Google Pixel 5a (2022)
Initial Checkout

• Battery technology: lithium polymer (LiPO). This means the electrolyte is mixed with polymer material that forms a gel, holding it in place. I have not been able to discover the exact battery chemistry: lithium cobalt oxide, lithium iron phosphate, lithium manganese oxide, etc. Nominal capacity when new: 4.68 amp-hours. Battery voltage (one cell): data not found. Service life: depends on the battery chemistry; quantitative data not found.
• Using the 3.0 amp USB charger that comes with the phone, which should have negotiated fast charging (20V 1.5A on the wire), charging took 87min from 32% to 100%, extrapolated to 128min (call it 2 hours) for 0% to 100%. Below 80% charge it added about 1% per minute. As usual the rate was less above 80%, like 0.5% per minute at 95%, and it took 10min to get from 97% to 100%. On Pioneer, Galaxy S5 and older phones with lithium cobalt oxide chemistry, the dropoff above 80% was a lot more, Another possible difference is that 100% could be defined lower on the Barbet, so the user would see it reached quicker. Also, it's well known that stopping charge below the usual 100% point increases the life of the battery. It's plausible that this would be more significant for lithium cobalt oxide (vs. lithium iron phosphate) because the former swells when charged, disrupting the crystal structure of the anode material.
• Battery life under excessive load, with all 8 cores maxed out doing SHA512 sums (at least 80% CPU utilization on each), plus playing a video (local file) in loop mode, which has priority to get on the CPU, so there were no freezes or dropped frames. The phone was noticeably hot. It ran the battery down from 100% to 33% in 284min = 4.7hr. Extrapolating to 7.1hr at 0%.
• Subjective impressions are that the Barbet is frugal with energy, even more so than the Pioneer which was already good. In my normal use it rarely goes below 80% in a day, but many people do a lot more energy-intensive activities.
• Overnight standby, with Wi-Fi on and with a SIM, on the stock OS, for 12hr 40min used 5% of battery. Estimated standby lifetime: 10 days.

• To boot from power off, hold down the power button (the textured sausage on the right side above the volume control), about 3 sec, until the Google splash screen appears, or the Fastboot warning that the bootloader is unlocked.
• Times to boot up, until the lock screen appears:
  • 21 secs (stock)
  • 42 secs (LOS-19) including 10sec in the Fastboot warning about the unlocked bootloader.
  • It takes longer after an update because it has to recompile all the Java bytecode. Stock: about 120sec; LOS-19: 60sec.
• Wake up: 1 sec (or less)
• Go to sleep: 1 sec (or less)
• Power off: 3 secs (stock and LOS-19)

• Dimensions: 1080x2340px, 146 x 66 mm, 157 mm diagonal with rounded corners, or 160mm as rectangle. Aspect ratio: 2.2 : 1
• Technology: AMOLED (vs. backlit light valve)
• 3 readability tests: black text on white background; colored test pattern (for a printer); Google Maps.
• Color rendition in indoor daylight, variable around 200 lux: Excellent in all directions even to 80 degrees off perpendicular. Similarly excellent in a dark closet, 0.9 lux.
• Readability in outdoor shade, 3500 lux: Completely normal, not distinguishable from the indoor experience, except for the blue sky reflecting off the screen. If you turn so you see your face (dark) reflected in the screen, you will see less sky brightness.
• Readability in direct sunlight. Lighting 108464 lux, at local noon, mid-May, air quality 46 (good, one of Los Angeles' best days). Appearance was completely normal, same as in outdoor shade.
• How many dead pixels? None that I can see.

• RAM: Mostly from /proc/meminfo (Gb means 1024 bytes times 10⁶; Gib is 2³⁰ bytes).
  • Total on phone: 6Gib
  • Meminfo total: 5.46Gb
  • Non-allocatable: 0.61Gb (kernel, maybe video RAM)
  • Swap total: 2.1Gb
  • Virtual address space: 263Gb
  • CMA: 295 Mb (expandable video RAM)
• Internal flash:
  • Total 128Gib.
  • In Android-11 it's hard to use df to assess how much of this is used. However, on stock OS, Settings-Storage reports 21.9Gb used, in these categories: 7.7Gb apps; 14Gb system; 0.25Gb photos; 0.15Gb other.
  • LOS-19 flash usage: 18Gb total, 4Gb apps, 14Gb system, others add up to 0.116Gb. Of course more was used a month later.
• This machine has no SD card slot.
• Filesystem formats per /proc/filesystems: permission denied.

