triple switcher

We keep running into the requirement to turn +12 or +5 volts into 3.3,

1.2 (fpga core), and usually a third voltage, 1.8 (for drams) or 2.5 (fpga Vccint). We'd like to consolidate this to save board area and general complexity. Any suggestions?

LTM4615 looks interesting, a bit expensive but very small.

formatting link

And other ideas?

John

Reply to
John Larkin
Loading thread data ...

.

How about LT3507?

LT3564 seems to get a lot of switcher into a small package.

I haven't been able to find a way to do it with either a PIC or a 555 but I'll keep thinking about it.

Reply to
MooseFET

Either's fine.

(view in fixed font)

U1 U2 ... U11 .-------. .-------. .-------.

+12v >--|gnd Vdd|--|gnd Vdd|-...-|gnd Vdd|--> +5v '-------' '-------' '-------'

U1-U11 = 555 or PIC.

-- Cheers, James Arthur

Reply to
dagmargoodboat

,
8...

That isn't 3 voltages per chip as the OP asked for. It also doesn't really regulate.

I have made a booster regulator with the LM555 that worked just fine. I have figured out the design for a bucker that looks like it would work but so far without adding a lot of external parts I haven't come up with a way to regulate more than one voltage.

Reply to
MooseFET

Lots of external parts, schottkies and inductors.

We already stock the LTC3407, 3411, and 3412, all similar. But it would be great if somebody did a minimal-parts, minimal-size triple synchronous switcher, ideally with internal inductors. I *told* LTC to do this, and, incredibly, they haven't got around to it. It's not as if FPGAs are losing popularity.

The Spartan 6 parts will run from two supplies (Vccint can run from

3.3) but we want to use a DDR dram, at 1.8 volts, so we're back up to three.

To keep the inductor and cap sizes down, you've got to switch at a MHz or two.

John

Reply to
John Larkin

...

something like a ISL65426 and an ldo?

-Lasse

Reply to
langwadt

John asked for a switcher with 3 regulated outputs, to save space. You wanted a 555 or a PIC. So, like our health care bill, I blended the two sets of requirements and came up with a solution that costs more, uses more parts & space, doesn't regulate, but does use either a

555 or a PIC.

That's called compromise.

The classic way is a flyback with multiple secondaries, close the loop on the main voltage, LDOs on the others.

But of course we all knew that.

-- Cheers, James Arthur

Reply to
dagmargoodboat

The LTM4615 has all that inside, including the inductors and LDO. That may be the most compact solution, for $24.

Still no magic bullet.

John

Reply to
John Larkin

On a sunny day (Sun, 15 Nov 2009 13:09:30 -0800) it happened John Larkin wrote in :

Voltages and how many change all the time, better have separate switchers, Havin ga 3 in 1 would reduce the market by a factor 9 :-)

You make small series with high profits. No reason not to use 3 switchers.

Reply to
Jan Panteltje

The LTC switcher voltages are usually programmed with one or two resistors per output. An SPI interface to a nonvolatile memory would be even better.

We're space constrained by the fixed size of a VME module. The current project is a 16 channel differential-input digitizer with BIST and a

32-bit interface (requiring two DIN connectors and lots of buffers to the backplane.) It's very tight, and the more features we include the more we can sell. Wasting 2 or 3 square inches on voltage regs is silly.

We commonly build products with 8 or 9 separate power rails, with the power regulators over a third of the board area. This is getting silly. SIPs would be useful, to use available height instead of valuable board surface area. The thermals might be better, too.

John

Reply to
John Larkin

That LTM4615 looks pretty stinkin' nice--I can think of cheaper ways, and smaller ways, but not cheaper, simpler and smaller ways.

I have some Power Trends modules, older models that drop into 78xx slots. Cute. I wonder if they've expanded their line...

James Arthur

Reply to
dagmargoodboat

traco has some 1A SIP modules ,

formatting link
they also have some 6A/10A sil modules

-Lasse

Reply to
langwadt

On a sunny day (Sun, 15 Nov 2009 14:08:59 -0800) it happened John Larkin wrote in :

Can you not make small vertical boards with each a switcher? The disadvantage of vertical is that it is more sensitive to mechanical vibration causing bad contacts or broken solder joints.

Reply to
Jan Panteltje

TI bought Power Trends (in 1999!) and, yes, they have expanded the line since then.

Reply to
Joel Koltner

ince

Hmmm, no luck. I went to the TI pages to look for power supplies, but all they had were power management solutions.

-- Cheers, James Arthur

Reply to
dagmargoodboat

Joel prodded me to scout TI nee PowerTrends. They have a bunch of single-output 5-SIP modules that take about the real estate of a TO-220.

PTR08060 -- 4.5-14Vin, 0.6

Reply to
dagmargoodboat

We looked at the TI stuff a while back. The external cap requirement is bad, as is the height. They run at 300 KHz!

John

Reply to
John Larkin

in

l

dules

That last one's 15x13x9mm, no output cap needed. The 1st one needs

100uF.

James

Reply to
dagmargoodboat

...

Yes so you need a really fast PIC or 555

The ADC on the PIC isn't fast enough to do cycle by cycle voltage checking so you would have to close the servo loop at way less than Nyquist. Unfortunately this would also make the filter capacitors big.

Reply to
MooseFET

I think you actually could do it with an AVR--their PWMs have ~40nS resolution.

-- Cheers, James Arthur

Reply to
dagmargoodboat

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.