Tweet
Originally posted by ivconic
View Post
-
I will look at the JLPcb web site my self to see what they offer. I know the amount of time it takes to lay out a complex PCB and still think you would be better of not doing that when something reasonable may be able to be supplied. I anticipate just collecting all the required parts can also take a considerable amount of time and undertaking this project takes a lot of dedication to that work. I fully anticipate that not many actually will choose to do so. -
I will need to update AGD24.4.A schematics and parts list due to renumbering of parts on the version 4 receive circuit boards. References in this note are for the to be updated schematic and parts list.
How to set the AGD24.4 decay end and TS-0 primary receive channel delay timing.
If you have test equipment available the delay timing becomes a lot easier but it can also be done with a number of small sub gram pieces of gold or other metal and air testing. The way to start if no scope is available is to pre-adjust the two delay trim pots to the values that will be specified in the future as notes on the TX board and RX board diagrams. But my advice however is to find someone who has a 2 channel scope with external trigger ability good to at least 50Mhz.
It is also entirely possible to make an assembly error and that test equipment may be required to find such the problem. Many surface mount parts have no markings and improper placement during board population may be hard to find after assembly is complete.
Adjustments are not hard.
1: To start select your most used coil favorite coil.
For best performance the coil of choice should always be used to set up the electronics timing parameters. I primarily use the CoilTek Elite mono 9 inch ones.
Info on TX pulse width:
I adjust the TX pulse to be 36us wide for all of my CoilTek Elite mono coils.
Going much higher than 38us at my supplied battery voltage makes the recoil voltage is greater than 600 volts and causes the body diode in the TX Mosfet to start conducting which is something to avoid since it will increase decay time. Repetition rate is set to just under 4000 times per second.
Timing adjustments: (Your coil may not be happy with 36us pulses).
2: Set the TX pulse width to 36us using the front control panel "PULSE WIDTH" potentiometer.
If a scope is not available set the pot resistance to the value specified by the note in the TX board diagram. Mark the front panel with a reference mark with a thin tipped permanent marker in case it gets changed by accident. Typically this is just above the 9 o'clock position on the knob.
3: Set "TS0 WIDTH" max counter clockwise position (2us pulse width) for the primary receive channel using the front control panel potentiometer.
4: Set "TS1 WIDTH" max counter clockwise position (~10us pulse width) for the secondary receive channel using the front control panel potentiometer.
5: Now set up the scope trigger input to trigger on the negative going edge to the TX pulse, TP2 on the TX board, or TP4 pin 6 on the RX board. Set the scope to 1us per division. Adjust as required.
6: Channel 1 scope input with 10:1 probe to: Decay end pulse, TP5 of the TX board. Adjust scope as required to display the pulse and adjust the pulse width to 3.43us using the DTD (Decay Time Delay) trim pot on the TX board.
7: Once the above is complete change to scope channel 1 and 2 inputs as below.
8: Channel 1 scope input with 10:1 probe to RX board TP2 pin 1. Use 0.1V per division. This will monitor time slot TS-0 Pulse and its Width and location as referenced to the coil decay signal. TS-0 Pulse width need to be set to minimum on the front panel control if not already set. Also leave TX pulse width at 36us.
9: Channel 2 scope input with 10:1 probe to RX board TP9. Use 0.05V per division. This will monitor the receive signal going into the sample and hold circuit for the primary channel and the TS-0 pulse will bracket the area of the receive signal that will be sent to the sample and hold circuit for the primary channel. It will also automatically adjust the secondary receive channel. The bracketed area will be moved around by the next adjustment.
10: Adjust TS-0 delay time by using T0D trim pot on the RX board to have a width of ~3.93us from the left start of the scope trace. This equals ~500ns and is the delay to time align the receive signal after leaving all front end circuitry and entering the sample and hold circuit. Your setup may require a slightly different of delay due to differences in the decay wave form presented. I have posted some pictures of this in the past.
