Recently, I attended the site of a near new installation to investigate the cause of inaccuracies in process measuring equipment and audible noise on an analogue telephone system (yes, it says analogue). The analogue telephone system made the job interesting, because instead of just seeing ‘jumpy’ measurements on a screen you could actually hear the noise for yourself (search “white noise” on youtube for many examples). Perhaps not particularly entertaining for Engineers and Electricians of ‘a certain vintage’, but for a child of the digital era, it was definitely a novelty.
The noise, audible as sort of a hiss, was present at neither microphone. So where was it coming from? The installation was filled with uninterruptible power supplies and more than 1500A of equipment fed by variable speed drives. For an analogue system that operates on the order of milliamps, in excess of 1500A of rapidly switched current presents a huge potential for pollution.
Still, with sufficient cable separation and noise rejection measures (such as shielding, and twisted pairs), it isn’t an unassailable problem.
In fact, the installers had done a really good job, see below:
Clearly, I jest. This installation is an abomination. How could it possibly get past the Quality department? Admittedly some attempt has been made to have signal cables perpendicular to power cables (good). However, signal cables are still being run directly on top of power cables. Additionally, on the ‘potential for pollution’ side of the equation, the power cables are three individual single cores. As can be seen in the lower right-hand side, the arrangement of the cores is red, blue, blue, black, black, blue (I think). When three individual cables in a balanced three phase system are run close together, the superposition of the associated fields mostly result in mutual cancellation. Running the cables as shown avoids taking advantage of this effect.
Curious, I made a rudimentary transformer to see if I could ‘catch’ some of the unbalanced fields. I took some of the same instrument cable (two core shielded type) and short circuited one core (as a current transformer (CT)) and open circuited the other core (like a voltage transformer (VT)). This device is shown below. I really didn’t expect to measure anything. However, the VT showed approximately 7V and the CT measured approximately 0.8mA, which was a significant fraction of the signal level. Grounding the foil shield resulted in a huge reduction in voltage across the VT. However, the current in the CT was unaffected.
My investigations determined that the in-situ cables were of the unshielded variety, so they were at the mercy of both components of noise. As the results of my rudimentary test device indicated, installing shielded cable would only partially ameliorate the problem.
The Client was unwilling to suspend operations at the site and re-route the cable (which isn’t particularly surprising). What to do? I couldn’t remove the noise sources, increase the cable separation distance or re-run cable with better noise rejection characteristics.
In the end, we installed common mode noise filters at the phone cabinet for each individual phone. This significantly improved operation of the facility but it was a suboptimal solution.
Given the opportunity to perform a greenfield installation, consider the following:
Buy high quality signal cable the first time – If you must replace poor quality cable, you pay for the labour twice, the wasted cheap cable and you end up buying the good cable anyway.
Create as much room as possible between power and signal cables. How much distance depends on many factors including imbalance between phase and magnitude of voltage/current (more correctly magnitude of dv/dt and di/dt for capacitive and inductive coupling respectively).
A metal divider or conduit surrounding sensitive cables can bring big benefits in really noisy areas. Additionally, a metal conduit gives great protection from mechanical and ultraviolet damage.