[SOLVED] Ultra low-power timer to make a power saving circuit

[inklen]

I need to build a circuit that should save power of the "next" (main) stage for 12 hours. So the 12 hours is the duty cycle.

I've run into TI nano timers but their max duty cycle is just 2 hours. And I need 12.
Also, I'd like to be able to set the "reference point" with a button click or some "pre-programming" or something else. Something that would not affect the current consumption.

Could you please suggest where to start with this?
Thank you.

 

[KlausST]

Hi,

Give specifications, like
* expected current consumption ... but please not "as low as possible"
* precision,
* duty cyle, or how the signal should look like
* supply voltage
... and so on...

Klaus

 

[inklen]

ok, the TI's nano timers consume ~35nA. So this is a target. Probably up to 1uA should be fine.

Precision - just regular watch timer to count minutes.
Duty cycle 12 hours. Not sure I understand "how the signal should look like".
Supply Voltage Range: 1.8 V to 3.3 V.


I thought if I can't find the nano timer with >2 hours duty cycle I could probably use some counter to use multiple duty cycles... I've just started reading about counters though...
So this is just a thought.

 

[BradtheRad]

A store-bought appliance timer could be adapted to suit your purpose.

To build your own long-period timing circuit there is the 4060 IC (or its cousin 4040 or 4020). It consists of a series of binary counters. Apply a clock signal from any pulse generator. Period can be once every 2.6 seconds. Outputs on the 4060 change state one at a time. You'll find that the highest (namely the most significant bit) does so about 12 hours later. Fine-tune the clock frequency as desired.

You don't need to wait 12 hours to see if the frequency is correct. Watch the lowest bit (2^4 is the lowest available pin). It should change state after 42 seconds. Then the highest bit should do so after 12 hours.

 

[KlausST]

Hi,

Definitions of "Precision" and "duty cycle" can be found in the internet.

A "signal" should be defined somehow...
(I assume it's a digital signal) ... low level, high level, ON time, OFF time..

Klaus

 

[inklen]

Currently I'm thinking about using nano timer + SN74HC161 4-Bit Synchronous Binary Counter with SN74AUP1G08 Low-Power Single 2-Input Positive-AND Gate. As far as I understand I can count every 2 hours time interval. In my case the power will be delivered to the next stage after 6 intervals (which is 0110 binary), which can be "monitored" by the AND gate.

I'm not sure this circuit will consume less then ultra low-power MCU though... But it should be fun to build for learning purposes anyway.

(I have not idea on the schematic yet at all... I'll continue reading).

- - - Updated - - -

A store-bought appliance timer could be adapted to suit your purpose.

To build your own long-period timing circuit there is the 4060 IC (or its cousin 4040 or 4020). It consists of a series of binary counters. Apply a clock signal from any pulse generator. Period can be once every 2.6 seconds. Outputs on the 4060 change state one at a time. You'll find that the highest (namely the most significant bit) does so about 12 hours later. Fine-tune the clock frequency as desired.

You don't need to wait 12 hours to see if the frequency is correct. Watch the lowest bit (2^4 is the lowest available pin). It should change state after 42 seconds. Then the highest bit should do so after 12 hours.

Interesting, let me see what is 4060 IC...

- - - Updated - - -

Actually there's a better counter - SN74LV163A with Icc = 20uA. So I guess total supply current would be 0.035uA + 20uA + 0.9uA = 22uA (probably a bit more).

Now, based on my readings ultra low power MCU's consume magnitude more at least.

 

[KlausST]

Hi,

20uA for a CMOS devuce is pretty high.
But it's not the typical current, but the maximum current.

Nexperia 74LVC163 datasheet says 0.1uA typ. quiescent current.
But every signal transition will add current. In either case...with either counter...

Klaus

 

[inklen]

Thanks for the tip on Nexperia 74LVC163! I understand that every transition will add up current - law of nature

With 74LVC163 the total Icc=~12uA. Not bad at all!

ok, now I'll try to make a schematic on this thing. There are still big blanks in my knowledge to fill...

 

[KlausST]

Hi,

Not clear how you come to 12uA...

There are low power microcontrollers. Like AVR picopower, maybe other brands draw even less power.
When running on a 32kHz clock XTAL ... and pushing them into idle mode most of the time ..
Benefit is:
* single IC solution
* programmable, most flexible
* easy to calibrate ... via software

Klaus

 

[inklen]

I added up 3 Icc's - for nanotimer, counter and AND gate:

Icc(tot) = 0,035uA + 10uA + 0,9uA = 11uA.

Is this correct? Please correct me if I'm wrong here.

Yes, I looked at the MCU's but I think it will consume more power. For example, in the datasheet for ATmega48A/PA/88A/PA/168A/PA/328/P MCU's:

> Active Mode: 0.2mA

This is ~18x more.

