charge pump/boost converter

My guess is that it would take 11 stages, possibly more.

Too much hassle, too many stages. I'd use one of the flyback transformers they sell for power over Ethernet (PoE) and use it in reverse.

At this roughly 10:1 ratio a boost converter will also work and they are often less noisy. If you can cloc-drive it from or synchronize it to an ADC clock or something.

You can get modules that do most of that for a buck and a half on eBay.

Cheers

Phil Hobbs

Two of these

with the outputs in series would work. They are fairly quiet, and you could add a little filtering to help. At light loads, their output tends to be a few per cent high. But then, 5 volts from USB may not really be 5 volts.

For really quiet, consider some sort of sinewave drive step-up transformer.

A diode-based charge pump would make less than 4 volts per stage, so that would take a lot of parts.

Interesting! Anyone know how quiet is the power coming out of a USB port (vs load)?

I don't think there is any spec on that other than the voltage tolerance of 4.75 to 5.25 volts.

Boost or flyback, and if noise is an issue then make 50V and follow it with a linear regulator.

A Baxandall Class-D oscillator - one step-up transformer and one inductor - would work well. MOS-FET transistors do a bit better than bipolar transistors, but Baxandall's paper rather pre-dates them.

Oh wow, those are great. They're cheap enough that I could just slap them into a "finished" product...

They do not have an isolated output. So connecting them in series to make 48 volts will not work.

Right, which is why I suggested the charge pump. Of course the OP could get a buck converter module (about the same dough) and connect it to make -24, then work off the difference.

Cheers

Phil Hobbs

Linears only reduce noise in their active bandwidth, often no more than

A shunt-regulated cap multiplier can do a _lot_ better.

Cheers

Phil Hobbs

That's a good idea.

Just a diode and capacitor need flipping.

And correct the feedback.

Not even. You ground the 'output' pin, and take the negative output from the 'ground' pin. It pumps its 'ground' down by itself.

Cheers

Phil Hobbs

That's usually good enough. When it isn't, there are things like TI's LP38798, which extend that another decade or more. After that, use a passive filter.

Oh God, You just couldn't keep that failed OSC out of th subject!

Jamie

Easier to use a different chip (LM2585 has a slightly different pinout, but almost exactly the same function as that XL6009, and goes to 60V).

ind

der

ND/4006941

or - would work well. MOS-FET transistors do a bit better than bipolar tran sistors, but Baxandall's paper rather pre-dates them.

lator1.htm

The 1959 Baxandall class_D oscillator isn't "failed". To quote from my web- site "The circuit is probably best known from Jim Williams' series of appli cation notes for Linear Technology, on high frequency inverters for driving cold cathode back-lights used in laptop computers (application notes AN45, AN49, AN51, AN55, AN61, AN65)." These apparently were Linear Technology's most popular application notes for quite a while.

It has been claimed that Jim Williams got the circuit from England, without the Baxandall label.

I do have a low distortion variant of the class-D oscillator, with a curren t mirror rather than the feed inductor, but that's not the low distortion o scillator I'm working on at the moment.

You are distinctly public-spirited in regularly reminding us that you are a n idiot, but you really don't have to do it quite as often as you do.