Page 2 of 2

Re: And still it keeps on going

Posted: Sun May 06, 2018 7:50 pm
by toml_12953
Russel wrote:
Sun May 06, 2018 6:49 pm
grsbanks wrote:
Sun May 06, 2018 7:17 am
[...]I was deliberately running the battery down to see how the machine behaved with such a low power supply.
Is 1.90 volts the lowest you have seen? Have you replaced the battery yet? You know what they say about the curiosity and the cat...
"How low can you go?" - Limbo Rock sung by Chubby Checker

Re: And still it keeps on going

Posted: Mon May 07, 2018 7:19 am
by grsbanks
LOL :)

Yes, I did replace the battery after that.

Not only did I want to see how the machine behaved with so little power available but I also wanted to see something that I'm surprised nobody else picked up on. Look at the battery icon in that screenshot... Having seen the source code for how the system chooses which icon to display depending on battery voltage, I knew there was a provision for a 7th icon that represents something more dire than the "empty" battery. So there is it, a sqeezed battery :)

This also uncovered a "bug" that only hit me yesterday while thinking of this... Operations that cause a write to the QSPI chip are all protected so that they don't happen if there's less than 2.2V across the battery. If the thing were to shut down partway through a write operation, it would cause data corruption. So, how come I was able to take a screenshot with there being only 1.9V there?

It turns out, that specific write operation is not (yet!) protected...

Re: And still it keeps on going

Posted: Thu May 10, 2018 4:53 pm
by Russel
grsbanks wrote:
Mon May 07, 2018 7:19 am
[...] Having seen the source code for how the system chooses which icon to display depending on battery voltage, I knew there was a provision for a 7th icon that represents something more dire than the "empty" battery. So there is it, a sqeezed battery :)
I noticed that...something to avoid when your calculations are mission critical! (I did suffer a fit of laughter when I saw the icon. :lol: )
grsbanks wrote:
Mon May 07, 2018 7:19 am
This also uncovered a "bug" that only hit me yesterday while thinking of this... Operations that cause a write to the QSPI chip are all protected so that they don't happen if there's less than 2.2V across the battery. If the thing were to shut down partway through a write operation, it would cause data corruption. So, how come I was able to take a screenshot with there being only 1.9V there?

It turns out, that specific write operation is not (yet!) protected...
Nice catch!