Linear vs Switched mode PSU

[anthonyTD]

I fully agree that it is preferable to have a linear supply. It's better to have very little noise in the first place than to have to go to all the time and trouble to filter it out. Also linear supplies are much more reliable than SMPS and vastly easier to repair if they go wrong. (SMPS generally best regarded as throw away if they go wrong unless an integral part of an expensive piece of kit)
However.... when we're talking £10 for the SMPS or £250 for the linear supply it is then quite an incentive to go to the time and trouble to filter and smooth the SMPS! And, if well implemented, the SMPS should then equal the linear supply for SQ.

The 60% efficiency of a linear (often less) V the 80-95% efficiency of a SMPS means we are increasingly finding SMPS the only choice, as the green bureaucrats increasingly get their way. Hence we will have to live with SMPS noise and get good at filtering it out well enough so it can match a linear supply.

[Arkless Electronics]

Apologies for going somewhat off topic....

..The Government has just announced the banning of diesel and petrol-driven cars from 2040 and thus we will all need electric cars to replace them. It takes at present about 8 hours to charge a car 'battery' and that this will take place mainly after the rush hour...ie in the evening when most of us, I assume, listen to hi-fi. Will the charging units be SMPS? Won't there be a significant drop in mains voltage? I wonder what effect, if any, it will have on SQ---always assuming that mains noise has a negative effect on SQ (!).

BW

Richard.

Yup they are already and will in the future be SMPS. VERY powerful ones! The cars themselves use switch mode control of the motors and do this at powers of 100KW+.... and as I've already mentioned many mains regenerators are SMPS and, in effect, a class D amplifier is a SMPS!

Then there is the SMPS that are used with domestic and industrial solar panels and wind generators to make DC or varying frequency AC into 240V AC and synchronise it with the grid.... at kilowatt and megawatt levels! They're everywhere!!

When properly designed though there is nothing to fear from them

[anthonyTD]

Hi Edward,
I meant the added noise generated by SMPS over a standard linear supply.
Sorry for any confusion.

What does 'additional generation of noise point' mean please?

I like to think pictorially and I found the block diagram in this article very useful in understanding things (even though I imagine some here will object to some of the narrative). Are you saying that less noise is generated by the transformer in a LPSU vs. a switching PSU?

@Jez, from your explanation it would seem that an optimum audio PSU would be an SMPS but with well built smoothing and regulation most often found in LPSUs. Correct?

If so perhaps I can simply just take the output from my existing SMPS and route it via the Paul Hynes smoothing/regulators I have. thereby skipping the transformer I have. Not sure what the rectifier circuit on the board would do with a DC input. Rectify it to AC?

I'm hopeless in understanding all this but I'm trying.

[Arkless Electronics]

What does 'additional generation of noise point' mean please?

I like to think pictorially and I found the block diagram in this article very useful in understanding things (even though I imagine some here will object to some of the narrative). Are you saying that less noise is generated by the transformer in a LPSU vs. a switching PSU?

@Jez, from your explanation it would seem that an optimum audio PSU would be an SMPS but with well built smoothing and regulation most often found in LPSUs. Correct?

If so perhaps I can simply just take the output from my existing SMPS and route it via the Paul Hynes smoothing/regulators I have. thereby skipping the transformer I have. Not sure what the rectifier circuit on the board would do with a DC input. Rectify it to AC?

I'm hopeless in understanding all this but I'm trying.

To hugely simplify things.. the faster the switching occurs the more noise it generates. The slower the switching the less efficient it will be. Linear supplies (we should really be calling them something else, such as conventional supplies, as they are far from linear! Linear regulator is accurate though) operate at mains frequency of 50Hz but switch at the top and bottom of each cycle so generate raw DC as pulses at 100Hz ie on and off 100 times a second. This is super slow motion in electronics terms! SMPS switch at around 100KHz ie 100,000 times a second and great effort is put into making each switching action as rapid as possible in order to get the best efficiency and the least waste heat.

Due to the above SMPS generate much more noise than a linear supply. However, it's at ultrasonic frequencies (above audibility unless you're a bat), whereas a linear supply generates noise at very audible frequencies.

I'm not saying that the "optimum audio PSU would be an SMPS but with well built smoothing and regulation most often found in LPSUs" no, but I am saying that with "well built smoothing and regulation most often found in LPSUs", correctly applied, they can be just as good as LPSU's.

Voltage regulators require a headroom of voltage above that wanted from their outputs. The minimum allowable being called the drop out voltage, as if it goes below this the regulator drops out of regulation. It will usually continue to give a voltage output but it will be all over the place! To use your existing SMPS you would have to be able to turn up its output to around 7-8V in order to regulate this down to 5V by a linear regulator. You would also have to bypass the rectifier as they only convert AC to DC, not the other way round. It could simply be removed and it's connections bypassed. If your SMPS can be turned up to 8V then it's easily doable but as you already have the mains transformer and case etc it makes sense to use it

[Edward]

To hugely simplify things..