Audio tone frequency and Indicator driver adjustments:
11: The idle audio tone frequency needs to set after this with the TOS (Tone Off Set) trim pot on the receive board. I set mine to 400 Hz with FEG (Front End Gain) gain and DTS pot (De Tech Sensitivity) set to minimum. These are on the front control panel. The VCO drive voltage should be about +1.0 volt at that 400 Hz.
12: The only remaining adjustment is for adjusting the zero point for the indicator driver. It is designed to drive a 50ua meter and connects to the meter movement it via a isolated miniature TRS jack on the rear panel. Make this adjustment using the IDO (In Dicator Offset) trim pot on the receive board. If no indicator is available adjust IDO for zero volts at pin 3 of connector IND.
I plan to place an order for the updated circuit boards tomorrow from my established source. This adds the parts required for the delay function, and consolidates some test points. Some parts did need to be moved around to make room for this which in combination with the added parts required renumbering. Timing jumper choices on page one of the schematic present in older versions were removed. The consolidated and removed parts will save on future assembly time.Comment
-
-
I have been working with "https://jlcpcb.com/ " for a year and I am very satisfied! Price + quality and speed! Check prices without registering on their site! Well of course they want Gerber files so they are not responsible for the end result and they are right!Comment
-
I did look at the site but decided to stay with my present board supplier. I'm very happy with the results and feel that the price I pay for each board is reasonable for the high quality of the circuit boards I receive from them. I'm not interested in changing board suppliers and intend to stay with Aisler, "https://aisler.net/".Originally posted by MLR View PostComment
-
This is just for info on my MONO/DD switch location issue that I ran into.
I was testing an AGD23.3 version detector modified to include the delay for channel one time alignment in my driveway which is mostly crushed mining tailings and has lots of small pieces of metal that can provide usable signals for testing. Most of the hard pack snow has melted and the ground has thawed so it makes digging possible if desired.
The detector I was testing was working extremely well until I put the detector down to go get some digging tools. I came back, picked put the detector and noticed right away that it lost a lot of sensitivity but did immediately know the reason. The reason for this that the MONO/DD coil switch on the back of the detector had toggled to the DD position which completely changes the coil decay slope seen by the receive board.
The layout for my rear panel is based on where the best place is for the various connectors and where the internal ferrite snap on core is be located and secured. All this has to be done in relation to where the connectors on the printed circuit boards are located. As it turns out the best location for the coil connector for my housing is on the top right of the rear panel with the MONO/DD switch under it. The receive board the coil connector towards on the top left and behind the rear panel.
When looking at the rear panel the circuit boards are mounted towards the left side of the housing while the right side holds the 4-inch speaker. Two of my detectors have an internal speaker, one is headphone only.
Conclusion:
Placing the MONO/DD switch under the coil connector leaves it a lot more exposed to being bumped into the wrong position and thus it needs some kind of guard around it. The switches that I use extend out from the rear panel about 0.6 inches and total toggle travel is about 0.42 inches total at the 0.6-inch point and do not need a lot of force to switch. I will consider changing to a switch that has a shorter toggle lever but right now I plan to machine a guard that will replace the switches nut.
I had to move the indicator connector to the right of the coil connector before. I added the DD coil switch about four months ago.
See the internal side of the rear panel in me next post.Comment
-
I have been testing AGD23.4 receive circuit board along with a AGD23.3 TX board. D2 on the TX board was changed to changed to NXPSC10650D6J Silicon Carbide Schottky diode. The silicon carbide type improves decay time slightly. The AGD23.4.A receive board has the delay adjustment for channel 0 time alignment I mentioned before. The blue trimmer pot towards the center of the receive board. It also shows the shield over the first gain stage of the receive board.
Total current consumption and supply voltage.
Comment
-
I thought that it may be useful to expand a bit in the functions of the logic pulses that open and close the analog gates at the proper times to route the receive signals to the next processing stage.
In order for the receive board to do more than supply the TX clock and reset pulses we need to get a signal from the coil decay end circuitry on the transmit board. The decay end detect circuit has a delay adjustment that sets how high we want to climb up the decay curve. If going up to high it will overload the signal gain stages so a point has to be chosen that is reasonable for the decay times of the items to be detected and is chosen as a certain amount for time in fractions of a microsecond from the end of the TX pulse. For the AGD23.4 I have chosen to set this time 3.52us.
