• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Testing and checking PSU for DAC

Nango

Active Member
Patreon Donor
Joined
Aug 6, 2018
Messages
167
Likes
98
#1
We all hear and read often that a better power supply always improve the performance of a D/A conversion. I dont know if your equipment @amirm is capable of detecting how this really works, if any. And I would really like to understand from a science/expert point of view why this works and how, whats is important and what is not, etc.

It is just a suggestion for further gear checking after the 30st test of the DAC itself.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
18,352
Likes
17,241
Location
Seattle Area
#2
I have done a lot of testing of this. Sometimes there is a measured difference, sometimes there isn't. When there is, it can actually make things worse, not better! I can do more tests but for now, this is folklore.
 

Nango

Active Member
Patreon Donor
Joined
Aug 6, 2018
Messages
167
Likes
98
#3
Ok, but what exactly makes a psu better (or even worser) than other?

The less ripple and noise the better, or is this also not that relevant? ....What else? Stability of the applied voltage, current ..... Is this today important, relevant?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
18,352
Likes
17,241
Location
Seattle Area
#5
Ok, but what exactly makes a psu better (or even worser) than other?

The less ripple and noise the better, or is this also not that relevant? ....What else? Stability of the applied voltage, current ..... Is this today important, relevant?
The main difference I see are how much mains noise or spikes from rectifiers bleeds into the output of the DAC. I would ignore any power supply specs by themselves. The only thing that matters is how much they change the output of the DAC.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
18,352
Likes
17,241
Location
Seattle Area
#6

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
1,206
Likes
1,867
Location
The Neverlands
#9
Ok, but what exactly makes a psu better (or even worser) than other?

The less ripple and noise the better, or is this also not that relevant? ....What else? Stability of the applied voltage, current ..... Is this today important, relevant?
With a very simple circuit you can actually listen to the power supply in a circuit.
This is educational.
Unfortunately there is pollution on power lines above 13-18kHz which you cannot hear but.. may very well be there, certainly in digital circuits.

For those digital circuits the quality of the PCB layout and proper usage of so called decoupling caps (selecting the best ones for the task is essential) is far more important than the used voltage regulator.
For higher speed amps local analog circuit decoupling is also more important than the used regulator.
Also the PowerSupplySuppressionRatio of the used circuits is important.

For switchers common mode 'garbage' and high (and low) frequency noise/hum/tones can be problematic in certain cases.
Most circuits fed by switchers as well as USB fed devices usually have local onboard regulators and decoupling which is more important than the power supply that feeds it.
As Amir already mentioned switchers usually have more spikes and nasties than transformers/linear power supplies but is no 'rule'
There are also switchers (SMPS) that exceed many linear power supplies in 'performance'.
 

trl

Addicted to Fun and Learning
King of Mods
Joined
Feb 28, 2018
Messages
545
Likes
328
Location
Iasi, RO
#10
Not sure how the below will affect the output sound, but worth posting:

Mains_outlet_sine_02.png

My original mains quality (after isolation transformer) during music playback

Mains_outlet+SMPS_sine_01.png

What happens after inserting a SMPS power brick in the same outlet (probably defective or drawing too much power from my 250VA isolation transformer)

Note: I used a battery-powered laptop ans a step-down 230V-120V transformer to be able to use my PicoScope there and do this measurement. Usually, for our safety, we should never use standard probes when measuring AC mains, instead differential probes should be used...and authorized technicians too.
Sometimes a really bad AC-mains might affect how internal PSUs from our audio equipment work like.
 
Last edited:

trl

Addicted to Fun and Learning
King of Mods
Joined
Feb 28, 2018
Messages
545
Likes
328
Location
Iasi, RO
#11
PLAY_PSU_12V.png

Switching noise that is usual for switching supplies (this is a low-noise SMPS brick).
Shouldn't bother us at all if these spikes are so low, definitely not doing anything to the output sound, especially because of their very high switching frequency.


