Problems with installations

ianm's picture

6. Problems with installations

The following question was recently sent around. Three responses have been garnered to date. Also, some other problems could be addressed, such as distance of sample to beam, lengths of fiber optic, distance of the spectrometer to computer, interaction of light from various sources with the fiber optics, and for liquids, bubbles and particles.

Installation problems and solutions

A topic that many (all?) of us have dealt with, yet would be good to discuss, is installations and the associated problems. For instance, at least one spectrometer firm says the power to the spectrometer should be conditioned, unless it is known that the power is well regulated. This holds for both lab and other places. This brings up the question: What are the problems each has faced in installations and how were they solved? When you respond, please indicate if the installation was in a controlled environment, such as a lab, or in a site without such protection. However, if the problem(s) is not peculiar to such a differentiation, please also indicate that.

From: David A Russell 695-2819 [email protected]

Generally, all instrumentation and computing equipment should be on power fed through an isolation transformer. Failing to do so can lead to fried computers.

Also, I prefer to provide a UPS for the analyzer host computer whenever possible. Some sites experience frequent short outages and you can't assume that the power you're getting is on the backed up supply.

Surge suppression on telephone and serial cables should also be considered in situations where a lightning "hit" is possible.

Dave Russell

From: [email protected]

Perhaps the most difficult installation that I faced involved placing an online instrument into an old plant with a rigid labor union environment. Each craft of the labor union had to handle its portion of the installation, but seemed unable to coordinate its work with the other crafts. Good communications was the key to maintaining progress, and most of the problems were solved by keeping all workers and supervisors continuously informed of the what, when, and why of each step of installation. The only problem that couldn't be solved was keeping the electricians away from the communications cable between the spectrometer in the field and the data system in the control room. After they used wire cutters to remove extra loops from the fiber optic cable, we had to bring in outside personnel to restore communications.

Jim Hermiller

From: [email protected]

Another installation problem that I faced involved using one instrument to make measurements at several locations in a process. If we multiplexed the instrument using fiber optics, then we could have taken advantage of the manufacturer's software to keep track of which location was being analyzed and to forward the correct information to the control system. In this particular application, however, we had to multiplex the sample using a series of programmable logic controllers to direct liquid streams at various times to the single instrument probe, and we couldn't figure out how to tell the manufacturer's software which sample was being analyzed at any particular moment. Our solution was to have the programmable logic controllers talk to the process control system, but this required much effort by plant personnel and is more complicated than it should be.

Jim Hermiller

From: Howard Mark [email protected]

Subject: Re: Installation problems

A couple of response comments here:

1) to reinforce Dave's comments: it's a general truism that the electrical lines in plants are much "dirtier" than what you'll find in your house, or in a separate laboratory environment. The reason is that when large electrical loads, particularly motors, are turned on or off, they induce large voltage "spikes" in the lines. This is because the effect of inductance (of which motors and transformers have large amounts) includes a term proportional to di/dt (i.e., change of current with time). When a switch is opened to turn a motor off, for example, the current goes from whatever it's previous value was, to zero, in a very short time (i.e., dt -> 0), so that this derivative term is large, and the induced voltage is large. Even if the controlled device is not a motor, there's always a transformer in the circuit, from the power company, so this samething happens.

An isolation transformer will help, but may not be sufficient. A UPS is good, but is probably overkill unless actual power outages are frequent. For intermediate situations, there are power conditioners available, which are designed specifically to protect the electronic equipment plugged into them from such spikes.

Incidentally, it's not a bad idea to plug the instrument into the power conditioner as well as the computer, after all, the instrument probably costs several times what the computer costs. Instruments are likely to have their own internal protection (you can find out from the manufacturer), but the "belt and suspenders" approach won't hurt.

Any other electrical lines (such as telephone lines) that are connected to the sensitive devices should also have surge and spike protection, as Dave says. Simple physical proximity to the power lines can create capacitive and transformer-type coupling and thus induce spikes in them also.

2) I find it most interesting that Jim's worst problems were the non-technical ones. It also makes me wonder if there was any relationship between the two - i.e., were the fiber optics "accidentally" cut by a disgruntled union worker? Also, I wonder how Jim was able to bring in outside personnel to fix the fibers, in the face of the union pressure? Was the use of (electrical) PLCs dictated by the union instead of the (non-electrical) fiber optics, or was that strictly a technical decision?