Synchronous Switchers

if

the

bootstrap

You didn't like my thermal IR sensor circuit that I posted a few days ago? I really like that one, for some reason. Now that I think about it, it has faint echoes of OTA.

You can see the products on my web site. They have circuits inside, and I doubt you can guess how any of the key ones work.

How can you whine about my not posting circuits when you claim to have killfiled me? That is primo stupidity.

You are apparently competent within a very narrow scope. You are pitiful outside of pushing transistors around on chips.

the

bootstrap

You have every reason to be reluctant.

square

Total blather. The spectrum shows that it's not a simple magnetically induced sheet current.

I can clip my scope probe ground lead onto the body of one of those grounded SMB connectors, and then probe that exact same ground point, and see tons of high frequency noise. Explain that based on magnetically induced plane currents.

layout

diode.

the

bootstrap

Why do you want to see another 555?

the

bootstrap

I was hoping to see something packaged, something that he actually built. I don't know if he uses PC boards.

rote:

quare

I just worked on a supply. The "shielded" inductor was, but wasn't. It made many 10's of mV in a ground "plane" whose edge it straddled (making some part of a half turn).

Cute!

er

Could be, natch.

square

Yes, my similar issue was a few years ago and I did some investigation with the probe described. (First time I had done that sort of thing, great fun experimenting and waving the probe around, and an excuse to fire up the old Tek 7704 + 7A22).

In my case I found it was better to run the switcher at the highest possible frequency, in my case 500kHz. This was enough that copper groundplane of the power supply board could provide some shielding against the field. More importantly the various passive filters in the system (RC and chip ferrite) were well into their stopband so when cascaded there was very good attenuation.

to

ass

the

ome

a square

I received this unit with the complaint that it was blanking out FM broadcast reception (a 500kHz switcher, that's the 200th harmonic!). The inductor-coupling was part of it, coupling those 10's of mV into the ground system-wide.

Like you, I appropriated the top-side plane under the inductor as a magnetic shield, and connected to a deeper layer for the actual ground. The top plane still has voltage across it, but it no longer matters. That cleaned the output up nicely.

square

So you created a magnetic shunt around the coil. Works ok as long as you can get the coil close enough to the surface and does not impede on the coil's operation.

I usually plan on coil orientation if I think it may be emitting RF.

Shield coils are at times are a better way to go but may require some space due to cavity requirements, so not to disturb coil operation.

Jamie

The usual probe-ground-lead loop is about 150 nanohenries. Could that account for the pickup?

some

a square

The coil was a shielded type. But, they still leak. That's why it's possibly a problem at the lowish levels John Larkin's seeing.

Sure, but there has to be some serious spikes-to-the-world hash on that connector body to induce a bunch of noise into that ground clip.

and

Your inital post reported nada at the fundamentals, but -75dBc @ 5th and 7th. An shoot-thru pulse should produce a relatively flat spectrum, and the usual LPF elements would roll that off, leaving more fundamental than harmonic.

So, whatever it is is getting around the filters. Often in r.f., that's magnetic.

square

Lots of these "shielded" inductors are sort of half pot cores with a cap on top. They leak a little flux out the top side. This is the 4.7 uH part we're using with the LTC3411s:

I really don't think the inductor mag field is a problem. Both the field and any induced circulating current in the ground plane are physically small and will drop radically with distance. The other two switchers are non-synchronous and aren't making anything like the noise from the two LTC3411s.

I have in the past had problems from unshielded drum cores which probably leak

We need to be more careful about EMI from our own switching supplies, when there's low-level signal stuff on the same board. Synchronous switchers can be terrible. Some of our boards have 6, 9, even 12 different power rails.

square

Of course, the body of the inductor can *capacitively* couple coil voltage spikes out to the world. That's usually not a problem if the coil is sitting right over a layer 2 ground plane with no signal traces nearby.

Something is clearly resonating, or at least highpassing, the spike spectrum. Maybe the board/box geometry. Analyzing that is past my pay grade, so I'll just spin the board and use less agressive switchers. Some LC filtering around the periphery of the power section would be prudent, too.

The two sync switchers are powered off +5, so are likely kicking trash back into the +5 plane.

if

the

bootstrap

if

regular

the

less

bootstrap

If Larkin were to show a schematic that's actually in manufacture the sycophants might laugh and choose a new role model >:-} ...Jim Thompson

square

ground

SMB