Tweet
Hi AMA,
I agree! The wheel is very old idea but this is not a reason to stop the using now days.
I agree! The wheel is very old idea but this is not a reason to stop the using now days.
-
-
I have been checking out the AGD24.1 prototype circuit boards shown below.
With the lowering of resistor values in front first stage i was required to determine new resistor values for everything that has do do with TX coil loading. I succeeded in getting that resolved and decided to try the 1W metal film resistors usually available on Amazon as shown in the bottom of the right picture. These are all standard 1% values but I do not know their temperature coefficient. I Usually I like to use 25PPM precision types which are a lot more expensive but will use those again in the future. I expect that the 14 higher quality resistors would likely cost about $42 or more but these are critical parts and for best performance the higher cost parts should be considered. They effect noise levels and also temperature stability.
Both of the prototype boards checked out OK without making any changes. During testing I did find that it was possible to improve the slow channel a bit by not sharing the output of the sample and hold driver with both channels. This lead to adding a second driver for testing to the bottom of the right board above. Picture below.
Old wire wrap wire comes in real handy. It looks bad but worked fine for testing.
To do this I cut up a older blank board and stuck it to the bottom of the main board with double sided tape. This added circuitry has already been added to the AGD24.2.A circuit board and both 24.2 version boards are presently in being manufactured. The B board just having some of its connectors move a bit.
Metal work was required for the new rear panel of the AGD24.2 version, and also the shield that covers part of the A board.
They will look nice once they are Cerakote'd.
I did arrange the D hole cutout for the coil connector so that the pins face the correct way for TX and RX wiring with no crossovers. The rear panels above were cut out from 4 inch wide aluminum flat bar which is about 0.130 inch thick. Or a bit thicker than sheet type.
Does anyone have experience with 3D printing for enclosures using filament that contains metal ? I'm curious if it offers reasonable interference rejection properties. Up to this point I have only considered aluminum enclosures but the right size can be difficult to find. 3D printing if suitable would be a good option.
Comment
-
Waiting for parts to arrive and thought about making some block diagrams to show signal routing. The first one is for the receive section of the AGD24.2.A board and it is posted here.
Comment
-
It was a bit of a challenge to get everything to fit in the old enclosure when I changed to circuit board and separated their main functions to make it simpler to add a micro controller if that is ever desired. I have the first detectors modification completed and this post show some pictures of how everything was made to fit and also some results. There are 15 pictures and also a .rar file that contains all pictures.
Remember that the AGD detector is a analog detector and was designed to see what level of performance could be obtained with a strictly analog detector.
This is a picture of the B board which could be modified to include a microprocessor if desired. It contains the audio amp, VCO, the CMOS timing circuits, and the important analog circuits that amplify and filter the demodulated signals from the A circuit board. These analog circuits are towards to top right.
The picture below shows how this board is installed it its mounting tray and also the wires that connect it to the front panel control board. The A circuit board plugs into the back side of this board.
The next picture show the A circuit board. This circuit board board contains most of the circuitry required to build a metal detector.
The below picture shows how this board is installed onto the mounting tray and also shows the shield that gets installed over part of the A circuit board.
The connectors that interface with the rear panel of the detector are the DDS, TXC and RXC. DDS being the Mono, DD, Concentric switch, and TXC being the output to the TX coil and RXC being the input to to receive circuit.
The above picture show how the rear panel connecting wires route to the A circuit board. The ferrite core needs to be toward the front so that sufficient room is available to plug in the three connectors. Starting from right to left.
The above is the back side of the rear panel. The wires are solid core not stranded wire so they hold their respective locations.
The rear panel of the enclosure. Note the guard that prevents accidental switching to DD coils.
The next picture show how the boards were stacked in the enclosure as seen from the rear with the rear panel removed.
The loose connector is a extension for the analog indicator. Signal from this cable are for a analog meter movement but could be run to a small portable scope with TFT display.
The picture below shows the rear panel connectors plugged in. I start with the RX connector on the right and then do the TX and coil selector switch. Its a tight fit but has enough space and not overly difficult. The ferrite size core causes most all the difficulty.
