LTC switcher, PADS

Oh man, switchers out of a can? Where's the sportsmanship? To me that always feels like wine out of a square box ...

To continue the food-and-wine analogy, we have other fish to fry.

Reminds me of the James Bond movie. He's in the dining car of a train, ordering dinner with a fellow British spy. The other guy orders fish and red wine, so James shoots him, knowing him to be an impostor.

John

If you want to still go to the race track, if only for watching Yugos or Minis race, National has had their Simple Switchers ICs for years now. Some of the magnetiocs manufacturers even make "drop in" inductors to match.

We mainly use Vicor power modules around here, and with a recent design, we're seeing exactly what Terry Givens had suggested we would: To actually get the rated power out of the modules requires far more heatsinking and air flow than one would expect. @#$!@#

Un bel giorno John Larkin digitò:

I've used them recently, and the PCB guy told me that he created the PADS symbol with the same wizard that he normally uses for BGAs (obviously you need to create the "complete" schematic symbol, with all the pins specified, as shown in the table at page 19). We've also noticed that some models have the same row/column order of the BGAs (A=left column, 1=bottom row), while others have the rows and columns inverted (1=left column, A=bottom row). LTM8023 is an example of the first type, while LTM4607 is an example of the latter:

On a sunny day (Wed, 25 Mar 2009 13:40:01 -0700) it happened John Larkin wrote in :

Yea, he should have ordered 'fish and chips'.

No, bangers and pohtah or bittah :-)

Same for many Maxim parts. Which I won't use :-)

At the end of the day I am always glad when I am done with another switcher design and it's all nice and cool. Don't know about Vicor but one of the issues I see with many commercial designs is that they try to get away with the smallest magnetics possible and then that gets hot.

My suspicion is that Vicor has felt the pressure to produce smaller modules (or more power in the same-sized modules), and it appears that the response has been more in terms of designing really good thermal mechanics rather than changing the electrical design with the aim of improving actual efficiency.

We ended up with a pair of those long, skinny fans (two rotors in cascade internally) that scream like banshees given the high RPMs they run at. But, it does work, and Vicor modules do meet their data sheet specs... just barely.

I'd like to try out some Synqor modules one of these days... but since we don't typically have cost (or noise!) as driving factors in designs, so since we've been using Vicor modules for years now it's hard to overcome the inertia.

---Joel

I'll swap him the part file for a nice bottle (or box?) of red wine.

John

I'd seriously look into other solutions. High RPM fans are just a problem waiting to happen. One fine day the noise will miraculously change or disappear ... tssssst ... poof.

We've used LT's LTM4601 series for quite a while. They work well - especially if you're tight on space and height.

Bob

That's as far as I've made it with the LTM8021 kit. BTW, this site might come in handy for the eval info:

I've done PCBs for these things in another package. Shouldn't be any different than making a BGA. For their LGA118 package, I used 30 mil square pads with 34 mil square solder mask opening. You'll need to use a 25mil square pad. Solder paste stencil opening was 29mil square. In the LT notes they say don't use soldermask defined pads, yet they also say to use pours under the input, output, and ground areas. I ended up using a pour with the equivalent of solder mask defined pads under those areas. Everything worked out fine. I talked to LT about how to deal with solder mask defined pads in the poured areas and they said you must break their rules and use solder mask defined pads in the pour areas. I made pin-A1 square and didn't use the funny chamferred pad. If you haven't figured this out, do the part in inches (mils) as that is the base units for this part.

BTW, I really like these parts. It's expensive, but they do work well and you only have one part to worry about regarding what voltage you're going to get. That's easy to test with an ohm meter before you fire up the thing and potentially fry your $500 FPGA.