Gadget Freak

The best companion to your music collection

Everyone loves valve hi-fi equipment. A fan of mixing the old with the new, Andrew Smith designed a very high performance valve headphone amplifier. Using a unique combination of thermionic and semiconductor technology, the headphone amp runs from a regulated 15V, 1A wall-wart supply for a complete, no-compromise power system. The headphone amp also includes a DC-DC converter providing the required 175V at 15mA for two channels which shuts down gracefully if the output current reaches 38mA, so there’ll be no chance of a burn-up while you’re burning up the dancefloor at home. So now you can take a step back in time a little while listening to everything from Beethoven to Britney.

Headphone Amplifier

View headphone amp circuit schematic

People love valve hi-fi equipment. Somehow the warm glow of the tubes and the nostalgia of past times never fails to enhance our enjoyment of good music; it is something like vintage wine by candlelight. So here is a design for a very high performance valve headphone amplifier which will find its place in any discerning audiophile system.

This amplifier is unorthodox in that it mixes thermionic and semiconductor technology. An EF184 pentode is used to provide voltage gain (x5) and a warm glow, while a Darlington emitter-follower provides a low impedance output. An active emitter load is used to improve linearity. The EF184 is a very low-noise device (it has a much lower equivalent noise resistance than most triodes), with a high and unusually linear transconductance; it is very suitable as an audiophile tube.

The design is capable of driving relatively insensitive headphones with a voltage of 700mV rms and a total harmonic distortion of 0.03% or better. It performs well with my Sennheiser HD25SP and AKG 701 phones, both of which combine low sensitivity with relatively low impedance, so the amplifier is likely to be suitable for most other high quality phones, many of which will be less demanding. On listening tests the amplifier gives the impression of the sonic equivalent of clear, bright spring water - transparent, detailed, refreshing. The only defect in performance that I have noticed is slight microphony from the valves. Some might regard this as an attractive glow-bug characteristic, but in normal use it cannot be heard at all.

The amplifier runs from a regulated 15v, 1A 'wall-wart' supply, and includes a dc-dc converter to provide the required 175v at 15mA (for two channels). The converter uses a Mullard LA1 Ferroxcube transformer core recovered from the junkbox. The converter has high efficiency (~85%), and the transistors run happily without a heatsink: the switching frequency is 25kHz, so no audible noise is contributed from the HT supply. The converter is housed in the same wooden box as the other components. The box is lined internally with self-adhesive copper foil.

Specifications

Output: 700mVrms to 85 Ohms (2.9mW)

Bandwidth: 15Hz – 25kHz @ -3dB

Voltage Gain: 5

THD: <0.03%

Power requirement: 12Vdc @ 800mA (two channels)

Input impedance: 47k Ohms

Output impedance: 10 Ohms

Circuit description

The valve stage has a gain of 5, and includes local cathode-degeneration feedback: there is no overall feedback which might compromise the transient response. The bias point of the EF184 is set by P1, which is adjusted to provide static anode current of 6.5mA: the bias is adjustable in order to allow precise setting of the anode current in relation to the characteristics of individual valves, ageing, etc.

The output stage uses a Darlington follower (T1, T2) with an active current source as emitter load (T3, T4). The dc bias voltage at the emitter of T2 is set to 8.0 volts by P2, while the bias current is set to 100mA by R10. The total dissipation of T2 and T3 is 1.2W, so a small common heatsink must be used. The semiconductors used in the output stage are not at all critical, and other similar components could be used, for example an IRL540 would be suitable for T3.

The valve heaters are connected in series to the nominal 12v supply. Ideally this should be 12.6v, which can be achieved by connecting a silicon diode in series with the common connection of a 7812 regulator chip.

A spectrum is shown below with the amplifier driving 1vpk into the Sennheiser phones (a very loud signal). Second and 3rd harmonic spikes are at the -83dB level, corresponding to about 0.01% THD; 10dB below the specification level.

This level of distortion can only be attained if the bias points of the valves are adjusted under test, but setting the bias current to 6.5mA will be a close approximation to the optimum condition. The setting of the static output voltage (P2) and the exact value of R9 also influence the linearity of the amplifier, so these can also be tweaked. Valve-orientated audiophiles might be interested to note that the amplifier can be set up to provide about 1 per cent 2nd harmonic distortion, should that type of performance be required.

Power system

The amplifier has two on-board 7812 regulators. One, fitted with a diode in the common lead, supplies 12.6v to the heaters and the output stage; the other supplies 12v to the inverter. The system is powered from an off-board 15v supply which is also regulated. This might appear to be over-kill, but the result is a completely no-compromise power system. Using a separate local regulator for the inverter is a very good way of ensuring that the on-board inverter does not cause any electromagnetic compatibility problems to other equipment.

170v DC/DC Inverter. This is a conventional saturating-core converter using a LA1 ferroxcube core. The efficiency of this converter is around 80%, depending on load conditions. The converter stops oscillating and shuts down gracefully if the output current reaches 38mA, so there is no chance of a burn-up if the circuit is overloaded. P1 can be set to minimum resistance or omitted, but it can be adjusted to optimise efficiency under different load conditions. It is not a critical adjustment.

HT Regulator. When working with an amplifier as good as this, it is difficult not to become obsessive about the fine details. Regulating the HT supply confers two minor benefits on the system, both related to the relatively poor power-supply noise rejection of the valve amplifier. Firstly, regulation improves the channel separation at low frequencies by lowering the output impedance of the HT power supply, and thereby reducing the cross-coupling between channels which occurs through this path. Secondly it reduces the effect of any residual noise on the 12v supply. Since the HT inverter acts as a transformer with a step up ratio of 14, noise on the 12v line can become surprisingly noticeable unless the 12v regulator is particularly effective. However, both considerations are of minor importance, so this section is strictly for the purists.

The HT regulator uses a GD150M/S gas discharge regulator which adds a third tube to the glassware on display. The current in the regulator tube is set to 5mA by the constant current circuit connected in its anode: this uses a high voltage P-channel FET (T1, IRF9610). Sadly, the mysterious purple glow which these tubes produce is not very evident at this current level.

Two high-voltage transistors - T3, T4, both 2SC2688 are used in a Darlington follower circuit to reduce the output impedance. Although there is no short-circuit protection, the limited current capability of the converter makes this circuit safe against accidents.

Power dissipation in T4 is in the region of 400mW in normal operation, higher during short-circuit conditions, so a heatsink is advisable. D1 is not used as a voltage reference - its function is to protect T1 from excessive gate-source voltage due to any accidental malfunction. The choice of device for T2 is not critical: any small PNP transistor can be used.

RS Parts list - Amplifier (One channel)

Regulator

Converter