Low cost high quality excellent performing LPSU for audio.

[One and a half]

Here's a good table that explains the differences between SMPS and Linear (regulated) Power Supplies. The rest of the texts comes from Acopian (not my words!). For audio use it's clear to use a linear power supply over an SMPS. The EMI at 60 kHz to 200 kHz is not acceptable, being responsible for high leakage currents from an SMPS which screw up DAC imaging plus a host of other reprehensible artifacts. It is that black and white.

 

 

Application guides

 

A linear regulated power supply regulates the output voltage by dropping excess voltage in a series dissipative component. 
They use a moderately complex regulator circuit to achieve very low load and line regulation. 
Linear regulated power supplies also have very little ripple and very little output noise.

General purpose use - including, but not limited to:

• low noise amplifiers
• signal processing
• data acquisition - including sensors, multiplexers, A/D converters, and sample & hold circuits.
• automatic test equipment
• laboratory test equipment
• control circuits
• anywhere that excellent regulation and/or low ripple is required

Noise (from page 6 of the pdf attached to this post) 

The linear supply is a continuous function unit with no discrete time clocking
or switching action. As a result, the output is virtually noise and ripple free,
and any noise seen at the load is due to pickup outside the supply itself in
the power wiring between supply and load. Using chokes and other filter
components as well as careful routing of the output cabling can attenuate
this noise. The linear supply itself does not generate any EMI or RFI.
In contrast, the switching supply is inherently a source of noise, with a
fundamental at its clock frequency as well as numerous harmonics. Typical
noise levels are on the order of hundreds of microvolts to tens of millivolts.
This is unacceptable for many applications where the output voltage is at
single-digit levels, or the load is sensitive to supply rail noise.

This switching-based noise can be filtered to some extent but is very difficult
to eliminate entirely. In addition to noise on the output cables, there is also
noise radiated by the supply, which can induce unexpected and frustrating
problems elsewhere in the system. Further, while filtering can attenuate the
output noise to an acceptable level, the problem of radiated noise is much
more difficult to manage.

Further, the frequency of the switching-induced noise may interfere with
other clocked signals, resulting in beat frequencies and other interfering
signals. In some cases, the switching supply’s clock frequency must by
synchronized with the system clock.

Further aggravating the situation, there are increasingly stringent
regulatory limits on how much noise a power supply can generate in
different frequency bands, both as a function of power supply wattage
and global zone. Some switching supplies meet the regulatory mandates
by using spread spectrum clocking, which spreads the noise energy across
a wide band. By doing so, the noise does not exceed allowed limits at
the clock frequency or its harmonics. While this technique works in the
“legal” sense to meet mandatory standards, the supply noise can still affect
internal system circuitry..

 

 

 

Linear-Vs-Switching-Power-Supplies-Whitepaper.pdf

[One and a half]

@airguitar there's no cheap linear power supply to buy off the shelf.

 

Please keep in mind that anything connected to a 100V-240V power network needs to comply with standards and certification to meet these standards is not cheap. If the LPS needs to be used in many countries, then the manufacturer needs to certify the use in each country that it won't cause a fire and behave as intended. Legal risk and all that.

 

It is not possible to obtain a GBP88 LPS 5V 2A without certification.  The ebay sample has no certificate of compliance, the insurance company will blow raspberries at you all day when your house burns down. Simple choice, far as I am concerned.

 

On that ebay LPS-25-USB PSU. Where's the isolation voltage to earth and input to output voltage rating, overvoltage protection on the output? High quality parts mean squat when the basics are missing.

 

Try Freddy Pardo, Sbooster, Farad, Plex, Paul Hynes, JCAT, few more if you scan these pages instead of shysters.

[One and a half]

On 1/10/2021 at 10:59 AM, Miska said:

 

It is not so clear.

 

Medical grade PSUs, like the one I referred to, have been specifically designed to have very low leakage currents, in µA range.

 

Even with standard SMPS wall-warts, it is entirely possible to design very low noise analog devices. I have measured enough many DACs running from SMPS that I know it is more about the design than the type of PSU. Overall, 60/200 kHz is much easier to filter out than 50/60/100/120 Hz. Especially since the fundamental and it's harmonics are outside of audio band.

 

Well, the measurements for so called low leakage supplies are usually in isolation, like the ifi iPower fabrications. Once they are connected to a load and actually supply voltage, the crap and mire they generate is audible soon enough.

