Audio Power Supplies - PART 1 - SMPS, LPS. SuperCap, Battery, DIY route. New devices opens up new options.
Sep 29, 2016 at 1:32 PM Thread Starter Post #1 of 9
Joined
Apr 12, 2013
Posts
5,930
Likes
507
After messing around with a whole host of USB DDC's and the new class of AES67 AOIP DDC's (like the Focusrite Rednet and BURL B2B DAC), I've found that low noise power is very important to digital sound quality.
 
From AC line filtering and balancing to ultra low noise DC regulators - the options today are growing.  Some expensive some cheap.  This thread is an attempt to explore and discuss what are those options.  And for me to relate some personal experience.
 
If you are in the camp that 'clean' power does not matter  - not the thread for you.
 
Many USB DDC's run off either and external DC power fed, or the +5VDC VBUS USB connection.  My own experiments with the new class of XMOS XU208 DDC's like the Singxer F-1 has shown me what a difference what better power supplies mean to SQ and how sensitive these femto digital clocks are to noise.  Further experiments with AOIP Audiante Dante DDC's and DAC's has shown they too respond well to PS improvements.
 
So to start a general discussion of various PS options - I am not an audio engineer - but on a steep learning curve.  I do enjoy discussing and exploring what the test bench measurements say - but in the end trust my ears to be the final arbiter of what works.  Enjoy!
 
OK first the two most popular forms of PS (99% of what most audiophiles use) - SPMS and LPS:
 
SPMS - Swtiched Mode Power Supply:
https://en.wikipedia.org/wiki/Switched-mode_power_supply
A switched-mode power supply (switching-mode power supply, switch-mode power supply, switched power supply, SMPS, or switcher) is an electronic power supply that incorporates a switching regulator to convert electrical power efficiently. Like other power supplies, an SMPS transfers power from a DC or AC source (often mains power), to DC loads, such as a personal computer, while converting voltage and current characteristics. Unlike a linear power supply, the pass transistor of a switching-mode supply continually switches between low-dissipation, full-on and full-off states, and spends very little time in the high dissipation transitions, which minimizes wasted energy. Ideally, a switched-mode power supply dissipates no power. Voltage regulation is achieved by varying the ratio of on-to-off time. In contrast, a linear power supply regulates the output voltage by continually dissipating power in the pass transistor. This higher power conversion efficiency is an important advantage of a switched-mode power supply. Switched-mode power supplies may also be substantially smaller and lighter than a linear supply due to the smaller transformer size and weight.

Switching regulators are used as replacements for linear regulators when higher efficiency, smaller size or lighter weight are required. They are, however, more complicated; their switching currents can cause electrical noise problems if not carefully suppressed, and simple designs may have a poor power factor.

 
 

Advantages and disadvantage:

The main advantage of the switching power supply is greater efficiency than linear regulators because the switching transistor dissipates little power when acting as a switch.

Other advantages include smaller size and lighter weight from the elimination of heavy line-frequency transformers, and comparable heat generation. Standby power loss is often much less than transformers.

Disadvantages include greater complexity, the generation of high-amplitude, high-frequency energy that the low-pass filter must block to avoid electromagnetic interference (EMI), a ripple voltage at the switching frequency and the harmonic frequencies thereof.

Very low cost SMPSs may couple electrical switching noise back onto the mains power line, causing interference with A/V equipment connected to the same phase. Non-power-factor-corrected SMPSs also cause harmonic distortion.

 
LPS - Linear Power Supply:
 
 

AC-to-DC supply

Schematic of basic AC-to-DC power supply, showing (from L-R) transformer, full-wave bridge rectifier, filter capacitor and resistor load​




Some DC power supplies use AC mains electricity as an energy source. Such power supplies will sometimes employ a transformer to convert the input voltage to a higher or lower AC voltage. A rectifier is used to convert the transformer output voltage to a varying DC voltage, which in turn is passed through an electronic filter to convert it to an unregulated DC voltage. The filter removes most, but not all of the AC voltage variations; the remaining voltage variations are known as ripple. The electric load's tolerance of ripple dictates the minimum amount of filtering that must be provided by a power supply. In some applications, high ripple is tolerated and therefore no filtering is required. For example, in some battery charging applications it is possible to implement a mains-powered DC power supply with nothing more than a transformer and a single rectifier diode, with a resistor in series with the output to limit charging current.

