In love with the LMZ12003 regulator

[KTP]

I needed a small efficient switcher for my LED project and decided to try the LMZ12003.

Wow.  It works so perfect and is tiny.  About half the size of a postage stamp and 4.5mm high, it takes 4.5V to 20V in and can output 0.8V to 6V at 3 amps.  For my application running on 12V, this gives me 4.5V at 3amps with a 12V 1.2 amp input (yes, I measured this thing at over 93% efficient!)

I thought about just using one of the many SOIC8 switching ICs and an external inductor but this thing is just too nice and easy (a tad expensive though at $6/quantity 100).

[Mint.]

Interesting, but I do not like the cost of it.

[blackdog]

Hi Mint,

Take a good look @ the datasheet, you do not need fets and any Inductors, these are included.
This is actualy cheap!
Think of the design time and the time to test your design...

Kind regarts,
Bram

[Psi]

I'm not really a fan of chips that have hidden pads under them like that, they're not the best for hobby usage.
I'd prefer a TO-220 style or d-pak

If the output range had gone up to 12V it would peaked my interest

[T4P]

These packages with ninja pads on the bottom also make it hard to desolder ... if you cock up even with a hot air gun

[KTP]

It was actually extremely easy to solder.  I put 16 thermal vias underneath the chip on the ground plane and left the solder mask off of that area (as suggested by someone on eevblog, thanks!).  A little solder paste under the chip, some heat on the backside where the solder mask was left off and presto...solder comes flowing through the vias and the chip is well bonded.  Even at 3 amps though the top of the chip barely was warm to the touch (I guess the 16 vias was overkill  )

I do agree that if you need more than 6V or more than 20V input this chip is not for you.  If space and/or hieght is a consideration though, I don't think you can get 3 amps any smaller.  And there is that nice thing about it working right out of the antistatic bag with no fuss.

[T4P]

*desolder
And not many people have hot air guns
------WebKitFormBoundaryuMImke8PLxonBBJo
Content-Disposition: form-data; name="message_mode"

0

[free_electron]

And not many people have hot air guns

Its 2012.. get with the modern times. Any serious hobbyist has a hot air gun  these days.

the days of the hot poker and thru hole is gone. filed under D for dinosaur, or E of Extinct.

[DavidDLC]

And not many people have hot air guns

Its 2012.. get with the modern times. Any serious hobbyist has a hot air gun  these days.

the days of the hot poker and thru hole is gone. filed under D for dinosaur, or E of Extinct.

Agreed.

I was going to comment something similar, get up to date.

David.

[free_electron]

besides ... SMD is waay older than you think. the firs tintegrated circuits like the SN501 ( Silicon Network 501 was the first commercially avaialble IC made by Texas instruments ) sit in a surface mounted ceramic package...

so Ic's were from the get go intended to be surface mounted ! only some dimwit decided to bend the leads and poke them in holes... and we were stuck with 50 years of 'doing it wrong'..

if you rip apart a cray - 1 computer form 1971 : all surface mount....
and that's 40 years ago ! if you still haven't got a hot air gun.. time to look for another hobby. Flower arrangements and basket weaving seem to be 'hip' these days. and minimal investment. a pair of scissors is all you need ( the rest are consumables )

back in topic.
Linear technologies have really nice integrated power modules as well. They call them uModule.
these things look like a BGA device. they have integrated magnetics ( they actually use a pcb with a ferrite core around it... so the transformer or inductor is eteched in the internal circuit board ) the whole thing is overmolded and looks like a chip. can be reflow soldered and everything.

we use em to power big FPGA's. slap one on the board next to the point of load and off you go.

[T4P]

Well of course i have a hot air station ... some idiots ripped me off on another forum just because i started to talk about BGA packages ...

[SeanB]

free_electron, you mean like this?


crDSC01482 by SeanB_ZA, on Flickr

[free_electron]

the bottom ones.
that was the original Ic package !
the mostek devices in the middle are DIl packages with their pins bent outward.
the devices surrounding the 2 in the center are the original ic package.

and, as you can see : smaller than a DIL ( half pitch ) and SMD !

