This smart speaker differs from its more diminutive sibling predominantly in its transducer allotment.
Back in May, I started noticing $70-off (the normal $99.99 list price, and also versus the original $129.99 MSRP) promotions for Google’s first-generation Home smart speaker at various retailers. I suspected (and the company later confirmed) that Google was clearing out existing inventory, either in preparation for a Nest rebrand or a more comprehensive next-generation replacement. I was able to snag one for $29 at Office Depot before inventory was depleted.
As I’ve mentioned before, mine is an Amazon Echo-centric household, so we don’t have an ongoing use for the Google Home, but I figured it’d make for an interesting teardown project, and so it was.
I’ll begin as usual with some box shots, accompanied by a 0.75″ (19.1 mm) diameter U.S. penny for size comparison:
Slip off the outer cardboard sleeve and open the box and those of you who remember my earlier Google/TPLink OnHub router teardown might begin experiencing some déjà vu:
Let’s put aside the Google Home for a second and see what’s underneath (note the “peg” in the plastic which fits into the unit’s power input port, helping hold the unit in place within the box):
First there’s some scant literature:
And underneath that is the power supply:
It’s worldwide AC voltage/frequency-compliant, and outputs 16.5 VDC at 2A:
No industry-standard micro-USB or equivalent connector here (which, come to think of it, was a wise design move on Google’s part, since the voltage carried over the connector isn’t USB-standard, either), only a “barrel” plug:
Now for the device itself, which measures 5.62” (143 mm) tall at its highest point (the top is sloped, if you hadn’t already noticed) and 3.79” (96 mm) in diameter, and weighs 1.05 lb (477 g). The front is pretty spartan:
Around back you’ll note the prominent-sized (and -labeled) microphone mute switch and, below it, the power LED:
The angled top surface contains the two microphones, in a far-field array configuration. It also includes numerous colored status LEDs (not visible in this photo, since the unit is off), and features capacitive touch controls to control music playback and more generally adjust volume:
And the bottom-side label information includes, among other things, the always-enlightening “A4RHOME” (believe it or not) FCC IC. Note, too, the DC power input connector:
The unit’s base is easily removable and can be swapped out for replacements with other colors and patterns to match the room décor:
With it removed, the 2” active speaker is apparent, alongside two 2” passive radiators:
And lo and behold, in the back is a largely-undocumented micro-USB port, akin to what we’ve already seen with other Google-branded hardware I’ve dissected:
Here’s what the foot looks like with the base removed:
Separating the top and bottom halves requires the removal of four screws with T6 Torx heads. Their difficult-to-access locations meant that I wasn’t able to use the screwdriver handle (with bit socket at the end) of my iFixit 64-bit driver kit, as I normally do; instead, I had to also employ the flexible extension between it and the bit (along with a pair of curved tweezers to extract the screws once sufficiently loosened):
With the deed done, the two halves disconnected straightaway, revealing a flex PCB cable between them. Recall that the top half contains microphones, LEDs, a capacitive touch surface, etc; presumably this is the power-and-communication channel between them and the main system board you can begin to see:
And speaking of which, here’s what’s on the other end of that flex PCB cable:
Notable ICs and other devices include Atmel’s (now Microchip Technology’s) ATSAMD21 32-bit ARM Cortex-M0+ microcontroller, a pair of NXP PCA9956BTW LED drivers, two InvenSense INMP621 MEMS microphones (the silver rectangular devices across from each other at the edges), and (in the center) a circular arrangement of a dozen multicolor LEDs. Also note the cluster of seven test points below one of the MEMS mics:
What about the other side of this particular PCB? According to iFixit, getting to it is near-impossible due to the hefty, robust dollop of adhesive between it and the enclosure. So instead of redundantly tackling this same challenge myself, I’m going to “throw in the towel” and use iFixit’s images instead.
Note, for example, the PCB holes for incoming sound to pass through to the MEMS microphones below, and for light from the LEDs below to pass through and out the top. Also see the capacitive touch structures surrounding the center:
Next step: remove the rubberized “O-ring” surrounding the inner chassis, which I suspect both dampens vibration and prevents sound coming from the speaker and passive radiators from leaking out the top of the device (interfering with the microphones in the process):
And here are some overview shots of that inner chassis post-O-ring removal.
Before proceeding further, let’s take a closer look at the “spring” mechanism that transfers external mute button presses into PCB-mounted switch activations:
With the “spring” assembly subsequently removed, the main board comes into full view:
In order to remove it from the inner chassis, we need to first disconnect the two cables at the bottom:
The purpose of the black-and-red two-wire twisted pair is likely already obvious to you; it heads inside the chassis and (I’m quite confident, though I haven’t yet confirmed at this point in the teardown) connects to the speaker on the other end. The flex PCB cable headed toward the bottom, on the other hand, stay tuned. Also note the aforementioned micro-USB connector.
