80-Bus News

This article explains how I managed to alter my CBIOS so that I was able to display the Date and Time on screen, using a Gemini GM816 I/O card which contains a REAL TIME CLOCK. Readers may already have an R.T.C. available on either the Gemini GM822 or GM816 card, or some other commercial or homebrew card. Many readers will also either have a ‘source’ listing of their CBIOS, or be using one of the ‘SYS’ CBIOS’s, in which case, the method described can be used with appropriate modification. If no listing is available, I have given at the end of the article some suggestions that may help, although I have not tried them.

The idea of having a real time clock on my computer has appealed to me for some time. I have used a couple at work. One was a simple scheme, using a 58174A hung onto the PIO of a Nascom 2 that is running all the time carrying out a control job. The other one, in a Gemini GALAXY 2, used the card produced by E.V. Computing that attaches to the Gemini GM812 IVC card. The latter solution is particularly nice as it does not interfere with the main CPU at all, and the time updates irrespective of any task that the main CPU is doing. The card offers several other features too, but the cost is of course higher.

Having an RTC in the machine is of course all very well, but it is obviously an advantage to be able to see the date and time displayed on the screen at all times, and also to be able to ‘call’ up the clock from user programs. A number of example programs have been printed in this magazine from time to time, describing how to build and/or program such clocks, but little has appeared on how to incorporate one into CP/M (versions 1.4 and 2.2 – version 3 supports a clock).

After some thought and a bit of trial and error, I succeeded in doing this with reasonable success. In this article I will describe how I did this with CP/M 2.2, using a Gemini GM816 I/O card and SYS version 16. Much of the methods used will be applicable to other software and hardware.

When considering the use of a Clock, there are several questions that must be answered before it can be decided whether a particular method is satisfactory. The first consideration is one of accuracy. How accurate must the clock be, and how accurate must the display be? The two are not necessarily the same. The Clock hardware used might be highly accurate, with errors of less than a few seconds in a year, but if the clock display is only updated every hour, then that is the limit of accuracy that can be obtained from the display. (A program might still read the clock much more accurately.) For most purposes, it would probably be acceptable for the clock to be accurate to better than a few seconds a day, with a display that updates at about once every second. This was my target, and it was easily attained, but with one minor restriction. There may be occasions when the Screen clock display will ‘freeze’ for a minute or two, although it will then be corrected at the next update.

This restriction can best be explained by investigating how a clock might be incorporated into CP/M. Most clocks are capable of causing an Interrupt, and the clocks mentioned are no exception. There is however a distinct problem in using an Interrupt. Firstly, none of the simple clock chip systems generate a vector (although one used via a PIO can use the PIO vector). Use of most of the RESTART vectors is also not available for the same reason, unless