London Power ad

Search the Forum

(Advanced Search)

Headphone amp for electrostatics
Hi Guys

The lowest-distortion headphones available are electrostatic types. The most famous brand offering these, Stax, from Japan, lovingly calls the headphones "ear speakers".

Electrostatic headphones require a high driving voltage, plus a high fixed bias voltage, none of which can be provided by a standard audio power amplifier. The bias voltage is between 200V and 600V and is current-limited by 5-10Megohms. The reference drive is 50Vrms + 50Vrms for 100dB of acoustic output. The listing of two voltages means there is differential drive, two inputs of opposite phase, which could also be listed as 100Vrms differential drive.

The headphone element has two fixed elements to which the audio signals are applied. Between these is a metalised film of thin plastic that has the fixed voltage applied to it. As the signal voltages move positive and negative, the positive-charged middle element is pulled towards one stator while being pushed by the other, then drawn and pushed to the opposite stator over the audio signal cycle. Since there is electrostatic force applied evenly over the movable film, it can move very quickly and as a rigid plane with very low distortion.

The stators appear capacitive to the amplifier, typically around 100pF or so each. This is not a particularly difficult load except that the charging and discharging currents can be high if there is no current limiting during fast transitions.

Stax offered many types of ear speaker drivers ranging from passive to exotic, setting the standard for others like Koch to follow. The passive driver uses a step-up transformer to raise the voltage needed for 8R speakers up to the requirement for the electrostatic panels. The secondary of the transformer is center-tapped, with the CT grounded, creating the symmetric anti-phase drive signals. A small transformer provides mains isolation and feeds a voltage multiplier to generate the bias voltage. Compared to the distortion room loudspeakers add, the distortion of the transformer and the very-low-THD of the ear speakers combined is much lower, and usually lower than a conventional dynamic headphone driven through the usual 100R dropping resistor from the amplifier output.

Dedicated headphone amplifiers for ES were introduced which allowed the user to turn off his large amp and not waste all that power, driven by the system preamplifier / source selector. The dedicated driver is always made as a differential amplifier from input to output, with dual feedback loops. Internal voltages are typically +/-400Vdc, providing an incredible signal swing approaching 280Vrms x2. However, considering that a combined 100Vrms produces 100dB of output, these rails would allow 5.6 times the total voltage swing, assuming no losses, which corresponds to only 15dB more sound. We are already way above the Human Scale of Loudness, so all this headroom gets us is freedom from clipping on signal peaks. We would also expect there to be better linearity when not using so much of the available swing.

The first diff-amp offerings had open-collector outputs, where there is active pull-down and passive pull-up. This was because the highest-voltage semiconductors were NPN. Even with this, cascoding was used throughout to distribute voltage stress and to improve high-frequency response. The next iteration added buffering with current-source loading for active up/down load control. This reduced THD slightly. A further variant was like the first with the high-voltage output transistors replaced with a dual-triode tube.

Stax offered a lower-output battery-powered driver that unfortunately used a switch-mode power supply. The hash from the SMPS was all over the output and the THD was poor and fidelity low. On a good analogue scope or any DSO, the SMPS noise at the output is terrifyingly ugly.

Personally, I have built drivers for Stax headphones using linear +/-100V rails (same as the Stax unit above) and have zero issue with clipping. THD is unmeasurable. The topology is a typical form using only inverting gain blocks and fully-symmetric circuits. Now that we have highly complementary semiconductor pairs, circuits like this are very common. As an experiment, I switched to 40V rails and still had no evidence of clipping when listening at sane loudness levels.

Many headphone enthusiasts have designed all-tube, hybrid and solid-state drive circuits for ES cups and report their findings. Some of the listening tests seem a bit scary to me, when they have super-high-voltage drive circuits and can hear clipping !

Forum Jump:

Come in where it's warm!
A warm welcome to tube amp modding fans and those interested in hi-fi audio! Readers of Kevin O'Connor's The Ultimate Tone (TUT) book series form a part of our population. Kevin O'Connor is the creator of the popular Power Scaling methodology for amplifiers.
Please remember these three principles: respect, sharing, community.
Not familiar with The Ultimate Tone book series? See discussion topics, or click here to visit London Power/Power Press Publishing.

Tube Amp Forum Hosted by London Power
London Power logo