low power microcontrollers, RTCs, ePaper and solar cells

[tszaboo]

IMHO it is odd nobody mentions TI's MSP430 series which are supposed to be very low power.

From the top of my head, excluding the oddball 4 bitters, probably the MSP430 FRAM or the 0.9V parts will require the least power.

[GeorgeOfTheJungle]

Cypress' PSoC 4 (and perhaps 5?) also have low power modes with figures in the few-ish nano amps IIRC and the dev kits are cute and cheap:

http://www.cypress.com/documentation/development-kitsboards/psoc-4-cy8ckit-049-4xxx-prototyping-kits

[ebclr]

You can use only one fet transistor with few passives, this application does not need a microprocessor

[nctnico]

Cypress' PSoC 4 (and perhaps 5?) also have low power modes with figures in the few-ish nano amps IIRC and the dev kits are cute and cheap:

That may be but in many cases you'll need to read the fineprint and that will tell you a completely different story. Some microcontrollers can only come out of the lowest power state with a reset 

[GeorgeOfTheJungle]

Cypress' PSoC 4 (and perhaps 5?) also have low power modes with figures in the few-ish nano amps IIRC and the dev kits are cute and cheap:

That may be but in many cases you'll need to read the fineprint and that will tell you a completely different story. Some microcontrollers can only come out of the lowest power state with a reset 

Huh, right you are!

The PSoC® 4 provides five power modes, intended to minimize the average power consumption for a given application. The power modes, in the order of decreasing power consumption, are: ? Active ? Sleep ? Deep-Sleep ? Hibernate ? Stop

Active, Sleep, and Deep-Sleep are standard ARM-defined power modes, supported by the ARM CPUs and instruction set architecture (ISA). Hibernate and Stop modes are additional low-power modes supported by PSoC 4. These modes are entered from firmware similar to Deep-Sleep, but on wakeup, the CPU and all peripherals go through a reset.

Sleep modes are a few µA, only hibernate and stop are in the nano amps.

[FrankBuss]

You can use only one fet transistor with few passives, this application does not need a microprocessor

How do you build an RTC with one fet transistor with few passives?

[BrianHG]

You can use only one fet transistor with few passives, this application does not need a microprocessor

How do you build an RTC with one fet transistor with few passives?

I think 'ebclr' is thinking you only want a 32Khz oscillator.  Yes you can create a 32khz oscillator with a fet and watch crystal, but, feeding that to a MCU RTC clk input would probably draw the same or more than it's internal RTC all within the IC's die.
With 3 fets, you may be able to make a 1hz multivibrator oscillator, yes, this would draw less feeding the MCU RTC so slow, but, accuracy will suck and it may fail to oscillate.

'ebclr' may be also thinking of using an MCU in ultra deep sleep (which require reset to wake up), setting an IO to charge a cap and fet amp to wake up the CPU through reset a second later, or even 10 seconds later.  This 'might' work, but, once again, a RC 0.1Hz reset clock would suck in accuracy with temperature and system voltage.  But it would be functional.  I'm not sure how much power you would save though.  You will be using something like 20 mega-ohm resistors to charge and discharge the cap & bias the j-fet to get that nano-amp range.  PCB leakage, even a stray fingerprint, change in humidity anywhere near this would throw it out of wack, though, this solution will always run unless your MCU code crashes or forgets to cycle that cap before going to deep sleep.

The stand alone clock chips in the previous posts was the lowest power I've seen, second closely by using ultra low power MCUs running them at their absolute minimum possible VCC and at room temp.  Combining those dedicated clk chips with a MCU is such a close call in power consumption, I would just use the best possible MCU running it on it's own.

[GeorgeOfTheJungle]

This processor needs only a few electrons to work

That was funny, hahaha

[JPortici]

Cypress' PSoC 4 (and perhaps 5?) also have low power modes with figures in the few-ish nano amps IIRC and the dev kits are cute and cheap:

That may be but in many cases you'll need to read the fineprint and that will tell you a completely different story. Some microcontrollers can only come out of the lowest power state with a reset 

Huh, right you are!

