SMPS in audio

Linear power supplies sag under load. So the voltage is high when it should be low, leading to unnecessary dissipation in the output stage. And the voltage is low when it should be high during peaks, causing a nominal 30V linear ps based amp to clip way before it's brethren with a 30V SMPS.
 
Oh but then you just use 40V and series regulators and waste the 10 volts@whatever current as heat. Oh and don't forget the big expensive heatsink you'll need along with the pass device. 😛
Just kidding - nobody really does that, do they?
 
No one sane does that in 2022, only people that haven't kept up with technology.

Of course it helps to make differences in topologies larger than they have to be by referring to classic designs with pronounced drawbacks. That is what the other side does as well. It also helps to confuse cheap consumer grade 2 years max. lifetime SMPS with industrial grade stuff practically never found in audio devices. Therefor the SMSP versus linear discussion begins to become a religious kind of thing. People denying the very short lifetime of consumer grade SMPS is an awkward item as well as I regularly see defective wall warts at other peoples places too.

Whatever can be said of both topologies, there is no denying that even a 30 year old linear regulated PSU wasting 10V@whatever current is way cleaner than the average consumer grade SMPS as found in low power audio applications. If one is after performance the choice is simple. If being cheap is the goal in life the choice also is simple.
 
Last edited:
  • Like
Reactions: kodabmx
Therefor the SMSP versus linear discussion begins to become a religious kind of thing..... People denying the very short lifetime of consumer grade SMPS .......If one is after performance the choice is simple. If being cheap is the goal in life the choice also is simple.
I am very happy that you make these points.

It is not a religious thing but an engineering decision to use an SMPS. I never had an SMPS fail. All the best amps currently in production use a switcher. You just get more out of an amplifier stage using an SMPS, both in terms of output per dissipated Watts of heat and in measurable quality.

It 's Kuhn all over again.
 
  • Like
Reactions: TNT and cowanaudio
It is also the same misunderstanding again. I try to make clear that consumer grade SMPS are definitely not the last word in low power audio applications. That is lower than 25 VA. Even 78xx based PSUs outperform recent cheap low power SMPS wall warts/adapters as delivered with audio devices.

For some reason power amplifiers and industrial stuff are taken by the hairs to the discussion. Low power = preamps/phono/streamers/DACs.

The choice for cheap OEM low power SMPS in audio is not made by engineering but because of costcutting/avoiding certification of PSUs.

No idea who or what Kuhn is and what he/she/it adds to this. I am glad that you're happy 🙂
 
Last edited:
The OP, who has not been involved in this discussion lately, specifically asked about SMPSs in power amplifiers.

"Thomas Kuhn argued that science does not evolve gradually towards truth. Science has a paradigm which remains constant before going through a paradigm shift when current theories can't explain some phenomenon, and someone proposes a new theory."
 
The OP opened this "SMPS in audio" thread 20-10-2021 and has not been active in this thread for a while but you are right. Being professionally and hobby wise involved and to see defects and performance differences on a daily /weekly basis (and repair them if possible) and then read that people never have had a defective SMPS 😀 Maybe Thomas Kuhn also had to say something about statistics?!

All has been said, enjoy your SMPS (and do NOT measure them)!
 
Last edited:
Output noise on an SMPS is almost all outside the audio band. Look at one on a CRO and you could quickly jump to the conclusion that the 100mV p-p hf noise is terrible. The noise spectrum below 20KHz should be very low on a well designed supply.

The >1V p-p ripple with components all within the audio band from an unregulated linear power supply are far more difficult to manage. Low frequency radiated magnetic fields will surround all your sensitive electronics.
 
Output noise on an SMPS is almost all outside the audio band. Look at one on a CRO and you could quickly jump to the conclusion that the 100mV p-p hf noise is terrible. The noise spectrum below 20KHz should be very low on a well designed supply.

The >1V p-p ripple with components all within the audio band from an unregulated linear power supply are far more difficult to manage. Low frequency radiated magnetic fields will surround all your sensitive electronics.
Apples are better than moldy bananas.
 
This thread was about SMPS used in a power amplifier application. Very few linear regulated power supplies are used with power amplifiers. That is why I compared a linear unregulated supply with a regulated SMPS. This is completely fair because a quality regulated SMPS will likely cost less than an equivalent output unregulated linear supply.

