inverting dc-dc converter circuit

They are only good for 18 volts, and their switch resistance is high,

John

John Larkin a écrit :

The real Groucho Marx generator uses relays.

What kind of regulation do you need on the -30V? Is that 30mA constant or does it move around a lot between 0 and 30mA?

And a doorbell button. ;-)

The big ones, multi megavolts, charge a row of caps in parallel using liquid resistors, and stack them in series using laser-triggered spark gaps.

John

Load is pretty constant and stability of maybe a half a volt would be OK. The +15 is well regulated, so a non-regulated charge pump would work if it was fairly stiff.

In real life, we'll probably use the inverting flyback with a very well-managed layout and an option for a deep-drawn aluminum cover in case it's needed.

I still want to do something with the Groucho circuit. It can also be a floating supply, up to the SSR voltage limits.

John

Put an LC behind it and you should be fine. If you pay attention to component height you can get away with standard shield covers, no fancy aluminum needed.

You're making life hard for yourself. Switch the wallwart for a 20 VAC version, and all the problems go away (including calibration sensitvity on the external regulated voltage). Linear or switching regulation of your

To some extent, Widlar and Gilbert were around when there was basic circuitry to discover. Not quite so for Camenzind who was more of an out of the box thinker. So maybe Widlar and Gilbert were there at the right time. Also Wilson, whom I met years ago. All great thinkers, but also there at a time when their intellect could be applied. However, I see your point. My opinion is people who constantly self promote really aren't even 10% as good as they say they are. It's right up there with the people who spend a lot of time telling you how hard they work.

There is always the philosophical question as to just how big is an inductor. If it is not shielded, well, it's influence can be felt way beyond it's borders. If shielded, less so, but nothing is perfect.

Slapping caps on a rail with a soft charge is probably the right approach, though you really don't have to get exotic here. You could go to a tripler scheme if it is "on demand" so to speak (i.e. regulated), If the flying cap to reservoir is sized appropriately, you can keep the step sizes small, then linearly regulate the output.

Here is a good question. How do they generate the phantom voltage in battery powered microphone line amps. Surely they have the same issues.

The ones I came across are just an oscillator and a Cockcroft-Walton cascade, or with a little ferrite core transformer. Then there are the really crude "solutions" like a stack of 9V batteries (seriously ...).

A brute force way I generated a low noise AUX voltage was to make a

Efficiency was about 40% but like you I only needed less than 1 watt.

Have you read Linear AN's 70 and 85, where Jim Williams presents a bunch of example designs for low current bias supplies with less than 100 uV noise? The flyback was the worst of the topologies discussed from a noise perspective.

nt

The Cuk converter is the best for inverted supplies. Still, the capacitance of the inductors lets some of the fast edges get through to the output capacitor. The layout of the windings can help by putting the moving end of the output side near the nonmoving end of the input side.

Typical. Larkin introduces a concept he swears by as "pretty quiet", then reverts to "real life".

That's because a charge-pump with significant load is a noisy bastard. See...

In his "real life", Larkin would use an inverting switcher. I'd steal a negative output from an already existing buck switcher as in...

Sure you will.

Personally, I'd call it a Rube Goldberg circuit ;-) ...Jim Thompson

We have the 15 volt Phihong switching-regulated wart in stock, and its international plug adapter kit, and we can send them anywhere in the world.

John

An85 is a simple boost, same as a flyback. Even a Cuk has fast currents being switched into different ground nodes. The trick with any switcher is to manage the layout so that ground-injected currents don't find their way into sensitive stuff on the same board.

Slowing down the gate drive (tweaking the gate resistor) can reduce spikes at the cost of some efficiency. Frequency modulation (spread spectrum) is helpful here too; we'll include hooks for that, just in case.[1]

ftp://jjlarkin.lmi.net/SS_LTC3411.zip

Here's the switching waveform from an LM3102 synchronous switcher,

ftp://jjlarkin.lmi.net/SwitcherRise.JPG

which clearly shows the substrate diode of the lower output fet acting like a step-recovery diode. This trashed opamps four inches away.

John

Probably could be "snubbered" but of course that costs efficiency, too.

Because nobody makes decent AM radios anymore :-(

But these are still large qty stock, 68pF:

FETs as variable resistors are nice though, if needed they can be servoed in a dual-package.