Looking for very flat SEPIC inductors

How much current do you need?

You could use a diode charge pump or my inductorless Groucho Marx Generator.

John

Someone (Fair-Rite?) makes a core that you can snap around windows in your PC board to make inductors. But the ones I saw were _much_ bigger than these critters, and it makes less and less sense the further you get away from a bazillion layers.

Two 5uH in series?

How about something like:

coilcraft has been very good at getting parts to me fairly quickly. Using a non-1:1 ratio could let you get better use of the core.

If you could live with single inductors, the rencousa RL-8300 fits in very tights space. They aren't bashful about prices though.

You could look at adding LTC660 power inverters to the design.

or

-----/ o------------+----))))))------ +out ! ! +---/ o---------- -out ! ! \ o ! GND

It requires some fancy logic to run the switches but it can make good use of a single inductor.

Right now the peaks go to about 2.5A when the switcher screams along pedal-to-metal.

It has to be low noise, there's imaging stuff going on in the same box. Swinging hundreds of milliamps for the negative side would be a bit much for inductorless transfer. I don't have hard numbers from the client yet but I know for sure the Coiltronic ones won't cut it. If I find a beefier one I can at least tell them "Ok, we can reach this goal but not more than xxx mA out".

That's what I am trying right now. Series rings a bit but in parallel would be ok. However, that becomes a sore spot for EMI. If there ain't nothing else I'll have to do it :-(

The 3015 and 4012 sizes in the PPD series is what I was looking at but they are way too wimpy for this app. Even two would be marginal and I really don't want to use more than two. Then I could as well design an inverting switcher in addition.

If it has to be two I'll try the Coiltronics first because they max out higher.

Makes it tough to get a negative voltage equal to positive. Also, when I looked at those the coil resistances weren't that mouth-watering ;-)

Thanks, but doesn't work in my case because it's +/-12V. Charge pumps are ok for a few ten milliamps but when you need more than 100mA it's tough.

Have you looked at Panasonic ELL4LG series? They are single inductors, 1.8mm height and 800mA saturation for 10uH.

As an aside, under low load your buck might go into discontinous mode, in which case there may not be enough energy to keep your aux charged if you use a dual core inductor. You might need to provide a minimum load.

Mark.

Thanks, have to look for it. Seems a bit exoctic though since Digikey only carries the ELL4LM singles. If it's 800mA per Winding for a grand total of 1.6A I might just squeak by.

It's a synchronous buck, so in essence a gigantic push-pull driver driving an "LC filter". The sims look great, so far. But the negative will always be loaded much less.

Using a pair also can cause the radiated field to cancel if you care.

Cramming a lot into a small space takes some trade off.

gh.

You just need I/10mA swithers :>

The LTC660 is able to do more than 10mA. It is the highest current down pumper chip in the universe.

If they are close enough. But it makes a mess out of the trace routing. Unless you can bury that it's like driving out the devil with Beelzebub.

:-)

I don't have the final specs yet but I'll definitely have to deliver at least 250mA on the negative rail.

That's an option, although that would add real estate. Thanks for the Wuerth link but it seems 4mm is the smallest. But even 2.7mm would put us a smidgen above, won't squeeze in.

Wurth goes to 2.7mm. WE-DD series.

RL

One regular choke, and one 100mA coupled choke.

RL