A project log for Lepton 3.5 Thermal Imaging Camera

Documenting my experiments with the FLIR Lepton 3.5 thermal imaging camera.

Dan JulioDan Julio 11/30/2020 at 15:540 Comments

A socially distanced Thanksgiving holiday finally gave me some time to work on tCam.  Specifically I finished getting a new version that I’ve been working in spurts for a while called tCam-Mini debugged and running.  I also put to bed some bugs and memory leaks in the companion desktop application.

My current idea is that there will be at least two models of tCam.  The full tCam with a local LCD display and file storage and tCam-Mini designed to be a small, simple streaming thermal imaging camera.  My work with the full tCam (now ported to the Fire Cam hardware for access to its 4-bit Micro-SD interface) shows that the ESP32 probably lacks sufficient horsepower to simultaneously display, record and stream over wifi the full 8.7 frames-per-second from the Lepton.  I plan to use two ESP32 chips for the full tCam and this will be the subject for a future project log.  However one ESP32 is sufficient for a small camera that forgoes the LCD and local file storage.

All tCam cameras will work with the desktop application.  It will grow to include the ability to graph the temperature of selected points in the image over time (I intend to use this feature to monitor temperatures on circuit boards).  I’ve also started to contemplate another computer application.  This one will be a recording web server, hopefully with the ability to recognize some thermal signatures so the cameras can be left running and pictures or videos recorded when “interesting” things happen.  Access will be from any web browser.

tCam-Mini acts as an access point by default but can be put on another network using the utility.  It can return single pictures or stream images.  It is based on the code from the full tCam code-base although some of the code was optimized for operation as a streaming camera.  The code base has also been ported to the IDF 4.0.2 release.  Although I am getting the full frame rate from the Lepton in one task, I can only stream 5-6 frames-per-second over wifi from another.  I hope further debugging will improve that number.

The prototype is based around a Sparkfun Thing dev board.  Initially I hoped that I could get by without extra RAM but found I couldn’t fit the necessary buffering in the limited on-board space so I bodged a PSRAM on top of the module.

Over the holiday I also laid out a circuit board for tCam-Mini and plan to mate it with a 3D printed enclosure.  It uses a WROVER ESP32 module and forgoes the Lepton breakout board to include the Lepton circuitry directly on board.  Aside from the Lepton, single unit price for the components and PCB is about $20.  I hope to build up a unit in December.

Ultimately - after further optimization and cleanup - tCam-Mini will be open sourced.  I may try to sell them on Tindie but pricing will be tricky as I currently pay around $200 for each Lepton.  That makes the camera fairly expensive with even only a moderate mark-up.  Research on quantity pricing for the Lepton is in order…