Teensy 4.1 vlf

[reetxeed]

Hi everyone! Has anyone tried building a VLF metal detector using a Teensy 4.1 microcontroller?
In my current circuit, the RX amplifier is an INA128U, the ADC is a 24-bit PCM1802, and it uses a Teensy 4.1 controller, a 240x240 pixel display, a rotary encoder, and a DRV8833-MD H-bridge module as the TX driver. Everything runs on 5V, except for the INA128U, which has a split supply via a TLE2426 virtual ground IC. Upon booting up, it plays a short reminder from the SD card to ground balance the machine.
I am currently testing two versions simultaneously: a standard one and one with decay measurement. However, I have run out of ideas on how to move forward. The code is currently around 11,000 characters long, so it's not small. It operates at 12kHz/96kHz and takes 8 samples live.

https://youtube.com/shorts/HLPLmnMbBYw?si=MzsPLQ53toEM18r7

https://youtube.com/shorts/irZPknO8K...PBssEKFq1RtVDP

[reetxeed]

deex_alap_v24_user_manual_EN.pdf

[Byko]

Thanks for sharing, but honestly, this isn't a metal detector yet. It's just a toy
Turning it into a real metal detector takes a lot more work, dedication and passion, and those have to come from you.

"The code is currently around 11,000 characters long, so it's not small."

For comparison, I'm currently writing alternative multifrequency firmware for the Quasar-ARM hardware, where the main constraint is 64 KB of flash and 8 KB of RAM. The project is about 72k lines of C code (lines, not characters). On top of that there are 47k lines of Python: validation scripts and test benches for DSP core simulation, debugging and data capture over a serial link. I won't share it until I've solved and tested salty and mineralized soil response, VDI correction by subtracting the ground vector, ferro check and much more, and until my perfectionism says it's good enough.

Today anyone can buy a detector with discrimination on AliExpress for $40+. Minelab's entry-level machines start at $250+, and the premium models go up to $2k. The market is saturated, with far more supply than demand. IMHO the chance that someone will build an unknown detector is very, very small, especially if the manual generated by Claude AI
How long did it take Claude to write those 11k characters, a weekend? Claude is a great tool, and I encourage everyone to use it

Over the years I've seen the same thing again and again: if the author doesn't finish the project, nobody will. Publishing the sources in the hope of contributions doesn't work either, because nobody contributes anything. Instead, if the project is any good, it becomes a gift to amateurs who put their own name on the splash screen. And if the firmware reaches commercial grade, it will attract our mass-production brothers, who will sell detectors at the cost of the components.

[reetxeed]

A factory machine might be much better, but there's nothing quite like being able to say that it's your own design and build—that feeling is worth everything. I might have been misunderstood earlier: this is not just a plan on paper, it is a working machine on my bench right now, though it could still fail in the field. I'm not looking to replicate an existing design; I've done that many times before. I want to build something unique that only a few people attempt. Since I don't speak English, Claude did a pretty good job describing the main points. I’ve been working on this build for about a year now.
In the end, I haven't completely run out of ideas yet. I have four AK4524 stereo ADC/DAC chips. If I drive the TX coil in push-pull configuration and read the signal differentially, I might still be able to squeeze some extra performance out of it. I've been an amateur builder for over a decade. You are absolutely right that if anyone builds a machine and publishes it online, it will get copied by people who have never even held a soldering iron before, and they will start selling it. Unfortunately, there is nothing we can do against that.

---

[reetxeed]

Elnézést az elírásért, durván 2400 sor és 110000 karakter, egy nulla lemaradt.

Sorry for the typo, it's roughly 2,400 lines and 110,000 characters, a zero was left off

[Byko]

Keep going, you're doing great.

[superspider1968]

Hi , i started a year ago thinking about pi with discrimination (long story and I will wrote it soon) - first with single mosfet and esp32p4 than teensy 4.1 and i have found if you really want repeatable results, you need truly precise control, as this will determine your ability to perform discrimination. and please don't take it wrong I’m not criticizing your choice— here’s what I’ve figured out so far...given my limited knowledge, I’m unable to achieve this using a Teensy (to make matters worse, I fried two of them). After a lot of testing (esp32 and teensy), I eventually settled on the STM32G474 (I have both the RET6 and VET6 versions for testing); I’m still at an early stage. Good luck with your project.

[reetxeed]

I haven’t had any issues with the Teensy so far, and since the ADC is external, dropping from 600 MHz to 170 MHz would be a step backward. Discrimination is perfect in the air; the VDI and the audio are different for everything. The control system can be written in a hundred different ways, and with a VLF metal detector, there's no chance of blowing up the controller—at most, the ADC will clip. The last time I built a PI metal detector was 15 years ago.