[rb2013]

Ok Now things well get more interesting - here I want to explore the LPS solutions available - focusing on the less expensive options (under $250).  But first a quick primer on the last main component of these LPS units - the filter capacitors.
 
WARNING – POWER SUPPLIES CONTAIN LETHAL HIGH VOLTAGE! NEVER OPEN THE CASE OF YOUR POWER SUPPLY.  SEEK A COMPETENT TECHNICIAN TO MAKE ANY ADJUSTMENTS OR CHANGES TO YOUR EQUIPMENT! I take no responsibility for anyone who ignores this warning!
 
 
LPS Capacitors:  There are many types of capacitors, but linear power supplies generally use electrolytics, due to their ability to withstand high voltage spikes and reasonable performance.  These caps are rated for voltage, and capacity, as well as expected life, temperature rating and physical size.  But other important criteria is the caps ESR rating, PSRR and impedance - more on this in a bit.  Generally the aluminum electrolytic types are preferred.
 
 https://en.wikipedia.org/wiki/Aluminum_electrolytic_capacitor
 
"The basic unit of electrolytic capacitors capacitance is the microfarad (μF, or less correctly uF)", that and voltage are two main parameters for use.
 
"Electrolytic capacitors are polarized capacitors and generally require an anode electrode voltage to be positive relative to the cathode voltage.   To minimize the likelihood of a polarized electrolytic being incorrectly inserted into a circuit, polarity has to be very clearly indicated on the case, see the section headed "Polarity marking".
 
Now for some details on why these capacitors are used in a LPS and how to select the better versions:
 
Impedance:

 

 ​

​

In general, a capacitor is seen as a storage component for electric energy. But this is only one capacitor function. A capacitor can also act as an AC resistor. Especially aluminum electrolytic capacitors are used in many applications as a decoupling capacitors to filter or bypass undesired biased AC frequencies to the ground or for capacitive coupling of audio AC signals. Then the dielectric is used only for blocking DC. For such applications the AC resistance, the impedance is as important as the capacitance value.

 

The impedance specified in the data sheets of various capacitors often shows typical curves for different capacitance values. The impedance at the resonant frequency defines the best working point for coupling or decoupling circuits. The higher the capacitance the lower the operable frequency range. Due to their large capacitance values, aluminum electrolytic capacitors have relatively good decoupling properties in the lower frequency range up to about 1 MHz or a little more. This and the relatively low price is often the reason for using electrolytic capacitors in 50/60 Hz standard or switched-mode power supplies 

It's my understanding that a lower impedance is generally better for use as a filtering cap.
 

ESR and dissipation factor tan δ:

The equivalent series resistance (ESR) summarizes all resistive losses of the capacitor. These are the terminal resistances, the contact resistance of the electrode contact, the line resistance of the electrodes, the electrolyte resistance, and the dielectric losses in the dielectric oxide layer.

ESR depends on temperature and frequency. For aluminum electrolytic capacitors with non-solid electrolyte the ESR generally decreases with increasing frequency and temperature. ESR influences the remaining superimposed AC ripple behind smoothing and may influence circuit functionality. Related to the capacitor, ESR is accountable for internal heat generation if a ripple current flows over the capacitor. This internal heat reduces capacitor lifetime.

Here a lower value is generally better.

Ripple current:

A ripple current is the RMS value of a superimposed AC current of any frequency and any waveform of the current curve for continuous operation. It arises, for example, in power supplies (including switched-mode power supplies) after rectifying an AC voltage and flows as biased charge and discharge current through the decoupling or smoothing capacitor. 

The rated ripple current of aluminum electrolytic capacitors with non-solid electrolyte corresponds with the specified lifetime of the capacitor series. This current may flow permanent over the capacitor up to the maximum temperature during the specified or calculated time. Ripple current lower than specified or forced cooling lengthen the capacitor's lifetime. 

Here again higher is better.
 
I have had good success replacing the stock capacitors on inexpensive LPS's.  I look for long life - 10,000 hours preferred (most of the cheaper LPS from China use 2,000 hr or less rated caps), low ESR, high PSRR and low impedance for the respective capacitor value.  I also try and find capacitors rated for 105C temperature, if possible.  The cost for these better caps is minimal compared to the overall cost of the LPS.  One note more and larger filter caps in a LPS will give a higher level of AC noise or ripple rejection - all other things being equal.  The larger number and size will slightly increase the noise fed back into the local AC mains (still orders of magnitude less then a SMPS). 
 
Here is a TeraDak X1/X2 with the stock Panasonic FC 25V 1500uf caps (rated at 5000 hrs, 105C Temp) - from Digikey (although they show a different color) Impedance: 30mOhm,
Ripple: 1.55A@120Hz, they do not show an ESR rating for these.

 
With Nichicon HW  25V 1500uf - 10,000hr (105C), low impedance: 16mOhm, Ripple 1.56A@120Hz.

Other possible caps I could have used that are excellent:
United Chem KZ (5000hr - 18mOhm, 2.7A) $1.00each
Pannasonic FM (7000hr - 15mOhm, 2.55A@120Hz) $1.40each
Pannasonic FR (10,000hr - 18mOhm, 2.08A@120Hz) $1.01each
 
I have the FMs and may switch the caps in this little unit at some point - more on this LPS later.
 
 
 
Existing LPS available for low cost  - under $250:
 
Just a note on linear power supplies - they do come in both regulated and unregulated versions.  Typically the unreg will be cheaper wall wart looking devices.  Do to the lack of regulation they are not recommended, their voltages can swing by wide amounts and even damage some equipment.  All the following are of the regulated variety.
 
