Intel MCS-80/85 manual Software

Software

The software (at end of section) for this application illustrates several features of ·the 8085A, such as the programming of the SOD line, interrupts and 8155 counter. Additionally, an example of a block search routine is illustrated.

Figure 21 is a flow diagram of the program. It has been cross referenced with program lines to the actual software for the reader'sconvenience. Following through the flow diagram it is seen that the interrupts are disabled in the beginning as the one shot is outputting a high level on its a output and interrupt pin 6.5 is high level sensitive. However, this high level will not be recognized until the level goes low and then high again. If the user would prefer a positive pulse interrupt the 8085A a dual one shot can be used with one triggering the other, or just a simple inverter. Starting and loading the counter is as described in the 8155 data sheet with the Port addresses being given in the previous Table (7). Code lines 18-23 repre- sent placing the counter in the counter mode (single terminal count pulse at the end of count) and starting the count, having the count clocked by the 8085A clock out pin. Reading the counter is not as straight forward and will be approached shortly. Code lines 28-32 are representative of programming the SOD line to output a pulse. This pin is intended for serial I/O interfaces such as a teletype, but as seen in this applica- tion, it can also be used as a single I/O port.

After the pulse is presented to the one shot, the interrupts enabled, the processor idles (lines 36, 37; Halt could have just as easily been used) until interrupted. Through the design of this application it was known that the down counter would never reach terminal count, as it is only being used as a pulse to digital count converter.

To read in the count value it is best that the counter is first stopped. The least and most significant bytes of the count length register in the 8155 are read using the same port addresses as was used during loading the counter, as seen in code lines 42-47. If one looks at this value and knows how many pulses occurred, he would come to the conclusion that there is a gross discrepancy! The reason for this is that the counter in the 8155/6 was designed to make its square wave function generation easy and when used in the counter mode, it counts by two's.For this application (where length of time is mapped into a temperature) and other similar event timing applications it is imperative to have an intelligible count re- turned from the 8155.

The counter in the 8155 is essentially a count down by 2 counter. After it counts down by 2 the initial value loaded by the user, it reloads the initial count (initial count -1 if odd) and counts down by 2 again until terminal count is reached. When reading the counter, the least significant bit of the counter does not represent the least significant bit of the count, but which half of the countdown operation you are in. If this bit equals 1, the 8155/6 counter is counting down by 2 in the first half, and if it is zero you are in the second half of the opera- tion. Because of this method of down counting there are two restrictions placed on its use:

1.The user can not use the initial value of 1 to detect only one pulse.

2.The user can not discern (through reading the counter) whether exactly one or two pulses on the timer input pin has occurred if he loaded in an initial odd count (does not apply to even). After three pulses the user can determine exactly how many pulses occurred. Note that this restric- tion only applies to reading the counter, the TO pin pulses correctly after the correct number of pulses regardless of what is read from the counter.

The first pulse to the 8155/6 counter (high level sensitive) loads the count length register, which says that the counter is not readable until a pulse occurs. If the user tries to read before a pulse is provided he will read a previous or old value. Now what is done with the value read?

Good question. An adjustment routine to convert this value read to an actual count can be summarized as follows:

1.Read in 16 bit count length register.

2.Reset the upper two bits (mode bits).

3.Reset carry and rotate right all 16 bits through carry.

4.If carry is set add 1/2 of full original count (1/2 (full count -1) if full count is odd).

In the software for this application is a general purpose rou- tine to do this; lines 179-199. To call this routine it is assumed that the lower order byte of the counter is in register C, higher order byte in register B and full original count is,in HL. Con- tents of H, L, Band C are destroyed returning actual count in BC register pair. To obtain the number of pulses that occurred, subtract this number from full original count and add 1.

Converting this remaining count to an actual temperature can be done by various methods but it was chosen to do a soft- ware map through the use of a block search routine. Table 8 presents approximations of what the remaining count should be for each temperature. To keep the software simple it was only necessary to compare the most significant byte to a list to find the appropriate temperature. This search routine is set up to find a "less than" match, incrementing the HL register as a pointer when a compare is made. The code for this search routine is in lines 118-144 and is optimized to be a fast 8 byte block search. This search routine can be made to search for a match by replacing all return on carry with return on zero. The performance of this subroutine is as follows:

Byte time = (11 + (166/8) N) CC/N = (11/N + 20.8) CC where: CC = microseconds per clock cycle

N = total number of bytes searched Byte time = time per byte searched

I.~

I~

I, i

I

I'I

Iii

!

I'

I

I

I

I

I

A1-28