Need help with _ _ _ schematic review _ _ _ for solar & supercap powered sensor

[doru.cazan]

Hi,
I am building a temperature sensor powered by a solar panel with a supercapacitor for backup. The main power part is heavily inspired by a Hackaday.io project: https://hackaday.io/project/175514-wireless-solar-powered-sensor-node/logs.
Will use a clock IC (below 1uA) to wake up the AVR MCU from deep sleep (also below 1uA); the MCU will set the clock IC timer alarm when initially powered then enters deep sleep. When the timer interrupt wakes up the MCU it will read the temperature then wake-up the radio part only for transmitting.
Are there any obvious errors in the attached schematic?
Many thanks!

[zilp]

Will use a clock IC (below 1uA) to wake up the AVR MCU from deep sleep (also below 1uA); the MCU will set the clock IC timer alarm when initially powered then enters deep sleep. When the timer interrupt wakes up the MCU it will read the temperature then wake-up the radio part only for transmitting.
Are there any obvious errors in the attached schematic?

I obviously don't know what exactly the application will look like, but a few things to consider:

- You are aware that the voltage supervisor eats probably half of your energy budget?

- You are aware that D2 leaks another quarter or so of your energy budget through R3/R4?

- Are you sure that the MTS01Z doesn't have any ESD protection diodes on the I²C pins?

[doru.cazan]

Many thanks @zilp for the points below!

- You are aware that the voltage supervisor eats probably half of your energy budget?

I was worried about this and the TLV809E data-sheet https://www.ti.com/lit/ds/symlink/tlv809e.pdf shows a typical 250 nA supply current.

- You are aware that D2 leaks another quarter or so of your energy budget through R3/R4?

I am open to select a diode that is better from this perspective (does better means lower reverse leakage current?); do you have any suggestions?

- Are you sure that the MTS01Z doesn't have any ESD protection diodes on the I²C pins?

Not sure. The data-sheet https://www.lcsc.com/datasheet/lcsc_datasheet_2210121700_Mysentech-MTS01Z_C5197400.pdf shows max 100nA input leakage current.

Can I assume that there are no ESD protection diodes?

Many thanks!

[Peabody]

What does R2 do?

[doru.cazan]

Hi @Peabody, not really sure, it is part of the Hackaday.io project design, link in 1st post. My best guess is that it helps limiting the supercapacitor charging current to even a lower value (30mA when the solar panel outputs 6V) compared to the shunt regulator capabilities, that can sink more than 60mA (according to the data-sheet max 100mA). Or, maybe to make sure the voltage drop on the D2 diode(depending on forward current) is low enough to allow the supercapacitor to charge to a higher voltage?

[zilp]

- You are aware that the voltage supervisor eats probably half of your energy budget?

I was worried about this and the TLV809E data-sheet https://www.ti.com/lit/ds/symlink/tlv809e.pdf shows a typical 250 nA supply current.

Ah, yeah,  I was looking at the TLV809 with 9 µA--though the 250 nA is only typical and for 3.3 V, up to 1.2 µA for Vdd=6 V.

- You are aware that D2 leaks another quarter or so of your energy budget through R3/R4?

I am open to select a diode that is better from this perspective (does better means lower reverse leakage current?); do you have any suggestions?

Reverse leakage current is the datasheet number you are looking for, yeah. One I have lying around would be the BAV116T (5 nA max at room temperature), but no clue whether that's a good choice for your application.

- Are you sure that the MTS01Z doesn't have any ESD protection diodes on the I²C pins?

Not sure. The data-sheet https://www.lcsc.com/datasheet/lcsc_datasheet_2210121700_Mysentech-MTS01Z_C5197400.pdf shows max 100nA input leakage current.

Can I assume that there are no ESD protection diodes?

I can't read chinese, but other than that there is no indication there that there are no protection diodes. Ususally, such specs are under the condition that all other limits are observed, and typically, one limit would be something like VIH <= Vdd + 0.3V, which would be exceeded when Vdd is 0V and VIH is 5V. Now, in this case, the datasheet doesn't specify an upper limit for  VIH, as far as I can tell. But that obviously can't be true (the chip certainly won't survive a sustained kilovolt applied to the pins), so we don't really know at what voltage those pins would not exceed 100 nA when powered down.

Also ... did you test that chip? I once tested a Mysentech DS18B20 clone, and that one wasn't great, though I don't remember details, IIRC quite a bit of offset from a genuine DS18B20+ from Maxim.

[Peabody]

Not an error, but it seems the circuit could be a lot simpler if you ran the ATTiny and the sensor at 3.3V.  Then you wouldn't need a 5V supply or any voltage translation.

Well on R2, it seems that it makes the regulated voltage vary with current.  The more current drawn, the bigger the drop across R2 even if the shunt works perfectly.  I just don't know why it's there, as opposed to making R1 200 ohms.

[doru.cazan]

Thanks @zilp, @Peabody!

everse leakage current is the datasheet number you are looking for, yeah. One I have lying around would be the BAV116T (5 nA max at room temperature), but no clue whether that's a good choice for your application.

The BAV116T looks good but will probably need to adjust the R3-R4 to be able to handle larger forward voltage drop; will look into it.
 

Also ... did you test that chip?

Used the chip on another project but nothing low power; have a nRF-PPK2 on order and will do as soon as it will be delivered.

it seems the circuit could be a lot simpler if you ran the ATTiny and the sensor at 3.3V

Due to supercapacitor backup for running overnight will need as much power as I can get; AVR, sensor and clock all run from 5.5V down to around 2V (will limit to 2.7V to be able to run the MCU at 10Mhz for smooth I2C communication and for reducing the processing time)

I just don't know why it's there, as opposed to making R1 200 ohms

I am not sure 100% either but the original design was tested by strange.rand so will keep it for now; depending on tests, I can replace the R1 with 200ohm and R2 with 0ohm later.