Shruti-1 internal signals

I bought a second-hand scope today to help me debug some stuff in the new project. I had some fun looking at the Shruti-1 signals. They are quite ugly indeed - some very low subharmonics ; and PWM meanies at around 30kHz. It takes some mental effort to realize that we shouldn’t really care, because there’s some stuff downstream in the processing chain that efficiently lowpass / hipass that. What struck me is that if I had a scope at the time I started working on the project, I would have never done it this way because the ugly waveforms would have driven me crazy.

i dunno, looks kind of pretty to me :slight_smile:

Come on, this is supposed to be the sine wave! (on a side note, soundcards have pretty sharp brickwall filters before their ADC, so when you record this on a computer, it indeed looks like a sinewave, because the sharp undulating ugly monster is at 31kHz).

My hears have a brickwall filter at 14kHz because I’m getting older - and it has alwyas sounded to me like a pure sine wave…

oh, dont get so scientific about it. If it sounds good, who cares. Did Roland obsess about the 303 square actually being a square (not even close)? 'round here, that’s what we call “character”

So, what does this mean? Do you think you’ll be reviewing the waves and making them more pure in future firmware releases? (I’m more of a “results-oriented” guy, so if it sounds good, it doesn’t matter what it looks like.) Maybe the skewed waves have their own idiosyncratic sound, exclusive to the Shruti-1?

EDIT: Altitude said exactly the same thing as I did, as I was typing it.

I won’t revise this because this is a hardware thing, not a software thing (the wave is pure, it’s just the way it is converted from digital to analog which is awesomely dirty). I had actually played with proper DACs and the difference (measured on a sound card) was barely noticeable. The point is that if I had seen this earlier I would have been scared and would have not gone in this direction at all (= no DAC, plain 1-bit PWM, relying on a hi-pass/low-pass filter somewhere downstream to kill the nasty harmonics). As said, on any device with a limited bandwidth (soundcard, human ears…), this doesn’t show.

Almost all the projects I have seen which are based on AVRs for generating audio are indeed using DACs - but I think I’ll stay faithful to high-speed bit banging and filtering for the moment :slight_smile:

So is this sample taken before the CEM, after the RC filter?

This is the sample taken before the CEM, after the active RC filter (which is a high-pass removing the 2.5V DC offset), and after another passive RC filter out of the circuit. Without this RC filter it looks even more ugly, since it’s really pure bit-banging…

i think it’s just cool that way!