Whilst many modern sources provide enough voltage to feed a reasonable signal to amplifiers, some, such as the Pass ACA, may benefit from a boost in gain, particularly if paired with inefficient speakers. As I had most of the parts, I thought I would try a variation of the CMoy headphone amplifier as a preamplifier and volume control.
As with the other CMoy projects I’ve completed, I wanted to incorporate some changes and learn a little bit more along the way. This build is a simple modification of the schematic as provided by the Tangentsoft website. It incorporates some of the tweaks outlined there, with perhaps the greatest change being to that of the virtual ground element of the circuit. The original CMoy creates a split voltage to power the op-amp by the use of a resistor-divider as shown in the schematic below. By using two resistors of equal value, and taking a wire from the midpoint, a 12V supply can be divided into +6V and -6V.
Tangent notes some problems with this arrangement, in particular the propensity for the divider to become unbalanced. This means that the rails may have voltages of +7.5V and -4.5V for example. The values may vary according to factors such as the type of headphones, op-amps used and resistor values. Whilst this imbalance may not particularly harm the sound, with low voltage power supplies the amplifier can be driven to clipping more readily in the channel with the low voltage. Increasing the power supply voltage can be one solution, whilst that used here is the replacement of the resistor divider with a precision virtual ground or “rail-splitter”. This splits a single supply into two rails and a ground and essentially replaces the resistors in the standard voltage divider circuit, whilst providing some internal buffering to keep the split stable. Texas Instruments provide the TLE2426, which comes in a 3 pin package or as an 8 pin IC. The version shown in the schematic below incorporates noise reduction through pin 8. Voltage is supplied to pins 2 and 3 and the ground is supplied from pin 1. Note also that only one capacitor is required before the IC.
The cost of using this precision virtual ground is minimal – the IC costs £2.14 from RS Components, although minimum quantity is five units but postage is free. The amended circuit with the original values is shown below.
I used the OPA2132 op-amp in the previous amplifiers as Tangent suggests this is the easiest to work with. However, this is relatively expensive and so for this build I decided to have a try with the OPA2134PA (£3.12 from CPC compared to £6.19 for the OPA2132PA which negates the etxra cost of the TLE2426).
Whilst looking at the datasheet for the OPA2134, I noticed in the layout guidelines that it is recommended to use a bypass capacitor in-between the IC’s power pins and ground. I searched around to find why this had been omitted by Chu Moy but couldn’t readily find an explanation. It clicked one day and I think it is because the original CMoy was powered by battery and therefore no ac should be present, but the inclusion of a power supply merits the addition of these bypass capacitors, as they will shunt any AC to ground before reaching the op-amp. Reading Tangent’s articles on op-amps provides some further insight in to why you might want to use these caps. Reading between the lines, I get the impression that the OPA2132/2134 ICs are relatively benign and won’t ‘misbehave’ without these caps. The datasheet calls for a 0.01µF low ESR ceramic capacitor, which is what I have used, but Tangent suggests a quality film capacitor (0.01-0.1µF) would be suitable as well – trading some of the ceramics efficiency at high frequencies for greater linearity. The leg of this capacitor should be placed as close as possible to the IC power supply pin.
The other changes to the original circuit are simply component values as suggested by Tangent. I have increased the power supply capacitor to 470uF, larger values can improve the bass. The high pass filter R2-C2 at the input to the op-amp blocks DC current and, according to Chu Moy, has a corner frequency of c. 15Hz with the standard values of R2=100Ω, C2=0.1µF. Using a 1µF capacitor lowers the corner frequency to 1.5Hz and again, this may improve the bass. These caps are physically larger than the standard recommendations and may be more difficult to accommodate in smaller enclosures, but this is not an issue for this build. The revised schematic, drawn in Tiny CAD, is shown below.
The final changes are the use of RCA connectors for source inputs and outputs to the amp, and an Alps RK27 “Blue Velvet” volume potentiometer. To make the connections easier, in the past I have soldered the pot to prototype board. Whilst this was easier to some degree than soldering wires straight onto the pot pins, I decided to look for a better solution and came across some cheap PCBs that accept three sizes of Alps potentiometers. I bought five of these for £3.40 including postage from China. As this pot does not incorporate a power switch, I have included a small switch to the rear panel.
The circuit was constructed on stripboard and I created a layout using DIY Layout Creator, the result of which is shown below.
I powered the circuit with a good 24V adaptor that came as a spare with my Graham Hill Solo amp. The preamp connects to the ACA and it works very well. There is no noise from the CMoy being passed to the amp that I can hear, even with my ears right next to the speakers. I now have a nice cheap preamp, with a good volume control in a cheap yet decent chassis.