STM32 power consumption

I'm interested in building a mini gadget that needs a fair amount of computing speed, and I'm thinking of using this Cortex M4 board, powered by a small lipo cell:

formatting link

I notice in the data sheet that the maximum power drain is 800 mA at 3.3 volts, i.e. around 2.5 watts. That sounds awfully high--even a fancy smartphone with a multi-ghz processor doesn't use that much power.

What's going on here? Is that just the amount of power available from the on board voltage regulator, to poewr off-board stuff? Any idea how much power a board like this really draws? That said, I don't yet know much speed I really need. It's a DSP-ish computation that will run continuously, but let's guess it needs only 50 mhz of the available max frequency (something like 170 mhz). What's reasonable?

Also, are there particular issues with running a board like this from a lipo battery (3.5-4.2v, more or less)? The 3-terminal regular seems to want 5v in....

Reply to
Paul Rubin
Loading thread data ...

No, that doesn't sound right. Most of these little widgett boards are little more than an MCU on a "carrier" of sorts -- maybe some I/O connectors. Is yours any different?

Are you sure you aren't seeing a spec that includes inrush current? E.g., most small MCUs are in the 1/2W ballpark due to thermal limitations of their packages.

If your board is "just an MCU" (i.e., if you can hand-wave away the other I/Os that may be on the board as having negligible power costs -- like switches, etc.), then go to the datasheet for the MCU. It will give you a better idea what the worst-case power consumption is (at various temperatures and with specific load characteristics). More importantly, it will also tell you what you can TYPICALLY expect to see. The difference will be dramatic (often a factor of 2 or more; in low power modes, often an order of magnitude!!!)

Regulator adds some inefficiency (esp if it is a linear -- as is probably the case on a tiny widget). So, if MCU is 500mW, regulator might throw away 100mW (depending on input voltage, etc.).

Power consumed in the MCU is related to how many signals are switching and how fast. So, power at 50MHz should be 1/3 power at 170MHz ASSUMING THE DEVICE IS SPENDING 100% of its time doing that "same thing" at both speeds.

Signals switching at the *pins* cost more because they have to drive "pins" ("C"). Toggling an output pin a X Hz costs far more than toggling a bit in a register at the same rate!

Many MCUs will run at ~2V. 4.2V might be high for them without an LDO regulator in front. Depending on expected battery life and your cost/size, you might just opt to drop the 4.2 down with a diode or transistor instead of a real LDO (assuming the I/Os that you want to drive can also handle the sagging supply.

Any "typical applications" you can review to give you an idea of how they expect it to be deployed?

Reply to
Don Y

The board looks like it has the MCU, a 3-terminal voltage regulator (don't know if it's LDO), a USB connector, three LED's, and a few other parts like R's and C's.

No idea, that's why I'm asking ;-).

Hmm, I guess that fits the picture, though I have the impression the SOC's on boards like the Raspberry Pi are in the 2 watt range (and they run hot). If this chip uses 0.5W at full speed then (per frequency scaling as you describe) it's in the 1/6W range at 1/3 speed, which I guess is acceptable for my purpose though I'd prefer lower.

Thanks, good point. I'll have to check this and I guess I can experiment a little.

Reply to
Paul Rubin

OK, I found it at

Nowadays, LEDs don't take any power (assuming they are just little status LEDs and not trying to illuminate a room!) Ages ago, you might dump 20mW into an LED and another 30mW in a series resistor driving it. Nowadays, I'd put that at 2mW (and only when it is on).

Regulator will dissipate power depending on efficiency. How that relates to input *current* can vary depending on the type of regulator used. Looks like the regulator on that board is just a linear: "It operates on a 3.3V power supply, and an on-board voltage regulator allows the board to be powered directly from a USB cable." So, that only comes into play when powered from that cable.

It also suggests that the board runs on < 500mA as that's all you can count on getting from a USB cable!

I see the 800mA reference: "Power consumption: depends on MCU state (max current into 3.3V pad is 800mA)" but, can't see anything that would *use* that sort of power! (also, the USB condition, above)

Schematic (in "User Manual") confirms this assessment. OTOH, User Manual also repeats the 800mA claim!

Datasheet for the MCU itself claims 117mA @ 168MHz, 3.6V and 105C with all peripherals humming away. But *typically* 93mA in those conditions. And, at 50MHz it looks like 32mA, typ (~60mA at the extremes cited above)

Of course, if you aren't using certain peripherals, you save. E.g., same conditions but with all peripherals DISABLED drops you to the 20mA typ range (almost half the power was dissipated in those peripherals).

Cost you $30 to find out! :>

All I can think is inrush current. I.e., if your supply can't deliver that "for some period of time", the board might not start.

AFAIK, rpi has lots of other cruft on board. E.g., it can drive an HDMI TV! Has *external* RAM, etc.

Remember diodes need current flowing through them to give you a drop! :>

Reply to
Don Y

Looking at the pics on this page, I see a SOT89-3 REG1.

I'd bet that this reg will handle 800mA before it cuts out. {searching LD29080DT33]

Schematic shows REG is:

formatting link

And its an 800mA regulator.


Reply to

Yes, but the regulator is *powering* the device when -- and only when -- it is drawing it's supply from the USB connector. That's like saying "the circuit that my TV is on is fused at 15A" -- and claiming the TV *draws* 15A!

Even if a user wanted to take advantage of the on-board regulator to power OFF BOARD loads (letting the +3.3V "power" pins on the "DIP" act to SOURCE power), you would still need to know how much of that regulator output is consumed BY THE BOARD! (and, would have to assume your USB host could source > 800mA if you were intent on using that capability off-board.

I.e., there is at least one piece of data missing in order to make

*any* sense of that 800mA spec -- what the board *itself* draws from a 3.3V supply (imposed on the pins of the DIP) or from a 5V supply (sourced by the USB connector) -- regardless of loads!
Reply to
Don Y

Don't know what else is there. But a 50MHz ARM M4 should draw 30mA to 40mA active, or even less.

Reply to

I am not familiar with that particular processor/board, but estimating power consumption can be tricky. The rating is probably based on the maximum usage under worse case conditions, i.e. the processor doing everything it possibly can. Under lighter loads, it can take a lot less power.

Looking at your cell phone example, compare the battery life for the unit just sitting there waiting (but display still on), and the battery life when playing a fancy game, these can be VASTLY different, and the difference can be primarily the processor load.

Reply to
Richard Damon

formatting link

This web page states:

"Power consumption: depends on MCU state (max current into 3.3V pad is 800mA)"

I bet that the marketing guy wanted to let people know that the board can supply 3.3v to outside circuitry.

Even if the USB port can not, the board can.

If a USB charger is used, then 800mA would be possible.


Reply to

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.