Headphone amps for dynamic standards

[K O'Connor]

Hi Guys

The every-day working man's headphone has always been the dynamic type, where the transducers are simply small dynamic loud speakers. These will be 8R up to 600R.

Traditional headphone outputs from power amplifiers were simply a set of 100R-1W dropping resistors feeding a stereo jack that had closed contacts to disable the main speakers. This meant that whatever the big PA was, it would be idling however hot it normally would, providing the usual voltage swing at much reduced current.

Preamps might have a headphone output that was fed from the main output, again with closed switches on the headphone jack, or from an auxiliary output dedicated to the task. If the latter, there would generally be a headphone Level control.

Dedicated headphone circuits were often just a couple of transistors before ICs became dominant, in which case a generic opamp circuit would drive the headphones. In all cases, there would be a low-value resistor added in series with each cup. This actually reduces THD and stabilises the driver from the cable impedance. At the lowest impedance, parallel opamps or a buffer help for high-SPL listening. 1mW at 8R is only 90mV at 9mA. 1mW at 600R is 770mV at 1m3A. These are RMS; peak values are 40% higher. Nested feedback circuits or discretes can push THD to inaudibility.

Most opamps can drive 1k loads, with more now capable and rated for 600R. The ubiquitous 5532 is rated for 600R loads yet it can drive 30R cups through 50R resistors quite well. For most people who only casually use headphones, this is adequate. The nice thing about opamps and most similar discrete circuits is the high power supply rejection ratio (PSRR), which is the ability to ignore supply noise. As a headphone driver with modest or unity gain, the PSU can be quite "cheezy" and still be hum-free.

Of course, with the low power requirement of headphones, we can be inventive and creative  and use tubes, transformers, or hybrid circuits as we see fit. Whatever your aesthetic is, just go for it.

[K O'Connor]

Hi Guys

A dedicated headphone amplifier should have one feature that a standard power amp driving room speakers does not need: a mixer.

The mixer is actually a stereo cross-feed bringing a bit of the right-channel sound into the left side, and the same amount of left-channel sound into the right. This blending achieves a sound similar to what we hear when listening to speakers in a room. In the room situation, each ear hears sounds from both speakers and we can localise instrument and voice sounds that were mixed more-left and more-right and right down the middle.

Using standard headphones and the common style of headphone amplifier, the sonic location of instruments will reflect the true mix designed by the recording engineer, and it will not be too musical unless the engineer mixed using headphones only. In most cases, small room speakers are used to monitor the mix through and the mix is inherently optimised for playback through room speakers.

In the cross-feed headphone amp, one can preset the cross-feed level or have it be variable. One can also decide whether the cross-feed signals are full-bandwidth or slightly "dulled" using a low-pass filter. The latter could be made adjustable, as well. The cross-feed could be switchable giving us the opportunity to hear the hard mix or the compensated "room mix". None of this circuitry takes much space and it is all considered to be low-level.

Solid-state headphone amps are typically just opamp ICs. There may be one per channel, which is functional and not excessively high-fidelity. Opamps can be paralleled for more output and until you get to seemingly outrageous numbers, the fidelity still leaves something to be desired. Also, with any of these options so far, we are limited to opamp supply rails which may not give the best headroom for high-impedance 600R headphones.

So, the next step after using a single opamp per channel is usually to add a discreet buffer within the feedback loop. This is typically a complementary pair of transistors. Pretty good fidelity can be achieved this way and the discreet stage can operate from higher supply voltages.

Opamps have improved remarkably since the 1980s and there are now some very good headphone amplifiers based solely on IC opamps. One factor in the popularity of using ICs is that the waste power is distributed and no heat sinks are required, easing the mechanics of the assembly. Many of the best opamps are only available as surface-mount devices. SM devices need large copper areas on the PCB to dissipate their heat, so sometimes this factor cancels the choice of using a tiny SM device over a much larger through-hole device.

Of course, discreet circuitry allows complete freedom of power supply voltages and topology. Going too simple always results in excessive THD, but the wrong direction of modest complexity can do so, as well. In any case, unmeasurable THD is possible via both the discreet and integrated approaches.

If we incorporate tube circuitry into the headphone amplifier, it will have reduced accuracy but may be pleasant to use.