Hotswap FET selection based on SOA with paralleled trench FETs

TI has a number of higher power hot swap designs using TI FETs. These have demonstrated very good robustness and reliability. A good example is the TPS2480EVM which is designed for a 12V, 50A application and uses a pair of CSD16401Q5's in parallel. TPS2480 contains TPS2490 and INA219 in a single package.

TPS2471x (x=0,1) and TPS24720 also employ TI FETs in high power hot swap applications. For higher voltage applications, TI now has 40, 60, and 100V FETs.

Just to make sure I'm reasoning about the SOA curve correctly, I want to run a hypothetical design by you.

Assume I use the Texas Instruments CSD18540Q5B.    http://www.ti.com/lit/ds/symlink/csd18540q5b.pdf
I'm assuming the hotswap timeout is 10ms and the max supply voltage is 30V.   One of the worst cases operating conditions is when the hotswap is enabled on into a hard short.   In this case, the FET will have about 30V Vds for 10mS.  The power limiting circuitry of the TPS2490/2 will have the keep FET current low enough during this period to keep the FET within its SOA curve.

Looking at the SOA curve (Figure 10),  at 30V and 10ms  the acceptable current for the FET would only be about 4.5 amps.   Thus the TPS2490/2 power limit must be set to something less than 135Watts (30V * 4.5Amps). 

The thing about the SOA curve is that if a look at a different Vds point, the power rating is much higher, say for a Vds of 10V and 10ms, the SOA current is a little higher than 20Amps.   This would be a power rating of 200Watts.

It seems that the SOA line is no longer a constant power, so I need to use  the SOA current limit at the worst case Vds voltage.   Does this make sense, or I'm I being too conservative?

Hi Derek,

In case it helps quick start your development, we have a calculation tool available under the product page in "Tools and Software"

http://www.ti.com/product/TPS2491/toolssoftware

This tool can help you design your component selection and SOA derating check.

Thanks,

Alex

Infineon has an appnote about the thermal stability line of the SOA that occurs at high Vds voltages in trench mosfets.    The appnote summary is that older planer mosfet's SOA curve only have a max current, max power, and max voltage lines.   While newer trench mosfets have a thermal stability limit that occurs at high Vds voltages.  This thermal stability limit is much lower than the expected power limit based on the package.

http://www.infineon.com/dgdl/Infineon+-+Application+Note+-+PowerMOSFETs+-+OptiMOS%E2%84%A2+-+Linear+Mode+Operation+and+SOA+Power+MOSFETs.pdf?folderId=db3a3043156fd5730115c7d50620107c&fileId=db3a30433e30e4bf013e3646e9381200

I believe this thermal stability limit for trench FETs is reason that the TPS2490 datasheet "Recommends against a "FETs whosesafe operating area (SOA) curves display multiple slopes on the same line (e.g. a line whose time parameter is a constant) in the region of high voltage and low current"

Finding a low Rds_on FET without multiple slopes in the SOA curve is next to impossible, so I started looking for a FET with an SOA providing a high current limit at the max supply voltage it wouldn't be operating at.    This wasn't easy, because many low Rds_on FETs seem to be optimized for switching performance, but have horrible SOA curves. 

I've seen this fact mentioned many times and found articles about certain vendors developing FETs specifically for robust linear operation and low Rds_on.   However, I couldn't find any actual products that were is production.
The FET series I did find was the NXP NEXTPOWER_LIVE series:

http://www.nxp.com/products/mosfets/standard_mosfets/family/NEXTPOWER_LIVE/#products

There is a pretty limited selection of FETs, but ones they do have a strong SOA and low Rds_on.

If anybody else has other FETs that have a similarly strong SOA, I would still be interested in knowing about them. 

Thanks, that was exactly what I was looking for.

Reviving this thread a few years later since it comes up in a Google search on the subject. 

It still doesn't seem to be possible to select FETs using SOA as a parameter using manufacturer's web tools.  Just the usual V(DS), R(DS,On), Q(G) type parameters.

SO, in addition to the above mentioned parts, I've observed that the Vishay SQ series of  "Rugged" Automotive FETs appear to have large SOA values with single-slope voltage-current curves.  Such large SOA values, in fact, that I'm almost dubious, but these appear to be modern parts introduced since the threshold -tempco/hot-spot phenomenon became widely known.

Cheers,

Ed