Linear regulator

The function of a linear voltage regulator is to convert a varying DC voltage to a constant, often specific, lower DC voltage. In addition, they often provide a current limiting function to protect the power supply and load from overcurrent (excessive, potentially destructive current).

A constant output voltage is required in many power supply applications, but the voltage provided by many energy sources will vary with changes in load impedance. Furthermore, when an unregulated DC power supply is the energy source, its output voltage will also vary with changing input voltage. To circumvent this, some power supplies use a linear voltage regulator to maintain the output voltage at a steady value, independent of fluctuations in input voltage and load impedance. Linear regulators can also reduce the magnitude of ripple and noise present appearing on the output voltage.

 
Since electrical noise is of critical importance to digital audio - so here is a comparison of the SMPS vs LPS general designs as it relates to various noise issues:
 
Radio frequency interference:
 
LPS:
Quote:
Mild high-frequency interference may be generated by AC rectifier diodes under heavy current loading, while most other supply types produce no high-frequency interference. Some mains hum induction into unshielded cables, problematical for low-signal audio.

SMPS:
 EMI/RFI produced due to the current being switched on and off sharply. Therefore, EMI filtersand RF shielding are needed to reduce the disruptive interference.

 
Electronic noise at the output terminals:
 
LPS:
Unregulated PSUs may have a little AC ripple superimposed upon the DC component at twice mains frequency (100–120 Hz). It can cause audible mains hum in audio equipment. 

SMPS:
Noisier due to the switching frequency of the SMPS. An unfiltered output may cause glitches in digital circuits or noise in audio circuits. 

Comment:
This can be suppressed with capacitors and other filtering circuitry in the output stage. With a switched mode PSU the switching frequency can be chosen to keep the noise out of the circuits working frequency band (e.g., for audio systems above the range of human hearing)

 
Electronic noise at the input terminals:
 
LPS:
 Causes harmonic distortion to the input AC, but relatively little or no high frequency noise.

SMPS:
 Very low cost SMPS may couple electrical switching noise back onto the mains power line, causing interference with A/V equipment connected to the same phase. Non power-factor-corrected SMPSs also cause harmonic distortion.

Comment:
This can be prevented if a (properly earthed) EMI/RFI filter is connected between the input terminals and the bridge rectifier. 

 
 
So the basics covered - moving on the common designs of the regulated LPS's.  Almost every audio device using a LPS is of the regulated verity.
 
 
First the different transformer designs for regulated LPS's:
http://www.audioxpress.com/article/Power-Transformers-for-Audio-Equipment
How Transformers Work: A Quick Review​
Grossly oversimplified, a transformer works by converting an AC current to a varying magnetic field, and then back into an AC current. Coiling wire into a “winding” and passing current through it produces the magnetic field. Conversely, the field passing through another winding induces a current in it, which is used to drive the load.​
In a power transformer, the “primary” winding is driven with AC line voltage — the power that comes out of the wall. The voltages required for the rest of the system are generated in “secondary” windings. All the windings are placed on a “core” made of an iron alloy. This is done because the permeability, or magnetic conductance, of iron is much higher than that of air, which allows a transformer to work much more efficiently.​
 

Core Type — The Many Different Ways To Build A Transformer​
You can build low-frequency power transformers in many different shapes and configurations. I’ll discuss a number of them in detail, starting with the most commonly used types, describing their construction and their suitability for use in audio equipment.​
Almost all power transformers share some common characteristics. Typically, they are wound on cores made from thin sections of an iron alloy, usually a type of steel made just for this application. Thin sections are used instead of a solid piece to prevent currents from being induced in the core itself—iron, after all, is an electrical conductor as well as a conductor of magnetic flux.​
Different physical configurations and manufacturing methods for transformers have evolved over the years in an effort effort to design a better product. One goal is to design a transformer that is nearly 100 percent efficient—meaning that all the energy in the primary winding is coupled to the secondary, and is not wasted by heating up the core or windings, or leaking magnetic flux outside the core. Another goal is to design transformers that are inexpensive to manufacture. As you might expect, these two goals are generally at odds, and the better the transformer is, the more expensive it is to build.​
 

 
E-I Transfomers:
The most common type of AC power transformer is called an EI-core transformer, because the laminated iron core that it’s wound on — before assembly — looks just like the letters “E” and “I.”​
 



