We all know that a healthy button/switch will bounce less than 20ms, right? I used to think that. I learned that 20ms number in high school and never really thought about it again. 20ms always worked... until one day it didn't. At first I thought there was something wrong with my project, the wiring, or the button. I checked everything and swapped out the button with the spare I had (same model, never used) and it also would occasionally bounce well over 20ms.
I did some basic research and came across Jack Ganssle's excellent guide to debouncing. I blame Jack for sparking my curiosity in button bounce behavior. I did more research and started thinking about designing a button debouncer that would handle my unruly buttons and also add the custom features I needed (tap counting, long press, ...).
While I really enjoyed Jack's article, it didn't have enough data to satisfy my curiosity. The few pictures in the article were great, but also very small and pixelated.
I wanted to see the good bounces, the bad bounces and the truly awful bounces! What percentage were good/bad/awful?
If I was going to design a debouncer that detected button stability, I would also need to know the longest pulse seen during testing.
I got out my digital oscilloscope and started testing the problematic button. I was using a PicoScope 2204A (10 MHz bandwidth, 100 MS/s, 8k samples). My PicoScope is one of the cheaper models (around $250 CAD) and delivers incredible value, but I found its 8k sample depth very limiting for this task. I could either capture high resolution data for a very short time or very low resolution data for a longer time. I was concerned that the low resolution captures would miss short pulses (confimed).
Using my PicoScope, I also didn't have an easy way to compare bounces or create a good summary. I could use the persistence feature, but it ended up being a bit of a mess of colors.
I thought back to Jack's article where he states: "Many of the switches exhibited quite wild and unexpected behavior. Bounces of under 100 nsec were common [...]. No reasonable micro could reliably capture these sorts of transitions, so I abandoned that plan".
I wondered if I could make use of an AVR/Arduino's edge counting peripheral. I hoped that if I could use cheap and common hardware (like an Arduino UNO) to capture button bounce data, maybe a few other people would help me build up an open source database of button bounce behavior.
Many hours later, we have a very capable tool for recording and analyzing button bounce data.
| Feature | Value |
|---|---|
| Sample period | 1.25 - 2.5 microseconds |
| Minimum detectable bounce | 62.5 nanoseconds |
| Maximum recorded/streamed data duration | 2 seconds |
See also ./faq.md#performance-accuracy-stuff
Pretty good! However, the Arduino Uno/Nano use a resonator clock which limits our absolute accuracy to around 1%.
See the bottom of this button test session for interactive graphs that compare our button recordings against an oscilloscope.
The below 12 minute video gives a brief demo, explains why it is better than oscilloscopes, and why you might want to use it. https://www.youtube.com/watch?v=jE4PtGqRFt0