I think apples have a greater variation in quality than bananas, and are also more time consuming to consume, but that is IMHO. 😉
 
I've fixed thousands of linear supplies in that time and many hundreds of SMPSs, despite....more SMPSs on that site than linear supplies.
Hi William

Don't waste your time with people who play "politics"🙄 lend me the benefit of your expertise instead😉
In your service experience have you seen any common failure modes that would help to inform the build of a hi-reliability SMPS?
Have you any views on circuitry, say Ćuk vs LLC or similar issues?
I tried to start a few threads on this, the latest is here https://www.diyaudio.com/community/threads/fundamentals-of-power-electronics-errors.387414/
I can send you links to the others if you want, but didn't bother because, despite a few helpful and informed contributors, there's a lot of the usual denialism (even some of the same deniers!)
I can't be bothered to contradict them because who cares about evidence when people have opinions🙄
But you could perhaps mention that the amplifier with the best measurements ever done on Audio Science Review had an SMPS,
as does the second best,
and the third best,
and, well I was too bored but all the quietest until I stopped, I couldn't even find a top amp with a linear PS, found bad ones however.


Best wishes
David
 
Last edited:
Hi William

Don't waste your time with people who play "politics"🙄 lend me the benefit of your expertise instead😉
In your service experience have you seen any common failure modes that would help to inform the build of a hi-reliability SMPS?
Have you any views on circuitry, say Ćuk vs LLC or similar issues?
I tried to start a few threads on this, the latest is here https://www.diyaudio.com/community/threads/fundamentals-of-power-electronics-errors.387414/
I can send you links to the others if you want, but didn't bother because, despite a few helpful and informed contributors, there's a lot of the usual denialism (even some of the same deniers!)
I can't be bothered to contradict them because who cares about evidence when people have opinions🙄
But you could perhaps mention that the amplifier with the best measurements ever done on Audio Science Review had an SMPS,
as does the second best,
and the third best,
and, well I was too bored but all the quietest until I stopped, I couldn't even find a top amp with a linear PS, found bad ones however.


Best wishes
David
Looking for reliability inside smps - some remarks.
Only a small amount of parts cause the most failures.
These are the ones most stressed.
Most common failures known to me are:
Secondary side schottky rectifiers often found inside TV-panels.
Worst case is some schottkies parallelled without thermal coupling.
Silicon-Diodes instead get slightly hotter, but live much longer.
Secondary electrolytics. These are exposed to high ac-ripple current,
worst case are flyback converters as found in 5V chargers.
The lowest voltage often is combined with maximum current and the tiniest bulk capacitor.
Do the math and you will place bigger caps there.
Primary switcher: Integrated monolithic switchers are not really rugged as myriades of failures indicate.
You better take a separate controller that drives a discrete MOSFET with specified over voltage surge rating.
All this applies to flyback converters - a simple topology with general high component stress.
Resonant converters like the LLC are more complex but work smoothly and are my choice for reliable smps.
Anyway you should focus on hot spots, a FLIR will give you some useful hints.
 
  • Like
Reactions: cowanaudio
G'day David

If you asked me two years ago I could have given you a VERY thorough answer! We kept a comprehensive failure history database that was easy to search and pull metrics out of.

The majority of failures were primary side. HV rectifier, start up supply (whatever form that took, sometimes just a resistor) auto voltage switching, surge supression and the main switching device. Mains power switch failures were not unheard of either! More than half these failures had an "event" that caused the damage. The power feeds were in many cases over 1Km away from the source of power in the power house. A lightning conductor was connected to the steel and aluminium structure the power supply was bolted to. A good storm would lead to many failures, even linear supplies. A greater percentage of linear supplies died from old age than the switchers, that were more likely killed than to die from old age. In our main equipment room that was close to the site earth mat and power station almost no SMPSs failed compared to the linear supplies, that still died from old age.

The general topology of the supply didn't appear to matter, but the small details did. A single constantly powered up resistor for a start up supply is a poor choice, for instance. Don't marginally spec your bridge rectifier and look after the care and feeding of the main switching device(s). If anything is dissipating more than half its designed power rating at the designed temp it will have a much shorter life. This is common practice for all aerospace and military hardware.

Cheers

William

PS. I just read the above post. It's interesting noting the differences between the failures in commercial Vs industrial power supplies. I've seen very few secondary side failures. I could almost count them on my fingers. I guess those supplies were run on the ragged edge. If the TV will pull 400W when operating, the manufacturer throws in a 420W ps.
 
Last edited:
  • Like
Reactions: Dave Zan