The PSoC® 4 provides five power modes, intended to minimize the average power consumption for a given application. The power modes, in the order of decreasing power consumption, are: ? Active ? Sleep ? Deep-Sleep ? Hibernate ? Stop

Active, Sleep, and Deep-Sleep are standard ARM-defined power modes, supported by the ARM CPUs and instruction set architecture (ISA). Hibernate and Stop modes are additional low-power modes supported by PSoC 4. These modes are entered from firmware similar to Deep-Sleep, but on wakeup, the CPU and all peripherals go through a reset.

Sleep modes are a few µA, only hibernate and stop are in the nano amps.

also some deep sleep modes may remove power to RAM, erasing its content

[Bruce Abbott]

As a sub-project of my Nixie clock, I plan to use a battery backup real time clock (or gold cap backup). First I thought about the DS1307, but it needs 500 nA

And this is a problem because...?

[BrianHG]

As a sub-project of my Nixie clock, I plan to use a battery backup real time clock (or gold cap backup). First I thought about the DS1307, but it needs 500 nA

And this is a problem because...?

Maybe he wants to have a 10 year battery backup.
Or, to backup the clock with a 10000uf cap.

[Bruce Abbott]

DS1307 datasheet says:- "A lithium battery with 48mAh or greater will back up the DS1307 for more than 10 years in the absence of power at +25°C". CR2032 is 200mAh.

There seems to be an obsession with ultra-low standby current at present, people wanting >10 years life out of products that will probably be thrown away in 5 years or less.

Like the charger for my electric lawnmower, which had two CR1225's in series to back up the PIC16F616. The manufacturer came up with a clever way to extend their life - instead of connecting them to the MCU via a resistor, they used a capacitor! So the backup battery had heaps of life left when the charger malfunctioned due to a power glitch, and cooked the SLA battery it was charging. Result:- charger went in the trash after only 2 years...


     
 

[rsjsouza]

Like the charger for my electric lawnmower, which had two CR1225's in series to back up the PIC16F616. The manufacturer came up with a clever way to extend their life - instead of connecting them to the MCU via a resistor, they used a capacitor! So the backup battery had heaps of life left when the charger malfunctioned due to a power glitch, and cooked the SLA battery it was charging. Result:- charger went in the trash after only 2 years...

Wow... just... Wow.

[FrankBuss]

Maybe he wants to have a 10 year battery backup.
Or, to backup the clock with a 10000uf cap.

Yes, with a capacitor. And not important for the clock project, but can't be wrong to know the best possible device for a given task for other projects, like my solar cell powered ePaper project, if I want to add an RTC to it (the integrated RTC of the ESP32 I plan to use is not very accurate, I measured a few percent wrong per hour).

[BrianHG]

Ok, for a cheap capacitor at 22x the one I mentioned, 0.22F 2.5v here...
(2 Ohm, fast charge and discharge, small and cheap...)
https://www.digikey.com/product-detail/en/eaton/B0510-2R5224-R/283-2775-ND/1026743

After charging to 2.5v, with a MCU which draws 25ua at 2.5v and dies at 0.9v drawing 18ua (example approximate PIC XLP MCU with 32khz clock crystal), how long can you safely retain keep time?

However, with solar charging, I see why the absolute least is the best possible solution.

[GeorgeOfTheJungle]

my solar cell powered ePaper project, if I want to add an RTC to it (the integrated RTC of the ESP32 I plan to use is not very accurate, I measured a few percent wrong per hour).

Frank, perhaps you knew already, perhaps you didn't, just in case:

Those epaper displays are 0 power iff you leave them alone but updating the display is a (quite) power hungry operation. If you pretend to update the display say every minute (a clock) I'd swear that small PV is not going to be able to provide enough power. During an update, in a kindle 4, I see peaks in the hundreds of milliamps (@ the lipo) and the updates last at least 0,4s (in black and white mode, in grayscale it takes much longer!) if not more, up to 900ms. Part of that is due to the increased cpu activity, but most isn't, I think it has to do with the +/- 20V power supply that has to be turned on to update, charge some filter capacitors, do the update, then shut it down again and loose all the energy that went into charging them, repeat for every update cycle. That plus the work/power required to move the eink in the capsules as well, of course.

