|
|
|
|
Deep ThoughtIt's easy when designing any item, be it mechanical or electronic, to get hung up on the pursuit of a single goal. 'Lower noise' is what the perceived wisdom was telling me mattered in a PSU, and that was what I was pursuing. Lower noise is important, but steady-state static noise is solely representative of the best performance you're ever going to achieve. The really important bit is maintaining that performance under all conditions, in a word (or four) DYNAMIC PERFORMANCE IS ALL. To that end the various parameters that one has a chance of controlling / improving in a regulator, all have a contribution to make. They are namely: -
You will not get a better final result, within the active part of the circuit's operating parameters, than the figure you obtain here. Easily measured using a good low noise buffer amp / spectrum analyser.
Really important, it affects the dynamic noise of the system, and is related to the small signal response (open-loop response) of the error amplifier used and it's surrounding circuitry. A bit harder to measure this one. The low-noise amp is needed to measure the noise, but we now need to construct an amplifier capable of driving the output of the regulator under test with an AC signal, whilst plotting the signal at the regulator output. With output impedances in the micro ohm (yes, 1 millionth of an ohm!) range, this is quite something to do. A relative calibration function is useful too, since driving those impedances, with maybe 100u of low-esr capacitance is a stability challenge.
The ability of the regulator to reject noise at it's input. For a good adjustable three-terminal regulator this looks good at first glance, in the order of 80dB rejection but that's only at low frequencies - it's MUCH worse higher up the audio band. To measure this a modulated PSU is required to feed DC and AC into the input of the regulator, and then measure the effect at the output. The input cap of the regulator under test causes similar problems to the above
Affects every one of the parameters above, ideally one wants the widest possible bandwidth, in order that all the parameters above are held (by the feedback loop) close to the steady state values. As a bare minimum we want sufficient bandwidth to hold the major parameters as low as possible over the audio bandwidth (or the bandwidth of signals being amplified), and ideally over the bandwidth of the circuit being powered. The circuits bandwidth is defined by design and component choice. In reality low impedance at HF is more efficiently and easily achieved using passive schemes. So how do we achieve these goals and what do we use to do it? Well, in true lazy engineer form, we plagiarise the best, and invent the rest*.
*This quote was plagiarised |
|
Send mail to
webmaster@alw-audio.co.uk with
questions or comments about this web site.
|
