Switch mode/linear hybrid power supply?

[BiOzZ]

OK so first off my brain is still not 100% from today's festivities so excuse me here

but im working on a project ... i have the FRONT half of a 2u half rack to include a precision digital 35v 6A power supply with noise ripple of no more than 2mv RMS

so with that spec and (kinda) plenty of space my brain said LINEAR! .. simple ... so i made the schematic, did my mathamaticals (that should be a word ...), than while laying out a PCB i was given my part of the rack ... this is where i learned i was given the FRONT half of the rack to work with ... not the back ... and i know they wont let me shove the heat out the front

after being told there is no way the other PCB could be moved to the front i ended up getting some aluminum tubing and routing air to the back ... BUT the airflow could not disperse the heat fast enough without a big loud annoying fan (well it could ... but not well enough for my liking) ... and no i cant find an XFR with more than 4 taps that are in spec

SO what i was thinking was take a switch-mode adjustable supply with a nice filter on it and a few switchable bleeders and have it lag behind a large power transistor liner supply ... less watts to dissipate and keeping it cooler!

i THINK i can do this with the 2mv RMS nose 2.5v-35v minimum spec spec and not have to spend a day fishing for tooooo many additional parts ... even buy an off the shelf switch mode ... BUT this is going to be made ONCE (maybe 2 ... long story) so i cant really afford to build this all up and it not working
and there is NOTHING worse than getting the calibration sheet back with a big red sharpie "FAIL" written on it *shivers*

you guys have been cool and not condescending in the past so i thought getting a second opinion here would be helpful!

im 99.99% sure im not the first person to have thought of this ... tho i have not seen any data on the performance
I HAVE seen ... quite a few times ... switch-mode power supplies with linear loads being used as a bodged together constant current source ... i dont doubt its fine under load but i cant see the voltage performance being any better than one without the load ...


---TLDR---

i need to know if anyone has made a powersupply with a switch mode following behind a linear supply and load ... if so what was the performance?

[Rerouter]

its generally termed a switching pre-regulator, as long as isolation isn't something you want to implement in the switcher then it generally performs better than a straight linear load, as instead of mains hum and all sorts of hard to filter low frequency spikes, you have 1 high frequency and a few even higher octaves that can be filtered out with a smart choice of current shunt value and input capacitance to the linear segment,

not to mention most op amps (or self contained regulators if your going that way) have noise immunity specked for those frequencies, further reducing its influence,

if your willing to share your design afterwards, a number of lurkers and some of us active users enjoy seeing a different set of eyes take on how to set up precision power supplies (especially the digital control)

[BiOzZ]

its generally termed a switching pre-regulator, as long as isolation isn't something you want to implement in the switcher then it generally performs better than a straight linear load, as instead of mains hum and all sorts of hard to filter low frequency spikes, you have 1 high frequency and a few even higher octaves that can be filtered out with a smart choice of current shunt value and input capacitance to the linear segment,

not to mention most op amps (or self contained regulators if your going that way) have noise immunity specked for those frequencies, further reducing its influence,

if your willing to share your design afterwards, a number of lurkers and some of us active users enjoy seeing a different set of eyes take on how to set up precision power supplies (especially the digital control)

Thanks m8!
and its good to know the term!

if i end up making it i cant post THIS schematic but as its just for someones lab
but with a few tweaks, use some more hobbiest  parts (like an atmel instead of this freescale ..) i see no reason why i cant post this design i made on my own time with my own software  O0

... ... i really hope i dont have to redesign this shunt ... its a bitch to get my hands on a good miliohm meter ... i wonder how much one of those costs ...
edit: *looked it up* ... $500 id rather buy a good agilent meter @.@

[SeanB]

Use a few LC segments between the switcher to decouple the output some more. This will both reduce the needed capacitance and help with regulation of the switching supply, as it isolates the load from the supply, giving it enough ripple to rapidly follow voltage swings. The inductors just need to be able to handle the full load current and a little more for current limiting to not have ripple, and have an impedance of a few ohms at the switching frequency to be effective. DC resistance as low as possible.