EI with cover:

 
 
C-Core:
C-core transformers are made on a core that is wound from a single strip of material, like a toroid. The core is wound with two straight sides, so it is shaped more like an oval than a circle.​


 
Toroidals:
The toroidal transformer, or “toroid,” is a familiar sight inside high-end audio equipment. The toroid looks like a doughnut, with windings equally spaced around the diameter of the transformer.  Toroids are also available potted inside metal or plastic cans, or molded inside plastic resin and​
equipped with pins to mount directly onto a PC board.​



Toroidal with cover:

 
R-Core:
A more recent development is the “R” core, which you can think of as a cross between the C-core and a true toroidal core. R-cores are wound from a continuous strip of metal and are formed into a shape with two straight sides, like a C-core.​


 
 
OK so the basics of transformer designs - let's get to the noise related issues with each:
 
Magnetically Induced Noise And Stray Flux:​
If you’ve built much audio equipment, chances are that at one time or another you’ve run into a problem with line-frequency noise getting into the audio signal signal path. Sometimes this problem is simply the result of not enough filtering or regulation of the DC power supply, or perhaps a ground loop in an input circuit.​
But often, magnetic coupling from the AC power transformer is the culprit. In a perfect transformer, the entire magnetic field generated by the primary winding would be contained entirely within the transformer and pass through the secondary winding(s). Of course, nothing is perfect, and there is always at least some part of the magnetic field that escapes from the transformer.​
This field, called “stray flux,” is a primary concern in the selection of a transformer for audio equipment. If an AC magnetic field that leaks from the power transformer intersects a wire (or PCB trace), it will induce a small current into that wire, just as if it were a secondary winding on the transformer.​
The resultant noise voltage produced is generally very small, but in audio equipment, even a few millivolts of noise in a sensitive circuit can be audible.​
 

Stray-flux problems usually manifest themselves as a line-frequency noise — more of a buzz than a hum — that is unrelated to power-supply ripple or filtering...The easiest way to see whether you have a stray-flux problem is to move the transformer away from the rest of the circuitry. Since the strength of a magnetic field drops rapidly with increasing distance from its source, often moving the transformer only an inch or two will diminish the noise dramatically.​
Sometimes — especially in power amplifiers — just keeping the transformer away from low-level circuits is an effective method of dealing with stray flux.​
But in other situations, as in a phono preamp, the design of the transformer itself needs to be addressed to provide acceptably low noise in the finished product.​

EI vs Toroidal:
Like the EI-core transformer, the core of a toroidal transformer is made of an iron alloy, but instead of being composed of multiple, stacked laminations, it is wound from a single strip of metal, much like a roll of tape. The fact that there are no discontinuities in the core makes the toroid very efficient and reduces stray flux to around 10% of that of a comparable EI transformer.​
Toroidal transformers still leak magnetic fields, mostly because the windings are not symmetrical — the wires are spaced farther apart on the outer diameter of the core than the inner.​
 

Toroidal vs C-Core:
The wound construction of the core also results in higher magnetic efficiency than a stacked core. When constructed with windings on opposite sides of the core, the symmetrical construction helps to cancel stray magnetic fields as well.​

Toroidal vs R-Core:
 
The R-core transformer is nearly as good as the toroid in terms of stray flux. It has an advantage over the toroid because the turns of wire are spaced equally around the core, since they are wound on a straight section of the core.​

That covers stray magnetic flux - but what about the ability to reject AC line noise - which is better?
 
Transformer and PSRR (power supply ripple rejection):
http://www.diyaudio.com/forums/parts/41967-r-core-vs-toroid.html
Toroidal transformers have good regulation and low output impedance, but OTOH high interwinding capacitance. High interwinding capacitance means wide bandwidth coupling between the primary and secondary windings. This may not be what you want in an audio power supply, as it implies that all kinds of noise as well as 50/60Hz can pass through the power supply. If you decide to use a toroidal transformer, I would specify an additional electrostatic shield between the primary & secondary windings, as this helps reduce the primary-secondary noise transmission issue somewhat.

