My AGD analog detector work

[Auto-Mation-Assist]

I was the closest to become one. Still am.
But I need video as last thing before I start.

Also, you used all the SMD, which i would like to avoid.
Bad sight, trembling hands, can't obtain all the original components from your schematics.
So I was thinking to make through-hole variant with ordinary components and available substitutions.
...
A strange phenomenon, which manifests itself in your last post... and to be fair; I notice it more and more often at myself too!
And that is: the older we are: the more stubborn we are and the easier it is to get offended.
As a younger person among two, I will give in here and say, everything is ok, no hard feelings.
But... I still need a video...

Thank you.

I understand why you would like need a video but I have no experience in making videos that are suitable to view on the internet. I had made a video as an introduction showing my work area, a bit about my equipment and thing I use and also showing the actual initial check out of the AGD23-2 board that I showed partially completed a few days back. That video was over 2 gigabytes in size. Tried posting it to my Github account even at reduced size and resolution it was about 120Mbytes and to large and would have locked up the AGD project page that I had setup about 5 months back and had not done anything with it since. Later I figured that the geotech1 forum would be better since there are so many knowledgeable individuals posting ideas, projects and sharing of information.

Now about using test equipment. If you consider that the heart of a metal detector is its electronics it becomes easy to understand that measurements need to be taking to insure that it is properly designed and meets all design expectations. Then also consider that this is the time potential short comings are detected, perhaps due to parts being placed improperly or unforeseen engineering issues. It is very important to make measurements of the electronics performance and that they adhere to the expected the set standard. The AGD23.2 receive board should be able to cause a 10 Hz increase in VCO frequency (400 Hz to 410 Hz) with a 1 micro volt input signal. That is about equal to my 0.015 grams of gold air test at perhaps a inch away form the Coiltek Elite 9 inch coil. This all depends on timing and other settings on the from panel and local RF levels. With improper timing settings you will not see 0.015 grams of gold.

I think that a through hole board instead of SMD would not be practical. I have been ordering 12 boards at time during updates and would make some available to you if you like. I use some 1 inch tall pieces for wood to rest my hands on while placing parts on the boards. This also helps prevent bumping parts already placed and moving them around before heating the boards for solder flowing.

[Auto-Mation-Assist]

I have the AGD23.2 board completed and tested it to make sure that it working as expected and no problems showed. Sensitivity was as expected with 1 micro volt signal. Since the circuit design is DC coupled it is fairly easy to test. The first stage can be tested by with 125Khz square wave at about 50mv input to the RX coil connector on the board. There should be no overshoot on the leading edge using its output test point on the board. The demod sample and hold function can be checked with a 2 Hz square wave input at a level of 1 Microvolt. The square wave allow you to be able to see what the positive - negative splitter does if you monitor the VCO input test point with your scope. Adjust input levels as you wish.

The below is a picture of the just completed AGD23.2 received board on the right. This replaces the AGD23.1 board with all the jumpers on the left and which made it mostly a AGD23.2 for testing prior to ordering actual AGD23.2 boards.

[Auto-Mation-Assist]

I have been checking the damping resistors values and reduced the values slightly for both the Mono and DD coil on the TX board. I will check to make sure that the updated values are on the diagram I posted before, if not I will post a updated one for the TX board. I have been checking my return receive signal with the Detech 8 inch round DD coil that quite a few people appear to be using to find small gold. While doing that I decided to rig up a swing thing using a golf ball and attach a 0.015 piece of gold to the bottom. It turned out to be close to the center but not quite. The distance to the coil turned out to be about 0.6 inches for this test. It could go a bit higher but I think this is likely sufficient for now.

The test ball. Just discovered that I cant upload .mp4 files that show the results and had to zip them up. The results are for my Detech 8 inch round DD coil. My Coiltek 9 inch Elite coils produce higher return levels. I hope that these files will be helpful.

[ivconic]

Thank you.

I understand why you would like need a video but I have no experience in making videos that are suitable to view on the internet. I had made a video as an introduction showing my work area, a bit about my equipment and thing I use and also showing the actual initial check out of the AGD23-2 board that I showed partially completed a few days back. That video was over 2 gigabytes in size. Tried posting it to my Github account even at reduced size and resolution it was about 120Mbytes and to large and would have locked up the AGD project page that I had setup about 5 months back and had not done anything with it since. Later I figured that the geotech1 forum would be better since there are so many knowledgeable individuals posting ideas, projects and sharing of information.