• Features: hardware floating point, SIMD instructions, engines for AES, SHA-1, SHA-2, CRC32.
• Detection of multiple cores: /proc/cpuinfo shows eight processors (correct) but not their nominal clock speeds. From the product page: 1 at 2.4GHz, 1 at 2.2GHz, 6 at 1.8GHz.

• Dial a call -- Works.
• Incoming call -- Works.
• Partner hears decent voice quality.
• Local user hears decent voice quality
• Speakerphone audibility -- Plenty loud (with volume turned up), and the partner heard reasonable voice quality.
• Available modulations: both offer 5G, LTE, 3G, 2G. None are identified as CDMA variants, though the USA model of the Barbet is supposed to be able to do CDMA. They recommend to leave it on 5G, and it will fall back to LTE until 5G is actually deployed. Find this in Settings-Network & Internet-Mobile Network-Preferred Network Type (on stock) or Settings-Network&Internet-SIMs-(name of carrier)-Preferred Network Type (on LOS-19). (What a strange place to put this setting! But Search Settings found it.)
• VoLTE (voice over LTE): The Pixel 5a (also 5 and 6) supports VoLTE intrinsically; there's no setting to turn it on or off. However, it doesn't work with some carriers and the phone will use voice over 3G. This is particularly a problem in non-USA countries where the phone is not sold, but is seen also sometimes in the USA. There is controversy whether the carrier can support it at all, or is using the wrong symbol in OTA configuration to announce it, or whether Google is failing to recognize a valid VoLTE announcement.
• How to prove that VoLTE is being used: Shut off Wi-Fi. Play a video via cellular data. Flick down notifications and verify that LTE (or 5G) is being used. Get a partner to make a voice call to you; answer the call. A well-behaved video player like VLC will pause. While on call, flick down notifications and verify that you're still on LTE, which proves that the call is going by VoLTE. And that voice calling will still work when T-Mobile's 2G and 3G service goes away on 2022-07-01. Tested on stock OS.
• Wi-Fi Calling: Find this setting in the same place you found Preferred Network Type. Turn it on, and in Calling Preference pick Call Over Wi-Fi. Make a voice call. Use traceroute on Holly (Wi-Fi access point) to verify that Wi-Fi is being used for the call. Alternatively, turn on Airplane Mode but turn Wi-Fi back on. Wi-Fi calling worked on both OS's. It used IPv4 IPSec to the carrier's offsite gateway. For my use case, few voice calls, I want reassurance that voice will work away from home, so I reverted Calling Preference to Mobile Network. It will fall back to Wi-Fi Calling if cellular voice is unavailable and Wi-Fi is OK.
• What size of UICC (SIM) does it expect? 12.3x8.8mm. referred to as a nano-SIM.

• Testing on the stock OS with Speedtest app by Ookla (speedtest.net). Units: Mbit/sec. Carrier was T-Mobile in all cases. It was run under these conditions, repetitions are 1 min apart:
  • 5G: Download 25.1 and 19.3 Mbit/sec. Upload 16.3 and 4.67 Mbit/sec.
  • 4G (LTE): Download 44.7 and 51.0 Mbit/sec. Upload 21.4 and 19.5 Mbit/sec.
  • 2G (EDGE): Download 0.08 Mbit/sec. Upload 0.04 Mbit/sec. Speed of arthritic snail.
• The theoretical maximum for LTE is 300 Mbit/sec, never achieved in the wild. These speeds vary depending on unknown factors, of which competing traffic is the most likely major culprit.
• On 4G LTE, IPv6 is used with a privacy enhanced random address.

• Text messages were sent out successfully.
• Incoming messages were received successfully.
• Both were received at the other end more promptly, 4 sec, than on the Pioneer using 3G, which sometimes had long delays in delivery, like several minutes. Partner was an iPhone 12 Pro.
• SMS can work via Wi-Fi Calling (see above).