In PART 1,2,and 3 we have explored the three main components of a linear power supply:  the transformer, the regulation circuit, and the power filtering capacitors.  So what to look for?  Well for me I really like the R-Core transformer for it's AC noise (ripple) rejection superiority over toroidal or EI transformers.  A high quality LDO with low DC noise and high current capacity (an issue with many of the ultra low LDO's available). 
 
You have to look at what your current and voltage needs are, when choosing a replacement power supply.  If you're powering a USB DAC or DDC - is 500mA enough?  Many draw in excess of this - the PUC 2 Lite for example with a 800mA draw.  Most external XMOS DDC's will draw less and abide by the 500mA USB std.  I would prefer a LPS that can provide at least 1A 5VDC(the std USB voltage).
 
Many LPS's also have adjustable voltage pots - so their voltage can be changed as needed.  Some of the external USB gizmos like the Regen and Recovery need a higher voltage then the std USB +5VDC, as they have their own on board LDO regulators that will then step down the required voltage to 5VDC - in the case of the Regen 7-9V is optimal and the W4S Recovery needs 9VDC.   Feeding a USB DDC like the F-1 can be done with a split USB cable (the data and power leads are separate), taking the power from the LPS and the data from the PC's USB.

As for capacitors - not as big of an issue for me as they are fairly easy and cheap to upgrade.  But it would be nice to have the better Panasonic, Nichicon, or Chemcon caps.
 
So here are a few I own, or look interesting to me.
 
First I want to discuss a neat little gadget:
 
iFi DC iPurifier:
http://ifi-audio.com/portfolio-view/accessory-dcipurifier/
 
This attaches to the end of the DC feed cable before the device and filters the DC power by many factors.  I have had very good success using one of these with a low cost LPS bought on Ebay.
 

 
iFi claims tremendous DC noise filtering from this unit - 316X under 1Khz, up to 100,000x above.  They claim they are able to achieve this with an active noise canceling circuit used in military radar.  I wonder if this type of circuit isn't being used on chip for these >1uv LDO IC's to get the DC noise levels down?

They are not cheap - around $80 from Music Direct:
https://www.musicdirect.com/power/ifi-dc-ipurifier-power-conditioner
 
But the big question is - do you need an ultra low noise regulator (>4uv) in your LPS if you use one of these?
 
As you'll see there are many decent and cheap LPS's on Ebay - some for less then $100.  If one uses the LT1083 or LT1084 LDO regulators (see Part 2 for a comparison of LDO's) with 180uv of DC noise - but that then gets filtered down by a DC iPur at least 317 times - that would leave less then .5uv of noise to the device!  This is the lowest noise level I have seen in a plain IC LDO. So for around $160 you can have a nice ultra low noise package.  That is if the DC iPur works as advertised.
 
 
OK On the LPS's:
 
Generic R-Core LPS:
 
This unit I own uses the LT1083 regulator mounted on a nice large heat sink - uses one 50V
3300uf filter cap, 4 of the new Vishay B560 Ultra Fat, Ultra Low Loss Rectifiers, and a 12V R-Core transformer.  The LT1083 does not have very good noise numbers rated at 180uv, and lowly PSRR.  But with an R-Core maybe high PSRR is not as necessary?  With the iFi DC iPur - the DC noise can be reduced to virtually nothing.  This is a high current unit - and not appropriate for a USB device - the lowest setting is 9V(3A) and can go as high as 24V(1A).  So if high current and voltage is needed this looks like a great low cost solution. Note the iFi DC iPur is rated to just 24VDC - and over heated when I tried it this unit at that voltage, but has worked well at much lower voltages.
 
$77 Delivered.
 
 
 
http://www.ebay.com/itm/30W-30VA-HiFi-Linear-power-supply-PSU-DC9V-12V-15V-16V-18V-20V-24V-for-choose-/161870356771?hash=item25b039a123

 
I will explore the DIY option in a later part 4.  But this could be the 'bones' for just such a project.  Just replace the LDO with some thing like this (the existing board is 75X50mm):
 
http://www.ebay.com/itm/142115498378?_trksid=p2060353.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT
 


Ultra low noise LT3042 .8uv - very high PSRR 79dB @ 1Mhz.  This board is rated at 6-12V input and 1.5A 5VDC output.  So the existing R-Core could be used.  The size is 69x47mm - so it would fit as well.
 
$60 shipped.  So with a little effort this could be a killer good USB DDC or DAC power supply.  The voltage is non-adjustable.
 
 
TeraDak X1/X2 LPS:
Neat little unit  - EI transformer - LT1084 LDO (180uv 30dB PSRR)- Panasonic FC caps 25V 1500uf (8 total), AC common mode filter coil, MOV surge protectors, etc... $80 shipped.
 
http://www.ebay.com/itm/TeraDak-TeraLink-X1-X2-DC8-5V-1A-USB-DC5V-port-Hi-Fi-Linear-Power-Supply-/181467993450?hash=item2a4055d56a:g:LzQAAOxyVVJR~9Nv

 
 

 
Upgrade the caps - as pictured - or use the stock caps and add a iFi DC iPur and you have a neat low cost LPS solution.
Nice feature - has two DC power ports - one a USB set for 5VDC and the other a 2.5mm barrel which has adjustable voltage.

Product Specification​

1. Output : 8.5V DC + USB-A 5 VDC Port @ ~ 1A

 
With a total of 1A 5VDC - I would not use both power ports simultaneously.  But 1A for 5VDC is more then adequate for XMOS USB DDC use.
 
TeraDak DC-30W (old design):
Nice copper shielded R-Core transformer, likely the LT1084 LDO's 180uv, 30dB PSRR, Panasonic FC caps 25V 1500uf (10 of them).
 
With the R-Core transformer and it's excellent inherent PSRR - the low LT1084 PSRR may not be as important as when it's used with a toroidal transfomer (see PART 1 on transformers).
 

