Exploring Whether Power Supply Type Affects Audio Quality

Among audio enthusiasts, there’s a common belief that switching power supplies are unsuitable for use in high-quality audio equipment. But is that really the case? Let’s find out! We’ll take a single amplifier and power it first with a switching power supply, then with a “classic” toroidal transformer. We’ll measure sound quality parameters — including distortion levels and output power — and draw some conclusions.



A Brief Historical Background

The transformer was invented back in the 19th century and became widespread in a rather “natural” way — simply because there was nothing more suitable at the time.

Switching power supplies were developed in the late 1960s, nearly a hundred years after the invention of the transformer. Their emergence became possible thanks to significant advances in the semiconductor industry.



A Brief Theory

In a power supply, the mains voltage is stepped down by a transformer, rectified by a rectifier, and then smoothed by a capacitor. The 50 Hz frequency of the AC mains requires the use of large and heavy iron-core transformers (lower frequencies demand higher magnetic flux density). Switching power supplies operate on a different principle: they convert the mains voltage into a current at a frequency roughly 1,000 times higher — around 50–80 kHz. This makes it possible to use magnetic cores made of efficient ferrite materials, fewer wire turns for windings, and smoothing capacitors of lower capacitance. As a result, the power supply becomes radically smaller and lighter — while delivering the same output power. And that’s not the only advantage.



Testing and Measurement Methodology

To place maximum load on the amplifier — and to spare our eardrums — we’ll be using the following device as a load on the amplifier’s outputs:

Fig. 1 – Load unit for amplifier testing

This is a high-power resistor block with 1Ω resistors, used to create 4Ω or 8Ω loads.

We’ll conduct three tests:

1. An analysis of power supply behavior under a sine-wave load applied to the amplifier,

2. Measurement of the amplifier’s technical specifications, and

3. A subjective listening comparison, alternating between the switching and transformer-based power supplies connected to the amplifier.

Test Subjects

Two Meanwell EPP-200-27 switching power supplies.

We’re using two because the amplifier requires a dual-polarity ±30V power supply, while each unit provides only a single-polarity output. The nominal power rating of each unit is 200W, giving a total of 400W. The combined weight of the two power supplies is 380 grams.

Fig. 2 – Switching power supplies

It’s worth noting that the switching power supplies we used are of reasonably high quality.

We deliberately avoided using generic no-name units from AliExpress, as it’s unlikely that serious audio manufacturers rely on second-rate components. As for the transformer-based power supply, it’s built around a TTP400 transformer (2×25V, 7.5A) with a nominal power rating of 400W, two Schottky diode bridge rectifiers, and a bank of smoothing capacitors — 40,000 µF per rail.

This is a classic power supply design that has been used for decades and is still relevant today. The weight of the transformer alone, not including the rest of the components, is 4.1 kilograms.



Fig. 3 – Transformer-based power supply



Part 1. Load Testing

To compare the performance of the power supplies, we’ll drive the amplifier with a 1 kHz sine wave and draw 350W of output power. This load level corresponds to an output voltage of 34.7 volts from the amplifier. Testing will be done both under load and without load — in idle mode.

We’ll start with the switching power supplies.Their output voltage in idle mode is:



Fig. 4 – Output voltage of the switching power supplies in idle mode



Output voltage under load:



Fig. 5 – Output voltage of the switching power supplies under load



As you can see, the voltage dropped by 84 millivolts, which amounts to 0.28% of the original value.

We’ll now test the transformer-based power supply using the same method:

Fig. 6 – Output voltage of the transformer-based power supply in idle mode

Fig. 7 – Output voltage of the transformer-based power supply under load



As we can see, under load the output voltage drops by 5.754 volts — which is 19.1% of the original value. That’s 68 times more than with the switching power supplies!

Why does this happen?

Let’s take a look at the waveform of the mains voltage and the current drawn from the grid using an oscilloscope. First up — the switching power supply:



Fig. 8 – Mains voltage and current waveform of the switching power supply (Yellow – voltage, Blue – current)



The yellow curve shows the mains voltage, and the blue one shows the current draw. As you can see, the current waveform roughly follows the voltage — energy is consumed evenly, in sync with the incoming power.

Now let’s take a look at the same measurements for the transformer-based power supply:



Fig. 9 – Mains voltage and current waveform of the transformer-based power supply (Yellow – voltage, Blue – current)



The current waveform only partially follows the voltage waveform.

Energy is consumed for roughly half the duration of the incoming power cycle; during the rest of the time, no current flows because the smoothing capacitor only charges when the rectified voltage from the transformer exceeds the voltage across the capacitor. For this reason, the instantaneous power drawn by the transformer from the mains is significantly higher than the average power it delivers to the load (While the amplifier consumes energy continuously, the transformer draws it from the grid only during part of the cycle — meaning it has to draw it twice as fast.)

