Snowdinger

by Dougal

If you have a Nascom 1, you will have noticed that certain processor activities cause a lot of on-screen disturbance patterns of apparently random white speckling all over the TV screen. You may, of course, have modified your machine by fitting the “snow plough” suggested in the first two issues of INMC news, or you may have a Nascom 2 (or a Video Genie, an Atom, etc., – they all seem to have similar problems). In that case you will still notice on-screen disturbances in the form of short black lines chopping up the displayed information. On a mainly black screen this can just about be tolerated, but with any sort of graphics capability all that the “snow-plough” approach succeeds in doing is to substitute black “snow” for white.

This is a great pity because a clean display is so much more satisfactory, and there are, in fact, several ways in which this can be achieved. The method detailed in this article is economical, and it has been success-fully implemented by a number of Nascom owners in my local club (even O.N.J. managed it).

The disturbances are, of course, caused by the need for two functionally independent units – the display logic and the processor – to access the same memory area: the V.D.U. RAM. The display logic cannot be kept waiting for the obvious reason that that the televison line scan just keeps right on going, so in the Nascom designs (and many others) the processor grabs the address and data busses to make its access, and whatever the display logic is trying to do at that time is simply not done. Either, with an unmodified Nascom 1, it is fooled completely, and proceeds as if the data that the processor has just accessed is the data that it was itself trying to access – in which case it looks up in the character generator ROM an “appropriate” bit pattern for the line currently being displayed, and merrily puts that up on the screen -or, with a Nascom 2 or a “snowplough”, it blanks out any information from the display for a period covering the disturbance, which is fine if nothing is supposed to be displayed at that position on screen, but not otherwise.

So, what approaches can be taken to provide a final solution – a good, clean display with no disturbances whatsoever? Three solutions come to mind, all based on the principle that if the display logic must not be disturbed, then the processor will have to be.

1) A software solution. For about 4 milliseconds in every 20, nothing will show on the screen because this is the frame-blanking time. All the screen updating can be done during this period, as follows. Two “monitor” routines are needed, one entered by means of a frame-sync interrupt, the other entered by a user call. The interrupt routine is given sale and exclusive authority to manipulate the contents of the video RAM, which it does by working down and emptying a list of demands. These are prepared for it by the other routine, which makes an entry in the demand list for every call from a user program.

2) A hardware solution. Use “wait-state” logic to permit processor accesses into V.D.U. ram only during the line-blanking time, which is 16 microseconds in every 64. Of course, during frame-blanking accesses need not be delayed.