Regulated supplies
To Regulate, or not to regulate........
I agree pretty much with the comments to date. Consumers won't pay for it, a lot of heat is produced, and there are other issues besides.
If you do decide to regulate the rails of a SS power amplifier, you must first choose series or shunt. Since shunt regulation consumes around double the power used by the amp module, most go for series. Of course, with vacuum tubes, where currents are much smaller and usually vary much less, shunt regulation is quite acceptable, though not much used in power stages.
Second, having settled on series regulation, you need to decide whether to use error correction (a feedback circuit) or emitter follower (no error correction at all). The former is complex, usually involves high levels of feedback with attendant problems of speed and overshoot, while the second is simple, but does not give very good regulation because the emitter resistance of the pass element increases with drawn current, giving a non-linear voltage reguation. However, this method has no issues of speed or overcorrection and is fine where current variations are small, for example in a preamplifier.
Like all these issues, design depends on environment and permissible levels of voltage sag. In a power amplifier, the current varies from quiescent, typically 50-100mA per rail, to an all out, head-banging 5A or more, depending on amp rating. This range is quite wide, and the regulation must be pretty good to avoid any kind of signal compression, which is usually audible as the regulation abruptly hits a brick wall. Given that global feedback PP SS amps can effectively compensate gracefully falling rail voltages within their feedback loop, you have to ask yourself if it is really necessary to incorporate an additional layer of complexity using expensive, high current transistors mandating even more expensive heatsinking. Good evidence of this is the perceived difficulty distinguishing sonic differences between a 300VA and a 1000VA transformer, given rectifiers, amp, source, speakers, cabling and filter caps which are identical.
OTOH, there are always exceptions, and in preamplifier circuitry I have found that series pass, emitter follower series regulator circuits give the best all round performance. I am heartened that Elso Kwak finds this too. Here, the sorts of differences I refer to have to do with imaging rather than dynamics; and this I cannot explain, except that the emitter follower series element would clearly offer capacitance multiplication, removing ALL ac from the line, thereby elimininating artefacts which might otherwise blur the delicate phase relationships responsible for good imaging.
I can see no advantage for regulating the rails of a Class A circuit; such amps have almost constant current draw, and a constant, heavy current on even a rudimentary supply gives very little voltage sag.
If you do go for regulation, it is important to design for speed to match the bandwidth of the amp. In this connection, an emitter follower has a natural advantage since it does not suffer Miller or overshoot effects, unlike the common emitter configuration. I liken this situation to a poorly designed power steering system on an automobile; an underrated pump, valving or ram system will not respond quickly enough to steering inputs; in a crisis, this could be fatal.
Lasty, one cannot overstress the reluctance of the consumer to pay for these electronic trinkets. Like it or not, most commodities in life accrue a dollar value in the eyes of the consumer; go over this, and risk selling nothing. When it is pointed out that in hifi the cost of manufacture is normally between four and six times lower than the retail price, it is obvious that every dollar spent at the factory is highly magnified at the consumer's hip pocket. This is a very powerful drive towards simplicity, no bad thing of itself of course, as long as the money is spent where it really matters.
Cheers,
Hugh R. Dean
www.aksaonline.com