|
|
|
near-success with a surface mount component
Saturday, January 10 2026
This morning our neighbor A came over (via a path through the forest) with her daughter W and their dog Henry to drop off a bunch of hand-written thank you cards from the elementary school class Gretchen had given a poetry lecture to yesterday. This was a nice gesture, and W in particular still seemed to be delighted by how things had gone (even if Gretchen had managed to get nearly her whole class crying). Gretchen had a date to meet our friend Kate for a telecast of an opera that they would later be abandoning during intermission, so she left before the others, and I ended up chatting with A about things like the mysterious presence of douglas firs along the west side of the Farm Road, presumably planted by the Brinniers (the former owners of her property, who'd bought it back in the 1950s). Meanwhile W was all snuggled in under the big fluffy blanket with our dogs Charlotte and Neville in a way that A just had to photodocument.
After they left, I went upstairs to the laboratory to resume work on my sprawling microcontroller system. Looking through my parts containers, I found I actually had a tiny 40-pin surface-mount Atmega1284, the eight bit microcontroller with a whole 16 kilobytes of RAM. Evidently I'd bought it when I went through a period where I was obsessed with eight bit Atmel microcontrollers (that would've been about ten years ago). Working with surface-mount components isn't easy and has been something I've shied away from. But I had that microcontroller and I even had a breakout board to solder it to which would run all its surface-mount pins to headers with pins spaced a tenth of an inch apart, the macroscoping scale I have been working in since I began tinkering with integrated circuits in the early 1980s. But how was I going to attach that board to my tiny Atmega1284 without soldering those tiny pins.
The pads where the those tiny pins looked like they were pre-tinned, so maybe all I had to do was place the chip on the board perfectly aligned and heat it all with a heat gun until the solder on the pads melted. I did this, and the chip even seemed to stick initially. But thinking about it later of course I should've put down some flux as well. In any case, it didn't take much force for the chip to come flying off, meaning that technique had failed.
So then I tried using a small soldering iron to solder the pins as best I could. This actually worked okay, and in cases where several pins were bridged, I could usually flow more solder into it and then fling the excess away and then use the tip to draw away any remaining, causing the bridges to collapse and disappear. But there was one place that was a bit stubborn and I ended up overworking it so much that the conductive traces for two of the pins delaminated, meaning those pins would never be connected unless I took heroic measures. By this point a lack of food and too much kratom tea had me shaking like a patient with Parkinson's disease, but by then I had things soldered enough to add pins for testing.
After adding some pins and wiring the board to a USBTiny programmer, I was even able to flash a firmware. But that only happened once, and after that the thing became unresponsive and I had to abandon it to salvage the day.
Later I added a feature to the ESP8266 Remote code to keep remotes from sending I2C pin-state commands to a slave if the pins were not changing state. In the past I hadn't seen what dfference it made it if sent superflous state information with every poll of the server. But now that I was implementing power-saving modes in the slave code (partly so I could use more energy-demanding slaves), it made sense to minimize I2C traffic. That traffic would cause the slave to wake up from a light slumber and cause it to use several times more electricity than it would in that mode.
The microcontrollers connected together to help me develop my ESP8266 Remote system beside the keyboard of my main computer. Click to enlarge.
For linking purposes this article's URL is:
http://asecular.com/blog.php?260110 feedback
previous | next |
