Noise in an analog audio line can be difficult to track down, let alone eliminate. Noise can be caused by such factors as pin 1 problems, inadequate common-mode rejection at an input, magnetic or electric field coupling, or shield current induced noise.
So it takes a methodical troubleshooting approach to isolate the source of the problem and get a handle on how to fix it.
Observe if the noise problem is system-wide or appears only in certain subsystems or lines—that will help guide your troubleshooting efforts. If the monitoring path has problems, it's a good idea to clean that up first, so that it can be used to listen for problems in other signal paths.
Bill Whitlock, president and chief engineer, Jensen Transformers, offered these helpful hints during one of his presentations at the New York AES conference last fall.
Be methodical. Don't keep changing things just to see what will happen. Get a notebook and write down everything you tried, and what happened as a result. A signal flow diagram can be very helpful here.
Before diving in head first, find out if the system or subsystem always had the problem. If all was okay at one time, then see what might have changed. Observe if the wiring or wire routing changed, cables or connectors swapped out, or equipment installed or removed. Look if any control settings have changed, or if the settings themselves affect the amount of noise.
Look beyond the racks.
Maybe some industrial equipment got plugged into tech power. Or maybe something in the environment changed. Be on the lookout for anything that could emit an electric or magnetic field.
In general, Whitlock advised, when troubleshooting, work backwards. Start at the end of the signal chain and move step by step towards the beginning.
Concentrate on two pieces of equipment at a time, a source box feeding a destination box. Then check each line in turn from source box to destination box.
Whitlock suggested a five-step process using a "Dummy" plug. For balanced circuits a dummy can be made from a barrel connector with a female jack or connector at one end and a male plug at the other end. Use connectors that will allow you to plug in and out of the equipment and cables that you'll be testing. You'll also need a switch on the barrel.
Fig. 1: The Balanced Dummy. Closing S1 creates a 10Ω source impedance imbalance, per IEC test for CMRR. (Actual imbalance may range from 9.8Ω to 11.6Ω due to tolerances.
For the female side, wire a 604 ohm resistor across the tip and ring of a phone jack or pins 2 and 3 of an XLR connector. The male side of the barrel is used to switch in a nominal 10 ohm impedance imbalance in line with the device that it's connected to (see Fig. 1 for details). Connect the shield connection of the female connector to that of the male connector. This is the only connection that goes straight through inside the barrel. The signal connectors are not carried across.
Using the Dummy, begin Whitlock's five-step troubleshooting process. Continue through the steps until you hear noise.
Unplug the existing cable from the input of the destination box and plug in the Dummy instead. Make sure that the impedance imbalance switch on the Dummy is in the open position (normal mode).
If you hear noise, the problem is internal to the destination box or further downstream. It's possible that the destination box doesn't use a star connection between the power supply common and equipment chassis, and its wiring is allowing high-frequency power line noise to flow in the signal reference ground.
Leaving the Dummy plugged into the input to the destination box, plug the existing cable between the source box and the Dummy. Noise under these test conditions indicates a pin 1 problem in the destination box.
Disconnect the Dummy from the input to the destination box and the existing cable. Connect the existing cable back into the destination box, and then unplug it from the output of the source box. Next connect the Dummy into the existing cable. Make sure the Dummy doesn't touch anything conductive.
If there's noise now, then look for a source of an electric or magnetic field (or both) that may be inducing noise into the cable. Observe the cable routing. Is it parallel to power cords, for example. Try re-routing the cable to get some distance from the field or change from a parallel to a perpendicular course. If possible eliminate the field at the source.
Leave the Dummy on the existing cable, and plug the open end into the output of the source box. If there's noise under these circumstances, the problem is shield current induced noise or SCIN.
SCIN occurs when there's a current flowing through the shield. This current creates a magnetic field that then induces voltages in the signal wires that the shield is wrapped around. If the shield is asymmetrically wrapped around the signal wires, then the induced voltages will be different.
Cables with drain wires are most prone to this, but these are also most commonly used due to their ease of termination. According to Whitlock, the best cables, from an SCIN perspective, are made with either braided shields or dual counter-wrapped spiral shields and no drain wire.
If you can't find a way to reduce the shield current, you may have to replace the cable with one with better SCIN performance.
Keep the same setup as for Step 4, but now switch the Dummy's switch to the CMRR setting. This creates a nominal 10 ohm impedance imbalance on the input to the destination box. If there's noise now, it means that there's not enough real-world common mode rejection at the input to the destination box. Contact the manufacturer of the box. If you can't replace or fix it, you'll need to increase CMRR externally, and here's where a transformer isolator can come in real handy.
If, after going through all these steps, you don't hear any noise, then look for noise problems at the output of the source box, or further upstream. If a pin 1 problem at the output of the source box can be ruled out, then go through the five step procedure again on the preceding stage. Make sure you remove the Dummy, and correctly reconnect the existing cable from the first test before moving on.
Mary C. Gruszka is a systems design engineer, roject manager, consultant and writer based in the New York metro area. She can be reached via TV Technology.