Now about using test equipment. If you consider that the heart of a metal detector is its electronics it becomes easy to understand that measurements need to be taking to insure that it is properly designed and meets all design expectations. Then also consider that this is the time potential short comings are detected, perhaps due to parts being placed improperly or unforeseen engineering issues. It is very important to make measurements of the electronics performance and that they adhere to the expected the set standard. The AGD23.2 receive board should be able to cause a 10 Hz increase in VCO frequency (400 Hz to 410 Hz) with a 1 micro volt input signal. That is about equal to my 0.015 grams of gold air test at perhaps a inch away form the Coiltek Elite 9 inch coil. This all depends on timing and other settings on the from panel and local RF levels. With improper timing settings you will not see 0.015 grams of gold.

I think that a through hole board instead of SMD would not be practical. I have been ordering 12 boards at time during updates and would make some available to you if you like. I use some 1 inch tall pieces for wood to rest my hands on while placing parts on the boards. This also helps prevent bumping parts already placed and moving them around before heating the boards for solder flowing.

Now I understand your reasons.
Problems with recording a video clip, problems with the size of the file that is generated on that occasion, problems later with uploading such a large file...
Yes, these are standard problems.
I don't know what smartphone you have, but every modern one has its own optimization of recordings. You need to start the camera on the phone and find where it is Settings and set a medium but still acceptable resolution there.
You transfer that recording to a desktop computer via a USB cable. And then you can do whatever you want with that recording. There is a "zillion" of free software that will optimize the video and adjust it for uploading to YouTube.
And you can directly upload the video to YouTube from your phone. You need to register on Youtube and that's it.
Here's how I do it, I use Youtube as a "repository" for my videos. I can easily send a link to a video, when I need it, via email, messenger, forum, etc.
I have an old Samsung S7 phone that has a solid camera. I take a video. I transfer it to the desktop with a USB cable. I have Wondershare Filmora software (downloaded from torrent, unlocked version, but you don't have to, you have Windows Movie Maker type software).
In that software, I load a video clip and just change its type and size, and at the same time set it to keep good quality (720p). And Filmora converts my recording into a file that is up to 10 times smaller in size. Then the upload is easy and fast. Then I upload to my YouTube channel and that's it.
You can do all this in really many different ways.
VLC player (free) also has the option of reducing the recording file. Lots of other software too. There are many that are free.
You will spend 2 hours in one day in your life to find out what suits you best and to put everything on your computer, once and for all.
...
I looked at those two videos and that's what I was looking for. To see at least part of the behavior.
Does the threshold have to be so pronounced? How about reducing the threshold to a minimum?

[Auto-Mation-Assist]

Now I understand your reasons.
Problems with recording a video clip, problems with the size of the file that is generated on that occasion, problems later with uploading such a large file...
Yes, these are standard problems.
I don't know what smartphone you have, but every modern one has its own optimization of recordings. You need to start the camera on the phone and find where it is Settings and set a medium but still acceptable resolution there.
You transfer that recording to a desktop computer via a USB cable. And then you can do whatever you want with that recording. There is a "zillion" of free software that will optimize the video and adjust it for uploading to YouTube.
And you can directly upload the video to YouTube from your phone. You need to register on Youtube and that's it.
Here's how I do it, I use Youtube as a "repository" for my videos. I can easily send a link to a video, when I need it, via email, messenger, forum, etc.
I have an old Samsung S7 phone that has a solid camera. I take a video. I transfer it to the desktop with a USB cable. I have Wondershare Filmora software (downloaded from torrent, unlocked version, but you don't have to, you have Windows Movie Maker type software).
In that software, I load a video clip and just change its type and size, and at the same time set it to keep good quality (720p). And Filmora converts my recording into a file that is up to 10 times smaller in size. Then the upload is easy and fast. Then I upload to my YouTube channel and that's it.
You can do all this in really many different ways.
VLC player (free) also has the option of reducing the recording file. Lots of other software too. There are many that are free.
You will spend 2 hours in one day in your life to find out what suits you best and to put everything on your computer, once and for all.
...
I looked at those two videos and that's what I was looking for. To see at least part of the behavior.
Does the threshold have to be so pronounced? How about reducing the threshold to a minimum?

Thanks for the info.

I cannot use Youtube since they banned me years ago for no good reason while I was a licensed federal firearms manufacturer and building custom precision hunting and competition rifles along with doing general gunsmith related work. My sister was the one who had notified me that I had been banned. I have no intention of ever using Youtube again due to this.