[FrankBuss]

The datasheet of the new ePaper display I plan to use says 8 mA, but I guess this is a bit low, or I read the wrong number. I've done some measurements with my 2.7" display and it was about 50 mA when updating, plus 20 mA for the 8266 module:



This will be more with the planned new ESP32 module, which needs 110 mA when running and transferring data over WiFi, and the bigger 4.2" display might need more power, too. But if it is less than 200 mA, it should be possible to update the display once per hour over the internet:

https://hackaday.io/project/20466-wifi-epaper/log/58513-solar-cell-tests-and-bigger-display

Maybe the RTC wouldn't make sense for the display then. But it could measure exactly how long it couldn't connect to the internet, if there is some problem

[GeorgeOfTheJungle]

Yes, once per hour requires 60x less energy than once per minute :-) But the updates in a larger display will also take longer => draw more energy. I love these eink displays. Can that one do grays or is it b&w only? I've seen there are new ones that can paint in red and black! How cool is that?

[GeorgeOfTheJungle]

I'm also having problems with the clock drifting too much while in sleep mode, I'm tempted to add a DS1307 for timekeeping.

[FrankBuss]

Can that one do grays or is it b&w only? I've seen there are new ones that can paint in red and black! How cool is that?

Yes, I'm planning to use one of those, it is this one:

http://www.pervasivedisplays.com/products/420

Click on the "Spectra (E4)" tab to see the technical details and for the right Digikey link, the other one is the black/white version. I have already one and when I bought it, there were like 10 in stock at Digikey, now 10 weeks lead time. My breakout board from OSH Park for testing it should arrive soon.

I think they have gray scale displays as well. But someone managed to display gray scale images on a black/white display, just with clever timing:

http://hackaday.com/2017/04/20/e-ink-display-driven-diy/

Maybe I can do this too with my display, the ESP32 should be fast enough.

[TJ232]

The datasheet of the new ePaper display I plan to use says 8 mA, but I guess this is a bit low, or I read the wrong number. I've done some measurements with my 2.7" display and it was about 50 mA when updating, plus 20 mA for the 8266 module:



This will be more with the planned new ESP32 module, which needs 110 mA when running and transferring data over WiFi, and the bigger 4.2" display might need more power, too. But if it is less than 200 mA, it should be possible to update the display once per hour over the internet:

https://hackaday.io/project/20466-wifi-epaper/log/58513-solar-cell-tests-and-bigger-display

Maybe the RTC wouldn't make sense for the display then. But it could measure exactly how long it couldn't connect to the internet, if there is some problem

"20 mA for the 8266 module" are you sure about that? I really have the feeling that you need way more for ESP8266 module. Like 200-250mA more at peak.

[FrankBuss]

Peak is 170 mA when sending and receiving over WiFi. When WiFi is disabled, the datasheet says 15 mA, so this looks plausible.

[TJ232]

Peak is 170 mA when sending and receiving over WiFi. When WiFi is disabled, the datasheet says 15 mA, so this looks plausible.

I am really looking forward to see a full ESP8266 boot process ending with no more than 15-20mA peak.

[GeorgeOfTheJungle]

Those red and black eink displays I'd bet come from http://www.good-display.com/ if so you can get them @ aliexpress too.

WRT the greys the kindle has several "waveforms" to choose from, the waveforms are the recipes (that come in a serial flash chip on the display itself) that tell the driver what voltage pulses to apply at a given ambient temp to get a given grey level in that particular display.

Another thing that will drain some energy is that every few update cycles you're going to have to redraw everything twice, first inverted then not to eliminate ghosting, so double the energy of a normal update.

We are lucky that you're the OP because this is going wayyy off topic hahaha.

Edit: add ghosting picture

[GeorgeOfTheJungle]

About those IXOLARTM solar cells, they are 0,089*0,055= 0,004895 m2 @ 22% efficiency @ 1 sun wouldn't that be 0,22*1000*0,004895= 1,0769 Watts ? Why do they say 22% and 714mW?