TPS62291: Efficiency curves at lower voltages that shown in the datasheet

Thanks for the quick response. The TPS62082 looks fine for my application.

But this brings up a general question I have about TI parts: There are often many (and sometimes many dozens) of TI parts that will work, but there is no guidance on the "age" of the parts or other factors that might push the decision one way or the other. Part numbers are not "incremental" and other than the copyright date on the datasheet, there is no indication of "age", or more importantly, parts that will likely have a longer life than others (i.e. I don't know that it's valid to assume that an older part would be discontinued just because a newer "similar" part was introduced). How should I go about making that determination?

PS - I'm also using the parts below in the design. Should I be considering alternatives to any of these?

TPS62130
TPS630252
BQ24298
BQ35100

I hear you on your having difficulty find the best device for a given application. There are always many ways to solve a problem and choosing an IC is no different.

In my experience, we don't discontinue very many parts at all. So, the reason to use a newer device is not this. Rather, newer devices offer better performance in several areas.

How did you come to find the TPS62291?

On each IC's product page, there is sometimes a list of 'recommended alternate parts' at the top, like here: www.ti.com/.../TPS62290

Otherwise, yes, the initial date of the D/S can be an indication of how old a certain family is.



I'm very curious about your mix of parts you have chosen. Could you post your block diagram here or send it to me privately in a conversation?

The TPS62130 has an A version, TPS62130A, which offers an improved PG pin functionality. But newer than this is the TPS62135/6 which is smaller and more accurate and with more features.

With all of these parts, which are actually being used on two different projects, I found them by using the parametric search tools on ti.com and studying related app notes and reference designs.

Here's a simple block diagram of the power supply "chain" for each of the projects:

#1: [3V-3V6 Primary Lithium Cell or LM46001@3V6 from external 12-48V power] -> [TPS630252] -> 3V3 regulated output to circuit

Notes on #1:
> There are two different power options: a) Primary Cell or b) External DC power. For the external DC power, something like the LM46001 looks good (peak output of around 800mA); but I'm not clear on what "99% Max Duty Cycle" means for this part.
> The TPS630252 is definitely too big current-wise, but I otherwise like the efficiency and buck+boost functionality of this part, in order to maximize the life of the primary cell.

#2: [12V Gelcell]->[TPS62130]->[BQ24298 charging 3V7 LiPo Cell]->[TPS62291 or TPS62082]->3V3 regulated output to circuitry

Note on #2: TPS62130 regulates 12V to 5V to supply BQ24298 charger. It's enabled as needed to charge the LiPo from the 12V Gelcell, which isn't always present. Ideally, I would have preferred to use a charger that could accept 12V (really up to 14V), but I could not find such a part.

Both of these projects are SoC-controlled devices that will sleep most of the time (<1% total duty cycle).

Regarding the TPS62130, I don't see any advantage for me to use the A version, and from reading the datasheets, it appears that the TPS62130 is more efficient than the TPS62135 (>90% across most of its load profile vs >80%).

Ah, it's two different systems. That makes more sense :)

For #1, there are many ways to do this depending on the required/allowed voltage at the load. A buck-boost has relatively high Iq, but if it can be turned on only when needed, this doesn't matter. If the load can accept a range of voltage and/or a higher or lower voltage, then either a boost or buck converter offers much lower Iq. This reference design shows a low Iq buck: http://www.ti.com/tool/PMP9753 Either the TPS61291 or TPS61098/9 are low Iq boosts. The TPS63051 is a lower power buck-boost.

When comparing the 12Vin bucks, the TPS62136 will be a little more efficient than the TPS62135. At 5Vout, these looks pretty close on efficiency.

For #2, don't you want a buck-boost there as well to operate the battery down to lower voltages?

Some 12-V chargers are listed below:

bq25606 - standalone

bq25601 - I2C controlled

bq25896 - I2C controlled

Thank you for the suggestions, I'll explore the alternative options.

For #1, I do need the buck/boost converter because there are multiple battery options, providing between 3V and 3V7 (different cost and battery-life profiles), so I have to put up with the higher Iq. And in this application, it is likely that the device will need to run until the battery is nearly or completely drained.

For #2, I can get away with using only a buck converter because the battery (a LiPo) will maintain >3V3 for 80-90% of its discharge cycle. And in this application, the LiPo will never actually get to that point (it's there only as a backup power source, when the 12V source is periodically unavailable for short periods of time).

Thank you. The BQ25896 looks very interesting.

By the way, the other two chips (and several others I found) show a Vin max of up to 13V5. At the start of a full charge, it is typical to see battery voltages of up to just under 14V (which the BQ25896 supports). That has made me reluctant to use the 13V5 parts for this application. Thoughts?

Most of our chargers were designed for no higher than 12-V wall adapter power, hence the lower recommended VINmax. The ICs will not be damaged by voltages up to 22V but we prevent operation until the input voltage drops below input overvoltage threshold. The bq25896 input OVP is slightly higher than bq2560x family.

Question about the BQ25896: In reading the datasheet, it's not clear if the V-SYS output is regulated, specifically when V-IN is 13+ volts. From my experience with the other BQ chip, I assume that V-SYS is regulated and will never be higher than V-BATT or V-CHARGE (i.e. ~4.3V). Is that correct?

Correct. V(SYS) is regulated to MINSYS when V(BAT)<MINSYS and then will be a no higher than V(BAT)+50mV typical.

Eric,

The HiZ bit is one of the bits that get reset to default if the watchdog timer expires. You will either need to disable watchdog timer and write a 1 to it before it expires. Default watchdog timer expiration is ~40s.

Never mind--it was the watchdog expiring. I had that idea originally, but since it was happening in less than 40 seconds (the watchdog default), I had discounted that theory. But I've just proven it by disabling the watchdog.