In reference to the fast response time:

I think that I adjusted the sample and hold timing to be fairly fast but not to unreasonable and to the point that it is to fast for the ear to respond to. It is not difficult to make it slower and only requires changing the time constant of the two sample and hold circuits. The best way to do that is the increase the value of the four 1.5K resistors to a personalized value.

[Auto-Mation-Assist]

While upgrading to my already built detectors to the version AGD23.2.A receive boards I decided to do just one since the cost of populating a board is fairly high and use the version 2 board to verify performance for my proposed updates for version 3. As a result I have completed updates for AGD23.3.A (Receive Board) and AGD23.3.B (Transmit Board).

I have ordered the version 3 boards and they should be here shortly after April 4, 2023. There are no basic circuit changes in this update. I eliminated on test point, added a test point on the first timer output to trigger a scope on and added a mounting screw hole. The added screw hole along with one existing screw hole allows for adding a metal shield over the first stage if desired. Also added a test point to aid adjusting coil damping resistor value on the TX board voltage divider output at 10 ohm resistor junction point.

The primary changes have to do with minimizing noise generated by resistors in sensitive areas that presently use thick film surface mount resistors which can be quite noisy. These will be replaced by lower noise types which of course are more expensive. The result of lowering this noise will make a difference in that it will detecting small objects will become more distinct because signals are not being modulated as much by resistor generated random noise as much. Parallel resistors are are also used to reduce noise since parallel devices reduce noise due to partial cancellation of random generated level and phase differences.

All resistors for the coil damping and voltage divider on the TX board are changed to through hole metal film types. I have chosen to use Holco resistors, H4P10RFCA, H4P2K7FCA, H4P7K5FZA and for the DD added load H4P1K8FCA, not my prefered value of 1K87 which is not available right now. Holco resistors are by TE Connectivity / Holsworthy.

On the receive board the three parallel resistors on the input to the first stage are also changed to through hole metal film types. I have chosen to use Holco resistors, H4P20KFCA or an alternate CMF6020K000FHEK from Vishay / Dale.

Also on the receive board all resistors of the first stage change from surface mount thick film types to the less noisier thin film types. I plan to use these resistors once the version 3 boards arrive and are available from Digikey, ERA-8AEB101V, ERA-8APB4750V, ERA-8APB5490V, ERA-8APB2552V, ERA-8APB3322V, ERA-8APB4992V, ERA-8APB5902V and ERA-8AEB105V which is used to provide a path to ground if the input plug is pulled from the board.

The only resistor change I made so far with my tests have been with the three parallel resistors on the input to the first stage others have not arrived yet.

I have been testing with 0.75 spacing on my golf ball swing with the 0.015 gram of gold with the Coiltek Elite 9 inch coils. I did suspend from the ceiling since testing with the DD 8 inch coil to give a wider swing. The DD coil I tested with before has a very knife edge response, and it would likely be easy to miss a small target in the field if the swing is to fast with the 8 inch DD coil. I have been using a 2us wide window to gate the primary sample and hold during my tests.


This shows the upcoming change to through hole parts for some resistors. On version AGD23.3 printed circuit boards these will have the correct pads. The picture show the CMF6020K000FHEK from Vishay / Dale parts which are less expensive than the Holco 20K 1 watt resistors and a bit smaller.

[Auto-Mation-Assist]

The updated receive circuit board (AGD23.3.A) layout is shown in the below pdf file and shows the resistors that will be changed in version 3 to metal film type. These are the three through hole foot prints on the bottom left side. The resistors will be installed raised off the board slightly and will have a shield above them at a distance that avoids excess capacitance between the shield and resistor bodies. The shield will be held in place with two screws on its bottom and cover and area just above the resistors and to just to the left side of the FEG connector (Front End Gain). The shield will have a notch for the RXL connector and have a clearance around it of about 0.023 inch. The added mounting screw hole that I mentioned in a prior post can be seen just to the right of the board bottom center.

The updated receive board:

AGD23.3.A-PCB.pdf

[Auto-Mation-Assist]

Thanks for the info.

I cannot use Youtube since they banned me years ago for no good reason while I was a licensed federal firearms manufacturer and building custom precision hunting and competition rifles along with doing general gunsmith related work. My sister was the one who had notified me that I had been banned. I have no intention of ever using Youtube again due to this.

In reference to the fast response time:

I think that I adjusted the sample and hold timing to be fairly fast but not to unreasonable and to the point that it is to fast for the ear to respond to. It is not difficult to make it slower and only requires changing the time constant of the two sample and hold circuits. The best way to do that is the increase the value of the four 1.5K resistors to a personalized value.

