Angela-Gilbert Yeung

I am of the belief that a big power supply is a key factor in the design of audio equipment. By big, I mean large amounts of capacitance. The job of a power supply is to be able to send sufficient amounts of energy instantly to the component parts of apiece of audio equipment to maintain accurate reproduction of a complex audio signal in real time.

Historically speaking, not a lot of audio designs have made use of large power supplies. In the 1970s, only a very few companies attempted to build relatively large power supplies into amplifiers. At that time, capacitor technology was not nearly as advanced as it is now, and capacitors were 5 to 10 times larger for a given capacitance and voltage. Space and cost were very much an issue. Today, it is much easier to build a robust power supply into a relatively small enclosure.

Way back when, most companies that were building what they considered to be a large power supply claimed that the benefit was mostly for low frequencies. This is a common misconception. All frequency ranges can– and will – benefit from a well designed and properly built power supply.

Capacitance– the more, the merrier…

We use relatively – sometimes hilariously - large capacitance values in the power supplies designed for our components. It makes a difference. The large banks of capacitors make it possible to store energy to meet the demands of any musical requirement.

The accurate reproduction of a musical waveform is a complex task. The initial attack of a voice or instrument is where we hear about 85% of the information that tells us what particular voice we are hearing. Notes start and stop. Other instruments play simultaneously. If we are to differentiate between the sound of a saxophone or a voice, an oboe or a violin, a piano from a xylophone, if we are to hear when notes stop, if we are to experience both large and small scale dynamics, in short – if we are to perceive all the wonderful elements that music contains, we must have sufficient energy being constantly and quickly delivered to the signal components in the audio chain. The absolute best way to achieve this is with capacitors.

The automotive analogy…

Not everybody is comfortable with audio terminology or concepts, but most people are familiar enough with cars that I can use some terms to help them understand some key elements of power supply design. Like any analogy it’s not perfect, but it does provide some insight. As I see it:

Voltage= RPM

Wattage= Horsepower

Capacitance= Torque

Too many power supplies are designed to deliver high voltage. Yes, thepower ratings look impressive on paper, but they are lacking in practical terms. You end up with loads of potential power that gets wasted in the drive chain and cannot be delivered to the car wheels.

There are also many amplifiers with high wattage ratings. But we have to consider those ratings carefully because they provide only a very small description of the performance of an amplifier. Essentially, wattage describes how loud an amplifier can play a single frequency at a fixed resistance. Basically, it’s like describing how fast a car can go on a smooth surface with no turns, changes in elevation, and no headwind. Unfortunately, this is nothing like musical signal reproduction.

Reproducing a musical signal accurately is more like off-road driving. The “road surface” (ie. musical signal) is constantly changing, the complexity of the waveform is constantly changing, and the speaker resistance (uphill and downhill elevation) is constantly changing. High RPM will not necessarily help, horsepower will not be of much use, but torque (applied force) is very much necessary to negotiate the uneven and unpredictable surface. Capacitance in the right quantities and values provides torque.

Not just any transformers…

We use transformers to our exact specifications. In a perfect world, there would be no voltage difference in a transformer running at no-load and a transformer running at full load. Of course, it’s not a perfect world – and it is not uncommon in some audio circuits to measure a voltage swing of about 20%. This is why many manufacturers use voltage regulators in their power supply designs. We do not, and a few paragraphs from now you’ll find out why.

Our transformers are designed to have a voltage swing of less than 2%from no load to 3/4 load. We never allow our transformers to run at more than a 3/4 load – that would be like running a car with the gas pedal constantly on the floor, which usually ends badly. If we use a common example from equipment that we design, a transformer with a 23.5 volt AC output under no load will drop to no lower than23 volts under a 3/4 load. It makes for a very stable operating voltage for the signal devices involved, and no regulator is required.

An extra side benefit is that since the voltage swing can be kept to a minimum, lower voltage capacitors with relatively high capacitance values can be used, which decreases cost and space requirements, and increases the amount of capacitance we can use for any given budget and space requirement.

