REGISTER | LOGIN
Breaking News
Design How-To

Cross-coupled output stages for balanced audio interfaces

NO RATINGS
Page 1 / 2 Next >
More Related Links
View Comments: Newest First | Oldest First | Threaded View
Les Tyler
User Rank
Author
re: Cross-coupled output stages for balanced audio interfaces
Les Tyler   9/6/2008 10:20:02 PM
NO RATINGS
Thanks, Haldor for raising some very good points. First, you're absolutely right that common mode noise rejection requires impedance balance. The source and load form a balanced bridge. Both the desired signal and any common-mode noise will be attentuated by the two voltage dividers in each half of the bridge. If the two dividers are not identical, the two dividers have different attentuation. This converts some of the common-mode signal to a differential one, which the receiver cannot reject. Bill Whitlock has published some excellent discussions of the balanced bridge formed in balanced audio lines. See http://www.audiodesignline.com/howto/196604119. Second, I agree that a far simpler way to accomplish impedance balance is to ensure identical output impedances at the source, and drive only one of the two outputs with a signal. The other one, as Haldor suggests, is simply connected to ground. However, doing this has a few drawbacks. First, you loose signal drive capability. Haldor states flatly that "the 'free' 6dB gain from the cross-coupled driver is not needed and the extra signal level makes it more likely that you will clip the downstream input device." But, in my experience, +24dBu is often considered the minimum required drive capability from a pro audio output stage. Normally, output voltage swing is going to be limited by the supply rails. Many pro audio products run from +/-15V rails. If a single-ended output stage runs from those same rails, it'll be very difficult to produce more than ~+21dBu from it. +20dBu is more likely, considering the build-out resistor and voltage drop in the output stage itself. Producing a second set of (higher) supply rails inside the device is expensive and complicated. Differential drive, especially when it comes in the form of a single, 8-pin IC, provides a more convenient way to get high output drive capability. Of course, if you don't need to reach over +20dBu drive, then Haldor's single-ended output approach can likely deliver that much signal from +/-15V supplies. However, in this day of power-conscious designs, supply voltages are dropping. That extra 6dB you get from differential drive can be handy to allow lower internal power supply rails, simpler power supply designs, and still provide adequate headroom at the output. Second, if you only drive one output, then you only have one phase of output to choose from. If the succeeding device is single-ended, then there's only one way to hook the device up. While most pro audio equipment has differential inputs and outputs, pro audio users must often connect to "seim-pro" and consumer gear which has single-ended inputs and outputs. An alternative to the cross-coupled drive is to simply drive both the hot and cold outputs in antiphase without the cross-coupling feature. In such a case, including identical build-out resistors in series with each output would again be the right thing to do. This provides high signal drive capability when driving balanced loads and delivers both output phases. But, single-ended loads are frequently found in the form of tip-ring phone jacks. In such a case, the "unused" output is frequently connected (deliberately or not) to ground. That causes the "unused" output to drive a short through the build-out resistor. It can be difficult to prevent the resulting high currents, which are likely to be large and clipped, from causing distortion that can leak into the desired audio channel. On the other hand, the cross-coupled output approach neatly addresses all these issues. For differential loads, it delivers 6dB more voltage drive than a single-ended output. It offers both phases of output. And, if the load is single ended, it won't matter whether one of the outputs is grounded or not, since the cross-coupled output prevents driving high currents into the grounded output. (Though, as I pointed out in the article, that can change for some of these designs when the driven output clips.) BTW, clever use of ASICC to draw the schematic! --Les Tyler

Haldor
User Rank
Author
re: Cross-coupled output stages for balanced audio interfaces
Haldor   9/5/2008 7:15:25 PM
NO RATINGS
I have to ask why on earth you would want to use cross-coupled outputs in the first place. Common mode noise rejection requires impedance balanced signals, not symetrically driven signals. A far simple way to accomplish this is to series terminate the hot and cold signals with the same value resistor at the source. Connect the hot signal to the output device, and ground the cold signal. Refer to the ASCII schematic attempt below (hope it comes through formatted correctly). R1 -|>--/\/\/\-------0 Hot Output R2 |----/\/\/\-------0 Cold Output | --- - R1 = R2 When you are dealing with line level signals, the "free" 6dB gain from the cross-coupled driver is not needed and the extra signal level makes it more likely that you will clip the downstream input device. Cross-coupled outputs are a solution to a problem that doesn't exist and provide no real advantages over impedence balanced outputs. Transformer based inputs however, are useful. They can be driven from an unbalanced output with only minor degredation in CMRR performance.

Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed