The AY-3-8910 Programmable Sound Generator Meets The NAS-BUS

Andy Tipler
Huntingdon

I am sure many readers are aware of the availability of the General Instrument’s AY-3-8910 programmable sound generator (”super-easy to interface to the S100 and other buses…”) from several suppliers currently advertising in the various computer magazines.

In my innocence, I bought one of these devices and, although it’s not too difficult, I found that interfacing is not as straight-forward as it first appears. I thought that my experiences and subsequent design of a working NAS-BUS compatible board (that’s the correct terminology isn’t it?) may be of interest to other Nascom 2 users.

In the first place any would-be user intending to interface and use the PSG chip with a Nascom 2 would be well advised to study the Nascom documentation (again) and to refer to several publications on using the PSG (see list of references at end of this article). In this respect I will not give detailed descriptions of the programming techniques involved, but rather concentrate on the hardware considerations specific to interfacing to the Nascom ,

What does the AY-3-8919 PSG chip do?

The PSG chip is a complex user-programmable digital to analogue converter specifically designed for audio sound output. The logic is TTL compatible and is thus able to be linked to the NAS-BUS. The chip contains 16 user-accessible resisters, each containing 4 to 8 bits depending upon its application. Two of these registers are dedicated to supporting the two 8-bit parallel I/O ports provided – useful for Joysticks, A to D converters etc (handshakes and interrupts are not catered fer). The remaining registers dictate the type of sound generated. Three analogue outputs are provided, these can adjusted independently for frequency and amplitude. A noise generator with variable pitch can be mixed with any of the above outputs. There is also an envelope generator, the profile of which can be programmed for frequency and shape. One of the apparent strong selling points of this device is that it will operate independently of the host CPU. This is certainly true for continuous sounds (boring) or sounds controlled by the envelope generator, but almost any “interesting” sound will require a lot of attention by the CPU. One way of improving this situation might be to use the Z80 CTC chip to interrupt drive the PSG at predefined time intervals leaving the CPU to get on with its business most of the time. However despite this, the AY-3-8910 does produce some amazing sounds and is very easy to operate.

Options available for interfacing the PSG to a Nascom 2:

a) Configure it as part of the memory. The articles in PCW and PE use this approach to link the PSG to a rival (?) computer kit, using PEEK and POKE to address it. While this is fine if you have odd memory locations scattered around your memory map, it does no justice to a system such as the Nascom 2 where every bit of memory may be put to more efficient use.

b) Attach it to the PIO port. This is very easy to do, however if you adopt this approach you will soon find that it does give a large software penalty over alternative methods: the PSG is controlled by pulses, the timings of which may be critical, so the PIO (which gives a latched output) must be repeatably set and reset, applying delays if necessary, in a specific order to control the PSG. This effectively more than doubles the the amount of software required to drive it, making life rather tedious.