Today I’d like to finish up on SMPS (switch mode power supplies) and how they work in a power amp as well as why we don’t want them in digital source equipment.
I’ve detailed how much better a SMPS is from the aspect of DC; we get almost no ripple, it’s clean and without the large gaps between cycles that are prevalent in traditional power supplies. I’ve also covered that power amps really depend on clean, low impedance DC to feed them and the closer this is to ideal the less the power amp has to work to make up the differences and the better it sounds. It’s a simple equation that just works. However, SMPS are not perfect, have their share of problems and aren’t great for everything. The careful designer has to choose wisely what works in any particular design.
I’ll give you a quick example of this; a DAC. When we were working on the PWD MKII we wanted to step up the technology inside the DAC and perhaps switch to a more modern SMPS rather than the traditional large power transformer we have in our flagship DAC. We sourced a really awesome SMPS that was way overkill, far gentler on the incoming AC then our big power transformer was and figured this should be great. We connected the SMPS up, did a careful series of measurements and actually found the noise floor dropped a tiny bit; looking very promising. That was until we hauled it into the listening room. Screech!
Strings were thin and steely, the entire presentation lost midbass, body was thinned out, it sounded digital; dreadful actually. I reasoned that we simply needed more capacitance and added that – better yes – listenable, no. We then tried using multiple analog regulators, fed from the SMPS as if it were just a power transformer/supply and this worked – but by that point it no longer made sense to incur the extra costs of the SMPS – especially since it offered no further benefit than did the less expensive power transformer – and we we stayed with the traditional supplies.
But power amps are quite different. As I have explained before, whether class A, A/B or D, the approach is all analog and fed from a proper SMPS, the results are quite opposite of what one finds with the last example involving digital audio. Low level detail with the SMPS is extraordinary, midbass blooms, strings take on body and richness – it’s almost as if night were day and day were night comparing the results between the two. Really 180 degrees apart and it doesn’t seem to matter what the amp class is, SMPS is better than traditional.
Lastly, many of you have asked me why prior use of SMPS and amps didn’t have such goods results. I can give you my opinion. SMPS is a crazy difficult design challenge and one not to be taken lightly. Switching high power at high speeds will result in radiation spewing out of the circuitry and into the adjacent circuitry affecting sound quality in every way. Containing that radiation without using physical means like chokes, filters and additions to the board is an incredible challenge – not to mention that SMPS aren’t great at instant power delivery beyond a certain limit. SMPS designers that are state of the art use layout, timing and “black magic” to produce designs that are ultra quiet and radiate less than even a traditional power supply (which actually radiates a great deal through its big transformer), coupled with greater energy storage for peak demands have created new SMPS designs that are really ground breaking.
I think the simple answer is that technology and design have finally come around to producing power supplies that, at least for analog products, gives the very best results – and results that are superior (finally) to analog/traditional supplies.
Ain’t progress great!