One of the problems I ran into due to the stacking of the circuit boards and the shield over the A board and the ferrite core was that there was no room for the 3.5 inch speaker anymore. As a result I had to change the built in speaker to a less efficient 3 inch one and move it forward.. I cut a filler plate to cover the old speaker hole and then cut a new hole. The round filler cannot be pushed in because it has a stepped back. I have done this for all three of my enclosures.
Comment
-
I have a few more pictures that are parts of the last post.
Its been a while since I posted a picture of the front control panel. It gives a lot of control. I usually leave the controls as shown which gives me a TX pulse of about 36us and a TS0 width of 1.8us. Never have a need to use the TS1 signal. Volume would be higher that shown. There are two markers on pulse width one of about 36us and the higher one for 40us. If a setting of higher that 36us is used the TS0 time should be increased since the coils decay time will be longer.
I wanted to show what the waveform looks like for the fast and slow channels. The top trace being the TS0 fast channel and the bottom being the TS1 slow channel. These signals are from the input of the two sample and hold circuits. The timing of the TS0 signal and the fine coil damping load adjustments have been set so that the left edge of the upper waveform goes down at the zero volt point, not above or below the flat line. This results in the best sensitivity with the least amount noise because timing of the analog gates is best. This nice to have trimmers that are able to set this point.
The above show the wave form when switching to a DETECH 8 inch DD coil. This signal was taken from the decay sample point that routes to the decay complete detector which actually detects a decay point slightly before the decay is actual complete. This allows for a variable additional delay adjustment to set any desired decay complete point via trimmer control. The TX coil inductance for the DETECH is a bit lower that my normal Coil Tech coils.
I tested the detector in my driveway by choosing six arbitrary locations at gave me some signals and dug up to see what it was finding. I think that the results were pretty good for this rusty metal as compared to a US 1 cent coin. Each one on these were from its own spot and I'm pleased with its performance. Really happy with the smallest one. These were hard to separate in my green plastic scoop from the other dirt.Comment
-
Excellent job! I'm impressed by the quality of work.Comment
-
Thank you. One of the things that I will work on is to determine the performance difference of the receive side by changing to FET switches and eliminating the resistors feeding the present diode clamps. So far I have used resistors to minimize the use of components that require switching on and off to minimizes the potential of generating IMD based on my prior experience when designing high dynamic range receivers. The present resistors cause some loss of sensitivity due to the noise generated by them. To reduce that to a minimum the total resistance of the parallel resistance has been adjusted downward from the original value and other resistors that effect damping of the coil adjusted to compensate for that change.
I have ordered some 950 volt MOSFETS to experiment with which should be safe to use since the peak decay voltage is clamped if it goes about 650 volts. If any one has recommendation on suggested components for these switches I would appreciate your input. I have sufficient test equipment left over to do the required two tone testing but no longer have a way to do spectrum analysis.Comment
-
Hello
I am interested in your household appliances project, the question is whether it is possible to reject small steel (Fe), how it reacts to small silver, is there any selection, to what extent does it react to soil, I live in the Netherlands, I want to build a detector for small silver coins, I have a GMAXX 2 detector, I want to build something stronger, best regardsComment
-
hunter63,Originally posted by hunter63 View Post
I designed the AGD detector detect small metal objects and it is not possible not to detect small steel. When I do my sensitivity checks, besides using small gold I also use a small steel wire that measures 0.2000 inches long and 0.0320 in diameter. It weighs in at 0.0250 gram and comes in handy at times for making comparisons to other detectors. The AGD detector has a front end gain control, a detect sensitivity control, and variable width sampling adjustment controls. All those can be adjusted to suit conditions and affect sensitivity to small objects. The two channels of the detector can be mixed in a additive or subtracted mode. It also adjust to surroundings condition automatically by establishing its detect zero reference point. I have not had any ground problems with my use here in Alaska and runs very quietly.
