POWER PLAY - PART 3
 

by Wayne M. Krakau - Chicago Computer Guide, October 1997
In this final installment of the trilogy, having already covered the basics of bad electric power from both outside and inside, I’ll conclude with one last real-world example and some possible defenses.
A couple of years ago, I was lucky enough to be referred to a client without a network. (It’s rare for me to find one. Usually I meet new clients who need an existing LAN fixed.) I sold the client a 10-user NetWare LAN, including all of the hardware (server, workstations, etc.), all of the system software (NetWare, DOS, Windows, shared fax system, various utilities), and most of the applications software (contact manager, database, word processor).
In addition, since the client’s system administrator had little computer expertise, my company became, in effect, their computer department. This situation is common for my company and imparts a great sense of responsibility for the client’s welfare.
The first problem was a report of defective monitors. Some monitors were frequently blurred to the point of unreadability. We replaced those monitors. That didn’t help. The new ones were also blurred. That led me to suspect some underlying power problem. The system administrator vehemently denied any possibility of a power problem. The building was only about two years old and he completely trusted the very experienced electrician who wired it.
Then the administrator reported malfunctions of the internal modem within their shared-fax machine (Essentially it was a nondedicated fax server - a configuration which I didn’t like, but the client demanded). The mode would completely lock up and even turning the computer off and on didn’t always revive it. It would remain locked up for random lengths of time ranging from a few minutes to several hours. Again I suspected power issues.
At the administrators request, I replaced the fax computer’s modem, then the motherboard, then the power supply, and finally the I/O card (which was also the disk controller). Nothing helped. Simultaneously, I began consulting with the top troubleshooters at APC (American Power Conversion, West Kingston, RI, 800-800-4APC). The file server (which had no problems) was protected by an APC UPS (Uninterruptible Power Supply) and the workstations, including the modem’s phone line, were protected by APC surge suppressors.
The APC engineer (and I do mean "engineer" as opposed to "technician") agreed that there were power problems. Even as I described the situation to him over the phone, I was receiving reports of random boot-up problems with other workstations on the LAN. He suggested that I try putting a UPS on the fax computer. The district manager for APC lent me a small UPS for this test. Even that didn’t help.
Meanwhile, I was examining the system under the direction of the APC engineer. Through trial and error experimentation with electrical devices, I found that I could upset the display of several monitors by turning on a freestanding fan. I also found that the desk light in one cubicle prevented another computer two cubicles away from rebooting!
Based on my description of the wiring, the APC engineer noted several NEC (National Electric Code) violations were apparent, and that an improper ground was preventing the UPSs and surge suppressors from doing their jobs. After talking to the engineer, the administrator finally agreed that there were electrical problems. He declined the engineer’s offer of a Power Audit, in which APC would send someone to this site with the appropriate test equipment to fully analyze the power problems and work with the client to see that they were fixed. The client said the Power Audit would cost too much and stated that he trusted his electrician to fix things.
A month later, I was called upon by the same client to inspect more allegedly defective PCs. The client had spent $3,500 on rewiring the building and was sure that the power system, including the ground, was now perfect. He took his electrician’s word for this. Now, all of the workstations worked perfectly - if they were turned on individually. If all of the workstations were on, the last couple that were turned on wouldn’t boot. The administrator couldn’t run all of them at once. At least one, and sometimes two, had to remain off.
The physical position of the malfunctioning PCs wasn’t an issue. The order in which they were turned on was the only deciding factor. The last one or two turned on invariably malfunctioned.
Although I replaced several PCs, retested all of their LAN cabling, and spent several days (and many hours on the phone with the APC engineer) trying to find a way around these problems, the administrator dropped my company as a vendor, accusing me of having cheated him. Through all of this, I had spent over 100 hours of my time on debugging their electrical problems, all without charging them. Oh, well - I guess there are some situations in which you simply can’t prove you are ethical. (By the way, the last I heard, they still can’t run all of their PCs at once.)
This episode leads me to my primary suggestion for attacking mysterious power problems. Arrange for APC to do one of their Power Audits. I have heard rumors that other companies may start offering similar services in the future, but, so far, I am only familiar with APC’s offering. Also, I have had several years experience using their products and accessing their technical support, so I know that they can get the job done.
In a Power Audit, APC personnel are sent to inspect the client’s site and test the power system based on the NEC, the IEEE and other industry and government standards for wiring and cabling. They use
equipment like harmonic analyzers, branch circuit analyzers, and various power monitoring devices to test the power system. Best of all, they use their experience to visually inspect each component of the power system, looking for standards violations.
Most of the time, they find multiple standards violations, including violations of the NEC, which, as I stated in Part 2 of this series, was issued in the 1930's and is the main code on which electricians’ base their work. I have no idea if these violations are caused by extreme cost-cutting in building and remodeling projects, by a lack of skill on the part of electricians, or if the same type of uncaring attitude that is the curse of the computer industry has affected electricians, or if there’s some other reason.
I just don’t have the background to comment in any serious way on the causes of these code violations. All I know is that some of these violations are safety related. It would be interesting to know the legal ramifications of an electrocution or death by fire traced back to this type of violation. I’m sure the personal injury lawyers would have a field day. (This is all, of course, in addition to any loss due to either downtime or inaccurate data from a malfunctioning power system.)
The main objection that I’ve heard against an APC Power Audit (or the equivalent service provided by some other company) is the expense. The cost varies so much depending upon the exact situation that I can’t quote exact numbers, but it is expensive to send a professional out into the field with a pile of fancy equipment. I have heard a rough estimate in the $3000-$4000 range for one 25-user LAN, but that was an off-the-cuff estimate specific to one particular client, so I wouldn’t be willing to budget based on that estimate.
Another, admittedly limited, option is to purchase a Perfector UPS from International Power Technologies (Orem, UT, 800-944-0356). It is a 1500 Volt-Amp digital on-line UPS. It runs completely without reference to a ground. It can take any voltage from 60 through 270 volts and can output your choice of any voltage from 90 to 270 volts at either 50 or 60 Hertz. It is the perfect choice for systems that are out in the field running off portable or otherwise nonstandard power systems.
At a $2,995 list price, it’s a bit expensive for day-to-day office use, but for situations where the underlying power infrastructure is not adequate, it could be just what you need. There are limits to its usefulness. First, you usually don’t find out about bad power until after you have already purchased a "normal" UPS. You would have to discard the old UPS or put it to use elsewhere.
Also, it can only protect up to 1500 Volt-Amps worth of equipment. If your whole building’s power system is inadequate, you might have to buy a lot of these Perfectors to run your business. I still haven’t figured out how to get around the network communications problems inherent in having multiple independent ground references from all these Perfectors. In the long run, it may be cheaper (and safer!) to just fix your power system.
Now, pardon me, while I check out this loose wire behind my computer. I’ll have to crawl under my desk and stretch out to reach it. Hmm, I wonder if it’s carrying any current. I’ll be right back to finish this article. Zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzt!
 

�1997, Wayne M. Krakau