EI (frame) core, C-core and R-core transformers usually have split-bobbin primary and secondary windings, which cuts down on interwinding capacitance and makes it considerably more difficult for non-50/60Hz noise components to pass in and out of the power supply. Here you can also specify an additional electrostatic shield between the primary & secondary windings for less noise transmission, but this is not as important as it is when using a toroidal transformer for power supply applications. OTOH, EI, C and R-cores are usually physically larger and heavier for a given VA rating (at a given flux) than a toroidal, and this can have a direct impact on the size and weight of the amplifier chassis.

At the end of the day, it is possible to make a good-performing and good-sounding power supply using toroidal, EI, C-core or R-core transformers. But the cost and implementation issues will differ if you are trying to make any sort of optimized design. 

http://www.head-fi.org/t/616494/whats-the-different-between-transformer-types
The top is commonly called a "R-core", semi-toroidal - the windings are in balanced sections on the two bobbins - the core has a round cross section is actually 2 "C" cut cores clamped together
 
the bottom is a fully toroidal transformer with the windings in layers evenly spread over the toroidal core - it has less external magnetic field leakage but much higher pri-sec parasitic C
 
the fully toroidal construction used to be more expensive but with automated winding machines its higher material efficiency makes them cheaper, widely available
 
a undeserved reputation for "quality" is a holdover from the early days of expensive fully toroidal transformers
 
the low radiated magnetic field is a real advantage if you want a compact build with sensitive circuits crowded up against the transformer
 
but the high pri-sec C couples more line noise, the core saturates hard and in driving material cost down most are working near the edge of saturation in normal operation
 
EI laminated core, split bobbin transformers have the least pri-sec C, better saturation characteristics - but do have larger external magnetic field leakage - need more room, or shielding - but isolate better from line noise
 
the R core is somewhat in between, its dual split bobbins double the pri-sec C  but partially cancel the magnetic field leakage

 
Part 1: Transformers
http://www.head-fi.org/t/821621/audio-power-supplies-smps-lps-supercap-battery-diy-route-new-devices-opens-up-new-options-part-1
 
Part 2: LDO's and LPS regulation
http://www.head-fi.org/t/821731/audio-power-supplies-smps-lps-supercap-battery-diy-route-new-devices-opens-up-new-options-part-2
 
 
I have these to finish in the next few days:
Part 3: Capacitors, and off the shelf low cost LPS's
Part 4: DIY LPS options
Part5: Supercap Power Supplies and Li ion batteries
 
Oct 9, 2016 at 5:48 AM Post #2 of 9
Benchmark has published 2 articles in this area...
 
July 2016 - on power supplies in audio D/A converters
https://benchmarkmedia.com/blogs/application_notes/inside-the-dac2-part-3-power-supplies
 
May 2016 - on the audio myth - 'switching power supplies are noisy'
https://benchmarkmedia.com/blogs/application_notes/152143111-audio-myth-switching-power-supplies-are-noisy
 
Oct 9, 2016 at 12:04 PM Post #3 of 9
  Benchmark has published 2 articles in this area...
 
July 2016 - on power supplies in audio D/A converters
https://benchmarkmedia.com/blogs/application_notes/inside-the-dac2-part-3-power-supplies
 
May 2016 - on the audio myth - 'switching power supplies are noisy'
https://benchmarkmedia.com/blogs/application_notes/152143111-audio-myth-switching-power-supplies-are-noisy


Good links - thanks. 
 
Just a couple of things they fail to mention or differentiate - switching power supplies can be made to have low DC noise.  But feed higher levels of noise back into the AC line - this high freq noise can wreck havoc with other components like the clocks in a DAC.
 
The iFi iPower SMPS are a great example of a low noise switching power supply - they claim 1uv on the DC output.
 
I have never heard a toroidal or R-core emit any mechanical noise.  But I do use AC line balancing and filtering before.
 
I would recommend separating the power supply in delicate circuits like DDC's and DAC's.  To elminate any micro-vibrations, EMI, or heat from interacting with the other components - like the femto clock crystals.  Unfortunately due to cost most DAC's and DDC's have built in power supplies.  And use toroidal transformers to save space (but pass through much more AC line noise (ripple)).
 
I cover this in greater detail in PART3 - but here are some interesting links:
http://lightharmonic.com/blog/2015/07/24/power-supplies-and-noise/
Batteries vs a Switched-Mode Power Supply (SMPS)
Noise generated by batteries is usually very high frequency noise, well into the MHz or even GHz range. This noise is very difficult to filter and is exactly the kind of noise we hate the most because of the artifacts it sends down into the audible range. Still, this noise is less than that which is generated by a switched-mode power supply. In comparison, batteries are the better choice, but still not a great choice.