Rated at 3Amps 5VDC and 1.5A for 12V and with an adjustable voltage  - more then enough for almost any DAC or DDC.  Great for powering the 7-9 VDC Regen or 9VDC Recovery USB reclockers.
 
 
TeraDak DC-30W (New design):
Nice copper shielded R-Core Transformer, looks like two NPN devices used for power regulation, mine had the Panasonic FC 25V 1500uf caps 3 total and two larger 25V 3300uf Panasonic SU caps (low 2000 hr life), EMI common mode filter and MOV surge protection on board.
 
I see the latest versions have the better Panasonic FC 25V 3300uf caps vs the SU.  (On mine I swapped the SU's for 25V 3300uf Nichicon HW's.  This made a noticeable improvement in SQ.)
$139 shipped.
http://www.ebay.com/itm/TeraDak-DC-30W-TOUCH-DC9V-2-5A-Hi-Fi-For-Audio-Linear-Power-Supply-/130906580297?hash=item1e7aa3e549:g:5UoAAOxyo4lRjkus
 
This new design has high current capability - rated at 2.5A @ 9VDC.  Also with adjustable voltage - so at say 5VDC plenty of power for any USB DDC and then some.  Plenty for the Regen and Recovery etc...
 
 

 
I could only find the marking of D44H11 on the regulation devices - there are two.  I identified them a NPN devices (OnSemi?) - but was not able to find any DC noise figures on this power supply.  Imagine less then 50uv with excellent PSRR.
 
http://www.onsemi.com/pub_link/Collateral/D44H-D.PDF

These series of plastic, silicon NPN and PNP power transistors can be used as general purpose power amplification and switching such as output or driver stages in applications such as switching regulators, converters and power amplifiers. Features • Low Collector−Emitter Saturation Voltage • Fast Switching Speeds

 

 
I use mine with a iFi DC iPur to excellent effect on both my modded BURL B2B DAC and with the Singxer F-1 XMOS USB DDC.
 
MEIYAN Ultra Low Noise - High Current LPS:
This unit I found on Ebay has me really intrigued - very low noise - 13uv @24VDC - low noise at lower voltages, very high current capability:
$192 shipped.

 
http://www.ebay.com/itm/131867485038?_trksid=p2060353.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT
We have following several kinds voltage for choose (Output voltage / current)

1. 5V /6A

2. 9V /5.7A

3. 12V /5.5A

4. 15V /5.3A

5. 18V /4.4A

6. 19V /4.2A

7. 24V /3.3A

 

Technical highlights

All-discrete topology.

Single-pass, series regulator design.
No IC (integrated circuits) are used. This allows complete design control over all operating points and parameters for superior performance.
Low noise, high PSRR

A constant-current source feeds a zener diode as a stable voltage reference. A low-pass filter (with a corner frequency of 1.6Hz) prevents zener
noise from being introduced into the error amplifier. This is an effective yet lower-cost alternative to expensive voltage reference ICs. The low-pass
filter also provides a soft-start characteristic.
The output noise (unloaded) is less than 13µV at 24VDC output (measured using a Tangent LNMP (low-noise measurement preamplifier) and a Fluke
187 50000-count DMM in ACmV mode). The output noise is even less when the output voltage is lower. This is much better than the noise of an IC
regulator based PSU tested under identical conditions.
The error amplifier is a discrete implementation of an opamp with a high open-loop gain of 102.5dB. The voltage supply to the error amplifier is isolated
with capacitance multipliers to boost its PSRR (power supply rejection ratio). This greatly improves the line regulation performance of the PSU.
A long-tailed pair differential amplifier with current mirror and constant current source forms the first stage of the error amplifier. The second stage is the
voltage amplification stage (VAS), also with constant current source load. The 3rd stage is comprised of the power MOSFET output devices configured as a
source follower.

High-current MOSFET pass transistors

Two paralleled high-current, highly reliable MOSFETs (rated at 18A each) serve as the "pass" transistor.
The high current rating provides a very high safety headroom against overcurrent damage.
The use of paralleled MOSFETs divides the heat dissipation, simplifying thermal management. Onboard heatsinks can be used which would allow the this PSU
to supply up to 1A continuous (with much higher peak currents). More sustained currents are possible by using larger, offboard heatsinks.
The negative temperature coefficient of MOSFETs prevents damaging thermal-runaway conditions that may plague conventional BJT devices.
No current-limiting.

 

Nice copper shielded R-Core transformer - excellent low noise.
 
One downside is the voltage is not adjustable.  But it is available in a range of voltages - I will likely order one at some point to try out.  Update:  The DC voltage is adjustable in a range of 5VDC to 24VDC.
 
 
So that completes the low cost LPS's I have found - now a few expensive ones:
 
Paul Hynes SR2 & SR3:
These highly regarded LPS's are excellent designs - but not cheap the SR2 is approx $250 ($195 English Pounds) and the SR3 $381 (300 English Pounds):
http://paulhynesaudio.com
http://hifipig.com/paul-hynes-sr3-power-supply/
http://www.paulhynesdesign.com

