• Become a member
  • Log In
The Institution of Electronics
  • Home
  • About us
    • Our Objectives
    • Our History
    • Governance of the Institution
  • The Electron Magazine
    • 2024
      • 2024 – Winter
      • 2024 – Spring
      • 2024 – Summer
      • 2024 – Autumn
    • 2025
      • 2025 – Winter
      • 2025 – Spring
      • 2025 – Summer
      • 2025 – Autumn
    • 2026
      • 2026 – Winter
      • 2026 – Summer
  • Members
    • Membership Grades and Fees
    • Members’ Resources
      • The Electron Newsletter
      • The Archives
  • Education and Projects
    • National Electronics Competition
    • Student Members’ Projects
    • Arkwright Engineering Scholarships
  • News
  • Contact Us
  • Menu Menu
Uncategorised

Inspecting a DOCSIS 3.1 cable modem’s internals: Part 1

The version publication ordering might not be as originally envisioned, but this two-part teardown of NETGEAR’s CM1000 still aspires to be technically (and otherwise) interesting.

Having recently wrapped up a four-part series on how I’d determined (and resolved) the underlying reasons for my flaky Comcast broadband service, with the concluding entry focusing specifically on the cable modem portion of the transmission-and-reception chain, I thought it’d next be interesting to take apart a DOCSIS 3.1 modem (roughly four years subsequent to dissecting its DOCSIS 3.0 precursor) and see what was inside.

That said, I wasn’t particularly motivated to tear down either my in-use primary or spare NETGEAR Nighthawk CM1100, for perhaps obvious service-downtime reasons.

Instead, I decided to tackle the CM1000 sibling, which looks identical from the front save for a metal tint variance (silver vs rose gold) and the absence of a second LAN connection status LED.

That latter variation is reflected in the inclusion of only a single RJ45 connector in the back.

An identity crisis

In researching the CM1000 in advance of surgery, I learned that two device versions existed, the original and a subsequent v2. I wasn’t able to uncover any specifics on the deviation(s) between them, however, save for vague references to “slight hardware changes.” And I couldn’t find an already published teardown of either version to see those “slight hardware changes” for myself, either.

So, I decided to tackle my own teardowns of both versions. I first purchased a (claimed) v1, judging from the published eBay photos (to reiterate: the blurry one that follows was not taken by yours truly). Absent a power supply and claimed “malfunctioning”, randomly rebooting every few minutes, it was therefore perfect for dissection purposes.

And then I bought a v2, complete with a PSU and in an eBay-advertised “untested” state.

The thing is, though, what was supposed to be a v1 ended up being a v2, too, and I didn’t realize the discrepancy until I sat down to start the teardown, after the eBay return period had expired, of course.

The eBay seller had multiple units available for purchase and (I realized in retrospect) didn’t explicitly identify them as v1 models in the description; I was relying solely on the post’s pictures. So, I’m guessing the seller had a version mix in inventory, and I was unlucky with the one randomly (presumably) shipped to me.

I’ve kept the “untested” v2 CM1000 intact as a potential additional spare since, as noted previously, Comcast doesn’t allow enable of the CM1100’s second LAN port anyway, and in the hope that “untested” doesn’t translate into “nonfunctional” or, for that matter, “blacklisted”. That said, here are a couple of closeups of the “untested” unit’s wall wart (which I also won’t be taking apart) so you can see what the power supply for a v2 CM1000 looks like.

I’ve ordered another v1 (I hope) CM1000 off eBay, which is en route as I type these words. I’m planning on taking it apart for publication next month. And for today, in reverse order versus my original plans, you’re getting the teardown of a “malfunctioning” (or not) PSU-less v2 CM1000, starting with some overview shots of the actual device versus the earlier “stock” shots.

Front.

Left side.

Back.

Right side.

Top.

And bottom, a closeup version of which you’ve already seen.

Heatsink profusion

Time to get inside. Have you already noticed the notably visible screw head in those earlier bottom views? Well, you won’t notice it anymore.

Here’s what the CM1000 looks like with the base now removed.

Next up, two more screw heads, this time around back (one at the top of the back panel, the other at the bottom).

And after releasing a few clips, the two halves of the device cleanly separate, revealing the inside of the left side of the enclosure (complete with a plethora of light pipes that route the PCB’s LEDs’ outputs to the outside world) and our first glimpse of the PCB itself.

Two more screws to go.

Ladies and gentlemen, the PCB is now free.

Before flipping the PCB over, let’s get the left side heat sinks off. The smaller silver one toward the top is the first to pop off after using a flat head screwdriver as a lever on it.

