Lenovo Battery Gauge wildly wrong X220 (WAS: erratic battery gauge) [ Edited ]

[JNavas]

---

ColonelONeill wrote:

Think of it this way, if you're mathematically minded:
P = power consumed in watt-hours.
dP/dt = instantaneous power consumption in watts
Knowing dP/dt, one can integrate to find watt-hours, and estimate total battery lifetime.
When dP/dt fluctuates based on usage, the estimated battery lifetime will fluctuate accordingly (sometimes wildly, as is the nature of the extrapolation formula).
The inaccuracy inherent in all estimates (which is why they're called estimates) varies based on the algorithm used to predict average dP/dt. Each algorithm may be good for some cases, and terribly bad in others.

---

Thank you, but I'm already quite familiar with how to calculate battery recharge time,

including the use of filtering and smoothing to produce more meaningful estimates.
If you consider all of my posts more carefully, you may see that this had nothing to

do with normal power usage fluctuations. It was likely the result of inappropriately

including time suspended in the calculation.

[mariol90]

when you bring the computer out of standby, chances are it'll tell you there's 48 hours of battery life remaining, because that was the power consumption measured while it was in standby more than likely. I've gotten my W500 to say ~7 hours left on 5% battery.

[JNavas]

---

mariol90 wrote:

when you bring the computer out of standby, chances are it'll tell you there's 48 hours of battery life remaining, because that was the power consumption measured while it was in standby more than likely. I've gotten my W500 to say ~7 hours left on 5% battery.

---

1. That's a bug.
2. I'm addressing recharge time, not life remaining.

[HiVolt]

Yeah I've had the 48 hours, then slowly decreasing when resuming from sleep. I've just learned to ignore it.

[ColonelONeill]

Are you charging with a 65W adapter? Could the system occasionally be sucking up too much power to charge the battery?

I've seen PWMUI.exe max out a thread just by being open.

[JNavas]

---

ColonelONeill wrote:

Are you charging with a 65W adapter? Could the system occasionally be sucking up too much power to charge the battery?

I've seen PWMUI.exe max out a thread just by being open.

---

I was (am) indeed using the standard 65W AC Adapter (as shown in the images I posted).

Here's a video that illustrates wildly varying estimates of battery recharge time even though the X220 is under steady high load (all 4 threads on both cores) with Power Manager 3.62 profile Maximum Performance:

http://www.youtube.com/watch?v=RWsW6oWgr8Q

I call that useless. YMMV.

[ColonelONeill]

Try closing everything at letting it charge at idle. The 65W may not be able to support a 35W CPU + assorted components to deliver a full charging wattage to it.

[JNavas]

---

ColonelONeill wrote:
Try closing everything at letting it charge at idle. The 65W may not be able to support a 35W CPU + assorted components to deliver a full charging wattage to it.

---

Indeed, the standard 65W Adapter does not appear to even have enough output to fully power the X220 (i7 version at least) even without battery charging, so CPU and/or graphics throttling occurs (a significant issue, to some people at least).

Thus a higher wattage Adapter seems to be needed for full system performance (see here), not to mention significant battery charging at the same time, and it's a pity that Lenovo does not have a modestly more powerful (and only modestly larger) alternative, like 72W, leaving only the much bulkier 90W AC Adapter or the much more expensive 90W Slim AC/DC Adapter, and then only as an add-on, not as a checkout alternative. (I've just ordered one of the latter to get full system performance with the least possibly bulk, something I shouldn't have to do.)

That said, the point here is that Power Manager does a crappy job of estimating battery recharge time, which is something that could be easily fixed (on the standard 65W Adapter) with a better algorithm. My Power Manager video shows battery recharge time varying wildly over a range of 01:14 to 48:00, which isn't helpful, and which could easily be addressed with (say) exponential smoothing.