Synchronous Switchers

to

ass

the

ome

a square

on top.

sing

Gap on top--that looks decent.

and any

will

and

leak

Here's the little bugger wot terrorized this guy I worked on:

of

x.

tcher

an be

It's getting to be a power supply circus--everybody wants their own (and then they want to talk).

Grins,

James

square

top.

using

any

will

and

leak

switcher

be

At least we know, now, that he studied *something*.

square

top.

any

leak

Super low profile. Hard to keep those little fluxy critters inside.

and

spectrum.

just

the

into

And I can do it monolithically along with CMOS... a great combination of worlds. ...Jim Thompson

and

spectrum.

just

the

into

Sound like you're *not* killfiling me. I never actually thought you were.

point, and

spectrum.

just

the

back into

I was, until today, I wanted to catch your insults first-hand.

Why is it you have to be a jerk?

I do quite nice ultra-high-speed things in SiGe. You're probably buying (and loving) some of my designs but don't know that I did them... like the Fairchild LVDS stuff ;-)

I'll return to killfiling you... you have nothing technical to say anyway, other than advertising your "wares".

...Jim Thompson

Also some rather pedestrian op-amps, except for certain specs.

Tim

point, and

magnetically

that

spectrum.

I'll just

the

back into

I don't barge into your technical threads with insults. Oh, wait, you have no technical threads.

But jerk? It was you who grossly insulted my wife, more than once. Only a redneck asshole would do that.

That's what's so stupid. You have been simultaneously claiming that I post no circuits, and simultaneously claiming you killfile me. No wonder you can't see my circuits!

Go back to lurking and pretending to read my every word second-hand. Geez, that is seriously goofy. But it does allow you to pretend that you don't see the circuit questions that I put to you, which must be mighty useful when you don't have answers.

e:

ly to

all

owpass

nk the

s some

ly a square

ap on top.

e using

ld and any

nd will

ous and

bly leak

It's actually pretty tall--the illustration's not accurate.

The significant feature though is that there's a gap and it faces down, toward the contacts.

It's better if the cup faces up, that way the fluxies won't spill on the board.

Nice of you to say so, but naah.

Mine was a multi-board setup, I was just using the power supply groundplane to magnetically shield the rest of the product. I never thought of an isolated shield under the part (and it was only a 2-layer board, with one of those the groundplane).

I am actually revisiting this design right now, with a totally new arrangement, where I could probably try it your way.

ote:

pply to

hey

n all

nd

lowpass

hink the

ig

is some

ally a square

cap on top.

're using

...

ield and any

and will

onous and

bably leak

..

That's debatable. If the gap is close to a solid ground plane, the leaking flux will generate current in the ground plane, which is bad, but the current circulating in the ground plane opposes the leakage flux, and the field decays rapidly as you move away from the inductor

- much more rapidly than if it is further away from the ground plane.

ge

Well, JL does use the unfortunate word "world", but I'm intepreting his comment as still discussing the parasitic coupling within his switcher board. There the coupling is likely near-field, predominantly capacitive or predominantly inductive.

What you say is true for the far field.

Reagrds, Mikko

We usually have ground plane as layer 2 of a multilayer board. A layer 1 copper pour under switching inductors might be prudent, to bounce leakage B fields back up some before they hit the ground plane. Costs nothing.

te:

supply to

they

RF

on all

ound

LC lowpass

think the

big

re is some

tially a square

t.

ed

a cap on top.

we're using

field and any

ll and will

hronous and

robably leak

-...

Well, you're right, it is debatable.

The construction is thus:

.------. .---------. .-----. | .--' '. .' '-. | | | |=======| | | | | |=======| | | | | |=======| | | | | |=======| | | | | |=======| | | | '------------------' | | | '--------------------------'

which concentrates virtually all the flux in the gap, which is further confined by the use of ferrite-loaded cement to seal the gap.

Here I'd rather have that gap pointed up so that the near field could have its fun up and away from my board. But, it's pointing down, inducing ground noise that was getting propagated (conducted and radiated) system-wide, through meters of cable and multiple assemblies.

It's not the case that this inductor induces an opposing, self- cancelling current & field in the ground underneath it, due to the fact that a conductive, electrically hot surface-mount part with contacts on the bottom makes a solid, uniform plane under the part quite impossible--topside you wouldn't be able to connect to the inductor's pads.

Here's the challenge--run minimum-length 3A traces to the following pads (on the top layer), without breaking a solid groundplane :-)

.--. .--. | | | | | | | | | | | | '--' '--'

Besides, as received, the deeper layers were already committed.

A gap-up part would've avoided all these epiphanies.

In another system it might be better to point the gap down, for example if there were signal-carrying coils or transformers nearby and you wanted to keep the switching noise out of those signal paths.

Yep. And I love to rub it in.

Wait till he posts another PCB. I'll read the part numbers and circle all the chips I designed, and re-post ;-) ...Jim Thompson

copper

back

Yes, I understand that is what James was suggesting. For my 2-layer I guess it would not do much good with inductor tracks on the top of the board breaking up the pour? Or maybe it would, excuse to try it and make some measurements!

Next version will very likely be 4-layer so will probably try it then.