In answer to your question I did not give him spec for the output other than from an isolated design I required as clean and regulated output as he manage.
So, not very scientific I'm afraid.
Bluh...
That's a dangerous spec! You might just end up getting little more than an automotive DC-DC converter. By that, I mean the kind of thing commonly used in car amps: 12V is chopped up, transformed and rectified. Nothing else: no regulation, no filtering (just a wad of capacitors), no limiting, no protection. It's a wonder they manage to even start up, but that's somewhat helped by their poor design (slow switching speed), as is EMI, I suppose.
The regulation of such a design is poor. The MOSFET Rds(on), DCR and leakage inductance of the transformer, and Vf of the rectifiers, all add up in series, determining the output regulation (i.e., change in output voltage over change in load current). The line regulation (change in output voltage over change in input voltage) is always 1. In other words: it's nothing more than a DC transformer.
That's perfectly possible as all he could "manage"!
It takes effort to do better: you must add a filter choke, current sense and feedback loop to provide regulation and limiting or protection.
Best case: the PSU is designed as modules (as suggested above), where each module has a control voltage input, power input (nominal 12V), and a power current output (Io = V(control) * gain). Note that gain has units of conductance, so this is a transconductance stage. Any number of transconductance amplifiers can be connected in parallel. Currents in parallel add, no sharing to worry about. Finally, an error amplifier sets V(control) based on output voltage feedback, thus regulating output voltage.
Because V(control) is always bounded (the error amplifier can't set any voltage outside its supply voltage range), Io(total) is always bounded. You get
implicit current limiting this way -- it's really quite fantastic, and it doesn't take much more effort than the alternative, which is significantly worse.
The main alternative is a voltage mode control. This also has to be wired in parallel, naively, because there's no way to sense current sharing. It's a very poor design, as the current sharing between sections is at the mercy of propagation delays and resistances in each section. If one delay is a little out from the rest, that transformer hogs current and melts transistors; cascade failure ensues. No matter the number of stages, voltage mode control is itself at the mercy of transients, as all it takes is a sudden change in source or load condition to cause an excessive current draw, and psst, bang, there goes all the transistors again.
So your options are: bargain basement, unregulated and unprotected; voltage mode, regulated and unprotected; and current mode, regulated and protected.
I need an attainable noise figure from you guys?
I am also aware that a supply rail with poor regulation will have a negative affect on intermod distortion from my amplifier.
Well... you tell us, it's your amplifier!
How much supply voltage variation is tolerable?
How much PSRR does it have? At what frequencies?
How sensitive is your receiver? Do you expect to be able to listen to conversations while this supply is running? In which bands?
If you don't even have any basic guesses, then test and measure! Vary the supply voltage and see where IMD degrades; couple RFI into the supply and see where it goes; test your receiver's dynamic range and discrimination; etc.
We are privileged to work in a field where every possible statement can be measured, tested and calculated. Don't let that opportunity pass you by!
Cheers,
Tim
Home
Help
Search
Login
Register
