Snowdinger 2

by G.J.Davies

In the first issue of Micropower (you did get your copy of the first issue??​–Ed), Snowdinger, a circuit to effectively stop on-screen disturbances caused by indiscriminate processor access of the VDU RAM, was published. The circuit had one major disadvantage, it was for a Nascom 1 and my own computer is a Nascom 2. Hence after reading the article, the project was forgotten. Now having just finished my exams at college, I was looking for something to usefully occupy my time with. Since mowing the lawn was completely out of the question, I decided to look at the problem of converting Snowdinger into Snowdinger 2. This article is the result of some two and a half days solid work.

For those who either didn’t get the first Micropower or have forgotten the theory behind Snowdinger, the basic approach is as follows:

Each line of the TV picture is 64 microseconds long and in every 64 microseconds, 16 are blanked (line blanking), thus if 64 characters are used for one display line, each will take up a time slot of 1 microsecond, Also 16 of the 64 will be lost due to the line blanking, leaving 48 characters to be displayed fully. Hence Nascom screens are 48 characters wide. For every character on a TV line there are 8 possible dots to be displayed, because each character cell is an 8x16 dot matrix, 16 TV lines are required to produce one row of 48 complete character cells. The VDU circuitry places an eight bit byte of data in the serialiser once per microsecond, where upon it is sent to the TV one bit at a time. When it is empty the next byte of data is loaded and the process goes on repeatedly for all 256 lines of the TV picture. If this period of 1 microsecond is split into two seperate halves, one half may be used to send the necessary data to the serialiser and the other half may be used by the Z80 to access the VDU RAM. Using hardware to synchronise the processor to these time slots of 500ns, no screen disturbance will occur.

The theory sounds simple enough, putting it into practice is however a little harder. The first thing that I noticed about Snowdingers circuit was that it required slight changes to work at 4 Mhz. If this could be avoided in Snowdinger 2, then the switches on the Nascom 2 could be used as normal. By drawing out all the clock waveforms on graph paper and looking in the Z80 technical manual at the timing for the read and write cycles, I soon discovered why Snowdinger needed circuit changes to work at 4 Mhz. Wait states are sampled on the falling edge of the 2nd clock period (T2) of both the read and the write cycle. Since there are twice as many falling edges in 4Mhz. a 500ns time slot has within it, two falling edges of the 4 Mhz clock, compared to one falling edge of the 2 Mhz clock. When no wait state is found, the following negetive edge of the clock terminates the cycle (whether read or write). In the case of the 2 Mhz clock this coincides with the access time slot, but for 4 Mhz the read or write cycle will be terminated before the access time slot occurs (see diagrams for details of the waveforms).