 

On the subject of iPower. For a resurrection of the iTbube2 for tuner use, while on the bench was a good time to measure the iPower.

 

Rated at 15V, the output was 15.23 V DC. For a line voltage of 115V AC, the 0V output measured to the Active or Neutral was 52V AC. To kill this voltage, the output 0V needs grounding (to hard earth) and the easiest path to do this is via the iTube2 RCA outer shell cases which are dead short to the 0V output. So the signal leads carry 52V of leakage back to the amplifier. Eeuuw.

 

For those interested in following this thread, the values of AC volts between frames of connected devices, ideally is 0 V. This means no leakage currents can develop between devices, such as a DAC and amplifier. No leakage currents is a cleaner signal, not added with bonus noise.

 

In the chain measured just now in the (non-critical) office setup with a Fluke 289:

 

Accuphase E-450 Integrated amplifier frame -> Grace Design m930 DAC frame, connected via RCA, the potential reads 0.08V AC, could be better, but that's what it is with standard unshielded AC cables and not very studious where they are routed. With the iPower connected to the iTube 2 the same measurement points  rise to 0.32 V AC. Now the itube 2 also needs a DC-DC converter to raise the tube's requirement B+ to 50V something +. Hey, there's another SMPS, they're everywhere. Low noise, hahahaha, not. 

 

 

 

[One and a half]

9 minutes ago, airguitar said:

Thanks for that link.

 

Interesting the reference to batteries and noisy regulators. But at least the battery cannot inject noise into other equipment because it is not directly connected to the mains.

And a battery is incapable of generating RF and EMI. Depending on its size, it can run out of puff for dynamic swinging loads that are repetitive. 

 

99.3% (my guess) of noise from linear supplies are as a direct result of connected to an AC network with:

a) noisy SMPS close to the audio equipment

b) large switching AC devices, heaters, air con and faulty fridge starters

c) wearing out light switch contacts 

d) less optimum earthing (grounding)

e) incorrectly designed audio equipment to begin with (look at a few Pro audio devices XLR wiring)

f) fluorescent lamps

g) Energy saving = bulk harmonics generator light globes (each one has an SMPS)

[One and a half]

On 1/12/2021 at 6:24 PM, airguitar said:

Just thinking about the dynamic swing load statement - if I am simply using a solid state streaming bridge like a Primare NP5 which just takes 5V (ideal for a phone powerbank charger) - where do dynamic swings come into play here? I mean it's just a circuit decoding digital and surely it's the pre-amp or power amps job to deal with dynamic swings where the signal is analog?

Oh, when digital circuits work, they don't 'pull' current in a fixed linear smooth load. They pull currents that are pulses and vary with activity, hence why I referred to them as dynamic. Paul Hynes makes mention of this, so does Acopian, the power supply needs to cope with this kind of load, that's their recovery specifications. An SMPS unless it's oversized has some difficulty in keeping up, but then we suffer the noise problem anyway.

[One and a half]

8 hours ago, guiltyboxswapper said:

The off-the-shelf medical grade PSUs  whilst have low leakage current, still can't compete subjectively with an OK grade linear PSU in practice.  

 

Hopefully Taiko Audio's switching DC-DC converter will start to put an end to the snobbery that prevents HQ SMPS (of low leakage) being developed for this application.  They too have designed their DC-DC switching frequency well out of the way of typical audio bands.

The main criteria for the power supply does the load require a tight or loose regulation of voltage. 

 

If you can get away with it,  a transformer,  rectifier caps will give you the same regulation for the mains ac supply, with the transformer regulation thrown in. Not precise calcs, this psu will be 5% of voltage out plus minus. That will be bare minimum of noise from the psu itself, since no IC or other chip does any regulation. A choke on the dc side will improve further opposing large sudden variations.

 

DC powered DACs should have some very basic regulation built in, obviating the need for the upstream device to be regulated.

 

Motherboards are a different story and AC powered dacs have their own reasons,  so best not to venture there too deep.

 

DC-DC supplies still switch, for MHz switching conducted noise could be filtered,  but introduces emitted noise from the switching devices and would need to be well shielded.

 

 

[One and a half]

4 hours ago, John769 said:

Are you able to notice any obvious anomalies with this one?  Limited viewing from the pics though.