Here is a picture of the delay pulse and its relationship to the received signal while triggering on the TX pulse and its expanded version.
Expanded view
On the receive board there are two analog gates that need to be switched on and off at the proper times by the pulses generated in the receive board logic. There are two separate channels in the receive board, these are numbered TS-0 and TS-1 and each has its own pulse width control that can be set via the front panel controls. The pulse width of TS-0 and TS-1 are added together and form time slot TS-2.
Time slot TS-2 is used to control what part of the received signal will be processed after the first front end gain stage via the analog gate it turns on and off. The signal leaving this is amplified and filtered before being broken down into separate channels later on via analog gates controlled by the TS-0 and TS-1 pulses.
To get the signal flowing let’s look at the time relation ship with the receive signal for TS-2 by looking at the below picture. The picture shows the TS-2 pulse on top and the typical receive waveform of the AGD detector below. The TS-2 pulse width is the total as set for TS-0 and TS-1 time slots.
TS-2 picture
If we look at the below pictures for the pulses generated for TS-0 and TS-1 we can see how they line up with the receive signal and thus what parts of the signal is routed to each channel via the channels own analog gate. The output of each analog gates drives its own sample and hold circuit and routes the signal to additional stages for further processing.
TS-0 Picture
TS-1 picture
When routing signals through electronic components it does not happen instantaneously. It takes time and each component has its own time specification and thus it is important to minimize the time required to route signals. Components have specific bandwidth limitations and circuit design issues can introduce overshoot which also adds to the time required to route signals. It similar to domino blocks standing on end in a row and each knocking the next one down in precession.
This brings us to time alignment of the receive signal at the demodulation stage, which in the AGD detector are the two sample and hold circuits, one for each channel. To have the ability to do a time alignment we have to have a reference point to adjust from. In the AGD detector the decay end signal is used for this purpose. As you may recall it was set to 3.52us at the start of this post and after routing through some additional components it is routed through a adjustable delay whose outputs start the TS-0 pulse. Completion of the TS-0 pulse triggers the TS-1 pulse to start. The TS-0 pulse time position very important for maximum detect sensitivity.
If we look at the picture below of the signal at the input of the sample and hold circuits, we can see that the decay end pulse occurs ahead of the signal zero crossing at about 4us and must be moved to that point. The two very small pulses, one positive going and one negative going are suspected of being front end overload recovery due to TX pulsing and appear to minimal and on the outside of our receive window.
Decay end expanded
Let’s look at the next picture below. It is and expanded version of the above and shows the TS-0 pulse after having its delay adjusted to the correct point. The two vertical dashed lines show the amount of delay adjustment that is required, and the text above for the width between the two lines indicates it being 0.610us. It could be adjusted to a bit less but it is very close to where it should be. The analog switches are very fast but do have a bit of turn on and turn off delay (110ns on and 85ns off).
If you were to do some air testing while configured for this test you will see the trace bending at or near the point where the TS-0 pulse should cross. Once this adjustment is made it does not need to be adjusted again. It will follow whatever the setting is for the delay end adjustment on the TX board.
Adjust the delay end adjustment on the TX board slowly as an experiment to see how moving up and down the coil decay slope slightly effects the signal wave shape within the TS-0 receive window. But do watch out for front end overload while doing this.
This is all I got for now.
Comment
-
Trying to understandwhat the scope pictures represent. Target or no target decay? I have been sampling signal when I see target decay, don't see target decay on your scope pictures. Including a spice decay and a scope picture of detector decay.
Comment
-
The scope pictures just show what area of the receive wave form is being processed at various times when the scope is triggered by the TX pulse output of the gate driver of the TX board TP1 and in and idle state. The pictures are simply an aid for verifying that gating is as expected and that TS-0 pulse is occurring at the best time.
