Too high power consumption on MSP430FR5739

Hi,

I took a look in the datasheet, and your power number matches the datasheet specification for 1MHz operation. If you look in the datasheet www.ti.com/lit/gpn/msp430fr5739 in section 5.4 Active Mode Supply Current into Vcc Excluding External Current, you'll see that it lists TYP current for 1MHz with ~66% cache hit ratio (which is a pretty typical application cache hit ratio) as 0.27mA. That is 270uA, exactly what you are seeing. This chart provides much more detailed information about current consumption than the 1 line highlight on the front page, so that you can better predict what you will see under your application's particular set of conditions.

So why does 81.4 uA/MHz not mean 81uA at 1MHz? The current consumption is not linear with MHz operation, so the front page has to show something more like what a typical application might see (which is why it is marked "Typical" on the front page) - but of course it can depend greatly on the operating frequency you use (which part of the curve of uA/MHz it is in) and what cache hit ratio your code has. For example, if you look in the same section 5.4 at 8MHz with 66% cache hit ratio (CHR), that is 0.58mA so 580uA/8MHz = 72.5uA/MHz which is better than the number on the front page - if you have a better cache hit ratio you can also see even better numbers. At 24MHz with 66% CHR you can calculate it is 91.6uA/MHz, still pretty close to that front page number even though you are running at the max speed of the part. 1MHz is actually kind of the worst case point if you look purely at uA/MHz.

For calculating the current budget in your application, instead you may want to plot in a spreadsheet the uA and MHz for some different points, as well as taking into account the duty cycle of awake vs LPM mode in your application. This can help you to find the "sweet spot" in terms of operating frequency to get the best current consumption for your particular use-case. For example, if you will do x instructions in active mode (to accomplish a certain task) once a second, and will be in LPM3 the rest of the time, your current numbers could look like this: (I assumed a 66% cache hit ratio for all of these, took wake time from LPM3, LPM3 current, and active mode currents at different frequencies all from the datasheet).

You can see that the average current depends on a number of factors then, and that in this back-of-the-envelope calculation that for 100 instructions 1/sec, it is better to run at 1MHz, but for 1000 instructions/sec it is better to run at 4MHz. Basically, it depends on how long you are going to be awake vs asleep to see if the incremental current increase of a higher frequency is worth it or not, and trying to find the best tradeoff to find your operating point for your app.

I hope this helps to give some insight.

Regards,

Katie

Startup from unpowered is difficult - if you are totally unpowered, the device will take longer to startup than it does when you simply wake from LPM - booting the part from nothing takes a lot more time than simply waking up. If you look in the datasheet at the section 5.19 Wake-Up from Low Power modes, if you are coming up from being off you're going to at least be seeing the time twake-up BOR which is listed at typical 1.6ms. Contrasted with wake from LPM3 which is much shorter.

So is this application needing to be entirely powered down between cycles - do you only get 1.5ms of Vcc periodically and have to get everything done in that time, completely powered down in between? How long is it between these 1.5ms bursts of Vcc? Can you tell more about what type of application this is?

The reason I'm asking these is that I think the timing will work out better for you if you'd be able to stay in LPM3 or LPM4 between these pulses - so basically (depending on timing) having a supercap or small battery that can be used in between these voltage pulses to simply let the part be asleep in LPM3 or 4 during this time - that way the startup time will be much shorter.

I also wanted to note, for that type of application I believe that MSP430FR5969 may be a better choice - it has a much shorter wake time from LPM3 or 4 (only 7-10us), and it also has lower current while in LPM3 or 4 (~0.5uA - see datasheet). The MSP430FR5969 is more optimized to have better LPM current whereas the MSP430FR5739 is more optimized to have better active current. If you're in LPM for very much of the life of your device, MSP430FR5969 is probably a lower power option. Note that the startup time from unpowered is similar between the two parts.
 
Regards,
Katie