I have to correct myself.

I discovered that I got Youtube and Facebook mixed up. It was FaceBook and 'not' YouTube that banned me.

[Auto-Mation-Assist]

I took some time to consolidate all my updated schematics for AGD23.3 including parts list and some notes, and attached them all as one file.

This is the final version that will be built by myself as soon as the updated circuit boards arrive early next month. I have received all the 1 watt metal film resistors for the TX and RX boards and just waiting for the 1/4 watt Thin Film Panasonic surface mount type for the first stage of the receive board. These should arrive by 29 Mar.

Hoping to update all three of my detectors to the 23.3 version by mid April and expect no further changes. I generated some Gcode to cut out and drill the small shielding plates and just need to think a bit about how tall to make the required spacers.

I believe that I have all the info on my project fairly well consolidated in the attached file and hope that it is found to be useful in some way.

[Auto-Mation-Assist]

I have assembled and tested AGD23.3.A and AGD23.3.B version boards and found them to work properly.



I did make a lot of initial;l measurements due to a initial loss of a lot sensitivity. I traced this down to the sensitivity control working in reverse because the connector on the circuit board got reversed in version 3 to make trace routing on the PCB and bit better. I moved the pins for the DTS plug around that connects the the front panel PCB and the problem was solved. I'm always looking for a way to maximize performance and while solving the plug wiring problem let me to improve two areas. The first one is the low pass filter that is found on page three of the diagrams. This filter is a bit slow and thus have a small overshoot on pulse response. My modification for this filter removes R153 and replaced it with a 0 ohm jumper , and also removes C144 on the prior submitted diagrams. It also changes R163 to 1.41K, R161 to 200 ohms, and R154 to 3.77K. This modification moves the -1db gain point up to 1Mhz and decreases pulse response time to about 360ns and almost overshoot on the rising edge. This filter drives the dual channel sample and hold circuitry on the circuit board and the modification makes the receive signal arrive at the sample and hold also about 400ns faster.

In the original AGD23.3.A diagram time slot T2 is used to close the first analog gate and let the receive signal pass. The width of the T2 pulse is the total of the time slots T0 plus T1. T0 being the leading and primary channel, and T2 being the trailing secondary channel. The analog gate U23 on page two passes only the desired signals present during the T2 time slot and routes those signals to the wide band variable gain amplifier U24. It is important to note that signal routing, amplification and filtering all add to signal processing time, which brings me to the second modification.

If I add up all the processing time required for signal processing before the signal enters the sample and hold circuits that it may be close to 0.8us and that this must be compensated for if maximum detect sensitivity is desired. In looking at the received signal from the output of U25.2 and comparing it to the analog switch U15.2 TS-0 pulse input it can bee seen that the gate on pulse arrives to soon and causes the first part of the signal passed to the sample and hold to have no valid data. This causes a slight decrease in detect sensitivity. My fix for this is to add a delay adjustment to the TS-0 pulse. I also noticed that on the diagram of page 3 that should read 0=On for the analog gates and not 1=On. The analog gates have built in inverters for this signal.

On page 1 of the prior submitted diagrams look at U6.2 pin 9 and its connection to U8.4 pin 9. This connection needs to be cut and replaced with a adjustable delay. This would be a 150pf cap from U8.4 pin 9 to -10V, and a trimmer pot between U6.2 pin 9 and U8.4 pin 6. The trimmer pot value of 5K is workable but I plan for updated circuit boards to have a 2K trimmer along with a 255 ohm resistor. On my present AGD23.3.A circuit board a trace has to be cut under C41 to allow for this modification by wired in components.

This new adjustment is used in combination with the DTD (Decay Time Delay) adjustment on the TX board. I will explain how these are adjusted in a future post.

On page 1 of the diagrams there are a number of jumpers that were used during the initial design to have the ability to select normal or inverted pulses that route to analog gates. These are no longer required and on future circuit boats they will be hard wired and the jumpers removed. This will save some time during board assembly.

[Auto-Mation-Assist]

On my AGD23 it is important to note that the detection level zero volt reference point is set by a multiple servo loops and is based on the shape of the signal present during the time slot TS-0. The zero volts detection reference point for each channel,
TS-0 primary and TS-1 secondary is at the outputs of the two 15hz low pass filters shown on page 4 of the prior submitted diagrams. The above picture is explained below.

Scope probes used were 10:1

The above picture show the effects of the servo loops on the receive signal at the inputs to the sample and hold circuits.

The scope reticle center horizontal line is the zero volt reference point.