Supply Regulation
		Many audio equipment manufacturers use industry standard solid state regulators in their products because they are readily available, cheap and easy to apply. They offer reduced power supply ripple breakthrough from the rectifier / energy storage capacitor. This allows a much smaller energy storage capacitor to be used, which in turn reduces component costs considerably, more than offsetting the cost of the regulator itself. Multiple regulator systems can also be applied cheaply. The main benefit of using these devices is essentially one of cost reduction. Whilst this is a laudable aim, most enthusiasts will generally prefer to look for performance improvement before cost considerations. So let us look at regulator performance with a typical industry standard, the 317 / 337 type adjustable regulator. Starting with power supply rejection (PSRR) of these regulators with respect to frequency. At 100Hz this is 60-70dB (60dB is a ripple reduction of 1000 times). At 1KHz, the PSRR begins to drop due to the regulator's internal frequency compensation reducing gain at higher frequencies, leaving less loop gain available for error correction. At 10 KHz, the regulators manage 50dB (316 times ripple rejection). At 100KHz they only achieve 20dB of PSRR (10 times ripple rejection) and at 1MHz only 10dB (3.16 times ripple rejection). At low audio frequencies both devices offer usable PSRR, but this situation deteriorates rapidly above 1KHz, becoming relatively ineffective at frequencies above 100 KHz where radio frequency interference and digital clocking load current transients may have to be dealt with.   Output impedance shows the regulator's ability to control the load with respect to frequency. The 317/337 graphs show 10 milliohms from DC to 1KHz. At around 1KHz the frequency compensation capacitor comes into operation to aid regulator stability, reducing the loop gain and negative feedback with respect to frequency. The output impedance is a function of available negative feedback and as this feedback reduces with rising frequency, the output impedance rises. At about 10MHz the regulator runs out of gain and is no longer functional. The internal frequency compensation capacitor used in these devices has another, more important, effect on their behaviour. This capacitor has to be charged and discharged by the internal circuitry before the feedback loop can apply error correction. Inadequate charging current causes Transient Inter-modulation Distortion (TID) and Slew Induced Distortion (SID) giving line and load transient settling times of up to 5 microseconds with these devices.

For the SR3, I have seen 5uv @10-100Hz quoted for the SR3 and PSRR of 80dB @ 0-300Mhz!
 
On his Z100A Shunt regulator he quotes 110dB of PSRR from DC to 20kHz! 
 
The SR2 used to be much more expensive - and maybe this is the best lower cost LPS to get.  The one issue is very long back order que from what I have heard - I'm sure with the recent devaluation in the British pound - maybe even more so today.
 
Note these PS's need to be built to spec as the DC voltage is not adjustable.
 
 
Teddy Pardo - Teddy Cap PS -
These made a name for themselves replacing the NAIM power supplies.  Available in a range of voltages - $369 for a 5VDC version.
http://www.teddypardo.com/powersupplies/underthehood.html
 

 

Linear power supplies, such as those used in Naim power supplies are based on a transformer, rectifier, smoothing capacitors, and a linear regulator. When the AC at the output of the transformer is rectified and smoothed, a ripple is still present on the DC rail. This ripple has the form of a saw tooth which translates into noise at all audible frequencies and has a negative effect on the sound quality. The better the power supply can filter this ripple the better the sound will be. To get an idea of how much the power supply can affect the sound quality suffice to mention that almost all Naim preamplifiers are based on the same amplification circuitry and differ in their ability to connect more or better power supplies.​

The amplitude of the ripple mentioned above decreases with the size and quality of the smoothing capacitors being used. Naim Audio uses a brute force technique for the HiCap; they use a 500VA transformer and rely on large low ESR capacitors. Not only that this technique is less effective in noise filtering, but over time, due to their chemical construction, the quality of these Electrolytic capacitors decreases (ESR increases), and as a result the ripple increases. This is the reason for which Naim power supplies require expensive re-capping every 10 years or so. ​

The most common way to reduce ripple is to use a linear voltage regulator such as the LM317. The core of a linear regulator is an operational amplifier which monitors the output voltage and correct it using feedback. In order to reduce cost, most monolithic linear regulators, including the LM317 used by Naim for their power supplies, make use of a simple operational amplifier which generates its own additional noise. According to the LM317 datasheet it has a typical output noise of 0.003% of the output voltage. For an output voltage of 24V the output noise will be 0.72mV, not negligible...​

In addition, all linear (feedback) regulators, including the superior Jung SuperRegulator, have a limitation which is inherent to the way they operate, they can only fix a problem after it occurs. Since the speed in which they can fix the problem (regulate) is constant, their ripple rejection capability decreases with frequency. The LM317 has a ripple rejection of around 80db at 50Hz, which drops to merely 20db at very high frequencies (note that a -60db difference means x1000 times less!!!).​

Switch mode power supplies, such as the small wall-wart power supplies provided with the Logitech Squeezebox or with many popular DACs, are based on switching technology rather than transformer/rectifier/bridge combination. The idea behind these power supplies is to charge a coil/capacitor at high frequency and use the energy they store. The advantages of this technology are numerous which makes them very popular. They are small, efficient, low cost, and they are readily available from many manufacturers (mostly in China). The main drawback of these power supplies is that they are noisy, very noisy... Not only do they have a high level of high frequency noise on the DC, but they often inject high frequency noise to the mains which may impact performance of other hi-fi equipment. In short, less than optimal for high-end audio equipment.​

The SuperTeddyReg​

The regulator is the heart of the power supply. The TeddyCap and TeddyXPS power supplies are designed around the SuperTeddyReg a regulator that has been developed especially for high end audio circuits and surpasses any regulator that we have tested so far. The SuperTeddyReg is an evolution of the TeddyReg and provides even better performance. The SuperTeddyReg is also offered as a small independent circuit to the DIY community and is being used for a large variety of applications.​

The innovation in these power supplies is the incorporation of a very effective low-pass filter that in conjunction with a voltage regulator provides a theoretical (impossible to measure with conventional equipment) ripple rejection of more than 120db over the entire audible spectrum and above 1MHz. This powerful filter allows using smaller transformers and smaller smoothing capacitors. Since it does not rely on large smoothing capacitors to eliminate the noise, it does not require re-capping if the quality of these capacitors decreases.​

Smoothing capacitors are usually in the order of 4700-20000 uF, the SuperTeddyReg filter stage uses much smaller capacitors. The advantage of using smaller capacitors is that capacitors other than electrolytic can be used. Electrolytic capacitors are limited in their ability to filter high frequencies above the audio spectrum, and have leakage noise. Our power supplies make use of a combination of Tantalum, and PPS capacitors capable of filtering very high frequencies (over 1MHz). Surprisingly, filtering high frequencies even above 100 KHz is clearly audible.​

I have not seen noise or PSRR numbers for these power supplies.
 
