Data Collection

Human Machine Interface ( HMI's )

Automated Data Collection Barcode Data Collection Wireless Data Collection

Any device or group of devices used to feedback machine data to an operator. Typically, when we talk HMI's, we're talking about an interface to a given machine or to a group of machines. This is the way in which the operator gets technical information from the machine or system and the way in which he controls the system, changing its operating parameters.

HMI's used to be push buttons, indicator lights, horns, buzzers and panel meters. Now, they are all that and also displays such as touch screens, monitors and dedicated single and multiline displays.

Choosing an HMI for a system requires some upfront thought. You have at least a couple of things to consider; 1) will the operator requirements of your system change? 2) is an more expensive HMI worth the extra cost.

Operator requirements. Is the machine you are building going to have only a few buttons and switches? Is the machine going to be reproduced and sold as a product? If there are only a few buttons and switches, i.e., this is a pretty simple machine, a sophisticated display may not be worth the extra money and time needed to make it work. If you're going to sell the machine as a product, buttons and switches are a lot easier to maintain, long distance, than a display. Everyone can change a light bulb on a panel light, but if a touch screen stops working, this is going to require a replacement from the factory, i.e., you. Is this the business you want to be in? Maybe yes, maybe no, but it does require thought.

On the other hand, if you think that there may be buttons/lights added in the future, or if you think the machines requirements will increase, then by all means a programmable display may be exactly what you need. If the machine is complex in the first place, then it might be more expedient to have a single display rather than an array of buttons, lights and switches taking up panel space and intimidating operators.

Expense. Installing programmable displays does cost money. The display itself may be kind of high and then it will almost always take a couple of days at best just to learn to program the display, plus time per screen to build the actual displays. Often people forget the time it takes to actually program the display.

Data Acquisition

Data acquisition has to do with the retrieval and storage of process data. This can be data from individual machines or from a group of machines. Typically storing this data for long term analysis and/or short term adjustments.

In the past 20 years or so, we've come quite a long way in this regard. In the early 80's and before, you could purchase computer systems with enough horsepower to collect some data for you, but there wasn't really much in the way of software to help you with the task of storage and retrieval. Another problem was that hard disk space was not the cheap commodity that it is today, but rather a premium resource, and using it up to store data that may not be looked at was unthinkable by many. The first data collection systems that I am aware of for process data were the PDP line of computers made by Digital Equipment Corp., then later the Vax, although I am also aware that HP and Data General produced some good machines as well. There were a couple of fledgling companies during the 80's who got the ball rolling on the data collection industry. Today we take products like Wonderware or The Fix for granted, but it wasn't always so.

So, during the 90's hard drive space cheapened and software became available to help with the data collection tasks and here we are today, ready to record temperatures, pressures, flows, weights, measurements and also discrete data. Yes, you can also record digital inputs if you need to, which can be a good way of diagnosing a long term machine issue.

When people think of storing process data, the first thing they think of is temperatures. Temperatures of molds, or possibly the chemicals that make up the mixture of whatever you're making, then they think of pressures and flows. I guess maybe that's why the first applications for modern data acquisition systems was recording parameters on furnaces and chemical processes, it just kind of fit. Data Acquisition was kind of a fluid thing, or a better word might be continuous and so was temperature. When you record data, you must by necessity record a snapshot of the data. So, you might record a given temperature, once a minute. That's great, you can plot it and when you have the computer draw a line between the points, you feel like you know what it did for the other 59 seconds that you weren't monitoring it, and basically you're right, depending upon the situation, you probably can interpolate between data points and be justified in your conclusions.

People have a hard time with the discrete data that is all around them, like alarm data, they just don't consider it. One of the most useful types of information, in my opinion, is the machine alarm data. You can draw some good conclusions from alarms. If you collect the number of alarm occurrences a day for a given machine and then they double, you've got a problem. Often the machine operators don't even report changes in the number of alarms, they don't think to until it gets really bad, after all they respond to these same alarms everyday, just not as many. Now armed with an alarm summary, we can be proactive, instead of reactive, and actually go out and fix the machine before it breaks completely, having the parts on hand to fix the machine when it does go down completely.

Another useful, and overlooked, type of data is counting data. By recording counting information at many points in your production line, you can get a feel for what a given count at a machine should be. Then, if there's less, find out why. You can also record defective counts and then trend those counts to see if the long term trend is up or down. A very straight forward and easy to use analysis tool, I am continually surprised by the number of companies not using a production counting system for process analysis.

One other thing you might want to consider is your network. Storing a massive amount of data from many systems back to a centralized storage server takes network bandwidth. Segment your network design so as to isolate the high traffic areas from the user areas.

programmable logic controller, versa view, panel view, versaview, panelview