DIY Audio....or how many solder burns have you had?

[Mr. C]

Being around soldering equipment every day, very, very rarely do I receive burns, thankfully. Although the odd cut around the lab in inevitable just a fact of electronics lab life.

With regartd to SMPS, I design a great many of these for various applications, the main pluses are PDF (Power density factory) high power for lower space, COST and for our US cousins less real estate to take up.

With the newer techniques involved with SiC & GaN switching and much better EMI supression and power distribution network (fancy way of saying psu micro controller lol) You see all of these are controlled in real time, with precise control on of input V&I, switching times, PWM lengths, current supply against output requirments. Coupled with the biggie these days efficiency if over 80%+ in many circumstances oh did I mention the cost.

This is why a few amplifier manufactures are heading in this direction, I mean even Sugden uses SPMS in their design.

Oh wait both Sugden & Pass are class 'A' that permantly on transistor current drenched electricity consuming monsters just what the world needs right now ?

[Macca]

Being around soldering equipment every day, very, very rarely do I receive burns, thankfully. Although the odd cut around the lab in inevitable just a fact of electronics lab life.

With regartd to SMPS, I design a great many of these for various applications, the main pluses are PDF (Power density factory) high power for lower space, COST and for our US cousins less real estate to take up.

With the newer techniques involved with SiC & GaN switching and much better EMI supression and power distribution network (fancy way of saying psu micro controller lol) You see all of these are controlled in real time, with precise control on of input V&I, switching times, PWM lengths, current supply against output requirments. Coupled with the biggie these days efficiency if over 80%+ in many circumstances oh did I mention the cost.

This is why a few amplifier manufactures are heading in this direction, I mean even Sugden uses SPMS in their design.

Oh wait both Sugden & Pass are class 'A' that permantly on transistor current drenched electricity consuming monsters just what the world needs right now ?

You can't leave a class A powered up all the time so I'd suggest they actually use less power than a B or AB or even a D if that amp is left powered up all the time.

Also world is moving to electric vehicles which in terms of power consumption make a big class A amplifier used say 20 hours a week look like nothing. Please don't encourage the 'green' virtue signallers who have to tell us they are abandoning class A 'For the planet'. They are so tiresome - and wrong.

SMPS have lower noise than an LPS assuming both implemented properly, is that not true?

[Barry]

SMPS have lower noise than an LPS assuming both implemented properly, is that not true?

My only experience with SMPS is with the little triangular 'wall warts' used to charge mobile phones. I have one plugged into a double socket along with the bedside radio (analogue). If the phone is being charged, there is a background noise on the radio. This does not occur if I use a charger plugged into a socket on the other side of the bed. I don't know if the interference is mains borne, or if it is RFI.

[Macca]

My only experience with SMPS is with the little triangular 'wall warts' used to charge mobile phones. I have one plugged into a double socket along with the bedside radio (analogue). If the phone is being charged, there is a background noise on the radio. This does not occur if I use a charger plugged into a socket on the other side of the bed. I don't know if the interference is mains born or if it is RFI.

Possibly the problem is with the radio's PS not filtering very well? Hard to draw any definite conclusion without testing the specific application.

I use a mobile phone charger to power a DAC and there's no noise I can hear even up close to the speaker. It's plugged into the same multi-way block as the pre-amp and transport. I can't discriminate it from the other DAC I own, which has twin built-in LPS.

[Barry]

Possibly the problem is with the radio's PS not filtering very well? Hard to draw any definite conclusion without testing the specific application.

I use a mobile phone charger to power a DAC and there's no noise I can hear even up close to the speaker. It's plugged into the same multi-way block as the pre-amp and transport. I can't discriminate it from the other DAC I own, which has twin built-in LPS.

Agree, it is difficult to ascertain if the interference is mains-borne, or if it radiated. It doesn't seem to matter how close the phone is to the radio, nor does playing around with the radio's telescopic aerial alter the effect. So that would exclude RFI, whereas moving the charger to another socket solves the problem, which suggests the interference is mains borne.

The radio is an old Roberts radio, designed and manufactured long before digital and SMPS became prevalent.

[Mr. C]

Well Martin

I spend maybe 30% of my week working on PDN's / IoT / medical and other switching power devices which the key criteria’s are lower more efficient power with a smaller footprint which is the correct way forward. However, to obtain a real quality low noise smps then the use of many methods is employed including specialist materials, careful high side layout and magnetics, low side with linear tracking regulators, in depth DC<>DC filtering (post smps).

Have tried many very well thought out smps designs on a few audio amplifiers (and RF ones lol) it works perfectly fine and with proper consideration can pass the UKCA/CE Cisper regulations.

90% of my lab bench power supplies are linear I have one half and half a Rohde & Schwarz smps with linear tracking regs, it’s the same size and the equivalent LPS but half the weight. and 2.5 times the amount of ripple noise.

only very, very few clients are specifying class 'a' designs these days for a multitude of reasons one being the physcial size, the weight of the unit, the power consumption and sheer heat they generate.

Lets look @ your Krell KSA50S it takes pretty much 345Watts of power @240Vac at idle which is roughly 450Va which for a quoted genuinely 50W class 'a' device is on the money at more realistic volume levels this isn't going to rise to much as the devices (FET's) are 100% duty cycle as in on permanently all of the time.