Uptone JS-2 -
$925  Wow!  This is a beautifully totl LPS - with multiple DC outputs that have various voltage settings:
http://uptoneaudio.com/products/js-2-linear-power-supply
 
R-core transformer!  The choice of the R-core here speaks volumes:

Just powering the computer, the difference—between the R-core and a toroidal transformer in the bass was shocking. And in comparisons powering a DAC or other audio-signal-handling component, the sonic benefits ranged top-to-bottom, cymbals and piano to deep bass.  Plus R-core transformers, due to the gapless construction of the core, are mechanically silent. 

  John Swenson on the benefits of a choke-filtered linear power supply:
The traditional cap only filter (transformer, diode bridge, big cap) produces raw DC with a sawtooth riding on top. That sawtooth produces lots of high frequency components that the regulator has to deal with. Traditional regulators do very well at low frequencies, but have lousy characteristics at high frequencies which means a fair amount of those high frequency components from the cap-only filter get through to the regulator. Fancy discrete regulators do well at blocking the high frequency components, but add cost and complexity to a PS. Our approach is to use a properly designed choke-based supply whose ripple is a perfect sine wave, no high frequency components, thus a traditional regulator works very well. The discrete regulator is not needed to deal with the high frequency components, since there aren't any.

 All diode types except Schottkys emit a burst of ultrasonic noise as they turn off. This noise can go forward into the load circuit AND it can go back into the AC line, and it can also excite the transformer resonance. The "slow" diodes still have this ultrasonic noise. Schottkys are the only type which do not have this noise. Schottkys also usually have about half the voltage drop of other diode types and are usually faster. Which type to use depends a lot on what your supply looks like and what you are trying to optimize for. 
With a traditional low voltage design with a large cap right after a bridge you get large current spikes, these produce a large amount of high frequency noise which needs to be filtered by what comes after the cap. In this type of circuit the slow diodes can help cut down on the extent of the high frequencies generated by the sharp high current pulse. BUT they still generate the ultrasonic noise.

This is another reason why we like to use the choke-based design. With the choke there is no steep high current pulse, so no disadvantage to Schottky diodes. You get the advantage of no ultrasonic noise, lower voltage drop (so lower power consumption in the diode) and no big massive current pulses.

Have not seen DC noise or PSRR numbers for the JS-2 but I bet they are very good.
 

HDPLEX 100W Linear Power Supply - ​

OK R-Core Transformer  - nice.  But...the LDO regulator used is the Linear Tech LT1083.  High noise of 180uv and only  30dB PSRR at high frequencies - not great for a $395 PS.  Says low ripple and 'void of high frequency noise' what ever that means.  Any numbers?
http://www.hd-plex.com/HDPLEX-Fanless-Linear-Power-Supply-for-PC-Audio-and-CE-device.html

Feature

Top of the Line Hi-End Audio ELNA Capacitors High Quality 160W Silent R-Core Transformer provides clean energy No Humming or Buzz noise for 50Hz and Peak load, 100% Silent Linear Technology LT1083 Low Dropout Positive Fixed Regulators Low ripple noise and void of high frequency noise 19V/8AMP Output is User Ajdustable from 15V-19V via XLR Connector, Supprt Thin ITX/NUC/Auralic Aries Mini/HDPLEX 250W DC-ATX Converter

I'd prefer going with a TeraDak DC-30W or the MEIYAN for much less money and adding a DC iPur.  Or get in line for a Hynes SR3.
 
Next Part - the new SuperCap PS's and Batteries.
 
Cheers!

 

[rb2013]

I received this in a PM good stuff:

  Hi,
 
I have read all your three threads on supplies (BTW, congrats for them), and I hope many people read about them to understand that really important part of their audio setup.
 
Supplies are many times considered as a separate part of the circuit, but it certainly is not.
 
There is one very important parameter, probably the most, that is missing in your presentation: power supply impedance.
 
It's quite likely the one that most relates to audio quality.
 
Yesterday I went have a look at the LT3042 datasheet, and there's no impedance graphic, and it's probably the same with all other low-noise LDOs. 
 
There's been some articles published in The Audio Amateur magazine, and the posterior renamed mag versions, that deal with these questions and also improving techniques on existing "old" chips (like 3X7) supplies. Particularly grounding techniques and filtering.
 
Most articles are found on line, but I recommend at least four: one from TNT audio, one from Walt Jung, one from Andrew L. Weekes and the other one from Tangent Audio. 
 
http://www.tnt-audio.com/clinica/regulators_noise1_e.html
 
http://www.pearl-hifi.com/06_Lit_Archive/14_Books_Tech_Papers/Jung_W/Regs_for_High_Perf_Audio_1-4.pdf
 
http://www.andrewweekes.talktalk.net/Circuits.htm
 
https://tangentsoft.net/elec/opamp-linreg.html
 
Most of them are a series of articles, and you should read all of them. In W.Jung's site you will find other articles, like the one comparing impedances from several supply types.

 

[rb2013]

I wanted to add some of the information I found about the latest generation of USB gizmos - like the Regen and W4S Recovery - the Schiit Wyrd as well.  Both the Regen and the Recovery make powering your USB DDC (like the Singxer F-1) easy with the simple DC external barrel plug usually provided with the LPS.
 