SMPS_15V_noise.png

SMPS ripple and noise after XL4015 boost-converter (AC RMS: 6.175 mV at inaudible frequencies)



Headamp_inaudible_noise.png

Headphones-out noise when powered from the above SMPS (AC RMS: 3.7mV at inaudible frequencies)

The headphones amplifier was somehow affected by the switching noise of the boost-converters, but this noise becomes barely audible on sensitive cans only, like 16 Ohms IEMs. It was probably related to to lower harmonics of the original switching noise (300-500KHz perhaps?). This was the first PCB revision 1.6 of PLAY. With ver. 2.1 and above there's no more switching noise present...at least nothing to look like in the above pics. Also, no more background noise present with the new boards.


LPSU_12V_noise.png

Linear PSU noise after swapping XL4015 with LM7815 regulator (AC RMS: 280uV)



Headamp_missing_noise.png

Headphones-out noise when powered from the above LPSU (AC RMS: 309uV)

Removing the boost-regulators from the board and adding a DYI dual-linear PSU instead improved the output noise of the amplifier by 3.6dB, as per below pic (compare columns 2 and 3 only):


Column 2: SMPS is powering the audio equipment, column 3: LPSU is powering the equipment.
About the same 3-4dB difference was found by Headfonia in their measurements between the first version PLAY V1.6 and the second version PLAY V2.1: https://www.headfonia.com/review-burson-play/3/. That means that the new boost-regulators from PLAY V2.1 do compete with my DIY linear PSU, which is a good improvement.

All my findings were already uploaded here: https://www.audiosciencereview.com/forum/index.php?threads/burson-play-mods.3346/. The linear PSU did improved the background noise with sensitive IEMs, but THD remained about the same (low enough anyway). The LPSU mod might be done for PLAY V1.6 only, as the newer revision has already a very low background noise, inaudible, so not worth the time spent (I own both revisions, I A/B tested and measured both).

My own thoughts and conclusions: if a SMPS is well designed and implemented, then it should be just fine (see 2'nd rev. of PLAY), otherwise LPSU might worth trying.

L.E.: Column #1 from the above RMAA test is the original PLAY playing with solid-state SS V6 Classic op-amps in I/V and LPF. Seems that the noisy SMPS boost-converters are affecting how these big opamps measures. However, after swapping the 1st mainboard revision with the newer one the issue goes away and RMAA numbers become similar with the ones from column #3. Same numbers were achieved after swapping internal boost-converters with linear regulators (without swapping the mainboard). Column #2 has the MUSES8920 and the same first ver. of mainboard (smaller opams from MUSES is not affected by the noisy boost-converters). Column #3 has the SS V6, but with the new mainboard (same results with MUSES8920).
 
Last edited:

trl

Addicted to Fun and Learning
King of Mods
Joined
Feb 28, 2018
Messages
545
Likes
328
Location
Iasi, RO
#12
A poorly implemented PSU could be found here: https://www.audiosciencereview.com/forum/index.php?threads/asus-essence-one-mods.2637/#post-75271; it has poor grounding on the PCB traces and ground-loops (on pin from -12V plug was not connected to GND). After resolving the PSU issues the RMAA didn't showed much differences in THD or noise, although jitter was improved and 100Hz hum decreased a bit. In A/B blind tests 2 people were able to find the modded DAC 10 out of 10 times (I had 2 identical units: one modified, the other one was original).

Conclusion: A poorly implemented PSU will affect how the DAC or amplifier will operate. If this will affect or not the output sound it's up to the measurements to prove that. I usually tend to measure a PSU and decide later if needs to get replaced or not. If AC ripple & noise is lower than 1mV, then I won't bother with any mods; today's opamps are having a CMRR/PSRR of about 100dB, so 1mV noisy PSU will inject nothing into the audio chain. Even 10mV of AC ripple & noise should not affect the output sound of an audio equipment.
 

trl

Addicted to Fun and Learning
King of Mods
Joined
Feb 28, 2018
Messages
545
Likes
328
Location
Iasi, RO
#13
Worth mentioning the main source of hum-noise (50-60Hz and it's 2nd and 3rd harmonic): the inside transformer, especially if too close to the analogue parts of the amplifier or I/V, LPF, buffer of the DAC.

