Filtering switcher noise

Yeah I would say an LC or a cap. multiplier, I've done both.

George H.

That circuit seems popular lately! A few opamps are c-load stable.

The easy way is just an RC lowpass, something like 10 ohms and a 10 uF ceramic cap. That won't resonate. Or a ferrite bead if you can't stand the voltage drop. Some beads have several uH of inductance, which would be 10s of ohms of reactance at 500KHz.

LT Spice has a ferrite bead part, and includes a zillion Wurth parts. What's cool is that the inductance is included. Lots of beads don't have a specified inductance.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

A series inductor with a reactance of 100 ohms, plus a cap to ground with a reactance of .1 ohm, would reduce the ripple by 1000:1.

But, a cap multiplier would be smaller...

Cheers, James Arthur

Or just hang a couple of 10 or 22uF ceramic caps on the output. Those little brick converters don't have much capacitance inside.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Two beads and two caps would do a nice job too--low drop, good damping.

Cheers, James

The NE5532/5534 has decent drive current, but I'm pretty sure it spazzes out if it has to drive too much cable capacitance.

+1 on the cap multiplier. If it's a positive supply, you can even put the cap multiplier inside the DC regulation loop. (You have to use split feedback, the way you do for driving capacitive loads, but it works fine.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

Why not just a resistor between the amp and the cable?

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Resistors in the power lines are nice... if it's only opamps and not high current.

George H.

Is it important that the LC network be tuned to the impedance looking into the the rest of the circuit to be approximately critically damped?

Not beads that you'd actually want to buy ...

Note that a ferrite bead/chip can have enough inductance to resonate.

If you use one in series with an electrolytic capacitor, the resonance is almost certainly going to be at a low enough frequency that series resistance of the electrolytic will be high enough to damp the resonance.

I got caught putting one in series with a decent-sized ceramic chip. The quick and dirty solution was to replace the ceramic chip with a tantalum chip, but this was too expensive for production, and I ended up squeezing a 22R chip resistor into the layout.

And do be careful with the ground returns around the switcher - switching power supplies seem to specialise in injecting loads of current into ground connections.

--
Bill Sloman, Sydney

That's one way of getting resistive damping, if the impedance looking into the rest of circuit is resistive, and well defined.

Making sure that any LC structure includes enough resistance to make it at least critically damped is a very good idea.

--
Bill Sloman, Sydney

That is what I would try first. Just put one cap on the output and see how well that works. You may have to do more, but this is easy and cheap and might be all you need.

Dan

It's being built on a piece of protoboard, so the plan is to have a single ground point at the negative terminal of the output filter capacitor, and "star" ground everything to that.

I use a pour on the component side of the board, connecting the input and output capacitors and the switch (or diode). For synchronous switchers, this gets connected to ground at the switch (under the package, normally). For asynchronous switchers, I've been connecting it at the input cap.