Rappel d'anniversaire valable 50 ans

Lit les registres d'alarme RTC et imprime l'heure et la date de l'alarme sur la console lorsque le "a" la commande est reçue.

Étape 9 :Tester

Ce serait la partie la plus intéressante de ce projet où vous saurez comment j'ai fini par fabriquer un appareil qui a fonctionné pendant 50 ans sur une pile bouton ! J'ai d'abord prototype l'ensemble du circuit sur une maquette et finalisé la conception. J'ai utilisé une batterie Li-Ion (3,6 V) à des fins de test afin d'économiser mes toutes nouvelles piles boutons. J'ai utilisé mon Fluke 87 True RMS multimètre pour les mesures de courant. Il a un 0,1 uA précision pour la gamme de micro ampères.

Voyons comment nous pouvons mettre l'Atmega168P en mode veille profonde et réduire considérablement la consommation de courant.

noInterrupts (); //désactive temporairement les interruptions set_sleep_mode(SLEEP_MODE_PWR_DOWN); // Choisissez notre mode de veille préféré :sleep_enable(); // Définir le bit d'activation du sommeil (SE) :ADCSRA =0 ; //désactive l'ADC power_all_disable(); //désactive tous les modules digitalWrite(LED_PIN, LOW); //éteindre la LED pour indiquer les interruptions de sommeil (); //réactiver les interruptions sleep_mode(); //s'endorme  

Comme je l'ai déjà dit, c'est la première fois que j'utilise les modes veille dans un microcontrôleur (MCU) car je n'en ai jamais eu besoin auparavant. La plupart des informations relatives aux modes de veille AVR ont été trouvées dans ce fil de discussion et dans la documentation de la bibliothèque AVR.

L'ATmega168P dispose de cinq modes de veille.

• SLEEP_MODE_IDLE – moins d'économies d'énergie
• SLEEP_MODE_ADC
• SLEEP_MODE_PWR_SAVE
• SLEEP_MODE_STANDBY
• SLEEP_MODE_PWR_DOWN – most power savings

More info on the sleep modes can be found here and in this video. We're using the SLEEP_MODE_PWR_DOWN mode as you can see there. At this mode, the current consumption at 3.6V is only around 0.2 uA See the below graph from the ATmega168PA datasheet that shows the relation between active current vs supply voltage and power down current vs supply voltage.

Here's the actual reading of the current consumed by sleeping ATmega168P @1MHz.

The value hops between 0.1 uA and 0.2 uA due to the lack of precision. But such a precise measurement isn't necessary but would've been interesting to see.

The power consumption of ISL1208 RTC at max is 1.2 uA . So if we add this with the power down mode current consumption of the MCU we get 1.2 + 0.2 =1.4 uA . My meter measured between 1.4 uA and 1.6 uA which justifies the calculations. The variations is only due to the lack of precision and our approximation or rounding of numbers.

Here's an unlisted video from my YouTube channel where I show the testing.

Now let's do the simple math to find how long we can the system on a coin cell. The CR2450N from Reneta has a nominal capacity of 540mAh . I have two red SMD LEDs on the system which consume about 6 mA (even with two LEDs) with when turned ON. So that's the peak current consumption of the device at worst. How long these LEDs light up can be summarized as,

1. As long as the time setting switch is activated while you're setting the time (but you don't have to do this every year)

2. The 6 second delay before sleep.

3. LED flashes on your birthday and stay ON for about 19 seconds.

Let's not be accurate here but just make an approximation. Let's assume the time setting requires 2 minutes (LEDs will be ON for 2 mins) and and additional 1 minute ON time for other things including the birthday flashing. So it's total 3 minutes or 180 seconds for which the system consumes 3 mA courant. Let's take the sleep or idle current as 1.6 uA , though it's lower actually. There's 31557600 seconds in a year of 365.25 days. If LEDs stay ON for 180 seconds in a year and OFF for (31557600 - 180) seconds, then the average current consumption will be,

Average Current =[((6 x 10^-3) x 180) + ((1.6 x 10^-6) x 31557420))] / 31557600 =(1.08 + 50.491872) / 31557600 =51.571872 / 31557600 =1.634 x 10^-6 =1.634 uAh 

If the average current consumption is 1.634 uAh, then the 540 mAh cell can run the device for:

Time Span (approx) =(540 x 10^-3) / (1.634 x 10^-6) =330477.3562 hours =13769.88 days =37.699 years 

Note than this approximation is do not consider self-discharge of the battery. It'll be taken into account later. You can also use the ElectroDroid app to calculate battery life. Here's a screenshot of the calculations we just did.

BUT WAIT...

Can we reduce the current consumption further ? YES WE CAN! I made further optimizations to my design to limit the average current consumption to 0.6 uA , yes I did that. Not let's see what optimizations I did,

1. To remove the extra red SMD LED to reduce the current when the system is active/running. Before it was 6 mA at max with two LEDs. So with one LED, it'll be reduced to half, ie 3 mA.

