System Throughput

Tech Note

"Hello, that looks like an interesting data acquisition system! How fast is it and how much does it cost?" Apparently, simple questions that unfortunately, require complex answers! Yes, there are no simple answers to how fast and how much!

Applications become immediately complex when we combine analog and digital (discrete) channels in any specific quantity not given on the manufacturers spec sheet! Moreover, we usually wish to service some channels more often than others. In addition, we would like to poll (or interrogate) the channels over a network!

So, number of readings (or writings for outputs) per channel per second (system throughput) depends on how many channels, analog and discrete, how the channels are grouped for more efficient reads and writes, and on what type of network – ethernet, serial, or proprietary. Equally perplexing is the question of cost. Of course, channel count effects the cost of the system as does the type or complexity of the desired network connection.

Most suppliers of industrial data acquisition products are sympathetic to questions of speed and cost. They normally provide you with a table that benchmarks some baseline performance guarantees. For example, Dataforth Corporation provides a System Throughput table in its Software Manual: Dataforth’s isoLynx Software Manual Page 21 System Throughput Table (34kb PDF file). I've marked the table to show various analog readings per second for different channel counts.

For example, a given analog input on one channel requires 2.08 msec of communication and transfer time and 0.25msec for the execution of the analog read command. Adding these two times together gives us a total time of 2.33 msec to acquire data from one channel. If you were to read this channel over & over again you could read it at a rate of 429 readings per second. You calculate this rate by dividing 1 by 0.00233 seconds. Moreover, this data is for a serial port set up to communicate at 115.2Kbaud. If you’re serial port is set to 9600 baud then your readings per second would be about 12 times less! Quite different!

This example benchmarks hardware timing for one manufacturer in one specific situation. These numbers could change when your hardware is combined in a real-world application. The software you use will affect overall throughput depending on how it’s configured. Say you need 10 channels with some monitoring at only 10 readings per channel that your software is comparing to some software set-point(s) and that at some trip-point you will output a discrete solid-state relay action! In this example you can see that software decision loops or case selections will affect the entire system throughput.

With such real-world applications, you need to rely on manufacturers’ product manuals and on the manufacturers Applications Support Engineers! Dataforth’s experience recommends that you describe EACH channels requirement such as type, timing and ultimate use for the information prior to selecting the hardware. Above all, try to keep the system simple! Dataforth’s isoLynx(tm) data acquisition system is a very simple analog and digital "front end" system. It is a low-cost, modular, reliable, and rugged for use in industrial environments.

Details on Dataforth's Industrial Data Acquisition System

Remember, our Application Engineers can assist you with signal conditioner selection over the phone or via fax and email. Call us at our manufacturing facility in Tucson at 520-741-1404 (fax 520-741-0762) or Email us at support@dataforth.com.

Was this content helpful?
Thank you for your feedback!
Menu
Top