Sound chips that shaped video games

Sound is present since videogames are such. From little beeps to the complex, positional digital audio system nowadays. Sound effects and music in the digital entertainment have come a long way, and so the necessary hardware has.

Among the electronic circuits that have made possible the sound aspect of the digital world there are really notable ones. Because the melodies they have been written for, the impact in the gaming (or general) culture, its capacities or simply because the leap they meant. As the dumb lists enthusiast I am, here is a list of some which, in my judgement, are the most interesting and deserve to be remembered.

The dawn of sound

First computers were huge machines devoted to military calculus and research, and were basically mute. Some could beep a little, but in 1961 a giant leap came and we could hear the IBM 7094 not just playing music, but singing too. It wasn’t until 1971 that, at last, commercial video games appeared as such with Spacewar! (Steve Rusel), which included among its circuits some noise generators to liven up the program.

Following this it should be noted that first video gaming began to be sold (staring with Magnavox Odyssey in 1972) were either mute or with a little buzzer as audio. In 1977 Atari presented its VCS, later known as the Atari 2600. Among its cuts we could fin the TIA, created by uncle Jay Miner, and later enhanced in the computers Atari 400 and 800.

Anyone who had one would remember its crude, off-key sound. It had mono output, with two oscillators with several modes, each one with 32 tone and 16 volume levels. It was a programmable sound generator (PSG) rather poor, but was one of the first recognisable ones, manufactured in huge quantities and with several versions. In the ’80s it derived in the POKEY chip, used in famous arcades like Tetris (Atari, 1988) or included in some console cartridges to expand its sound capabilities.

TIA, POKEY and many others weren’t really dedicated hardware. These chips usually had many purposes. POKEY for example worked as sampling hardware for potentiometers and other analogue signals, and TIA served also as video generator.

Texas Instruments SN

One of the first audio dedicated chips that we could hear in a plethora of console systems were the Texas Instruments SN family. Born in the early 80s, these chips had some modest capabilities: tree square wave generators and a white noise channel, at least with a more easily modulated tone and volume this time.

The list of system which used these SN is very big: Game Gear, Master System, Mega Drive, Sega SG-1000, Tandy 1000, IBM PCjr, Coleco Adam, BBC Micro, NeoGeo Pocket, a lot of arcade machines… and many others. SEGA consoles, apart from SG-1000, integrated this chip in their audio/video processors. They were used in so many places that became some of the most ubiquitous and recognisable 80’s sound. However it was easy to mix up with its direct competitor: the General Instrument AY.

General Instrument AY

This audio chip was born a couple years before SN, its sworn competitor. Its characteristics seem taken from them: three square wave oscillators and a noise channel. However, even when it wasn’t a intended feature, it was possible to play PCM digital sound through software techniques, at a high CPU cost. Its registers were bigger, too, so it was possible to tune frequencies and other characteristics in a finer way, and thus the audio output.

The main platforms that used the AY chips to play sounds were the Amstrad CPC, Intellivision, Vectrex and the Atari ST. Case of which was a deep sorrow among its user when compared to Amiga audio capabilities. 😊

Yamaha YM (OP)

The Yamaha YM family is varied and prolific, and has been used in an awful lot of computers and consoles, and some of the most common PC sound cards in the 90s. They are in general chips with frequency modulated synthesis and several PCM or ADPCM channels.

OP series was used in expansion cartridges in computers as the Commodore 64 and the MSX, but specially in PC sound cards. Starting with the AdLib and then on almost every Creative Labs Sound Blaster at the end of the 80s through mid 90s. In this case it was accompanied with one or several DAC, while retaining the particular Yamaha FM synthetic sound. In fact, a great majority of PC game of this era supported these chip to play music.

OPN family, on the contrary, was mainly used in video game consoles, although it was used in NEC Japanese computer line, known as PC-88. It was used in some keyboards too, with partial compatibility with Yamaha DX series. The main usage, however, was in video game consoles as Mega Drive and Neo Geo. In the SEGA console case with a single PCM channel, and in the SNK‘s with several ADPCM capable channels.

