Looks like it is time to go through this thread like a scythe through wheat...
non-isolated coupled inductor SEPIC topology
Bad suggestion. High stress on semiconductors and coupling capacitor. Prone to high circulating current and oscillation/ringing from coupling capacitor resonating with inductor. Very difficult to stabilize due to right half plane zero and a non-linear input/output transfer function.
So you're intending to use a buck converter? That will work fine for the higher any of the voltage range but will become challenging at low voltages. Why not shoot for a voltage half way between the two, about 12V to 15V, then use buck-boost? Despite the extra complexity it'll make achieving the large range of voltages easier.
The buck-boost (aka non-isolated flyback) also subjects the semiconductors, input and output capacitors to high stresses, and also suffers from the right half plane zero if the inductor current goes continuous. Without a transformer the output polarity is inverted as well.
Getting a buck converter to go down to near 0V isn't a problem, either *if* the PWM controller can skip pulses (usually this is intended for maintaining regulation at very light load but it can apply to very high step-down ratios as well).
The point is to use the switcher, as a pre-regulator, to reduce the voltage to a few volts above the linear regulator's output, so the power dissipation will be no more than 30W.
This, however, is good advice - the best in this thread, in fact. I would just reiterate that a buck converter would be the best to use as the pre-regulator, not a buck-boost or SEPIC or Cuk or whatever. Note to OP: Linear Technology has an App Note on doing just this with, IIRC, the LT1074. I can't remember the number but it should be easy enough to find with google.
I'd go for a synchronous DC-DC converter controller with integrated MOSFET drivers.
I agree in principle, but this is the first SMPS for the OP and so I wouldn't recommend a synchronous topology which is more challenging to get working because of shoot-through and operating in forced continuous conduction mode all the time.
...
FPGA is right direction or dedicated arm'ish processor
Microcontroller ADC is too slow to go above something 10W
I have no words... No one in their right mind is going to use an FPGA or an ARM to control a 0-30V/10A power supply. And the speed of a MCU's ADC is not the limiting factor for how much power can be controlled; if there is sufficient inductance present (regardless of topology) and the loop bandwidth doesn't need to be too high (ie - more than a few kHz) then even a crappy 8b PIC or AVR can perform acceptably quickly in multi-kW converters (I know, I've designed and built them). Just don't use an Arduino fercryingoutloud...
MOSFET gate capacitance is reason why driver square signal disapear. you can adjust driver first with 2nF capacitor insted of blowing MOSFET's
In BOOST converter you should be using higher voltage N-MOSFET than output voltage
In BOOST converter can't be used pwm more than 50% duty cycle
Yes, yes, NO.
Duty cycle can definitely go above 50% in a boost converter. The transfer function does get a bit twitchy as duty cycle exceeds 80%, and 100% duty is verboten, but pretty much every PFC stage on the planet needs to go to at least 95% duty near the zero crossings of the mains to achieve distortion requirements.
I won't even touch 494/594. There are so many modern controllers with integrated gate driver and some even with integrated power FETs, and the latest trend is to even put the inductor inside.
I can't see a reason why you want to use such a primitive controller unless you are counting beans.
BTW, pure digital is too hard to do and I don't recommend it, but a hybrid of digital and analog, as you've described, is a good way to go unless you want absolute best performance.
I mostly agree with this, except to reiterate once again that this will be the OP's first SMPS and so he probably shouldn't mess with MHz class switchers which is what an integrated inductor implies. Nor should he be messing with a hybrid digital and analog controller. A simple circuit using well-established controller ICs like, e.g., the SG3525A, would be much more appropriate and deliver excellent performance.
The problem is a simple (single) boost or buck stage doesn't scale up to larger power very well if at all. Also trying to make buck convertors work across a wide output range is difficult, so you're adding in multiple difficult requirements. If its a bench supply there is a wealth of information out there already, but why add the complexity of different switching stages if you only need 0-30V? Almost every design will start with a higher rail and buck down from there, because its simpler.
I agree with most of what you wrote except* that single stage buck and boost converters scale just fine up to nearly any power level. It is true that it is difficult to cover a very wide range of input/output voltage conversion ratio with a single stage unless pulse-skipping is employed, but that is where Hero999's suggestion of using a switcher as a pre-regulator for a linear regulator comes into fore.
* - forgot to insert this word in original post