Arduino Nano PI Main Discussion

[useus]

Found it -> https://sourceforge.net/projects/met...sed-on-rp2040/

It appears to be based on the Spirit PI.
Minimum sample delay is 20us, and TX pulse rate can be varied between 20pps and 160pps.
There is no Earth Field Elimination (EFE).
Interrupts are not used, as the timings are all generated by software delays.
Detection of a one Euro coin is claimed to be 20cm.

There may be a problem trying to implement the timings using low-level hardware timers and interrupts, as the Pico only has one 64-bit timer. Looking at the Pico SDK documentation, it seems that the timer peripheral is incremented every micro-second which could also be a problem, although there are 4 alarms tied to the lower 32-bits. Maybe this could be worked around using some assembler code, but (as I said previously) the Pico is much more complicated to use than the Arduino, and may not add any extra value in this application. I suspect the Arduino Nano will still generate more interest in the Geotech community due to its ease of use.

I understand the simplicity factor but, as I said before, working on it could become a very nice and advanced metal detector and very low cost being diy


however I would like to congratulate you on your project, soon I will start designing the pcb with DIP atmega328p.

[lc516]

Found it -> https://sourceforge.net/projects/met...sed-on-rp2040/

It appears to be based on the Spirit PI.
Minimum sample delay is 20us, and TX pulse rate can be varied between 20pps and 160pps.
There is no Earth Field Elimination (EFE).
Interrupts are not used, as the timings are all generated by software delays.
Detection of a one Euro coin is claimed to be 20cm.

There may be a problem trying to implement the timings using low-level hardware timers and interrupts, as the Pico only has one 64-bit timer. Looking at the Pico SDK documentation, it seems that the timer peripheral is incremented every micro-second which could also be a problem, although there are 4 alarms tied to the lower 32-bits. Maybe this could be worked around using some assembler code, but (as I said previously) the Pico is much more complicated to use than the Arduino, and may not add any extra value in this application. I suspect the Arduino Nano will still generate more interest in the Geotech community due to its ease of use.

PICO is not fit PI project and more power hungry. Port performance is not as good as stm32F103 series.

[Qiaozhi]

In a handheld detector the alternative is to make sure the mains frequency aliases outside the target filter bandwidth. This is usually set by the demod and the Nano has a BW = 1.5Hz (which is pretty low). Assuming 50Hz mains, if you set the detector to 1200Hz then the alias is synchronous and becomes a DC offset. But if the mains drifts by 0.1% then the alias is at 1201.2Hz and you hear a 1.2Hz beat frequency. However, if you set the detector to 1225 Hz then the 50Hz alias is at 25Hz which is suppressed by the target filter BW. And any minor mains variation remains suppressed. BTW, the demod is often followed by additional filtering that roughly matches the demod BW to offer better mains suppression. In the Nano, this would be R26-C15... setting them to 1.5Hz will help (they are currently 16Hz).

Interesting ...
The values used by the Nano are the same as those for Surf-PI. I hadn't noticed the difference in cut-off frequency between the two opamp stages, so I'll try changing C15 to 1uF, and setting the detector to 1225pps. Then I'll report back here with the results.

To continue this subject ...

I spent some considerable time today investigating this idea, and here are my findings:

First I tested the detector outside near a power line using the original setup, and there was some crackling noise evident in the audio as I got closer to the power lines. Then I modified the sketch to change the TX pulse rate to 1225pps by setting TXperiod to 816.3125E-6, and tested outside a second time, but there was still no detectable difference between 1000pps and 1225pps.

Next, I modified the sketch further by hijacking the Boost switch to allow me to switch between the two TX pulse rates. Again I went outside to do the testing, but still could not detect any reduction in external noise.

Finally I replaced C15 with a 1uF capacitor to set the cut-off frequency close to 1.5Hz. The main difference I noticed with this change was the increased stability of the audio threshold. Unfortunately this also had the side-effect of crippling the recovery speed. By waving a coin near the coil and then moving it away at the same time, the signal would completely disappear after being moved only a couple of inches. Continuing to then wave the coin at the new position would allow the audio sound to recover. Testing outside gave a very smeared out sound when passing over buried targets, and not the usual crisp beep with C15 set to 100nF. Also, there was no apparent reduction in power line noise as far as I could tell.

I have now reverted the detector back to the original hardware and software settings.

Question: Is the frequency adjustment control we see on some pulse induction metal detectors really intended more for eliminating general external RFI, rather than specifically noise generated by power lines?

[green]

To continue this subject ...

I spent some considerable time today investigating this idea, and here are my findings:

First I tested the detector outside near a power line using the original setup, and there was some crackling noise evident in the audio as I got closer to the power lines. Then I modified the sketch to change the TX pulse rate to 1225pps by setting TXperiod to 816.3125E-6, and tested outside a second time, but there was still no detectable difference between 1000pps and 1225pps.

Next, I modified the sketch further by hijacking the Boost switch to allow me to switch between the two TX pulse rates. Again I went outside to do the testing, but still could not detect any reduction in external noise.

Finally I replaced C15 with a 1uF capacitor to set the cut-off frequency close to 1.5Hz. The main difference I noticed with this change was the increased stability of the audio threshold. Unfortunately this also had the side-effect of crippling the recovery speed. By waving a coin near the coil and then moving it away at the same time, the signal would completely disappear after being moved only a couple of inches. Continuing to then wave the coin at the new position would allow the audio sound to recover. Testing outside gave a very smeared out sound when passing over buried targets, and not the usual crisp beep with C15 set to 100nF. Also, there was no apparent reduction in power line noise as far as I could tell.

I have now reverted the detector back to the original hardware and software settings.

Question: Is the frequency adjustment control we see on some pulse induction metal detectors really intended more for eliminating general external RFI, rather than specifically noise generated by power lines?