Comment
-
I have been working a a miner improvement when switching between a MONO coil and a DD coil. In my AGD24.2 version when switching to a DD coil my TX decay waveform corrector is disabled and a strictly resistive additional load is added to damped the TX coil and that resistive load was chosen to a very slight under damping. The problem with that method is that the DD coil TX decay time ends up being about 1us longer in time than with a MONO coil. To eliminate that issue I want to make the decay waveform corrector active when a DD coil is used and also replace the added resistive load with a simulated load that mimics the load that the normal RX first gain stage provides, which also includes the clamping diodes and input capacitance of the FET first gain stage and PCB trace capacitance. This change will decrease TX coil decay time to decrease by a bout 1us and thus increase DD coil sensitivity for small pieces of metal whose RX signal is weak and decay rapidly.
The waveform corrector circuit is basically a negative feedback circuit whose output is of the correct wave shape to null out the usual TX kickback spike when the RX clamping diodes stop conducting and can decrease decay time by the base width for the spike that is present which may be many mV in amplitude. The TX coil damping load that this provides is adjusted by a trimmer pot and can be thought off as being a time related adjustment and I have used this for awhile now and has proven to work well. It does also apply some negative feedback to the TX coil of the desired signal and thus causes it to be reduced it very slightly, a insignificant amount. A secondary benifit of the circuit is that it can reduce the potential of RX gain stage saturation due to the elimination or reduction of the amplitude of the spike it is removing.
Below is a picture of version AGD24.3.A main circuit board today that has the change. I should receive them in about two weeks and hope to have one populated and tested by the end of the month. I anticipate that resistor value changes will be required for TX coil loading since real life and simulations don't always agree. The adjustment trimmer pot is labeled DTA for Decay Time Adjust.
Comment
-
Hello
After watching the film, I decided to build your household appliance. Currently, I am at the final stage of the PCB design. I am a bit worried, however, that it reacts to small Fe. Can you give me some advice? In the Netherlands, I use GMAXX-2, the range for small coins ends at 30 cm. I would like to build a detector that would have a longer range. Maybe you have some documentation in your collection.Comment
-
hunter63, I have not posted schematics or parts list for AGD24.2 or AGD24.3 versions so I assume that you are using old diagrams which do not have the latest updates. You have to remember that the AGD detector was designed to detect small metal object, and while searching for gold it is usually "dig all" to see exactly what it is responding to. In order to detect those small objects it requires very fast decay time of the TX coil decay waveform and have reasonable ability to climb up the waveform without have a gain stage overload due to the higher voltage level that occurs there.Originally posted by hunter63 View Post
If we look at the Generic pulse induction signal available at lammertbies.nl we will see a voltage chart with typical clamping diodes with a 0.7 volt fully on rating as show in a copy of that picture below.
My AGD detector is usually adjusted to operate at roughly the 3us level or the upper end of the blue line on this chart and would never be able to detect the small items I want to detect because the clamping diodes have already clipped the RX signal. The AGD detector uses MMBD452LT1G clamping diodes which have lower clamping voltage than shown here. You can also see that as the curves go up the ability to detect phase changes decreases. Phase changes can also be considered to be time related amplitude changes and amplitude changes is what the AGD detector looks for. The AGD detector has two channels which can be mixed together in various ways and each has its own adjustable time span. This kind of equals the blue line for channel 1, the fast channel, and the green line for channel 2 the slow channel. I usually do not use the slow channel for anything but it does offer a way of adjusting sensitivity, but may not do what you wish to do. I think a lot of that will depend how trashy your search locations are.
If we look at the curves towards the top of the chart then at some point further up they would meet and make any kind of discrimination impossible. The choice is either to use a detector that is designed for detecting larger objects such as coins, or be potentially bothered by lot by trash using a detector such as the AGD which was designed primary to detect very small gold flakes. Digging up a lot of pull tabs or bottle caps is not fun. So your choice of detector is best based on your particular location and how trashy it is. Coin detectors will usually operate in an area of the chart that offers the widest distance between the curves to make it simpler to apply discrimination. A detector like the AGD may need to have its sensitivity adjusted to be lower to prevent overloads which have a recovery time. If I were you I would barrow a PI detector designed for gold and see how it acts while searching for coins before jumping into duplicating my project.
