by Simon Busbridge
I recently purchased the Sennheiser HD600 headphones second-hand and realised that their full potential could only be met with the highest quality amplifier. Hence this project! Before starting there were certain criteria I wanted to satisfy:
- no feedback
- low output impedance
- no possibility of nasty DC appearing at the output, which could damage the headphones
- safe operation
Output transformerless designs either use direct coupling or capacitor coupling. In the former case the possibility of DC appearing at the output is very real, either at switch on or when plugging the ‘phones in and out (when the common connection in the plug can short the right channel jack output). With capacitor coupling it is usual to use a delayed muting switch to prevent switch on and off “thumps”. Generally speaking, there are no turn on and off thumps on transformer-coupled amps because the current in the output valve tends to build up and decay slowly, a fade-in and fade-away type of effect. Muting circuits on other designs are fine, but a transformer does offer the ultimate in terms of protection (damage to the headphones and electric shock).
So that really decided the form of the output – it had to a transformer. To get a low output impedance I needed to use quite a high step-down ratio (20:1); after all, the amplifier may be used with headphones of lower impedance that the 300 Ohms of the HD600. The output valve is the 12B4A which has a anode impedance of about 1000 Ohms, so the output impedance of the amplifier is about 2.5 Ohms, which is very low for a valve amplifier. You can use other valves here but watch that the anode impedance is not too high or the output impedance and bass-cut off frequency will both rise.
It is usual with power amplifiers to match the impedance of the loudspeaker to the impedance of the anode circuit using the equation turns ratio = square root (impedance ratio). Under these conditions, maximum power transfer takes place from the amplifier to the load. With headphones, the power involved is so small that it is not necessary to operate under such conditions. It is possible to make the turns ratio such that the output impedance of the amplifier is much smaller than the load. Power transfer is not optimum, but the output valve is tending to run under constant current conditions which has the advantage of lowering distortion.
The output power will depend on the impedance of the headphones; the amp gives about 10 V output before clipping. The gain of the output stage is 0.33, so quite a bit of amplification is needed before it. I am a fan of SRPP because it improves linearity, lowers output impedance and allows for a large voltage swing. The first stage uses a double triode directly coupled to the output valve, giving an overall gain of about 20 dB. The circuit is shown in figure 1. The 12B4A cathode resistor tends to run quite hot so use a 7 W component here.
There is no feedback from the output to the input. The gain can be increased slightly by decoupling the cathode of the SRPP input, or alternatively try a ni-cad battery bias. Although I use the 6072A for this stage, suitable alternatives are the 5965 and 12AT7. The 12AT7 is slightly less linear. The 6922 can also be used with a change to the heater wiring but the gain will be lower.
The 2.2 K Ohms resistor on the output of the audio transformers is there to give the output valve some load if the headphones are disconnected. Without it, the transformer can ring off load. The output impedance of the amp is so low that its value is quite uncritical.
The Power Supply
The power supply, shown in figure 3, is fairly conventional, with valve rectifier and choke input filter for quiet operation. An extra stage of filtering removes any residual ripple and helps to separate the channels. All sorts of different power supplies can be used, but it needs to be capable of 60 mA at 200V. The 200V HT voltage is not critical. DC can be used on the heaters, but it is not really necessary as the gain is fairly low. Solid state power supplies can be used. They just don’t sound as good!
With my Sennheiser HD600, the amp will clip if you try to go too loud – 100 dB or so. I actually like it that way, because it is very easy to damage your hearing with headphones without realising it. There is, therefore, an in-built tendency to back off the volume control slightly. To increase the drive capability, raise the HT supply by 100V or so. This can be done by either changing the transformer secondary winding, or by placing a smoothing capacitor directly after the rectifier (figure 3).
In the former case the power supply remains choke input, but watch the maximum rectifier voltage or the valve rating may be exceeded, because you lose quite a few volts with this method. In the latter case, it becomes capacitor input. With a choke input filter, the output voltage of the supply is 2/p Vp, where Vp is the peak output voltage. With a capacitor input filter, the output voltage is just Vp or about 300VDC. I cannot tell any difference in the sound quality – take your pick!
The complete system is built on two separate chassis to avoid magnetic field induced hum (I recommend that you ask for a flux-band and earth-screen to be fitted to your transformer). If it is kept on two chasses (amp on one, psu on the other), then component placement is quite uncritical. The chassis frames measure 9″ x 10″ x 4″. They are assembled from hardwood boards (deal/pine wood) with mitred corners. I used a mitre saw to make the cuts, which was not easy. A circular saw set at 45 degress is how I do it now. The joins are with the small brackets screwed through from the inside.
The top plate 1.6 mm thick sheet metal, drilled or punched (with Q-max punches) in the right places, and then pearl anodised at a local anodising factory. On the chassis for the power supply, the top plate has strengthening bar across the middle; otherwise the top tends to sag. The wooden parts can be spray painted black and then polished for good effect. The finish is excellent as the grain of the wood shows through.
The mains switch is a rotary type. The shaft is a bit of old 1/4-inch brass rod fixed on the switch with a standard spindle coupler and held in the front panel by a bush, again standard off the shelf. The method is cheap and works well. The switch itself is mounted on a small bracket, just a scrap of metal bent to 90 degrees with the necessary holes in it. For T1 in the amplifier, I used 20:1 line output transformers (part no. pa106 from SJS Electroacoustics). The power supply transformer and choke were custom-made by Sowter of Ipswich. I paid 165.59 UKP for the two.
The lamp is an integral diode, resistor and LED in a chrome mounting. You might wish to consider fitting a fuse at the input. Some people say that they can hear fuses! Here in the UK, I use a 2A fuse. In the USA, the current will be higher so I suggest a 5A fuse. A 5A fuse is also OK for the UK. All UK mains plugs are fused anyway (unlike the US) with a maximum of 13A.
The input phono jacks are located round the back, along with the umbilical between power supply unit and the amplifier. The Switchcraft headphone jack is on the front (left) next to the volume control (right). The plan was to have a gold-plated Switchcraft jack, which I saw at an AES show, but I could not get hold of one, so it is a standard tin-plated Switchcraft.
I have only given the power rating of resistors which dissipate any appreciable power. For all others, 0.5 W resistors will do. The capacitors came from all over the place. The 1600uF big one was second hand in a junk shop; the others from HAM fests.
The amplifier should work with headphone impedances as low as 32 ohms; I have tried lower impedance Sennheisers and they seem OK. I would be worried at less than 10 Ohms or so. The sound quality of the amplifier is excellent. The lack of feedback produces an open natural, detailed, sound without any of the harshness sometimes experienced with solid state amplifiers. I can now listen in bliss well into the night when other members of the household have gone to sleep! This is definitely one of my best projects.
All of my hi-fi equipment is built along the same lines as this headphone amp. I plan to get around to designing feet which go over the corners of the boxes from top to bottom to hide the mitre joins. The top will have a nice domed finish and the bottom adjustable spikes. But like all things, it is the time…
Dr. Busbridge is a lecturer in the School of Engineering at the University of Brighton (U.K.).
6/7/99: Added paragraph discussing output transformer impedance when driving headphones.
3/25/01: Various revisions to article text. Added high-res images. The rectifier tube in the power supply (figure 2) was changed to a 5U4G. The GZ3 previously shown can also be used.
3/27/01: Changed value of first filter capacitor in figure 2 from 100uF to 1600uF.
c. 1999, 2000, 2001 Simon Busbridge.