Cray machines ( as well as AMDAHL and IBM ) were full of them. IBM had a BGA- style module in the mid 70's ...
So, i stand by my statement : anyone NOT on the SMD wagon is a dinosaur and has been doing it wrong all these years...

[SeanB]

Notice anything funny about the one on top in the middle?

[free_electron]

yeah, it's got its lid popped.

looks like someone wanted to probe it.

We still do this today.
We get prototypes assembled in equivalent ceramic packages. The lid is mounted on using a silver containing glue. a box cutter on the side will pop the lid off without problems and then you can stick it under the microscope

The mostek chips in the middle are so called side- brazed ceramic DIL packages. these are still in use today for military and avionics / space.

below a QFP80 and QFP64 in ceramic body with prototypes inside. these are beasties i helped make

the qfp64 is a multichip - stacked die ( there are 3 chips on top of each other they are bonded internally. top is an RF transceiver , middle is the analog frontend and bottom is the baseband.
Prototype of a bluetooth + wifi combo chip. Designed by three different teams as separate blocks. the final chip was single die.

[SeanB]

Actually it died from overvoltage. The power supply had an issue, and either the pass transistors, the driver, the error amp, the zener or one of the resistors was faulty, and gave the unregulated 14V rail out to the whole unit. Funny thing is that this was the only failure, and it was working, just "funny" and faster than usual. I had it running on the test rack for about 20 minutes before i pulled over the lead from the rack DVM to check supplies. -21V ok, -19V ok, -15V ok, 15V ok, 12V ok, 5V range did a unexpected range switch ( loud range relays on the Racal Dana meter) from 10V to 100V range. Looked at the display for about 3 seconds, saw the 14V, then pulled the power. Pulled PSU board and replaced all the parts on the 5V side, including the non working crowbar circuit, then replaced the pass transistors and powered it up again. Worked, now 5V and still "funny" in that self test light went out after about 1/2 second. Self test is basically a monostable fed from an internal power ok circuit. The PM5404 was part of that, and it was not happy at all. Swapped the card with a spare, tested it and all was ok. Plenty of those cards, as they never failed aside from dead caps. Simulated a couple of missions with it, and it hit the target every time. Did up the 100 capscrews and sent it out. It was one of the last I worked on.

[free_electron]

. Wrong topic ?

[BravoV]

. Wrong topic ?

I guess so, prolly this is what they called the disadvantage of too many opened tabs in browser. 

[T4P]

He was talking about the popped ic

[SeanB]

Yup, we jumped the shark again.........

[KTP]

Yup, we jumped the shark again.........

I was talking with someone the other day who used that phrase but didn't even know where it came from.

Heyyyyyy (combs hair back).

[SeanB]

Rathole time.........

[Stephen Hill]

You can also "Nuke the fridge."

[codeboy2k]

the qfp64 is a multichip - stacked die ( there are 3 chips on top of each other they are bonded internally. top is an RF transceiver , middle is the analog frontend and bottom is the baseband.
Prototype of a bluetooth + wifi combo chip. Designed by three different teams as separate blocks. the final chip was single die.

How do you go from a stacked design to a single die?  the stacked design will have a different layout because the interconnects are up and down, vias-in-silicon. 
When you have perfected the stacked design, then change it to planar, everything changes, it's like going back to square one (almost)

The goal was probably a stacked chip but then it was too hot, and you couldn't get the heat out.

Those are flipped dice too, right? the perimeter looks like a reinforcing rail to prevent warping of the dice, pretty standard for a flip chip.

[free_electron]

We had the teststructure and simply brought the signals to bondout structure. Thats it. Omce the design was proven the ring was stripped off and everything was integrated onto the same die. We were in a hurry and the process wasnt fully qualified for all elements. So we ran the thing that needed testing asap in an older process. Stacked the dies , bonded them and the firmware people could start writing and debugging the software stack. The second run had everything on the same die in a shrink technology.

These things did not have vias through chip. What you see is the glue holding the chips on top of each other. The chips are not bonded in the picture ( no bondwires present yet ) one of the dies had shifted during assembly and some bondpads were covered in glue. So this one was scrap. We built about 30 of these.