Comment
-
One of the things that I have paid a lot of attention to is the decay time of the TX coil and what standard to use to measure it. I have seen many posts that give decay times in microseconds but typically no info is given as to how the measurement was made which makes it difficult to evaluate different circuit designs. I know that my own designs can detect 0.015 grams at gold and thus have sufficiently short decay times in order to be able to do that. The AGD24.3 blank circuit board picture I posted before has its receive input stage opamps along with its changed to LM7171A types which requires some adjustments to the TX coil loading. In the original design of the AGD detector FET opamps were used along with higher value series clamping resistors which all increased noise levels but still worked fine to be able to detect the 0.015 gram of my test gold. These changes caused me to look at TX coil decay time again which brought me to look at my own method of measuring which I will explain below, but I cannot compare it to other designs because I have no idea how they were measured.
My TX coil decay time is measured like the simulation of the AGD24.3 as shown below which used a pulse width of 40us from a 14 volt source and 300uh simulated coil.
There are two vertical lines on the scope picture:
The left one is light blue and is at the location of the point of maximum recoil voltage. This voltage is 443 volts and way above the picture and thus not visible.
And the right one is set at my measurement point of the red trace which is the 1mV of the decay voltage as we slide down its slope.
The blue trace is the output of the clamping circuit feeding the LM7171A opamps. It show a negative offset due to their input currents which is compensated for later on.
Thus I measure my decay time from the peak recoil voltage to the 1mV decay level. So this shows that 443 volts decays down to 1mV without any over/undershoot of 2.673us. With a actual coil the recoil voltage and decay time to 1mv will likely be different.
In the AGD detector the recoil voltage is positive. The picture show that proper compensation was done after the clamping diodes stopped conducting by the absence of a negative going spike.Comment
-
Beside working on the above I have been thinking about discrimination between different types of metal and have looked at moodz MAGPI V3 for inspiration. I made the B circuit board of the AGD24.3 ready to be replaced with one that has a MPU a while back to make it easy to add some digital processing if desired. The AGD24.3 detector A circuit board already has two demodulated channel outputs and to feed DA converters which could be sampled at different multiple times and thus offer the ability to detect positive and negative going signals for both channels and then process that data as desired via firmware. The A board also provides tx-decay-complete pulse which can be used to measure any changes in decay time and uses a AD8561ARZ a very fast 7ns comparator and some additional logic to insure only one pulse per TX cycle.
For myself I don't care much about discrimination between different types of metal but it would be nice to develop some MPU code that works reasonable well for doing that for reasonable small items. The AGD24.3 A blank circuit boards that use the LM7171AIMX/NOPB opamps should arrive tomorrow and I should be able to build one up and compare its sensitivity to the AGD24.2 version which use the FET input ADA4637-1BRZ opamps. The LM7171A is not without issues since it has much greater DC offset and does not handle unequal input resistance on its input anywhere near as good as the ADA4637-1. Thus it will need a input compensation voltage injected into its negative input to compensate for the unequal input currents so that the DC offset control integrator can operate as in the past at near zero volt DC out with the RX input shorted to Gnd. The DC offset control is a gated signal and which is derived from the fast channels sample and hold circuit and the channels demodulated output. The slow channel has its very own equal.Comment
-
If you are talking about only the coil, then you can measure its L and C and have everything you need to predict settling. Or just measure its self-resonant frequency; the higher the SRF, the faster it will settle.Originally posted by Auto-Mation-Assist View Post
But that alone is pretty meaningless except as a way to judge the quality of a coil design. What matters is how early you can sample, and for that you need to look at the output of the preamp. I consider the effective decay time to be the earliest point where the preamp is no longer in saturation. Therefore, the coil, the flyback clamp/switch, and the preamp all play strong roles in the decay time. As you've probably seen, using a higher value clamp resistor can speed things up quite a bit, as can a lower preamp gain or maybe a 2-stage preamp. Many designs are speed-limited by the clamp and preamp so that making the coil any faster will not help.
Comment
Comment