Battery bypass @80% tests and issues

ikKPgtQqtSqaUVX

So I'm particularly interested by the battery bypass of pixel phone. I use a pixel 10 pro with the latest experimental release@2025/01/05.

This feature has been documented to turn on when:

80% battery charge limiting is turned on;
==80% battery capacity is filed (not more/less !).

Some sources on this:
https://discuss.grapheneos.org/d/18159-new-battery-bypass-feature-in-gos
https://www.androidauthority.com/phones-with-bypass-charging-support-3509066/

In addition, you should need the following hardware to get battery bypass:

A PPS (a USB 3.0 PD extension) charger (>=45W);
The proper cable (that can deliver >=45W);

Some documentation:
https://uk.ugreen.com/blogs/charger/fast-charging-protocols
https://en.wikipedia.org/wiki/USB_hardware#USB_Power_Delivery

Also, I used a device to measure the power going through the cable. This allows me to check that PPS is a thing, and that once the 80% of the battery are reached, the phone continues drawing power through the cable, bypassing the battery.

Some tools to reproduce:
https://www.chargerlab.com/v-1-9-7-the-new-software-version-of-chargerlab-power-z-km003c/
https://www.amazon.it/dp/B0D95Y8L33

(Note, a first order way to know if you have a PPS capable charger is if amps and tension can modulate outside the fixed rates; 9V etc.).

Now, I plug my phone at around 20% capacity in the evening, let it go up to 80%. It draws ~ 25-27W up until 65% capacity. (Then the powers is reduced, going near zero if not restricting to 80%, this is standard procedure).
I used the BatteryBot Pro apps (com.darshancomputing.BatteryIndicatorPro https://github.com/darshan-/Battery-Indicator-Support/blob/master/SourceCode.md) or com.glgjing.hulk to monitor the phone's status. When the battery reaches 80%, the cable's tension stays at 9-9.1V@_0A in idle state.

I use the camera/video as a load (takes as much power as a game this thing...3.6-5.4W device total when running), to check if bypass works. And it does work but there is a few weird things.

Looking at the Battery Bot logs, or other battery status apps, I observe constant "No charging (AC)"/"Plugged In (AC)" status switches, this seems to happen at lock/unlock transitions, due to GOS' USB restriction policies (charge only when lock, etc.). I wonder if it can use bypass when locked then. I didn't have a battery draining app to run in the background (this status switch behavior does NOT happens when charging [0-80) even if I lock/unlock the phone);
Before engaging bypass, the battery got to 80% then dropped to 79% before charging again and getting in bypass state. The first 80-79% transition was evidently not under bypass.

The status switch concerns me a bit, but apart from that it seems to work.

I then unlocked charging (disabled the 80% limiter), it got to 88% quickly, I then enabled the feature back again. The phone then registered its charging status as "Unplugged", (it was plugged though). Unplugging/plugging the cable didn't change the status, nor did rebooting. I suppose this is to keep the battery @80%, which AFAIK is also good for long term storage.

PS:
I'm looking for a background, battery draining app to continue testing.

jarell

ikKPgtQqtSqaUVX Looking at the Battery Bot logs, or other battery status apps, I observe constant "No charging (AC)"/"Plugged In (AC)" status switches, this seems to happen at lock/unlock transitions, due to GOS' USB restriction policies (charge only when lock, etc.). I wonder if it can use bypass when locked then. I didn't have a battery draining app to run in the background (this status switch behavior does NOT happens when charging [0-80) even if I lock/unlock the phone);

I'm curious about this too.

ikKPgtQqtSqaUVX

When charging the 99% percent, the wattage drops to 3W, then during the charging of the 100%, wattage amps drop to 0.1 and less while at 9.1V. When fully charged, the tension at the cable is 5.1V instead of 9.1V and the "Fully charged" status can be seen in Battery bot's logs.

Weirdly, when fully charged, I observe the same battery bypass behavior. That is, the power going through the cable is "proportional" to the activity, it responds quickly without lag. According to com.glgjing.hulk, the power is coming from the charger. Unlike when at 80%, the tension is 5.1V, not 9.1V, amps are thus higher to compensate.

SovereignCopper

jarell Same here. See https://discuss.grapheneos.org/d/20620-charging-optimization-80-charge-limit-recalibration-with-android-15-qpr2/62

jarell

Could you test the power draw with the screen off and a background process like music playing through the phone's speakers?

ikKPgtQqtSqaUVX

The phone was at 80% with bypass engaged. I used spotify. No GPS, BT etc. Only WIFI (high thgoughput low latency), no cellular data.