Other systems were backed by digital audio, but the Mega Drive, as well as SEGA 16 bit arcades, bid  for the FM synthesis in its Yamaha YM2612. This made its sound effect a lot less convincing, but in the other hand the music sounds like a proper synthesizer. In the adequate hands it could made some amazing tricks.

The rest of their chips, with various denominations (OPN, OPP, OPZ, RYP4, PCMD8, SCSP…) were used in consoles like Sega Saturn, but mostly they were heart for Yamaha famous keyboards and drum machines.

SPC700

Super Nintendo was a very capable system in its sound capabilities. This was due to a particular well designed and powerful sound subsystem, the S-SMP. This system included a DAC, a 64 KB SRAM, a 64 bytes ROM, a DSP and the SPC700 chip, brain for the audio system. It was designed by Ken Kutaragi, later known as the father of PlayStation.

Contrary to many contemporary audio chips it wasn’t a programmable sound generator, nor a FM synthesizer: it was a true wavetable engine with 8 digital channels rated at 32KHz, 16 bit. It was also a whole programmable system, executing these programs from its static memory. The SPC couldn’t read the SNES main RAM, but it was possible however to use the CPU to transfer between them, and some games took advantage of this to swap samples on the fly and overcome the 64K barrier.

In the midst of the four generation console war (link in Spanish only) it became a strong argument in favour for the SNES. With only one digital channel the Mega Drive couldn’t offer the same convincing sound effects and drums at the same time, something the Super Nintendo would excel in. At the cost, of course, of not having any audio synthesis capability, so the 64KB SRAM would frequently fall short for longer, modulated or dynamic sounds.

APU

1983 was a convulse time for the video game industry. Nintendo made a strong bid with its NES, including tasking Ricoh with a programmable audio chip at a very competitive cost, with superior capabilities compared to AY and SN. Five oscillators with triangular, sinusoidal and square waves, a noise generator and a channel in which it was possible to play PCM audio. Having no proper DMA capabilities, an intensive CPU use was needed, and the console couldn’t do much more when doing so.

In the beginning Ricoh was reluctant to comply. Nintendo asked for a very competitive price, and it was perceived as a dangerous move. But after Nintendo promising millionaire purchases they accepted. A great success, for it sold in many millions. Thus the 2A03 was born, the commercial designation for the Audio Processing Unit. (Just in case the intrepid reader has not realised, this APU doesn’t have anything to do with the Simpsons)

Even when Master System was a more powerful console, Nintendo took the prize and many people identify its particular APU sound as the 8 bit aesthetic. It is evident too that audio capabilities of the APU were superior, as it was more recent and had many desirable characteristics.

Paula

In the mid 80s it was the Amiga era. They were unrivalled  when it came to multimedia capabilities. In 1985 they competed even with professional musical machinery, because to their amazing audio features. Thanks to a multimedia oriented architecture and, of course, to a chip named Paula.

This chip was in charge of controlling the floppy disk drive, the parallel port communications and of course the sound system. Its capabilities were amazing for the era: four digital audio channels rated at 8 bit or two at 14 bit controlled by DMA. Thanks to this hardware the mod files became popular, which took advantage of all audio system characteristics.

Their principle of operation was more similar to modern sound cards than its earlier counterparts. Instead of the CPU polling and sending data in real time to the audio processor, it could just tell the Paula chip where in the RAM was the sound stored, how did we want to play it, at which volume, when to stop… and Paula would do its job on its own. I the meanwhile we would have the whole CPU for our games while digital, multi-channel music was playing, defeating anything heard before.

Other feature this chip had, even though seldom used, was using a channel as modulator for another, allowing rudimentary frequency modulated audio synthesis or using them as a filter. It wasn’t used very often as the results weren’t impressive. Something that was, however, was the software channel mixing, popularising 8 channel digital audio players. But this consumed too many CPU, so it was impractical for games and other programs.

