Example SMPS technical measurements

Example SMPS technical measurements

Switch Mode Power Supplies (SMPS) are ubiquitous in use with lower cost audio components. The one notable exception is Schiit which employs transformer wall warts and linear power supplies internal to most of their smaller audio components. How they differ in ripple and residual noise will be the focus of this comparative evaluation.

This post will cover select measurements for a few example SMPS. All will be 5Vdc output unless otherwise stated.

Post 1: Test setup and measurement limitations
Post 2: Soekris SMPS, iFi Power 2 and X, Meanwell GST25A05
Post 3: Several USB chargers
Post 4: dac1221 specific residual noise measurements
Post 4: dac1221 current draw vs. musical passages



On the left is the SMPS supplied with Soekris dac1221
Middle is an iFi Power 2
On the right is a commonly used Meanwell GST25A05



A test load designed for SMPS noise measurements will be utilized.
This test load draws 25 mA at 5Vdc.
Test load components:
200R 50W resistor
47uF 63V electrolytic capacitor
1uF 275V film capacitor



Test setup diagram
Oscilloscope is Picotech 5243B
AC coupling
Measurement bandwidth 10 Hz to 4.808 MHz

Measurement system noise floor without and with test load


Oscilloscope time domain residual noise with 75R terminator on input.
Measurement noise floor is 140 uVpp
This is the lowest this oscilloscope can resolve with 5 MHz bandwidth.


Oscilloscope time domain residual noise with test load connected to input.
Noise measurement increases to 150 uVpp.


Oscilloscope frequency domain residual noise with 75R terminator on input.
FFT is 256K with sample rate 1 MS/s
This is the FFT noise floor with which to compare subsequent measurements.
Observe the approximately -109 dBu spike at 243 KHz.
-109 dBu is equivalent to 2.75 uV.
At 100 Hz noise floor is approximately -125 dBu or 436 uV.


With test load added more noise is observed, mostly above 40 KHz.
Everything is shielded and grounded to actual earth ground with test load isolated carefully from potential noise sources.
Ambient environment EMI is the limiting factor for this measurement.
Noise measures approximately 3 uV and below from 40 KHz to 400 KHz.

 

Time Domain Measurements


Meanwell GST25A05 has a ripple specification of 80 mVpp.
In this measurement 14 mVpp is observed with approximately 525 uS period and smaller ripple with 50 uS period. Definitely better than specification with the test load.


Soekris dac1221 stock SMPS
6 mVpp ripple with a period of approximately 50 uS
As will be demonstrated in post 4 below, this is a very good choice to balance design tradeoffs of power supply noise with cost.


iFi Power 2 measures approximately 250 uVpp in this test setup.
Keep in mind the system measurement limitations noted in post 1 above.


iFi Power X measures approximately 250 uVpp in this test setup.
Keep in mind the system measurement limitations noted in post 1 above.


Meanwell GST25A05 with NoiseNuke™ measures 172 uVpp

Frequency Domain Measurements


Meanwell GST25A05 has notable components at
1 KHz ~1.25 mV
2 KHz ~2.50 mV
Above 2 KHz below 275 uV


Soekris dac1221 stock SMPS has notable components at
6 KHz ~275uV
22 KHz ~350uV
Everything else below ~110uV


iFi Power 2 measures close to measurement system limits of 3 uV


iFi Power X measures close to measurement system limits of 3 uV
A small oddity appears between 1 KHz and 4 KHz but below 1.1 uV


Meanwell GST25A05 + NoiseNuke™ near measurement system limits

 

A few USB 5V charging and utility SMPS were also measured

Time Domain Measurements


Anker two port charger 430 mVpp
Ok for charging but would definitely not use for audio!!


Original Apple cube charger ~4 mVpp
Possibly ok for audio use


SMPS supplied with FNIRSI oscilloscope ~5 mVpp
Components with excellent on-board power supply filters such as Soekris dac1221 would work fine with this residual noise.

Frequency Domain Measurements


Anker 2 port charger – stick to charging use only


Apple original cube charger – much better than Anker above


FNIRSI SMPS is also much better than the Anker charger above.

 

SMPS time domain and frequency domain measurements have been demonstrated to differ considerably. But what difference does it make to an audio component. That depends on the sophistication of the audio component on-board power supply filter design. If it is a minimal design then the external power supply noise can have a considerable influence on the audio component performance.

In the case of Soekris dac1221 which has a reasonably sophisticated on-board power supply filter design, the external SMPS noise influence is minimized.

Bandwidth of the following Dynamic Range measurements are 96 KHz and the DAC is receiving data from Dante via spdif at 88 KHz. Soft filter is selected.


dac1221 dynamic range with stock SMPS


dac1221 dynamic range with iFi Power 2

possibly 1 dB improvement.
I use the iFi Power 2 with my dac1221 but the listening experience with stock SMPS will be fine.


While making the SMPS noise measurements I began to wonder how much the current draw might vary during various musical passages. It became apparent that current does vary depending on music.

Measurements description:
Keysight 34465A DMM
NPLC of 10
Approximately 3 measurements per second
Approximately 60 second measurement period
y-axis is 500 uA / division

Below for the reader’s amusement and amazement are current measurements for Soekris dac1221.


Bloody Oil – Medeski, Martin & Wood


Little Walter Rides Again – Medeski, Scofield, Martin & Wood


Mami Gato – Medeski, Martin & Wood


New Planet – Medeski, Martin & Wood


Daybreak - Overwerk

Data collections were repeatable for each passage thus the differences are verifiable.
Kick drum and loud bass notes appear to have the greatest impact.

 

Reserved for additional data and / or corrections.

 

Very cool, many thanks for this. That stock dac1221 power supply is the same as the one that came with the earlier dac1321. Presuming the internal filtering of the 1221 doesn't differ much from what's in the 1321 (paging @soekris for confirmation ), the iPower SMPS and Mark Johnson DC filter that are now in front of my 1321 might go elsewhere in the system.

There is of course already a NoiseNuke in the most critical position.

 

Nice work.... I use Power Supplies from XP Power, not just some random Chinese ones, although XP Power also manufacturer them in China.... I learned long time ago to be careful, an old Power Supply vendor once sold me some Power Supplies that died quickly when used on 230V, I took it apart and found they had crappy capacitors,the input capacitor I was able to look up, it was specified at 1000 hours life at 85 °C, didn't know anybody manufacture them that bad.... Electrolytic Capacitor life is a product of capacitor design, temperature and load. On my boards I mostly use electrolytics specified for 5000-10000 hours at 105 °C. Life usually doubles with each 10 °C lower temperature....

And yes, the dac1221 have good filtering at the power input, although not exactly the same circuit as the dac1321.... When selling worldwide I like to have power supplies that work world wide....

 

Power supplies are 50% of an audio component design challenge.
Søren definitely takes this seriously. @soekris DACs perform very well as a result.
Since the dac1221 was mentioned prominently in this thread, see
this post
for my opinions about the dac1221.