Does the VCS have an analogue sound chip?

[PFL]

I've been listening to lots of chiptune stuff recently - gotta love Army of 2600!

 

Anyway, I really enjoy the sound of the VCS. It kind of reminds me of the 'warmth' that I feel in SID tunes as opposed to the 'harsh' tones of the NES or GB and that got me thinking. I had always presumed the sound output in the VCS was digital but on further listening I'm now not so sure... Maybe it's all psychological and I like the VCS sound simply because I'm familiar with it.

[SeaGtGruff]

Anyway, I really enjoy the sound of the VCS. It kind of reminds me of the 'warmth' that I feel in SID tunes as opposed to the 'harsh' tones of the NES or GB and that got me thinking. I had always presumed the sound output in the VCS was digital but on further listening I'm now not so sure...

 

No, the 2600 does not have an analog sound chip per se. It doesn't even have a dedicated sound chip-- the sound is generated by the TIA chip, which you could perhaps call a "dedicated graphics and sound chip," although that's not quite accurate.

 

Anyway, the audio signal that's produced by the TIA is digital in nature. However, the audio signal coming out of the TIA is then fed through some circuitry which (to my untrained eye) looks like it might be a high-pass filter, low-pass filter, or both (band-pass filter). I'm no engineer, and I can just barely understand certain portions of the TIA schematics, so I may very well be wrong about that-- but when I search the web for "high pass filter circuit" and "low pass filter circuit" there are some circuit schematics that (to my untrained eye) seem to bear a close resemblance to that section of the 2600 schematics (see below).

 

A year or two ago I created a simple program that loops through all 512 of the 2600's "built-in" sounds (16 control settings times 32 frequency settings) for about an hour, playing each AUDC0-AUDF0 combination for 7 seconds. Then I ran the program on my "heavy sixer," recorded the AV to a DVD at the highest quality and sampling rate my DVD recorder offers (96000 Hz), ripped the DVD's audio to WAV files, and analyzed the WAV files in WavePad. The waves look analog rather than digital, and I've posted some screenshots in other threads. I can't be sure that any high-pass, low-pass, or band-pass filtering wasn't introduced from somewhere else-- such as by the DVD recorder-- but I suspect it was primarily introduced by that circuitry between the TIA's AU0 and AU1 pins and the frequency modulator.

 

http://www.atariage...._2600_High.html

 

Here's the specific section I'm talking about:

 

Edited September 25, 2012 by SeaGtGruff

[Blacklight]

There's an Atari 2600 music kit here. Make sure you download the manual for it. It requires you to type out a text file and then compile it into a cartridge file that could either be played with an emulator or via Harmony Cartridge with a real 2600.

http://www.qotile.net/sequencer.html

 

Also, look at "Loopcart" on the same site.

[Rybags]

Like most sound chips it has a small analog component.

 

Each voice provides 4 bits of volume control, so there's 4 resistors per voice that generate the final output.

 

External analog components also have some bearing on what you hear - on the 8-bit computer the pure tone output is a square wave measured at the pin but at the final stage the wave has some decay but it doesn't change the overall sound much.

[SeaGtGruff]

I do hope someone with engineering knowledge and schematics-reading ability will verify whether all those resistors, capacitors, inductors, and transistor act as a high-pass filter and/or low-pass filter. It makes sense to me that Atari would probably want to filter the audio frequencies.

 

Anyway, the sound generation is pretty cool, and it looks like the waveforms don't always produce the frequencies you might expect. For example, the AUDCx=3 setting gives a waveform pattern that's 465 bits long, so you'd expect its base frequency to be around 67.5 Hz (i.e., about 31400 Hz divided by 465). But if you do a FFT frequency analysis of it, the dominant frequency is actually about 1080.4 Hz. I think that's because the AUDCx=3 setting outputs the 15-bit long pattern of the "4-bit poly," but clocks each of the bits by the 31-bit long pattern of the "5-bit poly," with the result that 16 different "stretched" variations of the 4-bit poly pattern are produced with lengths ranging from 26 bits to 30 bits, the average length being 29.0625 bits (465 divided by 16), so the dominant frequency is about 1080.4 Hz (i.e., about 31400 Hz divided by 29.0625).

[PFL]

So, like the rest of the VCS architecture it's a quirky piece of kit. I wonder if the unexpected frequencies found through FFT has something to do with making it sound a little more analogue than would be expected. I find that it's the errant sounds found in analogue that helps to give the sound its body...

[Rybags]

Poly output is sort of like a bunch of pulse waves overlayed, so interpretation might find all kinds of things.

 

Simple pulses are used for PWM audio - high frequency poly sounds can introduce that effect and e.g. start sounding like a lower frequency sawtooth.

 

The other thing is that often a soundchip can produce sound way above the frequency that the speaker in use is able to replicate accurately, so the speaker itself can become a kind of filter.

Edited September 26, 2012 by Rybags

[+batari]

The purpose of the circuit there is to modulate the audio to send it over RF to the television. However, it does appear that there is a low pass filter there.

[SeaGtGruff]

The purpose of the circuit there is to modulate the audio to send it over RF to the television. However, it does appear that there is a low pass filter there.

A low-pass filter lets low frequencies through and filters out high frequencies, whereas a high-pass filter lets high frequencies through and filters out low frequencies, right? What about a high-pass filter, does any of that look like a high-pass filter? I ask because I was thinking Atari might have wanted to include a high-pass filter to help prevent any possible damage to the speakers of TV sets. That is, if the AUDCx=0 and AUDCx=11 settings put out a steady stream of 1s, isn't that like an "infinitely low" frequency signal? Would setting AUDCx to 0 and AUDVx to 15 and leaving them like that have the potential of damaging or wearing out a TV's speakers? Wasn't Atari responsible for inadvertently damaging some people's TV sets in their earliest videogames (before the VCS) because the game images got burned into the phosphors from being left on too long, and that's why they made sure games included a color-cycling routine when left on without any input from the user? That experience may have made Atari all the more aware of the need to prevent any possible damage to the speakers. Also, would a low-pass filter and/or high-pass filter help in preventing RF interference from the VCS?

 

Edit: My question about RF interference may have been dumb (I freely admit my lack of knowledge in these areas). I was thinking in terms of the different ranges of frequencies used in radio or TV or whatnot. I guess a low-pass filter would help in filtering out very high frequencies from the harmonics created by some of the poly waveforms. My main reason for wanting to know if the VCS includes high-pass, low-pass, or band-pass filtering is because I want to understand how much of the filtering I see in my audio recordings of the VCS are coming from the VCS itself versus from the method I used to record them (VCS to VCR to DVD recorder, etc.).

Edited September 28, 2012 by SeaGtGruff