<snip>

Sorry, yes I did mean 'well built .... as often found in LPSUs'.

No I was not actually suggesting I use an SMPS and direct its output to the Hynes regulators boards - just theorising for the purposes of understanding things. Your answer was yes so helpful.

So, if I may sum up in a simple way?

A PSU that is well built and most importantly has quality smoothing/regulation will be just the same as another PSU of similar quality build regardless if linear or switching mode. However in practice switching PSUs are often not as well built as linear PSUs leading to the general observation that linear PSUs are 'better'.

Perhaps that simplifies things too much for your taste but are those two sentences accurate as they stand?

[Arkless Electronics]

Sorry, yes I did mean 'well built .... as often found in LPSUs'.

No I was not actually suggesting I use an SMPS and direct its output to the Hynes regulators boards - just theorising for the purposes of understanding things. Your answer was yes so helpful.

So, if I may sum up in a simple way?

A PSU that is well built and most importantly has quality smoothing/regulation will be just the same as another PSU of similar quality build regardless if linear or switching mode. However in practice switching PSUs are often not as well built as linear PSUs leading to the general observation that linear PSUs are 'better'.

Perhaps that simplifies things too much for your taste but are those two sentences accurate as they stand?

Basically yes. It takes rather more thought and effort to get the SMPS as quiet as the linear supply but yes it can be done by suitable smoothing and filtering and especially by then adding a linear regulator after that.

Most applications for power supplies are nowhere near as sensitive to these issues as hi fi equipment and most are intended for generic universal use ie supplying 12V for a printer or phone or router or lights etc etc. Unless specifically designed for very low noise etc by the addition of the means above, a universal linear regulated supply will be much quieter and better regulated than a universal SMPS.

If the PSRR of the equipment being powered is superb then it may sound the same whether the horrible noisy SMPS is used or the super duper linear regulated one.

Many/most items such as DAC's phono stages, etc that have an external PSU will have a voltage regulator built in and at least a little smoothing, meaning that so long as the SMPS is at least a decent one it should give the same results as a linear one.

El cheapo gear without even an internal voltage regulator will be very sensitive to the PSU quality.

Unregulated linear PSU's, especially of the wallwart variety (no room for really big smoothing capacitor), are the noisiest of the lot and instead of the SMPS's 0.005V of 100KHz noise will often have a volt or so of very audible 100Hz noise!

[bumpy]

So, if I may sum up in a simple way?

A PSU that is well built and most importantly has quality smoothing/regulation will be just the same as another PSU of similar quality build regardless if linear or switching mode. However in practice switching PSUs are often not as well built as linear PSUs leading to the general observation that linear PSUs are 'better'.

Perhaps that simplifies things too much for your taste but are those two sentences accurate as they stand?

Edward, you need to be careful that you have taken one persons opinion and turned it into a statement of fact.

[Arkless Electronics]

Edward, you need to be careful that you have taken one persons opinion and turned it into a statement of fact.

... 'cos obviously on the subject of electronics, the opinions of accountants, brick layers, bakers, solicitors etc etc who have read a few "Hi-Fi Answers" magazines and been on hi fi forums for a while are equally valid to those of electronic engineers specialising in audio electronics for decades who design and repair hi fi equipment professionally.... I guess if you were ill you'd enlist the help of some bloke who saw the first 3 episodes of "Holby City" rather than a doctor? Obviously their opinions on medicine would be equally valid....

You couldn't make it up.

[Gazjam]

The thing is....
Not all EE's agree on the best way to do things.

Is there perhaps an element of 'art' in the Design that goes beyond raw specifications?
Is this where keeping an open mind (engineering curiosity!) advances the established current best?

No doubt all current EE theory is sound, but is there an additional 'art' to design that advances the state of the art?
Your 640P phonostage is maybe a good example of this?

For myself, an illustration might be in having a Haynes Manual for my Car does in no way convey what its like to own and drive it?
Theres an art to design that goes beyond specifications.
As one who does it professionally myself I've seen this in a lot of different walks of life, hifi included.

Just sayin' fella

[RothwellAudio]

The thing is....
Not all EE's agree on the best way to do things.

And neither do all doctors, but there tends to be a general consensus.

Is there perhaps an element of 'art' in the Design that goes beyond raw specifications?

Yes, there's an art to it whether you go beyond raw specifications or not. There are many ways to skin a cat and there are many cats to be skinned in an amp design. There are obvious ways to do things and less obvious ways. In adds up to art in my opinion. In fact, one of the classic electronics textbooks is called The Art Of Electronics (by Horowitz and Hill).