The top waveform show the opening and closing times of the three analog gates. These test signals are available on the receive circuit board as TS-0, TS-1, TS-2 which are pins 1, 2, 3 of test point TP2 on the first page of the receive board diagram. The lower trace is the from the test point TP6 on the RX board and is the input to the sample and hold circuit and just show the receive signal form at the time. This part of the wave form is hard to see with simulations since this is an area that is just a few millivolts above the decay zero crossing point. Very fast sampling would need to be done to see these very small fluctuations or shown to what looks like ringing in the pictures. This is actually a very small fluctuation and may be caused by reflection of the metals items that I have in my work shop.
In notice the below issues while briefly looking at your simulation diagrams.
On your diagram I would remove D1 and D3 they have no useful function and only add to the capacitance of the coil driver M1.
Also, on the dark picture of the decay simulation on the right you have to range set to 880 mV. You need to set this to about 10mV or less to get to the point where you can see the part of the decayed signal that is being looked at for processing.
When you start looking at sub millivolt levels here you can start see how many microseconds it takes to get to the point of interest. Looking at in entire decay slope is for the most part not very useful and what looks good may not give the results desired. Remember that simulation speed needs to be very high to actually see what the signal looks like in this sub millivolt area and that simulators may not have the speed required to do this. Simulators are very useful but real time testing with actual parts will give true results.
In my AGD detector, and considering that the front-end gain control is set at maximum you will find that a maximum input level is just a bit more than 50mV before the final gain stage starts to overload before entering the sample and hold circuits. If my quick math is correct and we want to see a signal level 60dB below the 50mv level the signal level would be 0.000005 volts. Your simulation cannot show this area of the receive waveform simulation. Your 880mV level should be closer to 1mV to start seeing decay curve voltage levels in the area that are normally amplified by the receive circuitry of any PI metal detector.
Comment
-
I have continued to test the AGD23.4.A RX board and made PDF files of its diagram and parts list.
On the diagram and parts list there are some resistors that have type markings such as MF or TF. The MF designation are metal film types and are through hole parts with only three being used. These offer lower noise, higher voltage ratings along with better temperature %/C drift ratings.
The TF types are thin film surface mount types. Thin film were chosen for specific locations to minimize noise contributions also their improved accuracy better temperature %/C drift ratings than tick film types. All surface mount resistors 1206 size and those not marked as MF or TF are thick film surface mount types.
AGD23.4.A circuit diagram and parts list are attached.Comment
-
This last summer I ran into a issue with the Lipo4 battery back that I was using that apparently had a bad internal battery manager that was causing it to stop before it was ever fully charged. I discovered this when the low dropout regulators shut things down due to low voltage conditions while on vacation. I later verified the problem after taking the 12V 10amp hour battery pack apart. The battery had four series groups with each group having three individual parallel cell's. Fully charged voltage runs about 14.4 volts but it could never reach that level. The internal battery manager would shut down charging as soon as the charging current exceeded 1.4 amps. One group of three parallel always had a voltage of less that 1/2 a volt.
The reason I mentioned this unexpected problem is that It showed that I need to add a way be able to monitor battery voltage and which also leads to some other possible additions to reduce my component count by adding a MCU to handle some functions presently handled by individual components and those that would be added. I have also been thinking a bit of using a push pull center tapped coil.
I have started to do a little bit of development for appiding the MCU processor and I'm set up to due that work with a STM32G474 and the NUCLEO-G474RE development board. I choose this processor because it has a high resolution timers (HRTIM) and can operate at 3.3 volts. It can run at 170Mhz and the HRTIM can run at multiples of that frequency. I have about two weeks experience using the STM32CubeIDE and have done a bit of testing with the high resolution and others timers available in the STM32G474.
My initial goal is to get the four required timers working.
1. TX free running timer, initial set to 36us with a repetition period of 200us and with a adjustable resolution of 0.1us and test option to route to the RX timer input.
2. RX1 timer externally triggered and default set at 50us for driving the first receive signal gate.
3. RX2 timer externally triggered and default set to 15us to drive the receive signal gate for channel 1 signal gate.
4. RX3 timer triggered by RX2 timer pulse end and default set to 35us for the channel 2 signal gate.
The following signal routing time delays have to be considered for the RX timer start times.
