Voltage Stacking with energy harvesting STM32WB55

[methene]

Hello, I am attempting in voltage stacking of 2 STM32WB55 board, but unlike arduino I could not connect Top board's GND to bottom board's 3.3 V pin. I would like to request for help or insights for alternative method I could make it work. I would really appreciate the help. 

Problem

1. When I connect the Top STM32 GND to the VDD of Btm STM32, it discharged the voltage in Btm capacitor.

2. On one attempt I rearranged the circuit in similar stacked configuration, it did not discharged the Btm capacitor voltage. Instead the system goes haywire.

Thank you.

[Andrew Neil]

But, again, those (claimed) advantages are only obtained by modifying the internal architecture of the chip - at the transistor level.

You can't do that with a ready-made IC - let alone a full board !

I also can't see any sense in "stacking" two STM32WBs - if they're being connected by wires in series, why do you need radios at all? Certainly, you don't need two radios!

You would achieve far more real-life savings by just using a single radio - or no radio at all!

And the connection you showed is not series:

What is the purpose of that central connection?

That's what makes it not series! It gives you two separate current loops"

PS:

Did you mean this:

[methene]

Yes, they did on chip implementation, I do believe there are much easier and methods that made more sense. I am trying out different possibilities as well because I am not sure what is correct, but my goal it to implement and observe the halving of current. 

These are some of the other connection method I had done previously, which I am still stuck at connecting them in a configuration of top to bottom board.

[methene]

The PV will charge the capacitor, it will take time to charge the capacitor before the voltage is used. In open circuit voltage I think their datasheet suggest it is around ~5V

So, you are suggesting it is simply impossible to have top board's return connected to bottom board supply.

[unsigned_char_array]

---

@methene wrote:
my topic are related to voltage stacking of microcontrollers with energy harvesting capabilities"

---

What are "microcontrollers with energy harvesting capabilities"?
As far as I know this MCU does not suck energy out of the environment. It's not an energy harvesting MCU. You need to feed it power the old fashioned way. Which could be an energy harvesting circuit. An energy harvesting circuit usually takes a very low power power source (microwatts or lower) and stores this energy in a small battery or super capacitor and periodically wakes up an MCU which does it's thing (take a measurement, or send some data wirelessly) and then goes back to sleep. Here you seem to be connecting the solar cells directly to the MCUs. You need a battery.

You need to go back to first principles. An MCU consumes less current at a lower voltage. So if you give it the lowest voltage it needs you can run at a lower power (voltage*current). You can save much more power by lowering the clock frequency, using sleep modes or even powering down. A very low voltage might make it harder to interface the MCU with other chips, unless you use a different voltage for the GPIO(sometimes this is possible). You pick a voltage (or voltage range) and you stick with it.
You probably need a voltage regulator to achieve a voltage within the operating conditions of the MCU unless your power source happens to have just the right voltage (and stable). A voltage regulator is not perfectly efficient. Especially at low currents where quiescent current (no-load leakage current) is probably dominant.

Connecting two MCUs in series won't work as they require a specific constant voltage, not a constant current. They would need a shunt regulator similar to Lithium ion cell balancing circuits to keep the voltage equally distributed.
Voltage stacking in ICs is very complicated as it requires multiple voltage domains and therefore cross voltage domain interfaces in order to allow communication between different domains. In addition you would need some logic that would equalize the current consumption of both domains as to not have to waste some current using shunt regulators. Voltage stacking has been proposed for GPUs. That would make sense as parallel tasks could be synchronized to consume similar current. The GPU would then require double the voltage and half the current. I think current GPUs need about 1V and can draw hundreds of Amps! I see no point in doing that for low power MCUs.

Maybe you were thinking that if the solar cells produce more than twice the voltage an MCU needs you could lower the current when connecting the MCUs in series instead of in parallel. But a switched capacitor voltage regulator does that too internally (it charges capacitors in series and discharges them in parallel to half the voltage), so you could use that too. Or use a buck converter. Or connect the solar cells in parallel.

[methene]

You are correct, my phrasing is incorrect, I am connecting an energy harvesting circuit which compose of PV and capacitor to power STM32WB controller. The PV will charge the capacitor and once it reaches a certain threshold it carries out an action and then enter back a low power mode that consumes less power to charge again and the process repeats. 

Since in energy harvesting, the voltage are not stable instead it charges "increase" and discharged "decrease", does that meant that it is impossible to do voltage stacking for energy harvesting system.

I am running same tasks for the top and bottom domains, to make sure they consume similar voltage and current. My goals is to find out if it is possible for voltage stacking to work on a energy harvesting system.

[unsigned_char_array]

My advice is to optimize the voltage regulators for lowest quiescent and lowest shut down current. That's the bottleneck.
And measure current consumption (this is hard to do for very low currents, but there are devices that do this).

[Andrew Neil]

(oops)

[Andrew Neil]

---

@methene wrote:

The PV will charge the capacitor and once it reaches a certain threshold it carries out an action and then enter back a low power mode 

---

What do you mean by "it" here?

Does your PV have some "intelligence" to do this?

The STM32 can't do that, because it requires a minimum voltage to work - it will not work from a supply which ramps slowly up from below that minimum operating voltage.

Another thing that doesn't make sense: you have two PVs, and two microcontrollers. There's no benefit to connecting them in series, because each one will still be seeing the same current!

There is no halving of the current in this case!

You would only get a theoretical halving if one PV supplied enough voltage to drive the two MCUs in series.

[Andrew Neil]

---

@methene wrote:

These are some of the other connection method I had done

---

Only one of those is actually a series connection:

[Andrew Neil]

---

@methene wrote:

So, you are suggesting it is simply impossible to have top board's return connected to bottom board supply.

---

Not impossible - but of no benefit.

This connection should work:

But it's not a series connection. In this arrangement, each PV is powering one MCU board separately.