Now let’s look at the transformer’s waveforms with and without load. The yellow line shows the primary (mains) voltage; the blue line shows the secondary (output) voltage.



Fig. 10 – Transformer winding voltages in idle mode (Yellow – primary, Blue – secondary)



Fig. 11 – Transformer winding voltages under load (Yellow – primary, Blue – secondary)



As you can see, the mains voltage waveform remains unchanged, while the secondary winding voltage gets clipped at the peaks — precisely at the moments when current is being drawn:



Fig. 12 – Transformer-based power supply: Input voltage (yellow), output voltage (red), and current draw (blue) under load



The reason for this is energy loss due to heating of the transformer windings, which are forced to deliver pulsed current significantly higher than their nominal rating. As a result, a 400W-rated transformer is only able to supply about 350W. Attempting to draw more leads to noticeable voltage sag, which in turn causes audible distortion. To reduce this sag, the transformer must be oversized — typically by 1.5 to 2 times the actual power requirement — which means greater size and weight.

In contrast, switching power supplies with a nominal rating of 400W can deliver the full 400W without voltage sag. This is because they address the issue of uneven energy consumption using an electronic circuit known as a Power Factor Correction (PFC) module. According to regulations, all power supplies over 100W must include a PFC circuit. Of course, in practice that’s not always the case — there are many “cheap Chinese” variants without PFC or proper protection circuits, which can go up in smoke at the slightest provocation. Such units should simply be avoided.

Now let’s check how clean the power delivered to the amplifier actually is. Here’s the ripple waveform from the transformer-based power supply under load:



Fig. 13 – Voltage ripple of the transformer-based power supply under load



We observe ripple at twice the mains frequency — 100 Hz — with an amplitude of 800 mV, which is 2.6% of the total output voltage. To reduce the ripple amplitude, the capacitance of the smoothing capacitor must be increased. However, this shortens the current conduction period and requires the transformer to be even more powerful.

Now let’s look at the ripple from the switching power supply:



Fig. 14 – Voltage ripple of the switching power supply under load



The ripple amplitude of the switching power supply does not exceed 100 mV — that’s eight times lower than the ripple of the transformer-based supply. Moreover, the ripple occurs at a higher frequency and is more effectively filtered out by the smoothing capacitors.



Part 2. Measuring the Impact of Power Supply Type on Sound Quality

Now let’s see whether the type of power supply affects sound quality. To do this, we’ll measure the amplifier’s performance using both types of power supplies.

For measurements, we’re using a test system based on the Lynx Studio E22 professional audio interface and RightMark Audio Analyzer 6.4.5 PRO software.

During sound quality testing, we’ll set the output power to 60 watts per channel into a 4-ohm load. To achieve this, we feed a 1 kHz test signal from the measurement computer to the amplifier input and adjust the volume so that the output voltage reaches 15.5 volts

(based on the formula: U = √(P × R) → U = √(60 × 4) = 15.49 V).

We run the tests with both power supplies and obtain the following results:



Fig. 15 – Sound quality test results with switching and transformer-based power supplies



As we can see, there are no significant differences, but the amplifier’s performance is slightly better when powered by the switching supply. This is due to the voltage drop we observed earlier with the transformer-based power supply under load.



Part 3. Subjective Listening Test

Now for the most interesting part — building a button that, when pressed, switches the amplifier to one power supply, and when released, switches it to the other. This allows us to compare the two in real time, with actual music playing, and hear whether there’s any audible difference.

To switch between power supplies at the push of a button, we’ll use the following homemade devices — transistor-based power switches:



Fig. 16 – Transistor power switches for on-the-fly power supply switching



Here’s how our full test setup looks once it’s prepared for the subjective listening test:

Fig. 17 – Complete test setup



We listened to several different tracks, switching between power supplies during playback, and were unable to detect any audible difference in sound quality. This confirms the measurements and observations we obtained earlier.



Conclusion

Our research shows that the type of power supply used does not affect sound quality. The use of compact and lightweight switching power supplies by manufacturers is therefore completely justified — especially considering that SMPS units typically support both 110V and 220V mains input by default, whereas transformers require a tapped primary winding and a voltage selector switch.

That said, we must emphasize once again: this conclusion applies only to high-quality power supplies — ones where no corners have been cut in critical functional circuits.



Fig. 18 – The main components of the test, side by side

Thank you for reading!

Grigoriy Mozharovskiy and Roman Romashchenko

2020.05.20