Remove a few screws and the PCB lifts right off, with nothing but black plastic below it:
The PCB backside is now visible for perusal. Whatever’s underneath the Faraday cages is at this point still unknown, of course, but smack dab in the middle is a Texas Instruments TAS5720 class D audio amplifier, which drives the speaker (and is the same IC as the one found in my earlier Google Home Mini teardown):
Here’s another perspective, focusing (literally) on the DC power connector at the end:
And here’s another look at the main board topside, now standalone:
Let’s get those Faraday cage tops off, shall we? Underneath them we find soft, sticky pink goo:
And attempting to scrape it away didn’t dramatically improve IC visibility, so you’ll have to take my (and iFixit’s) word that underneath the frontside Faraday cage lid is a Marvell 88DE3006 Armada 1500 Mini Plus dual-core ARM Cortex-A7 media processor, the same as the one found in my earlier Google Home Mini teardown and the successor to the Armada 1500 Miniseen in my even earlier first-generation Chromecast dissection. Alongside it is a Toshiba TC58NVG1S3HBA16 2 Gbit SLC NAND flash memory (yep, you guessed it, also in the Google Home Mini):
Flipping the main board over, underneath the house-shaped Faraday cage lid there’s a Marvell (now NXP) Avastar 88W8887 dual-band 1×1 802.11ac plus Bluetooth 4.2 SoC. Alongside it are the 2.4 and 5 GHz antennas embedded in the PCB. I’m not sure if the 2.4 GHz Wi-Fi antenna does double-duty for Bluetooth or if there’s a separate standalone Bluetooth antenna somewhere in the design that I’m not seeing. Readers?
Underneath the rectangular lid there’s a Samsung K4B4G16 4 Gbit DDR3 SDRAM. And as previously mentioned, next to both of them is the Texas Instruments TAS5720 class D audio amplifier. Obviously, Google’s engineers did a lot of IC-sharing between the Home and Home Mini smart speakers designs, although in the interest of preciseness, the DRAM in my Home Mini teardown came from SK Hynix (same technology generation, speed bin, etc. however).
Remember that earlier-mentioned mysterious second flex PCB cable that I’d found connected to the main board, and headed toward the bottom of the system? It’s time to track down its destination; let’s remove the foot:
See those two tiny holes in the back? They, in conjunction with a paper clip, enabled me to get the foot label off. Its adhesive is VERY strong; it took me a fair bit of time, along with a bit of colorful language, to get it back offmy desk after taking this photo:
And what’s underneath it? More test points! I suspect they find use not only for assembly line testing purposes but also to program the firmware into the flash memory:
Last but not least, let’s crack open that inner chassis and have a look at the speaker and passive radiators from their backsides. The remainder of the disassembly involved the removal of six T8 Torx screws. They were so deeply seated within the chassis that I was unable to use the iFixit driver kit on them; the T8 Torx driver was too short, and the screwdriver handle (specifically the socket at the end) was too wide to fit in the holes. So I splurged on a $10 eight-piece Torx screwdriver set; its shipment to me from Amazon delayed the completion of this writeup.
Mission accomplished:
See, I told you that black-and-red two-wire twisted pair was headed for a speaker!
Befitting the passive radiator design, the remainder of the speaker cabinet is completely sealed (when fully assembled, of course):
As I previously mentioned, Google and its retail partners began liquidating first-generation Home inventory in May and officially announced the product’s discontinuation one month later. Its successor, the Nest Audio, was officially confirmed in July, but it didn’t officially launch until the end of September. It was apparently worth the wait, but I admittedly don’t “get” the lengthy generational gap, especially in this particularly “hot” product category rife with competitors from Amazon, Apple, and elsewhere. Then again, Google did pretty much the same thing with the multi-month delay from the cancellation of the Pixel 4 to the launch of the Pixel 5, so maybe this is the hot new trend in Silicon Valley marketing? In all seriousness, I suspect that COVID-affected production of both the old and new smart speakers (and phones) and their constituent components were to blame; if you have any insights here or with respect to anything I’ve discussed in this teardown, I look forward to reading your thoughts in the comments!
This article was originally published on EDN.
—Brian Dipert is Editor-in-Chief of the Edge AI and Vision Alliance, and a Senior Analyst at BDTI and Editor-in-Chief of InsideDSP, the company’s online newsletter.
Related articles: