What happens when you power an MCU with higher than recommended voltages

[mikael1985]

Hello everyone,

Could you please clarify a thing for me.

I'm working with an esp8266 on a nodeMCU board V3, in a battery powered project.

Right now power is delivered by a 18650 powerbank (usb output). I Have sleep function enabled so i only wakes up and connect briefly 4times every hour. But I'm not satisfied with the runtime from the battery - primarely because of the peripheral circuitry current consumption.

I would like to skip the peripherals and the powerband and power the esp8266 directly with the 18650 - giving a range of 4.2v to 3.2v. The esp8266 datasheet specifies 3.6 as max voltage. However alot of people has done this with no apperent ill effect - but all has also noticed significantly higher current draw at the high voltages.

It still appears to me Im better of that way than introducing a 3.3v regulator.

My question is what happens to the silicon in the MCU when you exceed the ratings. is it only heat issues because my esp8266 runs for 5 seconds every 15 minuttes so I dont think it will run hot.

Someone explained the excessive current drain from unintebtional revwrse/forward biased of transistors and diodes in the MCU, but as long as that is no permanent effect I dont care.

But can anyone explain in general terms what happens inside the MCU?

Thank you very much
Mikael

Sendt fra min SM-G900F med Tapatalk

[mikael1985]

I would like to add the lithium cell will be protected! Just in case anyone was wondering ?

Sendt fra min SM-G900F med Tapatalk

[rstofer]

If you exceed the Absolute Maximum Ratings, eventually, the magic smoke will come out.  There's not much negotiating with "Absolute Maximum".

That excessive current is also creating excessive heat.

[Jeroen3]

Latch-up followed by smoke.

[MK14]

I'd suggest you consider looking into Micropower voltage regulators. You should be able to get one, with a low enough current/power usage, to be viable for you.

Example only:
http://www.linear.com/product/LT1761

As others have said. Exceeding the datasheet specifications is a very bad idea.

[T3sl4co1l]

Once accidentally powered a PIC24 with a 5.0V LDO instead of the (correct) 3.3V version...

The programming interface worked fine, but USB wouldn't enumerate correctly.

Possibly, analog functions, like the USB's PLL clock generator, weren't working.  Over-voltage causes current to diffuse throughout the IC -- perhaps analogous to how your own brain feels woozy when it's running a fever!  Digital logic can usually deal with an environment like that, but analog stuff, forget about it.

The amount of current flowing was fairly small, under 200mA.  Nothing was getting particularly hot (including the LDO, because the whole board was powered from 5V in the first place).

Note that this was a current-limited case: the 5V LDO (supplied by 5V) was in dropout, so chewed up a bit of voltage itself.  Even if the MCU had failed shorted, it wouldn't draw more than an amp.

I don't think you're guaranteed magic smoke, but doing it 1. from an unlimited-current supply, and 2. to a chip with a lot of analog functions in it (i.e., oscillator, ADC, radio), I don't see it working very well.

3.3V LDOs are cheap and plentiful.  Use one!

Tim

[mikael1985]

Thank you - for quick replies

 i will look into the micropower regulators, they seems to solve the issue for a small cost.

Regarding the 3.6v it is not specified as absolute in the datasheet, just as the max voltage range (the datasheet is translated from chinese) so some absolute values might be omitted - but I get you point that exceeding the datasheet range  is the fast/eventual road to brick the chip,

but I was still curious if you could heatsink or otherwise artificially enhance the tolerated voltage depending on the what the point of failure was in silicon chips

Sendt fra min SM-G900F med Tapatalk

[daveshah]

I have in the past accidentally connected an ESP8266 for 5V for 15 minutes or so and didn't notice any problems (but I wasn't doing anything particularly sophisticated). However I wouldn't be surprised if nothing else if overvoltage causes RF performance to degrade in various ways - in the theoretical worst case, causing interference and violating EMC rules due to harmonics etc. Also there may be batch-to-batch variation in performance when overvolted and long-term lifetime issues; so I definitely wouldn't design a commercial product operating the ESP8266 above 3.6V.