• Does it connect to our access point? Yes.
• What modulation revision? 802.11 a-b-g-n-ac. N is tested and works. It can talk to the AP on AC (5GHz) but this is a repeater and functionality is limited, preventing thorough testing.
• Speed on 802.11n: 725Mb downloaded to the phone in 122sec, 47.5Mbit/s = 5.94Mbyte/s overall, 60Mbit/s peak. The phone was about 1 meter from the access point, no competitors.
• On IPv6 it has a link-local address (privacy enhanced) but did not do DHCP6 to get a routeable IPv6 address. My net sends router advertisements but they offer DHCP6, not RFC 4862 auto configuration. As a matter of policy, Android does not do DHCP6.
• How to find the MAC address (#1): Settings-About Phone-Wi-Fi MAC Address (near the bottom). The Bluetooth MAC and the IP addresses are also here. This works on both stock and LOS. In prior stock versions, About Phone was missing.
• How to find the MAC address (#2): Settings-Network & Internet-WiFi-WiFi Preferences (at bottom)-Advanced-2nd item from bottom. This also works on the stock image and on LineageOS.

Works. The client gets an IPv4 address, no IPv6. If you intend for the hotspot to export on cellular data you need to turn off Wi-Fi, because it's capable of associating with another AP in Managed Mode, and simultaneously associating with clients in Master Mode, sending its traffic via the other AP. (So why doesn't the client just use the other AP? Such mysteries are not for mere mortals to know.) Set this up in Settings-Network&Internet-Hotspot&Tethering-Wi-Fi Hotspot. Give the Hotspot Name (SSID), turn on Security: WPA2-Personal, and configure a (good) password. The Quick Settings page (super status bar) can be configured with a tile to turn on/off the hotspot (use the pencil icon to edit it).

Hardwarewise, Bluetooth worked out of the box on both the stock image and LOS-19. Pairing worked the first time, following standard procedures for the devices. Partners tested:

• 66 BT Sport headphones: paired, and played MP3 and Ogg Vorbis files using A2DP.
• Linux laptop: paired; either end can push a file to the other. It wants File Transfer Profile (FTP), not Object Push Profile (OPP). The receiving end needs the OBEX Push Daemon installed and configured; see the Bluetooth File Transfer app. The files end up in the daemon's configured directory; on Android the default is /storage/emulated/0/Download .
• iClever IC-BK08 Keyboard (with touchpad): paired; a cursor appeared and the touchpad could move it and do mouse clicks.

It's hard to prove that the principal audio codecs (Vorbis, MP3) are running on the DSP.

• Internal speaker: Working. Sound quality playing music is far from audiophile quality. There is a speaker on the bottom edge, plus the phone speaker at the top front, for a stereo effect (which I didn't verify explicitly).
• 3.5mm phone jack: Working.
• Volume buttons: Working.
• Bluetooth A2DP: Working. There is a separate volume setting for Bluetooth. I find that A2DP gives poorer sound quality than Vorbis or MP3 compression, and I wish I could test and use those codecs according to the latest Bluetooth revision, but I can't find the controls to do this.

The test media for the above were MP3 and OGG files via Chrome's builtin player on the stock image. Also tested: MP3 via Icecast; VLC on both OS's. For video, Ogg Theora, MPEG3 and MPEG4 were tested with VLC.

• The test was done indoors, not ideal for satellite signal strength. It was using 23 of 31 satellites. Phone was stationary.
• Does it work? Yes.
• Claimed position accuracy: started at 5.6 meters, was 3.8 meters after about 20sec of averaging.
• Stock: the first time it was used, it took only about 7sec to find the satellites and show a position. LOS first fix was in only 5sec (probably better space weather).

The magnetometer is in the upper center of the phone, 1-2cm above and to the left of the fingerprint sensor. Iron objects like a screwdriver or masonry reinforcement of course will affect the readings.

Uniaxial accelerations have one axis at its maximum and the others zero. To produce a positive value the indicated side of the phone must be pointing down: X axis is to the left, i.e. away from the power button; Y axis is down, i.e. toward the USB connector; Z axis is to the rear. All axes report 9.8 m/s² when so oriented. Resolution is 0.1 m/s². When the phone is flipped over, the uniaxial value becomes -9.8 m/s² on all three axes. Alignment of the axes with the phone's chassis is not quite perfect.

The sensor appears to be localized in the upper about 1cm of the display and ranging about 20% to 40% from the right edge. Forum posts and articles suggest that the sensor is optical and looks out through the top speaker grille.

It works, and influences the display brightness if turned on. The light sensor appears to look out through the grille of the top speaker, at the right end. It has been seen reporting 108464 lux in direct noon sunlight, down to 0.3 lux in a dark closet.

• Normal angle: 12Mpx, f/1.7, 83° field of view, 3000x4000px estimated, 1.26 px/arcmin.
• Wide angle: 16Mpx, f/2.2, 119° field of view, 4619x3464px estimated, 0.65px/arcmin.