Linear vs Switched mode PSU

[stairpost]

There is only one "measurement" that I apply to HiFi, and that is long-term listening enjoyment. Sitting down and critiquing a system for an afternoon is not that.

I have to agree with you again, I don't recall ever sitting and listening to a system that belonged to a friend or aquintance and mentioning anything I might not have liked about it. It's their taste not mine, it is much more enjoyable to discuss what I do like about it.

[MrRadish]

Overall a mixed picture I'm guessing

My take is that the quality of the design and the parts used is far more important than the type of supply it is - a good SMPS will be just as good as a good LPSU. Having said that I still filter my mains so all LPSU are on one regenerator and all SMPS are on another - just to cover all the bases.

[Marco]

...a good SMPS will be just as good as a good LPSU.

My concern is over what the 'S' stands for in 'SMPS'.

How much will the switching effect in any SMPS (even the best ones) inject interference/noise into the mains supply that my precious hi-fi components are ultimately sharing with it......??



Marco.

[Bonky]

How do we know if out SMPS units are injecting noise (besides using one's ears and guessing)?

Is there a cheap way of objectively measuring it? (I'd imagine one uses an oscilloscope - and there are cheapish ones on eBay -but if that's all one can use - are they any good?)

Richard

[Marco]

How do we know if out SMPS units are injecting noise (besides using one's ears and guessing)?

As a subjectivist, Richard, I just use my ears.

One such recent experience was simply having a quality linear PSU built for the Seagate hard-drives, suppling music files to my Raspberry Pi streamer, which replaced their stock (el-cheapo) SMPS units, and which had a significant marked effect on the SQ of the Pi, simply by removing the noise that the stock SMPS units were pumping into the mains.

Honestly, I was utterly shocked at the level of sonic improvement gained, which ably showed just how shit those mass-produced SMPS units are. Just think about what all the other ones are doing inside TVs, SKY boxes, fridges, computers, phone chargers, you name it, which are currently plugged into your mains...!

I also obtained similar benefit in replacing the SMPS unit for the RPi itself with a quality linear one, from Paul Hynes (his SR3DR-05). If switching noise could be measured, and thus confirmed objectively, then even better, but either way I'll always judge these things with my ears and experience

Marco.

[MrRadish]

Putting my router and NAS on the same regenerator as the rest of my system caused a dreadful clouding of the sound, so they got moved off again pretty quick! But then my NAD M51 never seemed to cause any trouble, and nor did the little SMPS that came with my Caiman, so either it depends on how much power they draw or some are less trouble than others.

[Marco]

...or the SMPS in your NAD (for example) was designed better, and of superior quality, to the ones used in your router and NAS? And that was why it didn't appear to cause a negative sonic effect.

That's why I'd like Jez to confirm how much cleaner SMPS units are in top-quality hi-fi equipment, compared with those commonly found on inferior consumer goods, and crucially, if they can ever be as clean [let's call it 'kind' on the mains supply] as a top-notch linear design...

However, the negative experience you've outlined mirrors my own, which shows that you should always keep the 'noisiest' components in your system well away from the rest!

Marco.

[Yomanze]

I think Jez is referring to the fact that his units have very high PSRR, so are pretty much immune from the effects of a SMPS. However, the impact on the entire audio chain may tell another story.

[anthonyTD]

. However, the impact on the entire audio chain may tell another story.

[Starterman]

May I ask what was your solution please?

Yes of course you can ask


Actually there are no secret or special parts in it...

An EI core transformer (mine is a 30VA pcb mount split bobbin type, but a C-core would probably be fine) - don't use a toroidal one, the sound is horrible.
Use centre tapped FW rectifiers (2 diodes). The diodes should have low slope resistance. Soft recovery might be a benefit, speed is not critical. The diodes are fine on the board below.
This is a good sounding reg:
http://www.ebay.co.uk/itm/Ultra-low-...8AAOSwo8hTnqIV
You will need to use the 0V connection to effect the 2-diode rectifier and lift a couple of the on-board diodes' legs.

The LT3042 board on ebay is also supposed to be good. I haven't tried it. I did try an mravica reg off DIYAudio - not to my taste.