I had this issues with my ASUS Essence One DAC because the toroid transformer was too close to the volume pot and the PCB analogue traces, but also with my MATRIX HPA-3B headamp. Under normal listening conditions the hum-noise was inaudible, instead when using sensitive cans and pumping the volume to the max. (audio inputs shorted) there was some hum-noise in my head.

Resolution was to shield the inside toroid transformer with GOSS; this hint was provided to me by an engineer from Petra toroid, that also gave me 1 meter long GOSS stripe, to check if this resolve my issue (Mu-Metal should do about the same). A good explanation could be read here, on page. no. 5: https://c4w4s3k2.stackpathcdn.com/w.../07/iFi-audio-Pro-iESL-Tech-Notes-June17-.pdf.

Also, a good read here: http://www.nuvotem.com/en/products/gossband.shtml and a very good representation of the undesired effect: http://www.nuvotem.com/en/products/stray_field.shtml. Based on Talema's representaion, the Mu-Metal is better than GOSS in shielding the transformer.

Again, seems that a good implementation of the PSU will keep noises and interferences away from the analogue parts, so the output noise in our cans or speakers will be non-existent.

P.S.: I bet that moving the transformers few inches away from the board (or rotating 30-45 degrees) inside Yggdrasil will cancel most of the low-noise spurious from the power supply: https://www.audiosciencereview.com/...ents-and-review-of-schiit-yggdrasil-dac.2358/ and https://www.audiosciencereview.com/...measurements-of-schiit-yggdrasil-v2-dac.3607/. Same cancellation effect will be achieved after shielding these transformers with GOSS or Mu-Metal (or at least with double-sided thick copper PCB tied to the ground, mounted perpendicularly to the mainboard).

P.P.S.: I had a hum-noise in my audio monitors, caused by a ground-loop, that was audible from 1 meter away from the speakers. Issue was resolved by lifting the ground safely with an isolation transformer (Hum-X will do as well). Also, using balanced interconnects resolves these ground-loops in most scenarios.
 

trl

Addicted to Fun and Learning
King of Mods
Joined
Feb 28, 2018
Messages
545
Likes
328
Location
Iasi, RO
#14
We all hear and read often that a better power supply always improve the performance of a D/A conversion.[...]
A good article here: https://hifiduino.wordpress.com/2014/12/07/akm-verita-4490-dac/. Seems that the analogue section of a DAC is more influenced by a better or worse PSU than the digital section. Hence, a lower AC ripple & noise will decrease the output noise:


Few years ago I asked TI:
- "What's max. acceptable ripple+noise for +3.3V and for +5V rails, so PCM1795 could handle the final PCM & DSD sound without changing it's output specs at all?" and their answer was
- "We do not have the data to provide you with a number for the max ripple+noise allowed on the supply lines before the performance suffers".
Can't find the original thread, but I got the cached one here: https://webcache.googleusercontent..../audio/f/6/t/471064+&cd=1&hl=en&ct=clnk&gl=ro

Who knows, maybe today's ESS DACs are measuring so good because of the new "1uV RMS of noise" regulators ES9311 developed by ESS especially for their DAC chips: http://www.esstech.com/files/3414/5193/1543/ES9311_product_brief_121715.pdf. I'm sure there's a good reason behind the appearance of these regulators, other than pure marketing, but perhaps @March Audio could jump in with few thoughts, please?

So, if someone has a test board with a DAC and has the possibility to actually "play around" with it, would be great. I personally tried to lower the background noise of my ODAC v1 by lowering the AC ripple, by increasing the capacitors value around the ES9023 DAC, but test was unsuccessful: same output noise was at outputs. Perhaps board layout and a very good grounding is more important sometimes than the power regulators.
 
Top Bottom