|Home | Articles | Forum | Glossary | Books|
By Joseph J. Carr
In the world of digital memory devices, line transients can cause catastrophic failure. Often past remedies used for analog circuits will no longer suffice. For a look at some newer approaches to "coping," read on.
A glitch is an unwanted pulse, or other disturbance, that raises hob with digital circuits. When one of these glitches gets into a system it can increment counters, erase memory, or reset everything back to square zero (or some other unlikely spot that is equally unuseful).
Now that digital devices are found in almost all walks of private and professional or business life, we hear more and more complaints about glitch witches having a field day. In the "old days," there was often the possibility of designing the location of the equipment so that glitches from external environment were all but eliminated.
Equipment, too, was designed to have fewer problems. But today, digital equipment is found in almost all types of "unprepared" sites, and competition has caused many manufacturers to leave out some of the design features that helped alleviate previous problems.
We can easily identify two types of external glitch: static electricity from our bodies and power line transients. Let us deal with these one at a time.
Static electricity builds up on our bodies, clothes and tools without any help at all! Just by ordinary rubbing, we can create many thousands of volts of electrical potential. This is viewed rather dramatically when you walk across certain types of rug, or other floor covering, and then touch a grounded object: ZAP! A spark will jump from your hand to the object, often rather painfully.
While most static discharges are not harmful to people, they can be devastating to digital electronics circuits.
Leaving aside the fact that sparks can blow out CMOS and certain low power TTL I.C. devices, there is the possibility of the spark getting into the circuit as a transient, and resetting some of the circuits. A small desk -top microcomputer used where I work (when not writing for ET/D, I actually work for a living) used to "go bananas" when someone would walk across the carpeted floor and then touch the keyboard, or the mainframe of the computer. The guy who had just typed in a 1000 step program usually takes a dim view of losing all of that work because a static discharge sent the computer into reset.
There are only a few things that could be done for this type of problem. One of them, of course, is to remove the rug and replace it with insulated tiles! No rug, no static. It's simple. Or is it? In some cases, the building management will not permit you to remove the rug. Another possible cure is to place a rubber, or plastic, covering over the rug. But be sure that the covering is anti -static, or you will just change one problem for another.
If the building manager does not want the rug removed, and the covering option is, for some reason, not reasonable, then you might want to try having a carpet cleaning firm, or the building custodial staff, apply an anti-static treatment to the rug. If there are no firms that do this, try it yourself.
Contact the suppliers of carpet cleaning materials. Note that this treatment must be repeated 3-4 times per year, despite manufacturers' claims.
The static problem will be worse in areas of low humidity, and will vary with the season over most of the country. Short of grounding everybody, which is tantamount to suicide in a service shop environment, there is little that can be done.
Power line transients
In technical school we all learned that the electricity that we get from the power company is real nice, clean, sinewave, right? Wrong! The electrical power mains often contain transients that average several hundred volts and may easily reach levels in the 1600-2000 volt range. Most transients last only milliseconds, but in that time they can shut down a major computer, foul up a transaction made on a point of sale terminal, or wipe out a letter stored in a digital word processor. An oscilloscope waveform used by G.E. in their MOV advertisements (a kind of transient suppressor) showed literally dozens of over -kilovolt transients recorded on a 110 volt a.c. power line in a 24 hour period.
There are many sources for these transients. Lightning, of course, is well known. But what is not well known is that it does not take a direct hit to induce kilovolt transients in the power line! Ever heard of induction? A bolt of lightning, anywhere near a power line (up to about 1/4 mile!), to ground or overhead, can induce mighty transients. This transient could be propagated along the line for many miles, but is usually most severe in the area on the same side of the transformer where the induction occurred. Transformers tend to attenuate (but not eliminate!) the transient, and cause it to spread out, even more drastic for a digital device! Sometimes the source of power line transients is inside the same building with the digital equipment. My own experience is in medical equipment, and consequently, my "war stories" revolve around hospitals and medical school facilities. In one case, sensitive scientific research instruments would give intermittently bad results. At other times, an optical scanning machine that was used to grade the mark -sense answer sheets turned in by medical students taking examinations would flunk out the whole class.
The trouble was traced to severe transients arriving on the a.c. power line.
The origin of the transient pulses was some power system switching equipment located in the basement of the building. We could get a premium rate from the power company if we used special equipment that would sense the lightest loaded phase (buildings this size use three-phase a.c power), and then switch the load of the building around to balance the drain from the three phases.
While this was nice and efficient, it also tended to raise hell with all of the digital equipment (seven computers!) and many of the analog instruments used in the building. Since this equipment was inherent in the design of the building, there was little that we could do to solve the problem at the source.
The final cure
The final cure was grounding of the equipment chassis through a redundant ground wire (not the power cord ground). Sometimes, a power line ground is not too good for a computer's needs, although it works fine for a.c. power. The redundant wire ground went to a cold water pipe made of metal, not plastic.
In another case, the glitch was due to X-ray equipment in a room a few yards away. A secretary at this suburban hospital was using a "word processor" device that digitally stores patient records. This particular model would allow the user to type in 256 words (averaging five characters each), until the cathode ray tube screen was full.
This data was held in temporary, semiconductor, memory. When the screen was full, or the job completed, then the user typed an enter command, and the material was stored more permanently on discs or magnetic tape.