All three devices filter the USB 5VDC power that then powers the attached DDC or DAC - the Regen and Recovery have external DC power ports  - the Wyrd is AC driven (by a linear power module - not sure if it's regulated or if the regulation occurs on the Wyrd board).  So here is the info:
 
W4S Recovery -



The Recovery uses a 9VDC input that is then further regulated to output 5VDC on the USB lines.  The low noise regulators in the Recovery product 13uv of noise ( I have been trying to find the exact one they use - will open mine up and take some closeup photos).  The Recovery prefers 9VDC - but can accept 7-10VDC.
 
Uptone Regen -



The Regen likewise has an ext DC power port - so makes using a LPS to power the 5VDC USB lines easy.  Like the Recovery it also has on board LDO regulators - filtering the inbound DC (from 7-9 VDC - 7.5 preferred) to 5VDC out.  Uptone does sepecifiy which LDO they use  - the TI TPS7A4700 - it has 4uv of noise.  Very nice!
 
Schiit Wyrd -



 
This may be a great lower cost power solution for USB devices - up to 500mA of clean linear power.  Using LM723 regulators - with 2.5 uv of noise - this unit for a direct power feed to a USB DDC may be a great solution.  I prefer the Recovery - with an external LPS - and a iFi iPurifier 2 in between (less the 1uv of noise).  But that is a more costly route.  Speaking of iFi
 
iFi USB3.0 Nano -


 
Another fairly inexpensive solution for USB audio.  The $200 iUSB 3.0 Nano - claims .5uv of noise when fed by it's included iPower 9VDC SMPS.  Being a SMPS - the iPower will produce much higher levels of noise back into the AC mains - possibly effecting other components - like your DAC.  But I have tried the iPower both directly and feeding the Regen and Recovery (and a older iUSB 2.0) to very good effect.  I'm sure this would be made even better with the Nano - a true bargain - as it's technology is the same as the iPurifier 2 - $109 on it's own.  For me I still prefer the LPS route through the Recovery and an iPur2 inbetween.  But certainly for $200 this is a very viable turnkey solution.

 

[rb2013]

http://www.diyaudio.com/forums/group-buys/167286-john-linsley-hood-ripple-eater.html
 
Found this interesting discussion of the 'John Linsley Hood Ripple Eater' circuit.
 

What is it?

The JLH Power Supply Ripple Eater is an add-on circuit that sits between a power supply and end user. The Ripple Eater measurably reduces power supply output noise and ripple. JLH demonstrated a reduction in noise and ripple on his own bench supply from a figure, measured over the range of 20Hz to 20KHz, of 300uV to 4uV. When he connected a good quality 470uF electrolytic capacitor across the same bench power supply, there was no measurable improvement in ripple. In a similar experiment, the output noise from a selected 7815 IC voltage regulator was reduced from 60uV, itself better than the maker’s specification, to 3.5uV, when measured over the same bandwidth. 

How does it work?

The circuit consists of two parts: a ‘ripple detector’ arrangement based on a long-tailed pair, and a constant current source. Any ripple voltage present across the supply line can modulate this current. The aim being that if the output voltage were to momentarily increase or decrease, the current drawn would automatically increase or decrease to oppose these output voltage fluctuations. The impedance of the circuit depends on the size of the capacitors used – particularly that of C4, which can be large since it only needs to be 1V working. With the values chosen in this iteration, the circuit generates the electronic equivalent of about half a Farad or more. 

Looks similar to the stage 1 in the MEIYAN LPS:

A long-tailed pair differential amplifier with current mirror and constant current source forms the first stage of the error amplifier.  

 

[occamsrazor]

Some other LPS options.....
 
Qualia Physic
http://www.qualiaphysic.com/
I read some good reports of these used with a Mytek Brooklyn DAC
 
Wyred 4 Sound PS-1 modular LPS:
http://www.digitalaudioreview.net/2016/10/wyred-4-sound-tease-ps-1-modular-linear-psu-streamer-at-rmaf-2016/
Just announced at RMAF 2016. Seems a bit expensive to me but a nice idea and design.

 

[wushuliu]

I Believe the Meiyan is really an Amb sigma 11 which can be bought custom built from ymb if you want to be assured of quality control.

 

[rb2013]

I Believe the Meiyan is really an Amb sigma 11 which can be bought custom built from ymb if you want to be assured of quality control.

Thanks for that- the AMB webiste gives more detailed information the MEIYAN LPS.
 
I would say for less then $200 shipped thefully assembled MEIYAN looks pretty good.  Nice quality of parts from what I can tell.
 
http://www.amb.org/audio/sigma11/

High performance but modest cost

1. Higher performance than IC regulators (such as LM317/LM337 or 78xx/79xx): Dramatically lower noise, lower output impedance, wider bandwidth, superior transient response, better line and load regulation.

 

Double ground plane

1. A ground plane on both sides of the circuit board, covering the entire board area (except under the heatsinks on the top side) provides a low impedance ground reference, shields against interference and allows optimized component arrangement on the board.

 

 
I like the attention to low impedance.

 

[rb2013]

  Some other LPS options.....
 
Qualia Physic
http://www.qualiaphysic.com/
I read some good reports of these used with a Mytek Brooklyn DAC
 
Wyred 4 Sound PS-1 modular LPS:
http://www.digitalaudioreview.net/2016/10/wyred-4-sound-tease-ps-1-modular-linear-psu-streamer-at-rmaf-2016/
Just announced at RMAF 2016. Seems a bit expensive to me but a nice idea and design.

I saw the W4S PS-1 announcement.  Nice they are using an R-core.  But no mention of noise or PSRR.  $499 for a single module version, $125 to add additional voltage modules.  Will be looking forward to more information on this one.  The power supply wars are heating up!
 