Just guessing. Adjustment control is for power line harmonics or maybe some other continuous frequency. Don't know if crackling noise is aliasing.

Some scope pictures of differentiator out, about 1500 pps(25 harmonic). Amplitude not the same at each harmonic, maybe do to EF sample not centered?

12V PS orientation makes a difference with amplitude

[green]

Just guessing. Adjustment control is for power line harmonics or maybe some other continuous frequency. Don't know if crackling noise is aliasing.

Some scope pictures of differentiator out, about 1500 pps(25 harmonic). Amplitude not the same at each harmonic, maybe do to EF sample not centered?

12V PS orientation makes a difference with amplitude

Tried some other harmonics. Odd seemed to be higher amplitude. Is there some reason odd is higher or just what it happened to be today?

[6666]

To continue this subject ...

I spent some considerable time today investigating this idea, and here are my findings:

First I tested the detector outside near a power line using the original setup, and there was some crackling noise evident in the audio as I got closer to the power lines. Then I modified the sketch to change the TX pulse rate to 1225pps by setting TXperiod to 816.3125E-6, and tested outside a second time, but there was still no detectable difference between 1000pps and 1225pps.

Next, I modified the sketch further by hijacking the Boost switch to allow me to switch between the two TX pulse rates. Again I went outside to do the testing, but still could not detect any reduction in external noise.

Finally I replaced C15 with a 1uF capacitor to set the cut-off frequency close to 1.5Hz. The main difference I noticed with this change was the increased stability of the audio threshold. Unfortunately this also had the side-effect of crippling the recovery speed. By waving a coin near the coil and then moving it away at the same time, the signal would completely disappear after being moved only a couple of inches. Continuing to then wave the coin at the new position would allow the audio sound to recover. Testing outside gave a very smeared out sound when passing over buried targets, and not the usual crisp beep with C15 set to 100nF. Also, there was no apparent reduction in power line noise as far as I could tell.

I have now reverted the detector back to the original hardware and software settings.

Question: Is the frequency adjustment control we see on some pulse induction metal detectors really intended more for eliminating general external RFI, rather than specifically noise generated by power lines?

Good experiment George, thanks for feedback.

[waltr]

Tried some other harmonics. Odd seemed to be higher amplitude. Is there some reason odd is higher or just what it happened to be today?

Short answer: Yes, typically the Odd Harmonics are higher due to half-wave symmetry in power systems.
The long answer is very long and requires much study and math.

[lc516]

Have you programmed the Nano?
You must do that before you can measure -5V at TP3, as the Nano generates the external clock for the 7660.

Hi,Qiaozhi

I have a question, Why does the 7660 not use internal clock?

[surfdetector]

Hi,Qiaozhi

I have a question, Why does the 7660 not use internal clock?

You want to synchronize the transmitter to the converter.
This is to avoid the switching transients that might interfere with the receive signal if you were to let the 7660 run on its own. This is covered on page 10 of the book.

[lc516]

You want to synchronize the transmitter to the converter.
This is to avoid the switching transients that might interfere with the receive signal if you were to let the 7660 run on its own. This is covered on page 10 of the book.

Thanks for your reply, That's same as I guess.
I have ordered this book,maybe I can recive it in two weeks. It need to travel half of the earth.

[lc516]

yes,I think I know what's wrong after excluding other possible problems.
I used a unmatch "MC7660", it can't runing on1k external clock.
I'll buy a new icl7660.


thanks for your information

I use a icl7662cpa replace the abnormal MC7660. But it have same problem, I can't get -5V on TP3 if use 1k external clock.
If use internal clock(disconnect R1) I can get TP6 output, just a lot of noise on there.
Do I should check C3 and C2, or elsewhere?




Any information you can provide, Thank you guys!

[dingbatca]

Two other MCU's to think about that both have Arduino IDE support.
72MHZ STM32 Blue Pill
600MHz Teensy 4.0

My goal (with the Voodoo) was to convert it to an Arduino platform. I am much better with Arduino's then the supporting electronics. I have always done everything I can to avoid hardware and put as much logic into software as I can. Just some ideas.

[lc516]

Two other MCU's to think about that both have Arduino IDE support.
72MHZ STM32 Blue Pill
600MHz Teensy 4.0

My goal (with the Voodoo) was to convert it to an Arduino platform. I am much better with Arduino's then the supporting electronics. I have always done everything I can to avoid hardware and put as much logic into software as I can. Just some ideas.

Maybe you don't need a better MCU on this project, unless If you want to get OPA output through ADC ?
But, the new idea are always good!

[Qiaozhi]

Two other MCU's to think about that both have Arduino IDE support.
72MHZ STM32 Blue Pill
600MHz Teensy 4.0

My goal (with the Voodoo) was to convert it to an Arduino platform. I am much better with Arduino's then the supporting electronics. I have always done everything I can to avoid hardware and put as much logic into software as I can. Just some ideas.

I'm currently working on an Arduino Nano-based adapter for the existing Voodoo PCB. The PI channel is working, and I'm just about to implement the DISC channel.
I will post details in the Voodoo thread when completed (Diptrace files, Gerbers, and Arduino sketch).

[Qiaozhi]

I use a icl7662cpa replace the abnormal MC7660. But it have same problem, I can't get -5V on TP3 if use 1k external clock.
If use internal clock(disconnect R1) I can get TP6 output, just a lot of noise on there.
Do I should check C3 and C2, or elsewhere?


[ATTACH]54938[/ATTACH]

Any information you can provide, Thank you guys!

I think the problem is that you're using an ICL7662 instead of an ICL7660.
Although looking at the ICL7662 datasheet, I cannot see anything which explicitly states that it won't run at 1kHz.