Clean off the thermal paste and…it’s blank. Not quite, actually; the markings that remain are quite faint, as you can see, but I hope you’ll take my word that they reference a Macronix MX30LF1G18AC (PDF) 1 Gbit SLC NAND flash memory, presumably holding system firmware.

Now for the larger black heat sink toward the bottom of that same side.

That’s a whole lotta not much, although I’d wager a bet that there’s heat-generating stuff on the PCB’s other side, which this heat sink (imperfectly, but better than nothing) deals with.

Let’s see.

Up top is an even larger black heat sink, held in place by not only a thermal pad but also mounting pins, and with Broadcom’s BCM3390 DOCSIS 3.1 SoC underneath it.

And location-associated with the aforementioned black heat sink on the other side is a sizeable Faraday Cage, containing an assortment of both ICs and discretes.

At lower left is the largest, square chip, Analog Devices’ MAX3522B DOCSIS 3.1 programmable-gain amplifier; directly underneath it is the copper-color heat-transfer “landing pad” on the PCB’s other side (see, told you so!). To its far right is a pSemi (formerly Peregrine Semiconductor) PE42722 SPDT RF switch. And above and to the right of the MAX3522B, in the other section of the Faraday Cage, is a chip marked as follows (readers?).

3432
KMLG 10
EE1852
0403 W (I “think” that last character’s correct, but I’m not positive).

All of this plus, as already alluded to, a whole mess o’surface mount capacitors, inductors, resistors, transformers and transistors, and such. An RF engineer’s dream (or maybe nightmare).

Zooming out on this same right side of the PCB, what else do we see?

To the left of the Broadcom SoC is an Elite Semiconductor Memory Technology (ESMT) M15T1G1664A 1 Gbit DDR3(L) SDRAM, interestingly with another (unpopulated) site above it. To the right of the BCM3390 is, unsurprisingly given its proximity to the RJ45 connector, a Bothhand GST5009 LF surface-mount 1000BASE-T magnetic module and pulse transformer.

That’s it for today; next month, we’ll see what if anything we can find regarding “slight hardware changes” in the v1 CM1000 design. Until then, please share your thoughts in the comments on what I’ve uncovered so far!

—Brian Dipert is the associate editor, as well as a contributing editor, at EDN.

Related Content

  • Debugging intermittent Comcast, part 1: Scenario-setting
  • Debugging intermittent Comcast, part 2: Remediation details
  • Debugging intermittent Comcast, part 3: Retrospective analysis
  • Debugging intermittent Comcast, part 4: Broadband diagnostics
  • Cisco Linksys’ DPC3008: A DOCSIS 3 cable modem that worked great

The post Inspecting a DOCSIS 3.1 cable modem’s internals: Part 1 appeared first on EDN.

16 November 2026
http://institutionofelectronics.ac.uk/wp-content/uploads/2022/12/IOE_LOGO.png 0 0 whdsolutions http://institutionofelectronics.ac.uk/wp-content/uploads/2022/12/IOE_LOGO.png whdsolutions2026-11-16 13:40:402026-11-16 13:40:40Inspecting a DOCSIS 3.1 cable modem’s internals: Part 1

Latest news

  • Building AI SoCs with a continuous hardware verification strategy17 November 2026 - 06:53
  • Inspecting a DOCSIS 3.1 cable modem’s internals: Part 116 November 2026 - 13:40
  • The shocking science (and art) of plastic welding16 November 2026 - 06:33
  • Physical AI: Where AI meets the physical world13 November 2026 - 14:36
  • The tchotchke13 November 2026 - 13:33
  • Google’s Fitbit Air: Another display-less option for tech you wear12 November 2026 - 13:11
  • MCU revitalizing SDVs with diverse compute, virtual prototyping12 November 2026 - 12:10
  • Automotive LED driver advances animated lighting11 November 2026 - 21:59
  • Isolated gate drivers serve power-switching needs11 November 2026 - 21:59
  • NXP simplifies robotics system development11 November 2026 - 21:59
IOE LOGO 2

Become a member

click here

Become a member

click here

Become a subscriber

click here

Become a sponsor

click here

© Copyright - The Institution of Electronics | Website by WHD Solutions
  • Link to LinkedIn
  • Link to Facebook
  • Link to X
Link to: The shocking science (and art) of plastic welding Link to: The shocking science (and art) of plastic welding The shocking science (and art) of plastic welding Link to: Building AI SoCs with a continuous hardware verification strategy Link to: Building AI SoCs with a continuous hardware verification strategy Building AI SoCs with a continuous hardware verification strategy
Scroll to top Scroll to top Scroll to top