I did come across the Qualia Physics 271 SV - but wasn't to impressed.  Use of a toroidal transformer - number 1.  They quote:

• Very Low noise and ripple 

But in their specs quote:

RMS output noise           10Hz «"f"« 10Khz: 0.0027%
Ripple Rejection           72dB IOUT = [Ifull_load 4.5A]

Well .0027% equates to approx 140uv of noise - not much better then the LM317 or LT1083 (.003% - 180uv).  And they only quote 10-10kHz not 10-100Khz like most other LDO datasheets.  As for PSRR - they claim 72dB but give no freq range (which usually varies widely - and drops significantly at high freq) - or supply a PSRR vs Freq chart like most LDO datasheets.  With the use of a toroidal the job of the regulation circuit is even more critical for PSRR.
 
I bet it's not cheap too.

 

[rb2013]

More info on the MEIYAN 100W LPS:
http://www.ebay.com/itm/131867485038?_trksid=p2060353.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT
 
Similar DIY board:
http://www.ebay.com/itm/Ultra-Low-Noise-linear-power-supply-LPS-PSU-KIT-5V-9V-12V-15V-18V-24V/121940137982?_trksid=p2047675.c100005.m1851&_trkparms=aid%3D222007%26algo%3DSIC.MBE%26ao%3D1%26asc%3D39856%26meid%3Da0c7bbb26c4a4f94a43bdf673b99b76b%26pid%3D100005%26rk%3D4%26rkt%3D6%26sd%3D121760253760
 

This is a high-performance single-rail linear regulated power supply (PSU). It is basically half the  dual-rail PSU σ22, and possesses the same low-noise, high-current, excellent line/load regulation, wide-bandwidth, and stability characteristics. S22 is ideal for DIY stereo headphone amplifiers, preamplifiers, DACs, network media players, and other applications requiring a single-rail regulated PSU. 

The circuit uses all discrete components (no integrated circuits), and features high-current MOSFETs as the output "pass" devices. 

this PSU is derived from the σ22, which is the result of a community service project for the audiophile hobbyist. The schematic diagram is open source. The circuit design, as well as the printed circuit board layout are the work of AMB with participation from the DIY community. It is designed to be relatively easy to build, but it is not recommended as a first-time project for the beginner.  

http://www.amb.org/audio/sigma22/

The σ22 regulated power supply

Technical highlights

All-discrete topology

1. Single-pass, series regulator design.
2. No IC (integrated circuits) are used. This allows complete design control over all operating points and parameters for superior performance.

Complementary regulators

1. The negative regulator is a complementary mirror image of the positive regulator, except for the voltage reference (see "Tracking rails" below).
2. Thematic resemblance to the topology of the β22 amplifier.

Low noise, high PSRR

1. A constant-current source feeds a zener diode as a stable voltage reference. A low-pass filter (with a corner frequency of 1.6Hz) prevents zener noise from being introduced into the error amplifier. This is an effective yet lower-cost alternative to expensive voltage reference ICs. The low-pass filter also provides a soft-start characteristic.
2. The output noise (unloaded) is less than 10µV at ±30VDC output (measured using a Tangent LNMP (low-noise measurement preamplifier) and a Fluke 187 50000-count DMM in ACmV mode). The output noise is even less when the output voltage is lower. This is much better than the noise of an IC regulator based PSU tested under identical conditions.
3. The error amplifier is a discrete implementation of an opamp with a high open-loop gain of 102.5dB. The voltage supply to the error amplifier is isolated with capacitance multipliers to boost its PSRR (power supply rejection ratio). This greatly improves the line regulation performance of the PSU.
4. A long-tailed pair differential amplifier with current mirror and constant current source forms the first stage of the error amplifier. The second stage is the voltage amplification stage (VAS), also with constant current source load. The 3rd stage is comprised of the power MOSFET output devices configured as a source follower.

Tracking rails

1. The positive regulator's output voltage is based on the reference zener voltage and the gain of its error amplifier.
2. The negative regulator's voltage reference is the output voltage of the positive regulator. Its error amplifier has a gain of -1, so that its output voltage is the inverse of the positive regulator's output voltage. The negative regulator dynamically "tracks" the positive regulator -- any small voltage fluctuations on the positive rail also appear inverted on the negative rail, improving the CMRR (common mode rejection ratio) of the amplifier being powered.
3. The tracking behavior means that the voltage on both rails rise and fall equally. When used to supply a fully-complementary amplifier such as the β22, no "thump" noise is heard as the power is turned on or off.

High-current MOSFET pass transistors

1. Two paralleled high-current, highly reliable MOSFETs (rated at 17A each) serve as the "pass" transistor of each rail.
2. The high current rating provides a very high safety headroom against overcurrent damage.
3. The use of paralleled MOSFETs divides the heat dissipation, simplifying thermal management. Onboard heatsinks can be used which would allow the σ22 to supply up to 1A continuous (with much higher peak currents). More sustained currents are possible by using larger, offboard heatsinks.
4. The negative temperature coefficient of MOSFETs prevents damaging thermal-runaway conditions that may plague conventional BJT devices.

No current-limiting

1. The high-current MOSFETs are not normally the limit of how much current the σ22 PSU could supply, as long as they are adequately heatsinked.
2. The maximum current limit is determined by the rating of the power transformer, the rectifier diodes (the specified MUR820 devices are rated at 8A), and the AC line fuse.
3. The AC line fuse rating should be selected to protect the power transformer from overcurrent damage.
4. There is otherwise no current-limiting circuit in the σ22, which allows it to supply peak currents of many amperes. High transient bursts of current are always available, which some amplifiers require to avoid clipping and distortion.

