This Flux Sucks

One time during my classes at university, the professor teaching our electronics project course said "if your circuit exhibits the behavior of a time variant system, then something is horribly wrong". This has been true for pretty much all of my experiences - a consistent circuit behavior which is independent of time. However one day, this was not the case, and it seemed I had a "time variant" circuit.

 

Needed it Yesterday.

At the time of this time variant circuit, I was working for a startup company. As you can imagine, time is of the essence. I could release a board design on a Friday and have a built prototype ready for a smoke test by the following Friday.

Digi-Key orders on a Friday morning would leverage time zones and weekend shipping. The PCB was made locally on a 2 day spin (I miss you CircuitLabs). When boards and components arrived, I'd pick-and-place the parts by hand with tweezers. Manual and expensive, but fast.

Not much was required for tools. For the bake, tweezers, paste, and a toaster oven. Rework was performed with hot air, soldering iron, and flux. For the designs I was working with, this method proved to be quite successful. But one day, I had lots of issues.

What Changed?

Perturb and observe - A classic debug method. Or flip it around - observe and find the perterb. I couldn't pin my circuit's change in behavior on a change. Maybe a bad power supply was to blame, they can cause inconsistent behavior. Surely this is it? Nope, it was rock solid. After some time, the microcontroller couldn't even be erased or flashed.

The only way I could solve this issue was to replace the QFP packaged microcontroller, but the relief was only temporary. After repeating this step a few times and not gaining more information, I was running low on extra microcontrollers. So I took a punt and reflowed the solder joints of the microcontroller.

Getting Warmer.

Okay so reflowing helped restore intended circuit behavior, but only temporarily. Wasn't thermal variance since the bad behavior began to appear well after the boards had equalized to room temperature.

"I wonder if it's moisture related" I thought to myself. I searched if flux is hygroscopic - it turns out some kinds are. This is also the first time using this pottle of flux. Sure enough, a multimeter resistance measurement showed tens of kiloohms dropped to hundreds of ohms after some time.

Moral of the story.

Avoid rosin flux. It sucks moisture from the air.