A. The signal transit time through the first RX gain stage to the signal gate following it. In the present circuit this time is adjusted by setting the decay end time ahead and setting the the trim pot in the decay end logic so that the desired part of the decay waveform is aligned with the firsts gate closed time.
B. The total signal transit time from the output of the first gate to channel 1 and channel 2 signal gates. This is presently adjust via trim pot on the RX board.
I can see a process where these delay adjustments can be automatically set by injecting an test pulse into the receive string and making a time difference measurements with one of the analog to digital converters in the G474. This would save on having to make manual adjustments and compensate for parts and bandwidth differences and thus optimize performance.
Comment
-
-
I was just looking at page 6 of the VCO
STM32 Blue Pill has a resolution of 12-bits. ... Voltage / Step = 3.3 / 4096 = 0.8056 mV / Step. It is giving 4096 steps and you have 1000.
Blue Pill has 4 timers which can do most of the job.
By the way, whot is sensitivity of your detector for coin and for cola can ?
Comment
-
Right now I'm just playing around playing around with the STM MCU's and learning how to use the ST development system along with figuring out how to use the various timers available in the STM32G474 MCU. I find the complexity of the high resolution timer somewhat difficult to work with but it is likely great for building very high speed complex pulses to drive coil drivers of various types. It is mostly designed for power switch mode applications and motor control. My issue with this has been starting three timers at the same time and then moving two of the pulses generated to the right individually via a count delay and trigger and thus to desired later time positions without affecting the prior generated pulse width of the two that were moved. This is not working for me yet. But It's likely just a learning curve and I will review the sample for generating a arbitrary waveform with the HRTIM timer to see if perhaps it will help out.
After learning a bit about how to use the STM32 programming tools over the last two weeks using the timers of the STMG474 its time to start working on a project that I have been getting ready for, which is adding a color display to the AGD detector.
Last summer when I was thinking about the size and weight of the analog meter which I was occasionally using to monitor the voltage going to the VCO while in the field I thought that it would be nice to do that with some sort of display instead. During that more or less the same time frame I also had the battery failure that I mentioned in a prior post but no way to see the actual battery voltage. This led to some confusion since the cause of why the detector shut down was not known and the battery had less that 3 hours use after it was supposed to be fully charged.
The addition of a display would make it possible to graph individual signal voltages from the two receive channels along with a graph of the voltage going to the VCO, which in the AGD is a mix of the two signal channel voltages and controlled by ratio adjustment pot and its associated polarity switch. The battery voltage could be displayed as plain text as well. As a result of looking at all benefits for adding a display some proposed use parts were purchased. Three BlackPill 8M boards which use the STM32F411 MCU were purchased on Aug 17th, 2023 along with some adafruit 2.2" 320x240 TFT LCD color displays. Now after having a little bit of experience on programming STM32 devices I'm ready to work on the project for the display addition. It will be a lot more beneficial then playing with the timer functions.
I believe that I may have some demo info somewhere on interfacing the display to the BlackPill and for graph generation. I will see if I can find that info but I'm always interested in finding out different ways for accomplishing the same end result. If anyone has some sample code or other information that they wish to share I would appreciate receiving it for use with the add display project to the AGD detector. I'm sure that such a display could be adapted to other detectors as well. In anticipation for this project I also purchased an Ender-3 V2 Neo 3D printer to print a suitable housing for the display.
In referance to:
"I was just looking at page 6 of the VCO
STM32 Blue Pill has a resolution of 12-bits. ... Voltage / Step = 3.3 / 4096 = 0.8056 mV / Step. It is giving 4096 steps and you have 1000.
Blue Pill has 4 timers which can do most of the job.
By the way, whot is sensitivity of your detector for coin and for cola can ?"
I'm not sure on how best to reply to the above:
The way I run the AGD detector is trigger the sample holds at a rate of 5000 times per second.
Perhaps you are saying that the Blue Pill is sufficient enough. It certainly is compact and easy to add and is why I bought the BlackPill version to play with for my proposed display project.
I don't use a coin or cola can for any type of testing. Every one is used to using their own way of testing.
Comment
Comment