Linear Power Supply (LPS) with a Super-Quiet DC Circuit
A linear power supply can be designed with a much wider bandwidth, a super-stabilized voltage and current capacity, a considerably lower noise floor, and much more power than a battery power supply. It does come with trade-offs, though.
A good LPS is usually very heavy due to the amount of metal needed in its transformers and the size of its capacitors. The more wattage you need, the heavier it is. Nobody in the last 30 years has been able to solve the issue of weight.
A good LPS is not inexpensive. Again, because of the components that are needed, cost is a major contributing factor into why users elect against an LPS.
Conclusion
There’s no perfect solution for noise-free power. Each one we’ve discussed (battery, SMPS, & LPS) come with their own pros and cons. In my opinion, based on how I use my audio system, I rank these solutions as #1: linear power supply, #2: battery power supply, and #3: switched-mode power supply. the beauty of this hobby, though, is that everyone gets to make their own decision. Hopefully now you have more considerations as you decide which method you like most.

 

 
From John Swenson on AC noise - from a discussion on their new SuperCap LS-1 and linear JC-2.  I will cover the SuperCap PS's in PART 4:
http://www.computeraudiophile.com/f27-uptone-audio-sponsored/mystery-revealed-uptone-audio-ultracap%99-linear-power-supply-1-a-28609/index13.html
 
On the issue of "noise injected back into the AC mains", it actually even more complicated, there are two types of noise related to the AC line:​
1) differential noise​
2) leakage current​
They are very different, behave differently and have completely different methods of dealing with them. ALL power supplies have a combination of both, but the relative intensities vary significantly from supply to supply.​
#1, differential, is noise between neutral and hot (or L1 and L2 if balanced). This is what you normally think about when the term "noise on the AC line" is brought up. Both​
SMPS
and​
LPS
produce it. The spectrum can be huge, from 100/120Hz up to over hundreds of MHz. Line Filters are designed to block various ranges of this type of noise. This is what various "power conditioners" work on. It gets into your audio chain by going through the power supplies of other equipment and onto the DC output (between - and +).​
#2, leakage current is caused by capacitances in the PS causing some of the AC line to couple directly into the DC output of the supply. This happens through the transformers of​
LPS
as well as the circuitry of the​
SMPS
. This is direct harmonics of the AC line (so 60, 120, 180, 240 Hz. etc.) Since it is caused by capacitances the lower harmonics are usually not very strong, it is dominated by higher harmonics. Leakage current does NOT show up between the hot an neutral, it shows up between the whole AC line as a group, and the whole DC output as a group. Thus most power filters, power conditioners etc do NOTHING to attenuate it.​
Leakage current does NOT show up as noise on the DC output, it usually shows up as noise between grounds of different boxes. As long as the DC output ground of the two different PS's are connected together (through cables) this current will produce noise on the audio signal.​
 

 
Oct 9, 2016 at 12:19 PM Post #4 of 9
One good solution to minimize AC line noise (and what I have had good success with) is to use separate Art Audio PB4X4Pro AC line filters and isolators.  One for my DDC chain - another dedicated for my DAC and a third for my PC Server.  I try and keep my DAC fed with as clean of power as possible.
 
They aren't to expensive - $85.  If anyone decides to try one - be sure to get the 'Pro' version, the regular 4X4 doesn't have the AC filtering.
 
https://www.bhphotovideo.com/bnh/controller/home?O=&sku=433046&gclid=CM_CipKPzs8CFcOBfgodh0YMFQ&is=REG&ap=y&m=Y&c3api=1876%2C91438732682%2C&A=details&Q=
 
All PRO SERIES models have an additional internal discrete module called APF™ (Advanced Power Filtering) which filters out digital and dimmer hash as well as any high frequency noise that is above the audio range. Some off-the-shelf add-on modules saturate and lose their filtering effectiveness as the load increases, but by using a high power discrete design we were able to create a filter that stays effective over the full operating range.
By using both Common Mode and Differential Mode topologies in series for the filter design we are able to block virtually all of the unwanted noise that is between the AC line and ground, and also the two sides of the AC line. This has the additional benefit of reducing ground loop problems in your system. High frequency noise currents in particular are highly attenuated in both directions so that any line noise that could be generated by one of your components is not allowed to get back into your main A.C. wiring so if you use a number of PRO SERIES Power Conditioners in your setup, you can distribute and isolate the noisy components from the sensitive components in your system. Additionally, any signals above 10kHz are filtered from the line with over 40dB of attenuation above 100kHz and beyond.
 

