A Balanced To Unbalanced Patchbay Interfacing Caveat

A Balanced To Unbalanced Patchbay Interfacing Caveat

by
Evan T. Chen
September 24, 1996

I discovered this problem during a mixing session in which the outputs of an Avid AudioVision (which is functionally a Digidesign 888 I/O interface) were routed from the newly-wired editing suite to the equipment room to the mix room all via three patchbays, a lengthy and noisy run. I noticed that my Avid calibration tone, which I had just painstakingly aligned with a fluke meter, slightly missed their mark on the SSL8000's accurate meters. After further investigation I realized that the small discrepancy resulted from Digidesign's implementation of their cross-coupled output stages on the 888 I/O interface operating in unbalanced instead of balanced modes. As it turned out, the Mackie 1604 unbalanced inputs to which the AudioVision outputs normalled were wired incorrectly (under these circumstances) on the brand-new patchbay, grounding the negative (-) side of the 888's outputs, effectively converting perfectly healthy balanced outputs into less robust unbalanced ones with 6db less headroom. Sending sixteen of these unbalanced runs through 50+ feet of conduit resulted in, as I found out later through careful noise floor measurements, approximately an extra 9db increase in noise per channel! The problem was easily solved by breaking the normal to the Mackie mixer with dead patches, reverting the Avid outputs into balanced ones once more. The funny thing is, upon subsequent visits to the studio, these long runs were still being used unbalanced, whether printing to DA88 or multitrack or mixing right off of the Avid! Shame on them.

I should mention that this method of grounding two out of three pins on the Avid outs and most cross-coupled outputs produces an unbalanced "unity" level on par with that of a balanced scenario, like transformer balanced outputs, that is, 1Vpp in equals 1Vpp out (that's the reason they were designed). Many compact mixers nowadays assume that this is not the case, that one of the signal pins is left floating, that is 1Vpp in equals 0.5Vpp out, thereby justifying two separate "unbalanced" and "balanced" unity gain detents on the faders or, more likely, knobs. With one of the signal pins grounded however, the unbalanced detents can't be trusted. Prior to their next-generation VLZ line which uses true balanced inputs, Mackie's compact mixer line used unbalanced inputs -- marketed to be compatible with balanced outputs -- with floating 1/4" TRS ring connections to avoid shorting the positive or negative leads (depending on whether pin 2 or 3 is hot) on balanced outputs and possibly damaging poorly-designed balanced outputs. The tradeoff here is that of gain for safety, and may be the source of the myth that balanced transformer and cross-coupled outputs lose 6db of gain when operating in unbalanced modes: cross-coupled outputs lose 6db in headroom, not gain, when used unbalanced. However, with balanced transformer and cross-coupled outputs, this is not a problem, so to avoid this loss of gain with the latter design, often times a patchbay of unbalanced inputs is wired to ground the negative terminal instead of floating it, as was this patchbay case.

Also, theoretically with cross-coupled outputs, the differential output should be the same whether one signal pin is grounded or not. This was not the case with Avid/Digidesign's 888 I/O box. Although the discrepancy was slight, it did lead me to investigate the cause of the problem. Well-designed outputs won't exhibit these anomalies.

Copyright © 2010 by Evan T. Chen. All rights reserved.