Hey all, I dropped a small but not insignificant firmware update adding some simple Infinite Impulse Response (IIR) notch and low pass filters, as well as an optional DC blocker if you only want to get the pulse waveform. You can use the 'F' command to toggle the filters (default ON for obvious reasons) and the 'X' command to add the DC blocker on. They will eat 50/60Hz AC noise for breakfast and ensure a smooth curve on pulse readings on the Delobotomizer regardless of power source. For the V1 HEGduino's I'm not entirely sure how well they work so I need some feedback, I'll be testing them within a day or two I just have to build another one real quick as I don't have any laying around. With any luck these filters will more or less rescue the V1 devices from their pretty serious signal quality problems.
Here's a SUPER noisy signal with and without the filters. I'm applying 3 of each 2nd order 50Hz and 60Hz notch filters (for depending on where you live), and 3 2nd order low pass filters. You can see how pulse is obscured by the significant high frequency noise.
For reference this is what the signal should look like without filtering if there is less environment noise. I'm pretty sure the above picture is due to my laptop battery crapping out.
Next is my new favorite and by far the simplest filter, the DC blocker. This will help me make pulse measurement much more accurate.
One thing that's kind of neat is that these filters are using some cool adaptive complex numbers math so you can see the filters scrub the signal in real time. I don't really understand them yet but they work like a charm. I'm stuck on the phase locked loop part which is a similar problem, which led me to integrating these in the first place. Here's what happens when you switch on the DC filter at first, the signals quickly trend to 0. It makes it plainly obvious how your pulse is even visible just with ambient light, and this is in my very dimly lit room. You can monitor pulse just with a phone camera especially if it's IR sensitive.
Otherwise, I have a pretty gigantic improvement for the software coming. If you check out this repo: https://github.com/moothyknight/eegpwa you can see what I've been up to the past few months on the software work. This is going to be a plug and play software for the FreeEEG32 and beyond, as I created a brand new javascript format to drive the app and make it much easier to develop modular features for it that should still carry over to the microcontroller's hosting capabilities pretty well, including a bunch of itty bitty libraries for high performance data visualization. I've basically refined everything I've learned about web software development into something I'm really pretty happy with and I can continually improve on and get others involved with more readily. Our friend Dovydas in the community here showed us a ton of tricks. Here's a WIP image from it using the EEG, before implementing similar filters there as well. The spikes are eye blinks, the lower chart is a messy coherence time series. There's a lot of cool features that will apply to the HEG nicely.
Yaaaaaay learning! To do it right you gotta do it yourself, so we're gonna do it all ourselves and kick BCI into a higher gear.
Hey Joshua,
When is the Crowd Supply campaign going to get launched?