 
Oct 9, 2016 at 12:44 PM Post #5 of 9
  One good solution to minimize AC line noise (and what I have had good success with) is to use separate Art Audio PB4X4Pro AC line filters and isolators.  One for my DDC chain - another dedicated for my DAC and a third for my PC Server.  I try and keep my DAC fed with as clean of power as possible.
 
They aren't to expensive - $85.  If anyone decides to try one - be sure to get the 'Pro' version, the regular 4X4 doesn't have the AC filtering.
 
https://www.bhphotovideo.com/bnh/controller/home?O=&sku=433046&gclid=CM_CipKPzs8CFcOBfgodh0YMFQ&is=REG&ap=y&m=Y&c3api=1876%2C91438732682%2C&A=details&Q=
 

I recently also bought one over here in the EU.
In the EU it comes with C14 connectors on the backplate and not Schuko (C7/7), so you will need to know beforehand and buy proper cable termination on your power cables or put in an adapter (C13 to C7/7)
 
I must say it does give a more quiet, relaxed sound. I'm using it on my Grimm CC1 master clock and Mutec MC-3+USB SPDIF reclocker.
 
Oct 9, 2016 at 6:14 PM Post #6 of 9
  I recently also bought one over here in the EU.
In the EU it comes with C14 connectors on the backplate and not Schuko (C7/7), so you will need to know beforehand and buy proper cable termination on your power cables or put in an adapter (C13 to C7/7)
 
I must say it does give a more quiet, relaxed sound. I'm using it on my Grimm CC1 master clock and Mutec MC-3+USB SPDIF reclocker.


Good to hear that - how much was it there?
 
Oct 10, 2016 at 1:04 AM Post #7 of 9
Good to hear that - how much was it there?


I had to get the Pro-version on special order as they only stock the non-Pro over here.
But the price for the Pro-version incl. 21% VAT and shipping was €102, so not too bad compared to the US-version.
 
Oct 31, 2016 at 10:54 AM Post #9 of 9
Found this interesting post about R-Core transformers - over on CA
http://www.computeraudiophile.com/f8-general-forum/keces-dc-116-linear-power-supply-unit-quick-review-17623/index3.html
 
Uptone describing why they choose an R-Core for their JS-2:
On a separate note, but still about transformers: R-cores don't seem to hum at all. Because of the way the cores themselves are wound, there is virtually no magnetic excitation. Check out this explanation of how the cores are made:​
Of course R-cores have all sorts of other advantages as well. But even at wholesale from China, they cost about twice what a decent toroid costs, and at least three times a cheap toroid. I know, because I have 25 custom 100VA R-cores en route from China right now. I compared it in my forthcoming​
JS
-2--versus a good 100VA toroid. In fact, this comparison took place with John Swenson here at may place just this afternoon. We went back and forth between the toroid prototype and the R-core in a pre-production chassis I finished assembling last week.​
We were hoping that my decision to spring for the R-core would prove worthwhile. Wow, did it ever! For my Mac mini it made the bass even more solid and detailed--R-core versus toroid. We were both surprised at how obvious it was (well my system does very good bass). But the knock our socks off moment came when doing the same switch with the​
JS
-2 powering (at 5V) a pre-production BottleHead​
DAC
board that John had brought along for tests. The effect on clarity and detail (oh, I'm a very unimaginative writer about what I hear) with the R-core was on much more than just the bass. It was a cymbal to vocal to piano to upright bass and horns attack improvement. I am REALLY happy about this as I now feel that the supply could find a even wider market that for computers, drives, and headphone amps. Not that there are that many top-flight DACs that take just a single-ended (+DC) supply.​
Here are some pics to show what were were doing (click pics for larger version; also, my production R-cores will have the pretty copper electrostatic shield--and some fancy labels I am making):​
 

 

Users who are viewing this thread

Back
Top