Two capsules, same code, one eats battery, the other not

#1

I’ve been running two cubecell capsule’s with HDC1080 attached.
One monitors my shed, the other my greenhouse.
They both run the same off shelf code, the HDC1080 LoRAWAN example, with the id’s being the only difference. All runs unconfirmed, ADR on, AT off, RGB off.

The one in my shed, runs perfectly, has been running for months now and eats almost no battery. The one in my greenhouse, eats a 3000 mAh Li-on battery each month.

As the code is the same, the difference shouldn’t be there. The HW is slightly different. The on in the shed is the capsule where the battery soldering points are next to each other at the top, the one in the greenhouse is the one where the negative is a bit to the left.

I have been debugging and what not. The troublesome Capsule seems to go fine into sleep mode, all is good.
I have checked Vext on both, to see if it remained on. Here is where I found the only difference, but in the opposite of what I expected. On the shed capsule (the good one), Vext remains around 0,8 V while in sleep mode, where as on the Greenhouse one (the bad one) it goes to 0,00 V .

Anyone that has any ideas or any suggestions on where to look?

#2

Have you tested the current in sleep state?

#3

Sadly not. I don’t have equipment that can measure current, only the voltage :frowning:

But I imagine that is where the issue is. I mean, I just put in a freshly charged battery in the greenhouse capsule, here 2 days later, the voltage of it is already well below that of the shed capsule, which has been running for about 4 months now.

Is there a way I can troubleshoot if the correct deep sleep is happening? I have tried putting serial.print lines around the LoraWan.sleep() call and it appears (after 6 cycles or so, which I hear is normal) it goes quiet, until the sleep time is over and then it goes again.
But is there a way I can get debug info, to really confirm is deep sleep is happening?