 

[KlausST]

Hi,

Yes, I looked at the MCU's but I think it will consume more power. For example, in the datasheet for ATmega48A/PA/88A/PA/168A/PA/328/P MCU's:

I recommended
* 32kHz --> 25uA @ 3V active
* most of the time idle --> 15uA @ 3V

For sure, if you don't follow my recommendation, then you get different values. No wonder...

Klaus

Added:
Your values are confusing.
For "your" ICs you use the typical current, for our recommendations you use the maximum current.
For our IC you ask for "Precision - just regular watch timer to count minutes." which is in the range of 50ppm, but your TI nanotimers have an accuracy error of 10000ppm (excluding R tolerance).

For me it seems you measure with different scales.

Klaus

 

[inklen]

I actually googled "AVR picopower" and ended up with "ATmega48A/PA/88A/PA/168A/PA/328/P" link. I just looked into Microchip site again, I selected "low power" microcontrollers and choose the simplest one (I think) - ATtiny104. On the page 218 - 220 they have different charts. On the "Active Supply Current vs. VCC (Internal Oscillator, 128 kHz)" (Figure 23-5) it's about 40uA. So yes, this is comparable.

Please tell me which one exactly you look at.

Now, the values I use from IC's datasheets are just Icc (MAX), so I'm not sure why it's a different scale.

Anyway, if there is an MCU with ~11uA active I'd be glad to use it.

Alex

 

[KlausST]

Hi,

In post#10 you have chosen Atmega48.
--> Thus in post#11 I referred exactly to Atmega48 datasheet.
Now you choose another AVR ... and another clock frequency.... why?
Please stay focussed.

Icc(tot) = 0,035uA (typ TI) + 10uA (max LVC163) + 0,9uA (???) = 11uA.

Klaus

 

[inklen]

I haven't use any MCU yet (well, like any of these IC's in this thread), so I'm not familiar with any of them. That's why I googled and picked the ATmega48 first.

yeah, you're right. With max values:

Icc(tot) = 0,05uA (max nanotimer) + 10uA (max LV163) + 0,9uA (max SN74AUP1G08) = ~11uA.
Same value anyway. I guess it's final.

 

[crutschow]

Here's a reference for a CD4060 circuit that will give up to a 24-hour delay.
If you omit the transistor, LEDs, and relay, the average current draw should be only a few microamps.
That could likely be reduced, if needed, by using higher values for R3 and R4 (with a proportional reduction in the size of C3).

 

[inklen]

Hm... thanks a lot! This is probably what I need! And this is the single IC circuit. Nice.

Let me see how I can use MOSFET here...

- - - Updated - - -

The CD4060 datasheet is not clear to me. Can you please tell me if I can use 3V source voltage? I see only 5V in the datasheet...

- - - Updated - - -

CD4060 datasheet.

 

[KlausST]

Hi,

The CD4060 datasheet is not clear to me. Can you please tell me if I can use 3V source voltage? I see only 5V in the datasheet...

Why do you ask?
Page 159 specifies supply voltage range....and it starts at 3V.

But if you want lower supply voltage, then just choose another logic family. Like 74HC4060. It starts at 2V.
But there is no typical supply current specification.

Klaus

 

[inklen]

I missed supply voltage part for some reason.

Icc for SN74HC4060 is 80uA max @ 6V. So it should be slightly lower @ 3V.
Not 11uA though... Decisions, decisions.

Ok, I'll see what the external circuit will look like. If I ever able to create it

- - - Updated - - -

I wonder if I can use the nano timer as a clock for SN74HC4060 ... It would be very low frequency clock. So it would count 2 hours cycles.
Sorry, if I write completely noob things...

- - - Updated - - -

ok, using nano timer as an oscillator probably not a good idea as crystal probably consumes less power anyway.

 

[std_match]

For this application, you should not look at the "active" current for MCU's. Some of them have a low power oscillator that can wake the chip from "sleep" after a certain time. The software can then do some quick operations and go back to sleep.
You can wake up once every second or so, and still have a very low average current.
One task at every wakeup is to increment and check a software timer. With this technique you can easily create any delay, 12 hours, 12 days, 12 months or whatever.

 

[inklen]

I spent hours today reading about different options and I decided to go with TI ultra low power MCU + D-type Flip Flop + MOSFET.

TI MSP430FR2000 MCU has RTC and can be put to very low power mode and then be waken up using its RTC interrupt. At this mode the power consumption is ~2uA.
If I set RTC wake up interval to 12 hours I will get what I need.

Now, I plan to connect MCU's I/O to D-type Flip Flop in order to keep it H/L while MCU is in stand by mode. This will keep open/close state of MOSFET.
Flip flop Icc = 0.9uA max.

This solution is optimal as far as I can tell... and I like it