SID

Finally, as usual, my superfavourite. The Sound Interface Device, better known as SID, is the audio chip for the best selling compute system ever: the Commodore 64. It isn’t a sophisticated audio digital player, or anything near it, just a humble programmable audio synthesizer, with mono output and three voices that sounded like hells. It’s considered one of the reasons the demoscene was born on this computer.

Each of the three voices had its own programmable oscillators, each one with a independant ADSR (attack-decay-sustain-release) curve, CPU controllable if necessary. It included external analogue parts as low-pass, notch filters or ring modulation, channel independant. This would make it a true modern audio synthetiser; everything in a 1982, mid priced computer.

This chip had also more traps than rodents for trapping. For example, modifying some of the registers at the adequate frequency could seemingly play digital sound, albeit at low resolution. Unfortunately this was “fixed” on later revisions, making this audio almost inaudible. Fortunately another trick was discovered for both versions, although it required a dedicated channel, sensibly reducing the available audio channels.

This chip doesn’t seem a big deal on paper, in fact in early years is wasn’t as exploited as deserved. Along the years, authors as Jeroen Tel or Ron Hubbard squeezed the SID potential in games, providing some of the most memorable soundtracks of the era. The demoscene did the rest, pulling a sound that would have rivalled with professional synthesizers. All this would contribute to consider a cult sounding machine, and very valued among musicians and collectors. It also helps to its collectible status that it was one of most delicate parts of the system, requiring a dedicated 12V AC (9 in the second version) and getting screwed beyond repair at a drop of a hat if the power source wouldn’t supply very regular values.

Author for this chip founded Ensoniq company, devoted to synthesising keyboard. The second was the ESQ-1, built to be what its authors wanted the SID to be back in his day.

Some other here and that

It was very interesting that PC-Engine, a modest 8 bit console, had such a great graphic and sound capabilities back in 1987. Thanks to its CPU (which served as audio chip too) it had near wavetable audio capabilities, mixing PSG capabilities and digital audio. Having its size, pricing and scope is well worth tipping our hat.

In PlayStation we could hear the might of the CD digital audio, but we mus not forget the PSU, Playstation Sound Unit. This chip allowed an incredible 24 ADPCM digital audio channels at CD quality, sequenced by MIDI streams. Well yes, CD audio was great, but you suely remember musics in Final Fantasy and the like… Nintendo and its powerful 64 bit CPU were not in position to compete. PlayStation 2 had a enhanced version of this chip with more channels, more quality, more effects and… well, more of everything.

There was a worthy contender however in the Sega Saturn‘s YMF295 chip, that would allow this and more: FM synthesis, DSP filtering and a dedicated RAM. Everything controlled by a Motorola 68EC000 dedicated CPU. Unfortunately this complex piece of technology didn’t had audio uncompressing capabilities.

Endnotes

Something odd happens in the world of audio synthesis. I love the warm sound of analogue synthesiser as the Minimoog, and the might of the modern digital synthesizers. In the 80s and 90s however there is a void of uncool pop music. Even then, one of my greatest hobbies is video game music of this era, using precisely this kind of technology.

Well, this could have helped to video game music stereotypes (classics at least) being outdated and childish melodies. Us veteran players however mostly agree that they are an essential part of the digital culture.

I didn’t want to brought the topic of sound cards (Sound Blaster, Covox Speech Thing, etc) or general purpose system dedicated to audio generation and processing (MT-32) as they didn’t were dedicated audio chips, but they deserve however an own article at length. I’ll leave it for other some time.

I tried to gather and discover some of the best examples (and those which I like the most) of chips that shaped videogame sounds at its best moment, although I must recognize that it’s strange doing so in this acronym jargon and uncomprehensible technologies. Oh and, of course, we lack a lot of chips here in lots of forms, colours and flavours, but that is left as an exercise for the reader.