The issue with digital interfaces such as found on computer terminals etc. is that there is nothing heard or felt, so when data is suddenly corrupted it is put down to "software bugs" or other such more believable reasons. Stating it could be a grounding problem is often overlooked. What is also not realized is that this could be time related as the voltage rises and falls as the load varies during the day. There are a few methods of data transfer. The common ones are RS232, RS485, and 10/100BaseT. The first is an unbalanced signalling system, the latter two are both balanced. All are prone to corruption as shown in the main part of this section. Although RS485 was designd to overcome such noise issues through being balanced, it is extremely limited in common mode input voltage, usually no more than plus/minus a few volts. In this world there is also this unbelievable desire to make data flow faster, regardless of the data being trasferred. If this were the short bursts of data from a supermarket check-out, and it could be successfully transferred at 2400 Baud, then why is there this desire to push it to much higher limits. What is forgotten is that for every doubling of speed the chances of data corruption under these circumstances is quadrupled. Sometimes the solution is simply to reduce the speed of the data link. There is one word of warning, and this is directed more at the inexperienced and badly taught IT "experts" than PQ investigators. Be careful of the signal ground being shorted to the chassis (usually to the case of the connector) on computer equipment. This causes ground loop currents within the device and through the motherboards (and others) and can catastophically influence the microprocessors on these boards resulting in crashes. PCs are the most vulnerable (although this fault has been seen on other types of equipment). Next time you look inside one, have a look where the power is fed to the board, where the comm port is taken off the board, and remember the primary earthing point is in the power supply. Now work out where the loop will exist when you short out signal ground to the casing. A fault like this was discovered in an emergency control room where the PCs, fitted with touch screens and used to control all the facilities of the building, would 'hang' and could only be cleared through a power off-on cycle. It was later discovered an over-zealous IT technician had decided to solder Pin 5 to the casing of the COM port connector. Be warned, this above fault was not a PQ issue!
© 04.12.01 |