Charge times, rising edges, and why we don’t use voltage regulators…

In our power supplies, we have a 120 Hz wave coming from the transformer, which charges the capacitor bank every 8 milliseconds. The capacitor bank has to be able to hold a charge long enough to sustain a continuous series of rising edges until being charged again by the transformer. A rising edge frequency of 1 kHz passes in 1/3 of millisecond, so the capacitor bank in the power supply needs to be able to reproduce more than 24 rising edges of a 1 kHz signal. Real music places a much more complicated load on a power supply than a simple 1 kHz signal, so you can begin to understand the demands that a complicated musical signal can place on the power supply.

Can the same result be achieved by using less capacitors and a voltage regulator? Yes, but you pay a penalty. A voltage regulator samples the output of the regulator and feedback to the input in order to adjust to the voltage differential and maintain a constant output voltage. This takes time – time to sample, time to feedback, and time to adjust. By the time all of these things happen the output signal has already passed and a new adjustment is required. This delay only takes milliseconds, but in those milliseconds part of the initial waveform is lost, lessening the accuracy of the musical content.

A second disadvantage of using a regulator is that it wastes voltage from the transformer. The regulator has to drop voltage in order to maintain constant voltage at the output. In doing this it generates heat and wastes energy. Using high capacitance avoids all the inherent problems with time delay and energy loss that a regulator creates. There is always plenty of available power that can be delivered instantly.

Getting the energy from here to there – THE MOST IMPORTANT ELEMENT…

In terms of food preparation, if ten people started with a recipe, the resulting dish would undoubtedly taste a little different in every case. In cooking, the chef makes a huge difference… in many cases, ALL the difference. The same applies to audio. Too often we forget who the gear is being built by, or how it is being built. We go crazy over a name brand capacitor, and never stop to consider how the capacitor is being implemented, or how well it is being connected to the circuit. It is possibly the most important thing to consider in audio and is generally completely disregarded.

Moving back to audio… even if we have properly designed the power supply that has all the energy we need to adequately power the signal devices, it’s not much use unless we construct that power supply in a way that will transfer the energy as quickly as possible to where it is needed. The industry standard is to use PC boards. While PC boards look neat and tidy, the thin traces and small solder joints definitely do not benefit the transfer of energy. By far, the best way to build a power supply for maximum energy transfer is to solder thick copper wire directly onto the capacitor taps. When you consider the thickness of the copper wire as opposed to PC board traces, copper wire has somewhere between ten to twenty times more capacity to conduct electrons than a PC board trace.

Some would argue that using thick wire, ring terminals, and screw mount capacitors is equally effective, but this is not the case. The total contact surface area between the ring terminal and the screw mount isa mechanical contact. No matter how flat the surfaces are, there will be tiny gaps. You could help alleviate this issue by filling up the gaps with solder and pushing all the air out of the contacts, but it would take much more time and labor to construct, and since screw terminal capacitors are not meant to be soldered they would likely be damaged by the heat involved in soldering. Additionally, it’s simply not as effective a solution as just using directly soldered copper wire.

Another argument against soldering would be that solder is less able to conduct electrons than copper, silver and gold. This is absolutely true. For this reason, we make sure that every solder joint is bigger than the wire to ensure that so that the solder joint is capable of Maximum conductivity. The solder joints are as big as I can make them without making multi layered cold joints – cold solder joints are terrible for energy transfer and are not tolerated. Using large solder joints ensures that the ability to efficiently transfer energy is maintained.

To  Summarize…

There is a lot of information here to take in, but there relatively few key ideas to keep in mind. First and foremost, a well-built and properly designed power supply is a key element in producing the best musical results in a piece of audio equipment. I achieve these results by using transformers designed to tight tolerances – important, using much larger amounts of capacitance than normally would be found of in a component – very important, and most important of all – hand built construction using large copper wire conductors and advanced soldering techniques.