Arch Linux

wast3

Member

Registered: 2012-01-26

Posts: 27

Website

Differences in power draw with different batteries

TL;DR: I've got a small and a large battery and my laptop draws substantially more power when using the small one. Why Why Why??

Not really an Arch-related issue. Maybe not even Linux related, and possibly not OS related at all. Nevertheless I have this curious behaviour in my laptop and maybe somebody here more knowledgeable can help me understand what's going on.

My laptop takes one of different batteries: 4,6 or 9 cells. I have one with 4 cells rated 35Wh (14.5V) and one with 6 cells rated 60Wh (11V). With the smaller battery the laptop draws systematically more power. I made numerous empiric tests measuring idle power draw using upower and powerstat and I noticed that, while the absolute minimum power draw is more or less the same, with the bigger battery power draw is more or less constant towards the minimum, whereas with the smaller one there are spikes in power consumption, and the minimum is reached rarely.

Recently I discovered this small utility, gnome-battery-bench, and did more systematic tests, which confirmed my suspects. Here are my results. The small battery draws 23% more power at idle and almost 30% more under light load!

Has anybody seen anything like this? What do people think the causes may be? Could it be the difference in voltage? Are there other tests that I could do? I only use arch on this laptop and I wouldn't want to install Windows to test this stuff

Edit: I have two different 4 cell batteries and it happens with both. In other words, the battery is good.

Last edited by wast3 (2018-04-29 21:01:57)

ewaller

Administrator

From: Pasadena, CA

Registered: 2009-07-13

Posts: 19,785

Re: Differences in power draw with different batteries

I question your instrumentation.  What run times to you actually achieve with the 35Wh battery versus the 60Wh battery?

Before I continue, your first link is 404.

In DC systems, Power is the product of voltage and current.   Modern switching power supplies are pretty good at reflecting power from the load to the source with about 90% efficiency.  In other words, for 10W output, the input will be roughly 11W.  To deliver the same power, a battery with a lower voltage must provide a higher current.   The 11V cell should be drawing 14.5V/11.0V times the current drawn from the 14.5V cell (a ratio of about 1.32).   But... The power remains constant.  That 1.32 smells an awful lot like your 30% increase.

It sounds like your instrumentation is making assumptions about the battery voltage.  I am guessing that it is assuming the voltage of both batteries to be 11V. That erroneous low voltage times the real current would represent a power figure that is about 30% less than the actual power for the larger battery.

Another possible explanation would be the efficiency of the switch mode power supply.  Often you can only get the peak efficiencies of 90%+ at the "Sweet Spot" where the output current draw and the input voltage are at the ideal values.  Moving away from that operating point (say, a higher input voltage) will cause that efficiency number to drop.  But 70% for a 3.5V delta would seem more than a little extreme.

Last edited by ewaller (2018-04-30 13:56:44)

---

Nothing is too wonderful to be true, if it be consistent with the laws of nature -- Michael Faraday
Sometimes it is the people no one can imagine anything of who do the things no one can imagine. -- Alan Turing
---
How to Ask Questions the Smart Way

wast3

Member

Registered: 2012-01-26

Posts: 27

Website

Re: Differences in power draw with different batteries

I question your instrumentation.  What run times to you actually achieve with the 35Wh battery versus the 60Wh battery?

Run times seem consistent with the power draw measurements. I never exceed 5 hours of moderate use with the small battery, but get more than 10 with the 6 cell,  although I have never drawn it 100% to 0% in a single session. Anyway the 6-cell depletes less than 10% in one hour.

Before I continue, your first link is 404.

Sorry, the correct link is https://aur.archlinux.org/packages/gnom … -bench-git

In DC systems, Power is the product of voltage and current.   Modern switching power supplies are pretty good at reflecting power from the load to the source with about 90% efficiency.  In other words, for 10W output, the input will be roughly 11W.  To deliver the same power, a battery with a lower voltage must provide a higher current. The 11V cell should be drawing 14.5V/11.0V times the current drawn from the 14.5V cell (a ratio of about 1.32).   But... The power remains constant.  That 1.32 smells an awful lot like your 30% increase.

It sounds like your instrumentation is making assumptions about the battery voltage.  I am guessing that it is assuming the voltage of both batteries 11V. That erroneous voltage times the real current would represent a power figure that is about 30% less than the actual power for the larger battery.

Thanks for the explanation. I must admit that I am not familiar with these concepts. I don't know whether it is the case of bad measurements: the only tool that reports battery voltage is upower, and it correctly displays the 4-cells as 15.765 V and the 6 cells as 11.85 V.  These values are not constant though, they may vary of a 0.5V in between measurements, but the ratio between the two is kept, I'd say.

What really puzzle me are those spikes in power consumption with the small battery, whereas the big battery depletes at a more constant rate, which is what one would expect in idle conditions. Could these be the product of bad measurement alone? They really look like some form of malfunction. I should try and isolate the subsystem that produces them. Could it be bad circuitry?

Thank you very much for your help though!