 

https://www.aliexpress.com/i/32966779449.html

The Leo looks like a well made and neat unit. The R core transformer is a nice touch, these are not cheap. The overpressure level at 0.4V is really overVoltageProtection and that figure is useless unless it is accompanied by a time value (ms usually).

The regulation is 90% chance 3 terminal regulators which do a decent enough job, but since they don't mention LDO (low drop out) regulators, then they would be the standard 78xx model or LM317T. 

The specified use is a router, yes, good choice, but not for audio like a DAC or anything in the signal path, like USB fixer.

 

There's no data sheet or compliance plate fitted to the unit, so it would not comply to any IEC let alone country specific standards, therefore not recommended.

[One and a half]

Often wondered what the differences were for the medical type grade EMC filter and the industry type, so here it is from TDK.

This filter has the built in switch and IEC320 male input as you find on appliances. The data sheet attached to this post.

The standard version has two caps after the choke, their centre point connected to earth (ground), whereas the medical filter does not so its leakage would be far less than the standard version.

Standard would remove quite a lot of Common mode noise, whereas the medical more suited to reducing differential noise. Same with some SMPS falling into the medical grade, how to have low leakage when connecting the two caps to ground is not effective. Need to point out that EMC filters typically reflect the noise back to the source, rather than absorb energy as is.

 

For all TN type AC systems (where the neutral is joined to ground) the E' and N' voltages are very close to one another, depends on how far away from the N-PE occurs. This means the cap from PE' to N' does hardly anything, since the voltage across it is so small. If the L to N were a 60-0-60 supply, then the cap N'-PE' would do a lot more work, this is the main reason for creating a balanced supply was created in the first place.

 

 

B84776.pdf

[One and a half]

2 hours ago, guiltyboxswapper said:

I take it you've not heard one in practice..... 

 

I had the revision before of the LEO in 19v format.  It was in a similar ballpark to the Farad Super3 of 19v (which everyone seems to adore on here) powering an Intel NUC, for approx half the cost.  

No I haven't listened, how to when there's no certificate of compliance? Hmm, that goes for Farad as well.

Anybody can build what they like, it will sound great and work for years, when it comes to conformity and build to rules, and then having it tested, well that's a different story and $$$.

[One and a half]

1 hour ago, guiltyboxswapper said:

 

They do, but they still prefer a cleaned voltage input to work from.

 

Sure, which then renders most products out of scope if you play entirely by those rules. 

 

So what does an audiophile do your case, stick to SMPS only?  

For my setup, I need 9V and 5V and both of these are provided by Acopian linear supplies which have accreditation. Not in the same leagues as Sean Jacobs or Paul Hynes as cost goes, they sure do sound better than the ifi smps they replaced.

[One and a half]

5 hours ago, jabbr said:

Common mode noise can be transmitted through the USB cable twisted pair. D+/-. twisted pairs are useful to reduce differential mode noise.

Any papers found on the transmission of common mode noise over USB? Don't need full blown thesis, but practical measured. I'm not in the industry, but have noticed few USB type medical sensors, even blood sugar monitors use IR for uploading data.

[One and a half]

14 minutes ago, asdf1000 said:

 

But WiFi is an RF source and can be right next to potentially sensitive analogue source ?

 

Although like anything, with good Engineering the issues can be significantly reduced at the DAC/amp end by the designer.

You raise a good point. 

Frequencies over 30MHz stay as radiated noise and can only be a problem when they become not radiated noise, but conducted noise (like through wiring (any type of wiring). This is very difficult though, since to be come conducted, the RF needs to be connected like an antenna, have the right components to tune to that frequency. That's how an AM radio works, jus hold one up to a computer and you can certainly hear the activities of a mouse and keyboard very clearly. An audio circuit is too low in frequency to induce these RF frequencies, since the components are very lossy to RF, and the RF would need to have lots of power and these devices are strictly regulated (except Russia where any thing goes I suppose).

 

Leakage currents, common mode noise exist in a conducted medium like a wire, and it's frequencies below 30MHz we should be worried about and keep them the hell out of a DAC or anything that is on the receiving end of USB, Ethernet, coax. Adding an SMPS is creating a common mode noise generator, NBG!

[One and a half]

8 minutes ago, plissken said:

 

I heard, whether Luddites realize it or not, digital, non-realtime, playback systems 100% play out of buffer and not off the wire directly like analog.

So does the buffer also playback the noise?