2. To reduce the current consumption of RTC, we need to disconnect the VDD pin of the ISL1208 RTC from the VBAT pin. Previously the VDD pin was connected to the VBAT pin where I supplied the 3V from the cell (you can see this in the schematic). In that configuration, I also had the LOW_POWER mode bit (LPMOD ) of the RTC set to 1 activating the low power mode. So now you might think if the low power mode is set, then the chip might be consuming the lowest current. But that's not the case when we have the VDD tied to VBAT. Because low power mode bit is only useful if we have VDD> VBAT all the time. At such situation, the RTC's internal power switch will select VBAT as power source reducing the current further by 600 nA when VDD>=VBAT (from typical 1.2 uA which I've mentioned before). But if we can run the RTC in VBAT only with VDD =0, the current consumption can be reduced to the minimum ie, 400 nA as per the datasheet. So what I did is, first I disabled the low power mode by setting LPMOD to 0. Then added a jumper to the RTC breakout board to disconnect the VDD pin from VBAT when I don't need it. Why need the jumper is because, the VDD pin must be greater than or equal to VBAT in order for the I2C to work. So I can connect the jumpers when I need I2C while I'm setting the time, and can disconnect it after. This will let the RTC to consume the targeted 400 nA current. Tada ! We did it!

Now that we have reduced the current consumption of the RTC from 1.2 uA to 0.4 uA (400 nA), we can do the math again!

System Active Current =3 mAh max

System Sleep Mode Current =0.4 uA (RTC) + 0.2 uA (MCU) =0.6 uAh

System ON time =60 s (time setting) + 19 s (birthday flashing) + 6 s (before sleep) =85 seconds

System Sleeping Time =31557600 s - 85 s =31557515 seconds

Total time in a year =31557600 seconds

Battery capacity =540 mAh

Here's the current consumption test video after the optimizations and mods.

If we put those new values to the ElectroDroid's battery life calculator, we get, 101 years and 136 days. A theoretical operating time of more than a century! The average current consumption is now only 608 nA . Here's the screenshot.

Okay, What's the actual operating time ?

Batteries aren't perfect, nor anything we design. So let's also consider the 1% self discharge of the cell into account.

1% of initial capacity of 540 mAh CR2450N =5.4 mAh

Self-discharge current =5.4 mA per year or 616.4 nAh (5.4 mA / hours in a year)

Adding this 616.4 nAh with the 600 nAh sleep current =1.216 uAh

Expected operating time with average current of 1.224 uAh =50 years, and 131 days.

That's the actual operating time if the cell will be fine

Here's a table of actual operating times of different types of coin cells with the 1% self-discharge of initial capacity every year.

The main practical concerns associated with running the device for such long periods are,

1. Will the battery retain the charge and voltage for that long ?

2. The effects of environmental variations on the circuit and battery performance.

3. And you screwing up things! XD (don't drop it, chew it, sit on it, run your car through it or launch it into space!)

Coin cells are usually made to last for 10 years, which is their shelf life , and some even last longer than that. But that doesn't mean they'll start to disintegrate to an useless state after that. If the cell is physically fine, it can still power things. As per Renata datasheet, the self-discharge rate of CR2450N is less than 1% of the capacity every year. As per this datasheet from Energizer, that 1% figure is of the fresh capacity. Below is a chart that shows the standard discharge time in years (this doesn't consider the self-discharge or other exceptions). It clearly shows the theoretical expected operating time is way too longer than 10 years.

Self-discharging not only reduces the capacity but also reduces the voltage. Both ATmega168P and ISL1208 are designed to be operated fine at voltages as low as 1.8V. So the reduction in voltage might not be a problem. You can learn more about running systems with coin cells here.

To ensure long operating span, we must make sure the device is properly concealed against much of the environmental changes such as temperature, humidity, corrosion etc. These are some things you can do to protect your circuits,

1. Coat the PCB with long lasting conformal coating.

2. Place a pack of silica gel inside the enclosure.

3. Seal the enclosure with less temperature conductive materials and make it air tight.

4. Place it somewhere safe from naughty hands!

Step 10:Building

I used a perfboard to solder everything as per the schematic. I used berg connectors for the battery, switch and LED so that it'll be easy to remove them if needed. Below are the some images of the PCB.

To build the enclosure I used a 4" x 4" switch box which I bought from an electrical shop. I made two rectangular holes for the switch and USB. You can 3D print an enclosure if you want; sadly I don't have one. The dome was snatched from a cheap LED light and used super glue to fix it on the box. I painted it with silver spray paint.

Use your ingenuity to build it.

What missing is some decorations . I'm not good at decorating things. If you are going to gift this to someone, you know what to do.

The final output is satisfying to the extend of my hardwork. I might find someone else to decorate it.

Step 11 :Improvements

There's always room for improvement. Some of my suggestions are,

1. Using a Nokia 5110 LCD with or instead of the LED bulb. The LCD controller only consumes a couple of 100 uA at normal operating modes without the LED backlighting of course. Monochrome LCDs only consume extremely low currents. So using it would be better than a flashing LED, where you can actually print the happy birthday message on the screen itself. I might do this in future because I have couple of them lying around here.

2. A software that runs on the computer that'll set/sync the time accurately and automatically. I'm thinking of developing a simple one using Processing.

3. Flashing the LEDs to indicate the current age - for example if it's your 5th birthday (OMG are you're reading this ?!), it'll flash a sequence for 5 times. Currently you can not set the current age in the system. You may add this.

4. Designing a dedicated PCB in eagle (planned).

5. If blinking LED isn't your thing you can actually make this more exciting with for example using an opto-coupler you can turn on an AC alarm, music player, lights or anything you want to blink, move and scream on the birthday of the one you wish. You can even exclude the microcontroller and only use the interrupt from the RTC. Everything's possible!

So what are you waiting for ? Make one, decorate it with stickers, color papers, glitter or anything and gift to your loved ones or even yourself! And tell them to wait for this magical device to wish them happy birthday.

What you see here is actually a prototype of a device that'll refine and release in future. So stay tuned. Please feel free to share if you have found any errors with this documentation or have any improvement suggestions. Happy making :)