.

The second point I took some pictures of are at the output of the third gain stage which acts as low pass filter ad also drives the sample and hold circuits for both channels. The second variable gain stage was set to maximum gain which is 18 dB. The low pass filter has a gain of 6 dB and thus the total system gain up to the input of the sample and hold
circuit is about 52 dB or about x398.

The 'MyNormalSignal.png' picture shows my normal receive signal after the 52 db of gain and just prior to entering the sample and hold circuits. The signal is mostly below the horizontal zero volt line by design. If it was not pushed negative it would have severe clipping on top and not usable. Minimizing the amount of negative push is desirable
and this is where the flatten-er comes into play which can be either an analog circuit or be generated by digital techniques.

The results of the addition of the flatten-er can be best seen in picture 'MyNormalSignalFlattened.png'. It shows that now it mostly shows only the difference between the reference signal and the incoming signal. Even with my quick and dirty circuit it can be seen that the benefits of this are substantial since for the most much smaller signal
levels are now involved which gives increased signal handling range.

I also added a picture of the output signal of the waveform corrector which spends most of its time being overloaded with the exception being in the time slot when it is needed. Ref picture: 'CurveCorrectorOutput-10v-Div.png'. It circuit is described in a prior post.

I decided to order some small 300uh surface mount inductors for further testing and will post the result of using those after they arrive.

[Auto-Mation-Assist]

I have been testing with a number of homemade inductors to generate a reference waveform while waiting for the ones I ordered to arrive and found that they are only usefull for preliminary testing since they all picked up way much external noise.

While looking at my parts stock I did find some 330uH surface mount ones that are rated at 65mA and have 13-ohm resistance. They measured to be right at 330uh at 10kHz and showed a Q of just a bit over 1.9 on my tester. See the picture below for size in comparison with a 1206 SMD size resistor. The very small size took care of the noise that the homemade inductors were picking up.



In the pictures below I show the waveform generated for this SMD 330uH inductor by my current drive circuit at 10 volts per division. The both the inductor damping resistor and series resistor feeding the 1st gain stage were adjusted to give the results shown. The waveform correction circuit is also used. If you look at picture 'InitialSignal.png' below you will see that the peak voltage at my final RX gain stage is 8.150 volts with maximum RX signal gain and getting close to the clipping point of around 9.8 volts for the normal pulse width and gating periods used. The signal waveform amplitude in the picture is 5 volts per division.



I the second picture below I show the wave form at the same point but with the flatten circuit active. It shows that the peak positive RX signal voltage is now reduced to 1.440 volts or a decrease of 6.71 volts and it can do this with loss of sensitivity. This represents a gain of signal handling capability of just under 15dB with present settings. Using an inductor with higher current carrying capacity would allow for a increase the reference curve peak voltage and thus give a larger adjustment range.



My last picture shows the output of the analog gate driving the fast sample and hold circuit using my present timing and show a voltage peak there of 1.320 volts reference to ground potential. The sample period is about 1.8us. When I use this method and everything is adjusted properly it appears to me that I see good improvement in sensitivity and clarity of weak signals.



The inductors that I ordered should arrive some time today and have a current rating that is more than 100mA higher than the ones in this test and should thus be able to provide a larger adjustment range.


If using a digital waveform generator instead of an analog based one it would be simpler to generate a reference waveform for different coils and reading a coils ID to pick the best reference wave form to use. There a many one wire ID parts available on the market to do this in a detector system that has a built in CPU. Voltage levels would of course be expected to be much lower thus circuit layout will be different but functionality will be the same.

[jeep6789]

good