Last friday I mainly worked on getting the new marbles (the ones that hadn’t undergone any heating process) to the desired magnetic field values. Remember, on this post I had already manage to have the marbles we already heated at the desired value. It needed a heating time around 5 min, so I instinctively thought that for marbles that had undergone no heating, it would need a longer time. I was wrong. And not slightly wrong, but more like completely wrong.
So I started heating at 300°C for 6 mins, it was too much and the magnetic field almost completely disappeared (barely noticeable). I naturally thought at this moment that I justed heated a tad to much, and so I adjusted to 5 min 40. Wrong. 5 min so ? as we used to do it. Also wrong.
I went on decreasing this time, going crazy the further down I went, until I reached the heating time of 3min40 (which wasn’t enough to have our marbles directly to the desired value, but at least it wouldn’t render them useless.
I don’t actually know why there is such a difference, but I have two hypothesis :
- although the marbles are labelled identical and sold with the same reference number, there’s a design/structural difference that makes them behave differently at those high temperatures (which are outside the range of use)
- the first times I use the oven, I used to close it in a different way that the intended one, but the oven security wouldn’t be triggered. I suspect that the way I used to close the oven let open a little slit for the air to engulf, and as such would make the temperature profile in the oven heterogene. For the last marbles, I closed the oven properly and that would explain the difference. However there’s two flaws to this explanation. First, I also closed the oven properly for some of the marbles that were already heated and it still needed around 5 min. Second, the very first marbles heated by Henri were also heated with the oven properly closed and for 5 min.
This day I also tried different sanding for the holes of our box with the Dremel, without any sucess alas.
The last few days I’ve been working on the SPI communication between STM and ESP. I’ve been reading quite a lot, about SPI communication in general, about STM SPI, ESP SPI and to catch up on how freeRTOS works. I fetched some example code from ESP-IDF and tried to get it work.
The code example was SPI communication between two ESP with a handshake GPIO line. It was working fine, but some message would be dropped. I spent quite some time trying to figure out why the example code would not run at the specified example speed (5 MHz). As of today I still haven’t found a way to repeat the bug, and it’s now working fine at the specified speed. I tried to use saleae logic analyser, but I didn’t identify much clues about this communication problem.
At lower frequency there was no errors, and no packet would drop.
I suspect this problem comes from the fact that the example use the GPIO matrix, which introduces a considerable delay, but I have yet to confirm it.
Another problem I have when running this example, is that the master is completely silent on the MOSI line, or outputs only empty chars. I have actually no clue regarding this behavior.
Anyway, because the PCB test won’t have SPI communication between ESP and STM it’s not an urgent matter. I’ll get back to it if the UART communication happens to be unfit.