Wide bandwidth

1. The all discrete topology allows the σ22 to be optimally tuned for the best combination of wide bandwidth and solid stability. Since the σ22's output impedance is much lower than even the best low-ESR large aluminum electrolytic capacitors, having wide bandwidth allows the σ22 to respond to fast changing current demands better than a large capacitor (or a bank of capacitors) ever would.
2. σ22's bandwidth extends beyond the audio band, and maintains supremely low output impedance in the µΩ range. (in fact, the hookup wire will dominate the output impedance).
3. As such, only a 1µF decoupling capacitor is used on each output rail onboard the σ22. The PSU can supply an amplifier with little additional capacitance for very fast response.
4. σ22 is also stable with a large capacitive load (tested to 10000µF), making it suitable for use in a wide variety of applications.

 

 

[astrostar59]

   

HDPLEX 100W Linear Power Supply - ​

OK R-Core Transformer  - nice, LT1083 (meh) 180uv noise, 30dB PSRR - not great for a $395 PS.  Says low ripple and 'void of high frequency noise' what ever that means.  Any numbers?
http://www.hd-plex.com/HDPLEX-Fanless-Linear-Power-Supply-for-PC-Audio-and-CE-device.html


I'd prefer going with a TeraDak DC-30W or the MEIYAN for much less money and adding a DC iPur.  Or get in line for a Hynes SR3.
 

I can vouch for the HDPlex unit. I have used mine for over a year and great performance. I nearly bought the Uptone Audio unit but as they told me, they couldn't make the HDPlex for double the price just in parts. The R-Core in both are the same unit. And the HDPlex has a 3 year warranty, you won't get that from many of the others.
It also has 4 outputs not one.

 

[rb2013]

 I can vouch for the HDPlex unit. I have used mine for over a year and great performance. I nearly bought the Uptone Audio unit but as they told me, they couldn't make the HDPlex for double the price just in parts. The R-Core in both are the same unit. And the HDPlex has a 3 year warranty, you won't get that from many of the others.
It also has 4 outputs not one.

Uptone said that the JS-2 uses the exact same R-core transformer?  The four outputs is a nice feature - but the LDO regulator they use has very high noise, and poor PSRR - it's an old design.  You'd expect better for the $395.
 
But if it works for you - great!
 
Here is the R-Core in the HDPlex:

 
And the JS-2:

 
They sure don't look the same.

 

[astrostar59]

 
Uptoner said that the JS-2 uses the exact same R-core transformer?  The four outputs is a nice feature - but the LDO regulator they use has very high noise, and poor PSRR - it's an old design.  You'd expect better for the $395.
 
But if it works for you - great!
 
Here is the R-Core in the HDPlex:

 
And the JS-2:

 
They sure don't look the same.

Here is the correct HDPlex photos. The unit you show is the wrong model. See the larger amount of Elna caps and the R-Core is the same apart from the copper shield. 
 

 

 


Not sure if the regulators are 'noisy'units, but that is part of the story, the caps also impact the noise and ripple reduction. The unit comes with CE approval and 4 other standards.
 
Where does it say LT1083, I can't see that quoted on the HDPlex site?

 

[rb2013]

  Here is the correct HDPlex photos. The unit you show is the wrong model. See the larger amount of Elna caps and the R-Core is the same apart from the copper shield. 
 

 

 


Not sure if the regulators are 'noisy'units, but that is part of the story, the caps also impact the noise and ripple reduction. The unit comes with CE approval and 4 other standards.
 
Where does it say LT1083, I can't see that quoted on the HDPlex site?

Thanks for the photos!  
 
See the beginning page of this thread -I'll repost it again.  Taken directly from their website under 'SPECS'
 
http://www.hd-plex.com/HDPLEX-Fanless-Linear-Power-Supply-for-PC-Audio-and-CE-device.html

Technical Specifications​ ​ Wattage 100W (Max 160W) ​ Feature Top of the Line Hi-End Audio ELNA Capacitors High Quality 160W Silent R-Core Transformer provides clean energy No Humming or Buzz noise for 50Hz and Peak load, 100% Silent Linear Technology LT1083 Low Dropout Positive Fixed Regulators Low ripple noise and void of high frequency noise 19V/8AMP Output is User Ajdustable from 15V-19V via XLR Connector, Supprt Thin ITX/NUC/Auralic Aries Mini/HDPLEX 250W DC-ATX Converter 12V/8AMP Output via XLR Connector, support Legacy Wavelet DAC/Qnap Nas/PicoPSU or 12V based CE device. 5V-19.5V/3AMP User Adjustalbe Output support SoTM PCIE USB card/MicroRendu/Upton USB Regen. 5V/3AMP Output support PPA (Paul Professional Audio) USB Audio card and Squeezebox. USB 3.1 Type C 5V Output

​

 

[rb2013]

It does have a nice bank of caps - but the ELNA's are not the best at PSRR.  They are fine caps for SQ - I have used the SILMIC II ELNA's inmy DAC60 mod project and in a pre-amp mod that I did.
 
Wouldn't use them in a PS filtering  - the Pannie FC's and FM's are better  - same for the Nichicon HW's.  Lower impedence and higher PSRR.
http://tech.juaneda.com/en/articles/electrolyticcapacitors.html

 

[rb2013]

See my PART 2  - I go over the more common LDO IC's
 

 LT1083: http://cds.linear.com/docs/en/datasheet/108345fh.pdf
 
180uV, PSRR 30dB (10-100K)

Not only do they have high noise - but poor PSRR.
 
It's the same ones used in this $80 LPS:
http://www.ebay.com/itm/161870356771?_trksid=p2060353.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT
 
IN contrast the MEIYAN has 13uv of noise at 24VDC and lower at lower voltages.  Very high PSRR as well.  Can deliver 5.5A at 12VDC - copper shielded R-Core transformer.
 
$152