The two vertical lines show the DTD (Decay Time Delay) on the AGD23.3 TX board was adjusted to 3.7us.

The negative going pulse to the right of the right vertical dashed line shows the start and end of the 2us receive window for looking at averaged receive signal level during that time.

The signal remaining after the 2us pulse is negative and is the receive signal present in time the slot TS-1. It will be brought back to a zero ref level in its 15hz low pass filter and will go positive on detecting a signal during its time frame.

The effect of the servo loops is that the major portion on the coil decay voltage is initially reduced to a negative voltage. How far negative goes is determined by the averaged voltage that the sample and hold circuit sees and the servo loops. It is also determined by the coil decay voltage level which is controlled by TX pulse width, coil characteristics and damping. Lots of unknowns.

Part of the automatic zero reference adjustment is done within the primary TS-0 channel 15hz low pass filters and part is done by controlling the front end offset. The benefit of this is of course the to automatically set a zero volt detect reference point, but it also gives the ability handling higher input signal levels in the front end amplifiers when climbing up the decay slope to very detect small pieces of metal that decay very rapidly after the TX pulse stops. Think of it as automatically re-centering the receive signal withing the available +/- 9 volt signal range of the front end amplifier stages based on what segment of the receive signal is being looked at.

The below show the the same signal but just the signal area of interest at 1us per division sweep rate.

[ivconic]

Very neat and nice work, I said that before.
I forgot if you posted the gerber files yet?
If you are not; do you intend ?
I see that most of the components are of such sizes that I might dare try to solder them by myself.
Because I like the concept, I would like to try to make one for myself.
BTW you have a very good oscilloscope, what model is it?

[Auto-Mation-Assist]

I had not intended to upload gerber files. I have been ordering 12 boards at a time and plan to use 3 boards for our own use. At 12ea the 4 layer RX boards cost $24.08 US per board and stencil $24.79 US. The TX board is only two 2 layer at the present time and for 12 boards was $8.19 US per board and stencil $14.09 US. I willing to supply left my overs at my cost plus shipping to a builders location if asked.

When I initially started with hand assembly of the circuit boards I purchased aT12 soldering station with various heating elements and a quick 861DW hot air station along with 45 degree angled tips. Also a set of 9 tweezers from Amazon and size "T4" Chipquick solder paste. I build it section by section using the hot air when I started this project. The T12 soldering station was used for some rework and soldering the through hole parts. I feel that those are all required items to save a lot of aggravation. A head band magnifier is a great help. Cut three or four 3/4 inch by 3/4 in wood of proper length to surround the boards to rest your hands on to prevent shaking while placing parts. Don't wear a long sleeve shirt it may catch place parts and move them around.

There are quite a few small parts mostly diodes of various types.

My scope is one that I bought perhaps 20 years ago. It a Iwatsu SS-6122 4 channel 100Mhz scope.

[Auto-Mation-Assist]

I have completed the required schematic and circuit board changes required for upgrading the RX circuit board to AGD23.4. Version AGD23.3 RX board does not have space available to add the fore mentioned 150pf capacitor, added 255 ohm resistor and the 2K trim pot. Other than that it is completely useable if the added parts are hand wired in and one trace is cut on the circuit board.

Schematic drawings and parts lists for version AGD23.4 is attached. Parts have been re numbered and a number of resistors have been specified to use thin film types with TF on parts list and the schematic. I will check the AGD23.4 circuit board layout again for proper placement of the top silk screen and plan to submit it for production by April 13.

[ivconic]

I assume you didn't make the pcb in China judging by the prices?
I imagine the prices would be significantly lower if the pcbs were made by JLPcb or some similar company in China.
That's why I asked about the gerber files.
I have been making my pcbs in China for several years and mostly at JLPcb and PCBWay.
The prices are more than favorable. However, I have never ordered a stencil until now.
But now I see that it is a very important and useful aid in the placement and soldering of components.
And your additional tips are good and valuable, I will remember them.
It remains for me to draw the pcb myself. At least to try.
I never drew more than 2 layers.
That will be a problem, in case 4 layers are very important in this detector.
Of course when I say "I will make"; that does not mean right now these days, but later when the conditions are right.
The idea is good and interesting enough and deserves a place in my "to do" collection.
The prices of your pcbs are not high, given that they are very well made, as can be seen from the pictures.
A much bigger problem is the shipping costs and the arrival time of such a shipment.
With those additional costs, it would become a serious figure.
If I had gerber files; I could upload them to JLPcb's server and see the production price right away and I bet it would be many times lower.