Misc sound-info on the NES (Nintendo Entertainment System). About: This document is based on simple sample analasys, some of the info may actually be correct (but donīt count on it). It was done on my PAL-NES (board from 1987) with a "Turtle Beach Montego" sound card. Everything is sampled directly from leg 1 and 2 on the CPU in the NES (to get past the filters and to separate some of the channels from each other), all samples were made in 48kHz 16 bit. The 2 audio out from the CPU contain 5 channels in total. leg 1: Square wave 1 and 2. leg 2: Triangle wave, noise and PCM. Square Wave (from leg1): ------------------------ No information available. Triangle Wave (from leg2): -------------------------- The triangle seem to have only 16 steps for the amplitude. Reason for this is, after analysing the samples from several games I could clearly see the steps and smal noise, it was easier to see on the low frequency tones then the high ones, but with a high enough sampling-rate it should be possible to measure it there too. Another thing I took notice of was that the top and the bottom step was twice as long in time as the other steps. When counted on a low tone, the difference was 14 to 28, so i think itīs quite accurate. In ASCII it would look something like this: __ Top --> __ _ _ _ _ | | _ _ _ V Silence V _ _ ________________ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Bottom _ _ | _ _ | _ _V _ __ Well you get the idea. Noise (from leg2): ------------------ No information available. PCM-channel (from leg2): ------------------------ The sample data used for the DMA method is 1-bit delta. The actual format of the data is currently unknown, as is the latency of playback. When I studied the output (DMA version) from some games it looked like this: /\/\ /\/ \/\ \/\/\ It looks very analogue compared to the triangle wave or the noise (not to talk about the square waves ;).