Loudness was turned up to the maximum. Spotify was using the very high setting (~ 320 KiB/s).

I played Dire Straits from Dire Straits...

The first track starts low and is intermittent (intro). Then It starts playing. So two phases. During the intro, it draws little power (<0.5W) with some spikes, maybe due to spotify network activity. After the intro, during phase 2, it was drawing 1.8-2.8W@9.1V.

For the 2nd and 3rd tracks, the power was oscillating between <0.5W and 2.73W.

Overall, the average power draw is low. I see spikes that seem correlated to loud phases and network activity.

jarell

ikKPgtQqtSqaUVX Looks like battery bypass is working even with "charging only when locked".

jarell

ikKPgtQqtSqaUVX Was the screen off?

ikKPgtQqtSqaUVX

I watched a 4k video (https://www.youtube.com/watch?v=DVTRklHhEsU) out of firefox & youtube with sound and brightness turned up to the max.

2.7-7.2W@9.1V and averaging at ~ 4.5W.

Consumption seems to be tied to:

video movement/decoding;
network activity.

I paused on a frame of the video, keeping the brightness to the max, the power was low, averaging ~ <0.7W. So brightness seems to be less of a factor ??

Well playing the video using the lowest brightness setting, all other things being equals, it drew between 2.7-7.2W@9.1V averaging at 4.02W.

As soon as I start scrolling though a web page, my apps, use the camera or the network heavily, then it starts drawing a lot of power.

de0u

ikKPgtQqtSqaUVX Depending on the video format, there may or may not be hardware decoder support, but either way decoding involves a fair amount of matrix multiplication etc., and a fair amount of updating video memory. Scrolling quickly runs the GPU and involves a fair amount of updating video memory. Cellular data reception involves a bunch of matrix computation for DSP. Almost all network traffic these days is encrypted, so receiving-then-decryption hammers on the DRAM. A static screen will likely stop some of the GPU clocks and may well suspend some of the CPUs too. All in all, those numbers seem plausible. Thanks for providing the measurements!

Quantum

Just adding some cable specs.

Any cable that is USB-C to USB-C will be at least 60W. Any cable rated at 100W or more will have an E-Marker chip which is needed for PPS charging.

Personally I havent seen a E-Marker chip in a cable under 100W. I could always be wrong though about all of this since USB can be convoluted.

ikKPgtQqtSqaUVX

Yea so, we have this post, https://discuss.grapheneos.org/d/20620-charging-optimization-80-charge-limit-recalibration-with-android-15-qpr2/60 and https://discuss.grapheneos.org/d/18713-charging-the-battery-to-80-does-not-work/4 reporting the same issue on different devices, on different OS version (aosp 15/16).

@GrapheneOS I'm not sure who sould I ping, contact or where to search for a technical anwser, any idea ?

ikKPgtQqtSqaUVX

I did a few tests related to power draw and tethering.

The phone was locked when tested. When wifi was used, cellular was off and same but reverse when testing cellular.

All tests where indoor a concrete building.

Wifi, great connectivity, 12 MiB/s, 1.8-3.6W@9.1V
5G, 2 bars connectivity, 7.8 MiB/s, between 2.7-6.4W@9.1V
4G only, 2 bars connectivity, 5.4 MiB/s, between 1.8-4.5W@9.1V.
3G, 2 bars connectivity, low throughput (240 KiB/s), between 1.8-2.7W@9.1V.

Like the tests above, take it wit a grain of salt,

jarell

ikKPgtQqtSqaUVX So it does seem like bypass charging works with the screen off.

jarell

Thanks for all the tests @ikKPgtQqtSqaUVX

jarell

I purchased a power meter and can confirm at 100% the measurements suggest bypass charging is working. I'm tempted to find a phone with no bypass charging support to see what measurements it provides at 100% as a comparison.

ikKPgtQqtSqaUVX

@jarell Great, would you mind sharing if you reproduce other experiments ? I'm also currius about other phone's behavior, I may be able to check an iphone 16 pro max.

jarell

Did some testing with an old Samsung S8+ running Android 9. It shows the same behavior at 100% as the Pixel at 80% with charging optimisation; wattage drawn from the wall varies depending on what actions I'm doing (video recording, opening apps, video playback, etc.).

This makes me think power draw is not a good indicator of whether battery bypass is working.