So, in 20 hours you will have used 0.35 of Kwh x 20 so around the 7Kwh @ the capped rate of 0.34p per Kwh gives you £2.38 for the amplifier total electrical use (plus whatever the rest of the system uses in that period) so compared to a EV being charge at bollock charge between £40-£70 yes its relative indeed.

A class a/b amplifier is around 55-65% efficient so you could reduce that outlay by a further 25-30% and with a class 'd' style amplification by at least 100%, this is not about which one sounds better purely on consumption.

Also you have a small amplifier in terms of most of the current class 'a' units which are out there, recently visited a cleint who use the older big krell mono's with the meters on them, you could literally wtach the electric meter jump to warp when he switched them on.

Although the big Gryphon Colosseum’s idle around 3.2Kw each and double up as house heaters but do sound superb.



Personally class 'a' has never worked for me, plenty of others yes most are overly safe and somewhat nice which is why the big fad for class 'a' was kicked off in the mid 80's due to the slew of forward, hard sounding glut of average solid state amps but a multitude of manufacturers.

Back to would I use a correct smps in any audio products like amps, maybe one or two designs would lend themselves to this for sure (would it sound better or worse or yes, but as I mentioned cost is the key here), however for myself and the designs we create not yet Martin.

[Barry]

only very, very few clients are specifying class 'a' designs these days for a multitude of reasons one being the physcial size, the weight of the unit, the power consumption and sheer heat they generate.

Lets look @ your Krell KSA50S it takes pretty much 345Watts of power @240Vac at idle which is roughly 450Va which for a quoted genuinely 50W class 'a' device is on the money at more realistic volume levels this isn't going to rise to much as the devices (FET's) are 100% duty cycle as in on permanently all of the time.

So, in 20 hours you will have used 0.35 of Kwh x 20 so around the 7Kwh @ the capped rate of 0.34p per Kwh gives you £2.38 for the amplifier total electrical use (plus whatever the rest of the system uses in that period) so compared to a EV being charge at bollock charge between £40-£70 yes its relative indeed.

A class a/b amplifier is around 55-65% efficient so you could reduce that outlay by a further 25-30% and with a class 'd' style amplification by at least 100%, this is not about which one sounds better purely on consumption.

Also you have a small amplifier in terms of most of the current class 'a' units which are out there, recently visited a cleint who use the older big krell mono's with the meters on them, you could literally wtach the electric meter jump to warp when he switched them on.

Although the big Gryphon Colosseum’s idle around 3.2Kw each and double up as house heaters but do sound superb.



Personally class 'a' has never worked for me, plenty of others yes most are overly safe and somewhat nice which is why the big fad for class 'a' was kicked off in the mid 80's due to the slew of forward, hard sounding glut of average solid state amps but a multitude of manufacturers.

One of the reasons why I stopped using the Mark Levinson ML-2 monoblocks (into Quad 57 ESLs). They are big, heavy, very inefficient and run hot (so much so, the black anodising of the heat sinks had started to bleach).

In theory class A amplifiers can never be more than 50% efficient. The ML-2s are true Class A designs, wherein the mean audio output power doubles for each halving of the load impedance. Specifically the ML-2 provides 25W into 8Ω; 50W into 4Ω; 100W into 2Ω, and 200W into 1Ω - that is, provides a constant rms voltage of 20V, although given the power supply current capacity of 7A, the power into 1Ω should be taken cum grano salis. The power consumption from the mains (described as "idling" power consumption) is "around" 400W, or 8x the audio power into 8Ω, that is an efficiency of 12.5%.

So for the Krell KSA50S 50W/channel amplifier, the idling power of 345W implies an efficiency of 29%; more than double that of the Levinson design.

I now use a pair of Quad 510 monoblocks (or a pair of 'monoblocked' Quad 405 amplifiers (upgraded to Keith Snook level 3) and with improved power supplies). Whilst legend has it that the combination of Levinson ML-2 amplifiers with the Quad ELS was a "match made in heaven", I can say that the replacement of the ML-2s has not resulted in a degradation of sound quality.

[Mr. C]

SMPS have lower noise than an LPS assuming both implemented properly, is that not true?

For very small psu for IoT and wearable tech work I would say yes, simply due to the nano size, we have devices that 'sleep' using pico (One million millionth (10-12) of an ampere.) or femto amps, (One quadrillionth of an ampere. An ampere being the SI base unit of electric current. 1 Femtoampere = 10-15 amperes. 1 pA = 0.000 000 000 000 001 A.) microscopic amount of currents.

Yes you can easily measure the amount of both radited and conducted rf noise from any electrical device with a quality spectrum analyser/ A/C power analyser/ H & E field probes and for current a good DVM with mA and pA ranges all calibrated. You may also require emc tent/ testing chamber etc.

All of this has to checked and tested by an approved CE / UKCA testing facility for the relevent standrards according to each product being sold.

For higher currents and lower noise then LPS is the way to go if you wish to spend the money. remeber with a traditional LPS then you only have here in the UK 50 cycles @ second in which to fill that capacitor with a smps you can have upto 250K times a second to fill it, hence why they are much smaller caps on a smps but MUCH more RF noise. You have to kill the noise for both conducted and radited emmissions along with some form of dc filtering.