Applies To:
  • CitectSCADA 1.00, 1.01, 1.10, 1.11, 1.20, 2.00, 2.01

When an operator changes the display period (also known as timebase or time resolution) to display a long portion of historical (past) data, they may reach the upper limit for the display period (that they can enter). When you reach this limit, the trend period will not change. The size of the collected (raw) trend data before calculating the displayed data is :

raw data size = (display period * the number of trend samples) / trend sample period

The upper limit for the size of raw data is 64kilobytes or approximately 32000 points. When you display a trend, Citect must read all of the raw data into memory before it will start (to average it to get the final trend data). As an example, say, sampling period is 10 seconds, the number of samples to display is 500 and the display period is 60 seconds, the raw data size will be (500 * 60) / 10 = 3000 points or 6 kilobytes. If you change the display period to 20 minutes, the raw data size will increase to (500 * 60 * 20) / 10 = 60000 points or 120kilobytes, which is over the 64k limit.

The reason for this 64K upper limit is purely the speed consideration. The main criterion of the trend system design was fast performance. Citect tries to get the data from the disk on the trend Server onto the Trend page on the Citect Client in less than 1 second. In order to achieve this performance, the raw data size has been constrained to 64K. Since the maximum raw data size is 64K and the number of trend samples displayed is constant, we can only change the ratio between the display period and the sample period. If the above formula is rewritten to calculate the maximum display period it will read:

max display period = (32000 * trend sample period) / the number of trend samples

As can be seen from this formula, increasing the sample period will result in higher display period and therefore a longer time span of historical trend data. For example, if the user changes the sample period of the trend in the above example from 10 seconds to 60 seconds, the display period will increase to (32000 * 60) / 500 = 3840 seconds or 64 minutes.

Based on the above discussion, you can configure two trends with different sample periods for the same data point (or the expression) - one to examine the recent data with a fast trend, and one to examine very old data with a slow trend. 

Configure two trends for the same data point. Use the one with a fast (actual) sampling period to display the real time data, and the data in the near past. Use the slow one to examine the historical data over a long period of time.