But every now and again, usually after the secretary had typed in 255 of the allowed 256 words, a power line glitch would erase most of the data (sigh). The problem was traced to nearby X-ray equipment. An X-ray generator is a high powered device. It might require 125 Kv at 60 mA, which is 7500 watts, for a brief instant. This could throw a transient out onto the a.c. power line. It is this transient that will cause problems.
In this case, the problem was solved by using a special type of isolation transformer. These transformers (see Fig. 1) attenuate the pulse much more than ordinary transformers and help keep glitches from the equipment.
These transformers are actually a.c. line voltage regulators, and are used for protection against brownouts, transients and other forms of a.c. power line problem.
Start up transients One of the most difficult to suppress problems is the so-called start-up transient. When a heavy duty motor, or other mechanical device, is started up, it momentarily requires a large surge of electrical power in order to overcome the resting inertia of the equipment. But when the device is running at full speed the power requirements drop to a much lower level. This phenomenon is seen most easily, perhaps, by watching the lamps in the room when a window air condition turns on. They will dim almost completely for an instant. In fact, it is likely that the average residential a.c. power line will experience 25 percent voltage drops lasting less than one (fatal to digital) second three times per day! It is also probable that a 75 percent, split-second, drop will occur at least once a month. Overvoltage transients, in which the line voltage goes up by 2000 percent for 10 milliseconds, occur more than two dozen times per day in the average installation. Commercial and factory installations, where additional heavy machinery or transient-generating electrical devices exist, can be expected to be worse.
Of course, the best cure is to tackle the transient at the source. Bypass capacitors, in the same manner as used to reduce noise in mobile radio installations, and series inductive filters can often reduce the level of noise to a livable level. But be sure to contact the manufacturer of the specific equipment before applying any "corrections." They might have some objections, related to design problems, or might be able to offer some suggestions of their own. A good review of mobile noise suppression might be in order for any technician who contemplates this type of service.
If it is not possible to suppress the noise at the source (often the case), or if the noise trouble -shooting problem turns into a sticky mess much akin to trying to nail Jell-O to the wall, then consider using one of those transformers. There might be a little customer resistance because of the initial cost, but they will provide many benefits. Not the least of these is the fact that they can now use their expensive digital instrument, where before they were being reset to square -zero every time some clown down the hall turned on the air conditioner, or took an X-ray.
Sometimes, modification of the equipment being interfered with is necessary. Either the modification alone, or in conjunction with a line isolation/regulator transformer, is often the real key to eliminating the problem.
But before jumping into some modification scheme, it might be best for you to contact the manufacturer about the problem. It might be that they have already seen this problem and know the cure. But be sure to contact the in-house service technicians first, the engineers second, and the salesmen last. Keep in mind, that corporate pressure is heavy on sales people (most salesmen are liars, and sales managers are first kin to Satan) so they will not want to tell you the defects of their equipment (it is also likely that no one will tell them!).
The engineers have pride of design and the same corporate pressure as the salesman. But that kid technician in the repair shop, whom nobody pays any attention to, will have some good smoke about the product's defects. There is something about being nailed to the wall that gives one a special insight. Note that the service guys have much to say, so much in fact, that many general managers refuse to permit the customers to talk to any one in the service shop. I recommend that my "customers" refuse to buy anything from companies who, in the past, have followed this policy.
Also, be sure that the problem is not in the peripherals connected to the digital device. My own personal microcomputer would reset every time I turned on the Teletype machine, or switched the function from local to loop.
The problem turned out to be the interface between the computer and the teletypewriter. The +5 volts d.c. and the -12 volts d.c. used in the type of circuit that this manufacturer designed to supply the 20 mA current loop for the teletypewriter would throw a glitch into the d.c. supplies of the computers. The solution was to connect the teletypewriter to the computer through optoisolators and use separate d.c. power supplies for the 20 mA loop.
One of the most common forms of equipment modification involves installation of an LC filter (Fig. 2) in the 115 volt a.c. power circuit. This is best done right at the cord entrance to the equipment so that reradiation will not occur. Several manufacturers offer balanced LC filters built right into a 115 volt a.c. chassis receptacle. These could be installed in place of the power receptacle already on the equipment.
Alternatively, the filter should be installed in a location very close to the receptacle that already exists. It is my own opinion that the filter should be fused, in the event that the capacitors become shorted, you do not want an unprotected short across the a.c. power line.
The other way to eliminate the problem is place a transient suppression device, such as the General Electric MOV, across the power line. The MOV is a varistor-like device that remains inert in the presence of ordinary 115 volt a.c., but will develop a very low resistance to a high voltage transient.
This causes it to shunt the transient between the lines, but a.c. flows in the ordinary manner. The G.E. MOV is actually one of a family of devices, and it is wise to read the applications literature in order to select one for any given application.
There are also a number of similar-acting devices that operate on different principles. Some, for example, are high power Zener diodes that have a Vz value of greater than 200 volts. The a.c. power line potential will not cause them to break over, but transients will. Others are selenium stacks.
A method that is not used quite as often is the use of RC "snubber" networks. These act to reduce the a.c. potential, however, so their use is limited to other types of lines entering the cabinet (that could also admit transient signals picked up by capacitive or inductive coupling).
Of these methods, I tend to prefer the use of a line conditioner isolation transformer (ala Topaz and Sola), as the least effort with the greatest likelihood of success. The use of transient suppressors, LC filters, or snubbers could, conceivably, void your customer's warranty. So it is advisable that you contact the maker of the equipment prior to modification if the equipment is still in warranty.
(source: Electronic Technician/Dealer, Feb. 1980)