Sort of. Unfortunately the constant load means some of the advantages of being able to optimize the feedback/compensation network for a specific load pole placement go unused. And the overall efficiency is still low 'cause, hey, it's class A. Helpful for line regulation, though, and the constant load makes passive filtering of switching noise quite a bit easier.
In the watt or so range I've found class AB with a linear supply tends to end up equally or slightly more efficient. A 2.5+W class A amp off a buck converter would be a nice little project. Another class A application where switching works well is tube heaters.
Looking forward to seeing it when you have the chance.
Substituting a digital camera for a scanner might work.
Nelson uses them in his Amp Camp Class A amp.
http://www.diyaudio.com/forums/diyaudio-com-articles/214808-amp-camp-amp-1-a.html
As he says..... 'What? An audiophile component with a switching power supply? Get over it – it works fine.'
http://www.diyaudio.com/forums/diyaudio-com-articles/214808-amp-camp-amp-1-a.html
As he says..... 'What? An audiophile component with a switching power supply? Get over it – it works fine.'
Outside of the Audio (exclude pro audio) amps, SMPS's are used everywhere, both large and small on board supplies for various digital voltages. They are used almost exclusivly over linear because not only are they cheaper but more efficient and these days that is a concern when power consuption is a design consideration.
Nice to see Mr Pass using them, it may help dispell the myths that abound regarding SMPS supplies.
Nice to see Mr Pass using them, it may help dispell the myths that abound regarding SMPS supplies.
"Uncommon" as in *all* PCs, *all* cellphone chargers, *all* TVs , *all* CD/*DVD players , *all* monitors, etc?
Well, besides the 5 to 10 SMPS present in every household, yes, *maybe* they could be considered somewhat uncommon
Hi ! title changed to be more specific
Regards,
gino
Hello ! if I understand well SPS have difficulties when the current draw is variable
It is really interesting that they can be good for class A amps
Small power high quality amps with high efficiency speakers ... interesting
Thanks to all for the very valuable informations.
Kind regards,
gino
It is really interesting that they can be good for class A amps
Small power high quality amps with high efficiency speakers ... interesting
Thanks to all for the very valuable informations.
Kind regards,
gino
It depends of the topology choice. Flyback naturally has the largest dynamic range and then fast to response upon load variations, and all buck derived are the worse, because the large low pass filter at its output generates large time delay in the regulation feedback loop. I repaired lots of SMPS, and flyback's can support large overloads, and a range of 100:1 of load transients are well supported by them, but it also depends on loop characteristics.
Yes and no, today you can design multiphase SMPS's so each phase runs at optimum load, and you switch phases on or off to go with the load requirements, add spread spectrum to the design and you can get some interesting supplies, that are efficient over a wide range and have minimal noise, done 2 to 8 phase layouts recently.
spread spectrum PS – could be considered a stupid trick to game a bureaucratic limit
get the real EMI power down for real quality, minimal effect on associated equipment
then spread spectrum should be evaluated in the system context – whether a (hopefully weak) spectral peak causes more problems in the system than a lower level, broad spectrum – that may excite circuit resonances that the narrower line could be designed to avoid
in general good ground up sw mode ps design is considerably more complicated than most diy audio linear amps
it is suprizing that we have gotten away so long with exemption/non-enforcement of line harmonic limits in consumer audio with linear supplies
get the real EMI power down for real quality, minimal effect on associated equipment
then spread spectrum should be evaluated in the system context – whether a (hopefully weak) spectral peak causes more problems in the system than a lower level, broad spectrum – that may excite circuit resonances that the narrower line could be designed to avoid
in general good ground up sw mode ps design is considerably more complicated than most diy audio linear amps
it is suprizing that we have gotten away so long with exemption/non-enforcement of line harmonic limits in consumer audio with linear supplies
Last edited:
100:1 is a bit marginal for class AB but workable if the phase margin can be tuned to be good at the low end of the load range and acceptable at the high end (class D might be a little easier due to the quiescent dissipation often being a bit higher). I wouldn't mind having another look.
