Suggest your favourite buck switching regulator, 12v-5v in 3v3 1v2 out

[chickenHeadKnob]

Help me please. I am suffering option paralysis. Way too many choices.

Which is the switching regulator do you like to use for board level down conversion. I want to supply an  mcu  and spartan 6 (xc6lx9)  FPGA. Price is not the  biggest concern. Should supply up to 1.5 amps with low heat production. I doubt I will need that much current but would like to have some headroom. Regulators which can supply 1 amp or more are still in consideration. Does not need to be dual output! That is I don't mind  separate chip for each fixed output. Minimum external components  a plus. Low noise also a plus. I don't think it matters if the regulator generates a lot of hash because of  cycle skipping at low output.  I prefer tsop and so  packaging over QFN. Don't want any BGA types.

[zenerbjt]

linear.com sync bucks sound like your best...they have the free LTspice sim with them....(its now owned by analog.com)

[poorchava]

I routinely use TI parts. They have a nice online tool that can usually give you a good starting point. Pricing is usually reasonable too.

I tend to stay away from LT parts of any kind, as they quite often have various quirks (like an SPI with an active-high CS, and polarity mode 1.0 or 0.1, config being latched during power up based on voltage at a pin, which later has some completely different function, etc)

[thinkfat]

LMR14030SDDAR. Not a sync buck so you need an external diode, reasonably high switching frequency for small components (I've used a XAL5030 inductor with good results). Can be integrated on a 2 layer PCB np. SOIC-8 with power pad, maybe a bit larger than what you'd like.

[exmadscientist]

Sync bucks are really nice because a big chunk of the hot loop ends up inside the chip. They're cheap enough these days that I rarely use anything else, at least for low voltage.

I also really like TI's "HotRod" packages, QFNs with bars under the chip. They're usually marked "VQFN-HR" or something similar in TI's search tools. The little bar-shaped pad often lets you make a really good layout with excellent performance. The downside is that you have to rework a tougher-than-usual QFN when something goes wrong. Something like TPS62148 would be my preference.

If I'm not selling you on those little beasties, take a look at LMR33620 and family. It's a big family with many parts available in either SOIC or fancy QFN. I haven't used every variant, but the ones I have used have been solid for me.

[razvan784]

+1 for the LMR33620.
I used a LMR33630 in a project recently, the 400 kHz SO-8 variant. Runs slightly warm at 2A (but with plenty of thermal vias and a 2oz ground plane).

[rfbroadband]

LT8609:
VIn 3 to 42V
IOut cont. up to 2A, peak up to 3A
low noise, EMI
few components
power good pin
efficiency was between 80%  and 90% (see 3.3V measurements attached)
overall excellent performance

I have used it in several designs, (VOut 3.3V, 5V, 2.5V, 1.8V, IOut 2A, Vin between 7V and 24V)  various configurations...excellent performance

only draw back : price (you will find cheaper ones) the trade off works for me...

[chickenHeadKnob]

Thank you gentlemen. After going bleary-eyed looking at datasheets I decided it would be better to pick a current "King of the mountain" and then compare each additional competitor to it. My problem is that beyond understanding that higher frequency equals smaller external components I don't have a good background to evaluate one over the other.

So for comparison purposes I have chosen the ST1S06PU12/ST1S06PU33 as the preliminary default winner
  Pros - relatively low price, high freq (1.5 Mhz), low part count incl external
  Cons - only 7V max input and tiny DFN package ( I can live with this)


LMR14030 - more muscle than I need, but that is OK. I do prefer TI datasheet over ST. Especially since TI gives PCB layout.
LMR33620 - closer to my requirements, I like this one, particularly as it comes in power pad soic type package
TPS62148 - on first pass  I can't tell if it has a major advantage over the others
LT8609   - while I am not price sensitive I don't see the value trade-off here. It may be low noise, but if I prioritize low noise then
           I would rather revert to linear regs and deal with the heat

Still evaluating

[T3sl4co1l]

TPS562201DDCR
Haven't used this exact part but another one in the family; seems well enough behaved.

FYI, this is a regulator, but you asked for a controller (external switch) -- which is at odds with the "minimize components" request.  I'm assuming a regulator (internal switch) will be fine.

Tim

[chickenHeadKnob]

TPS562201DDCR
Haven't used this exact part but another one in the family; seems well enough behaved.

FYI, this is a regulator, but you asked for a controller (external switch) -- which is at odds with the "minimize components" request.  I'm assuming a regulator (internal switch) will be fine.

Tim

Yes my bad. I meant regulator not controller. At the power level I need was expecting internal mosfets. I will edit title and original post.

looking at TPS562201 cheap! tsot 23-6  and low parts good!

[thinkfat]

I'm looking at the footprint and sample layout of the TPS562201, surely they could have switched the GND and SW pins? That'd have allowed for a much smaller SW node area and tighter current loop and eliminated the need for vias in the switching node.

[exmadscientist]

I used the TPS562208 recently (same as the '2201 except at light loads) and I really did not like it. As thinkfat points out, the layout is difficult due to poor pinout choices. (That's the real reason I personally like the TPS61247: good layouts made easy.) We found output ripple to be unusually high, but that was probably related to another problem. This regulator is exceptionally fussy about output capacitance, all three of its value, its type, and its physical location. The datasheet really does not make clear how important getting this right is with this regulator, though it is, technically, all in there, in that if you do exactly what the datasheet says, the part will work, and if you deviate more than their allowed bounds, you might be very sad.

The failure modes we observed with this annoying part were interesting. It likes to drift away, presumably as it heats up. So if you bring up a board and test its power rails, it might pass. Leave it running and come back in an hour and it might be failing. And give it to the software guy to program, whereupon he puts it near a pile of warm computers, and it might have run off into the woods, possibly never to come back. Or, possibly, when he dumps the broken piece of junk on you to fix, you don't get to it until the morning, when it's cold and doesn't misbehave at all... yet.... Less rantingly and more usefully, there were four symptoms we observed when the output capacitance is not perfect: regular old high noise on the output, large sawtooth waves (in the 2V p-p range) on the output, loss of regulation (5V output idling at 5.8V), and total loss of regulation (running off toward the input rail).

These behaviors are not really all that surprising if you know what's in the box, but I feel like this poor thing ought to have been at least a little bit better behaved than it was. Then again, it's the cheapest thing in (this section of) the TI catalog, so what did I expect....

Honestly any of the parts in this thread will work for you. If you're cost-sensitive, avoiding LT will help your wallet. The LMR36xxx and TPS56x20y families are nice if you need pin-compatible part upgrades for more current. If you're just powering a small Spartan-6, that's not too bad. I'd offer to share an old design of mine that does that, but there's really not much to it. Showing off that I can make four buck converters and the decoupling scheme from the datasheet is not worth scrubbing the identifying information.... If I were you, I'd pick based on:

• Ease of assembly/rework
• Total solution size
• Ease of layout
• Price
• Output noise
• Whichever one makes you feel warm and fuzzy inside. If any.

In whatever order is most important to you. This isn't a particularly challenging application and so any of these parts should be suitable.