Gentlemen,
May be a dumb question but I need to ask:
If the FrontEnd supply is so important but with lower and steady consumption why we don't use more complex regulators like the ones designed from Andy, Teddy, Russ, etc.
I suppose they can be tuned or re-designed to give higher output voltages or am I missing something?
Thanks,
Ivo
Another good question and a slightly tricky one to answer.
Regulators can do two things, produce an output that doesn't vary as much when the input voltage varies (Line Regulation) and produce an output that doesn't vary as much when the the thing they are driving has varying current demands (Load Regulation).
We started out in the basic NAP circuit with the power rails of both the front end and the output stage using the same noisy high current supply and from the NCC onwards have been separating and improving the front end supply. We started out with very limited line and load regulation. The front end has an almost constant current draw and therefore needs less load regulation than the power stage with its large variations in current draw.
The first change was a resistor capacitor filter as per NCC which gave increased line regulation and depending upon how good the cap, increased load regulation. The next logical step was a separate transformer and cap board for the front end which further improved line regulation.
The HackerNap has introduced a cap multiplier which gives amplified line regulation and a little bit of load regulation defined by the Vbe characteristics of the transistor used. The load regulation comes about because there is a feedback in an emitter (or source) follower, it delivers more current when the output voltage drops relative to the base (or gate). The transistor in the Cap multiplier has a high bandwidth (the Ft of the device used - into the 10s of MHz) and as it is just a single transistor stage it is unlikely to go unstable. The bandwidth is important as there are a great number of high frequency components flying around a power amp after all that rectification.
Now a Teddyreg is a version of the cap multiplier with an improved follower output stage and a voltage reference at its base. It doesn't have any more overall feedback and retains a highish bandwidth so it may well be the equal of or better than the cap multiplier. It might be worth a try though its a lot more complex than the cap multiplier with a lot more to go wrong.
I do however doubt there is any benefit from using high feedback regulators for the front end, such as the Superregs. I would argue that their high feedback and limited bandwidth makes them both less good at filtering out the highest frequency components from the raw DC supply and less good at regulating these same high frequency changes to the load. They are also even more complex still than the Teddyreg.
There is one thing that needs a little more consideration though. The positive rail of the NAP and NCC circuits front end does have small variations in current demand (the negative does not). These small variations increase with frequency as the open loop gain of the NAP circuit decreases. There might be some benefit in having better load regulation here on the positive rail, perhaps a simple low feedback high bandwidth regulator after the cap multiplier would provide an audible increase in load regualtion to deal with these current demand changes.
I'm not sure where the law of diminishing returns cuts in on this improvement process. It may well turn out to be more effective to bridge two HackerNAP circuits and thereby clean up the earth line, just as Naim did with their NAP500.
John