Any suggestions on a design approach that would work with off the shelf magnetics? Or for low cost, low minimum order custom magnetics? The ones I've plugged audio design requirements into all come back with application specific windings, which seems to mean one has to buy a couple hundred US dollars/euros of sample run magnetics to do a hello world on the supply. Even if the design's good out of the gate, no tweaking is needed, and one finds buyers for the unused parts that kind of kills the cost structure of a typical DIY project where you're going to build one or two supplies (I don't actually listen to the living room rig all that much so when I worked it out the cost riser for custom magnetics to supply the small amps I use was enough to pay the power bill on a linear supply implementation for the better part of a decade).
Challenge is switching the phases on and off fast and cleanly enough to track the current demand without startup or shutdown transients bonking the rails. This is essentially the class H amp problem, just with current tracking rather than voltage tracking. Performance with real time tracking is usually not so hot, with the usual fix being to lag the audio signal to give the supply some lead time to get to where it needs to be. Fine for audio playback, can be problematic for video, and the variability of music signals can clobber the efficiency gains. For example, if you have a small class AB amp that goes between 10mA quiescent and around 1A peak the optimal multiphase implementation is nominally a 10-100mA phase and a 100mA-1A phase. Home listening levels tend to peak right around that 100mA breakpoint, with the result being both phases end up wanting to always be on.
A maybe more viable approach---depending on personal listening levels, speaker efficiencies, and so on---would be a composite chip amp with a two phase supply. Those are usually around 30mA quiescent so one could have a 30-300mA phase and a 300-3 to 4A phase that would go up to the chip amp's current limit (these numbers are for two phases per channel but one could have higher current phases serving more channels). I've toyed with this idea but never found an implementation I was exited about enough to want to build. A linear supply with the control loop regulated and the output devices unregulated provides the same sort of scaling at lower cost since the incremental cost of increasing the size of the output devices' heat sink is usually pretty low. One concept I keep coming back to is amps with a loud mode. Idea is you flip a switch and control circuitry boosts the supply output and increases the amp's gain for, say, a 20dB increase in output power. This aligns with a two phase SMPS but, similarly, I've never found an implementation elegant enough to be worth building.
The spread spectrum, multi phase were in a money no object design, I forgot there was also a control card, an two filter cards with active filters. So probably not a DIY type design at the moment, but it was an interesting insight into where power management is going, with the ever increasing demands for efficiency, battery life for mobile devices and keeping the EMC noise to the absolute minimum.
The advantage of spread spectrum is as pointed out, spreads the noise, and more importantly avoids peaks where a single frequency is used.
Multipahse allows for a compact design, with several small inductors instead of one huge one, same with caps and other components. Thermal issues are more easily controlled and catered for (from a manufactrung point of view, all SMD componets, no large heat sinks to bolt thing to etc) with the heat spread out, but also allowing the base of the PCB to be bolted to the case with thermal pads for heatsinking.
Probably not suitable for dynamic audio reproduction, the current requirements were not as severe as bass transients
The technology is not new, I was looking through my collection of documents on SMPS,s today and found Linear Technology AN29 from 1988, clocks mention include 30kHz split into two 15kHz clocks, probably not the best operating frequencies for audio.
The advantage of spread spectrum is as pointed out, spreads the noise, and more importantly avoids peaks where a single frequency is used.
Multipahse allows for a compact design, with several small inductors instead of one huge one, same with caps and other components. Thermal issues are more easily controlled and catered for (from a manufactrung point of view, all SMD componets, no large heat sinks to bolt thing to etc) with the heat spread out, but also allowing the base of the PCB to be bolted to the case with thermal pads for heatsinking.
Probably not suitable for dynamic audio reproduction, the current requirements were not as severe as bass transients
The technology is not new, I was looking through my collection of documents on SMPS,s today and found Linear Technology AN29 from 1988, clocks mention include 30kHz split into two 15kHz clocks, probably not the best operating frequencies for audio.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.