Inspecting an Insteon (and X10) controller

Article By : Brian Dipert

The Insteon powerline modem this engineer tore down back in July is only part of the implementation puzzle; an intelligent home automation controller is also necessary, and this one’s cloud independence came in handy recently.

Back in July, I disassembled and detailed the internals of a SmartLabs 2412S Power Line Modem (PLM). While PLMs provide a bridge between a home automation controller (hub or other) and various X10-and-Insteon powerline (as well as Insteon wireless, whether directly or via a powerline-to-wireless intermediary) devices, they contain only minimal native processing and memory facilities. The bulk of these resources alternatively reside in the controller and/or (with a defeatured hub) the cloud server to which it’s Internet-connected.

In that prior writeup, I mentioned that SmartLabs had (only temporarily, as it turns out) abruptly turned out the lights, locked the doors, and powered down the servers in response to a fiscal crisis, in the process crippling the home automation networks of the customers of its cloud-centric hubs. In response, alternative standalone controllers such as those from Universal Devices garnered even more interest than they’d had previously. My Insteon network had been mothballed for more than a decade, since I’d moved from California to Colorado, but I still had (among other things) a never-used and rare Universal Devices ISY-992i/IR PRO controller (the successor ISY-994i series was more common) which I sold on eBay.

Today’s teardown victim is the ISY-992i/IR PRO’s predecessor, an ISY-99i/IR PRO. This unit was the nexus of my personal Insteon network for several years, where it performed error-free and otherwise robustly. The “IR” in the product name references the fact that this particular unit is capable of being controlled not only from a networked computer via its Ethernet port but also via an infrared remote control that supports Philips RC-5 codes. “PRO” references the fact that this device integrates enough memory to support up to 1,024 device/scene combinations and 1,000 programs. I’ll as-usual begin with some overview shots, in a couple of cases with the device accompanied by a 0.75″ (19.1 mm) diameter U.S. penny for size comparison purposes.

While the ISY-99i/IR PRO and ISY-992i/IR PRO look identical from the front, the back sides (therefore the top side markings) are notably different:

The ISY-99i/IR PRO includes one RJ-45-based port (A), intended to connect to (and, depending on the PLM model, potentially also capable of being powered by) a PLM, and one DB-9-based port (B), intended for initial configuration for users without a DHCP-enabled network as well as for advanced PC-based troubleshooting. The successor ISY-992i/IR PRO offers two RJ-45 ports, with the DB-9 connector instead labeled “Console”. In both cases, there’s also a barrel plug power input (for when powering via the PLM isn’t supported) and an Ethernet connection.

Whereas the embedded SD card, which we’ll see shortly, is user-accessible via a left-side cover on the ISY-992i/IR PRO, it’s not on the ISY-99i/IR PRO:

And not only is Universal Devices based in Encino, CA, manufacturing also takes place in the United States:

Time to peek inside, a straightforward undertaking after removing the two Philips screws on either side of the unit:

Completely removing the PCB from the chassis is equally straightforward, subsequent to the extraction of four more screws (only two of which are immediately obvious from a glance at the prior photo; revisit the earlier bottom overview image to discern their locations):

This is, I must say, one of the cleanest PCBs I’ve come across. No Faraday cages, no epoxy “blobs” or other tricks to obscure IC identities, only straightforwardly laid out and labeled chips. At bottom is a Samsung K4S641632K-UC75 64 Mbit SDRAM. Above it are an LCX244 octal non-inverting buffer/line driver (manufacturer unknown) and an SST (now Microchip) SST39VF040 4 Mbit EEPROM-derived flash memory. And to their right is the “brains” of the device, a Freescale (now NXP Semiconductors) MCF5270CVM150 ColdFire v2 32-bit RISC SoC.

Continuing toward the top of the PCB, in the upper right corner there’s a Davicom DM9161 physical layer transceiver, implementing the device’s 10/100 Mbit Fast Ethernet facilities. And in close proximity to the RJ-45 and DB-9 connectors, both supporting serial port connectivity, is (unsurprisingly) a Texas Instruments MA3243 (PDF) 3-V to 5.5-V multichannel RS-232 line driver, curiously alongside a Maxim Integrated MAX3078 RS-422/RS-485 interface IC.

About that SD card; it pops right out (and back in) via a standard spring-loaded-and latched eject-and-insert mechanism:

It’s a fairly conventional 128 Mbyte storage device, as it turns out. Quoting from Universal Devices’ documentation:

The ISY has two different memory storage locations. First is the base memory [EDITOR NOTE: the already shown SST flash memory chip soldered to the PCB) where the ISY has some of the basic code like its boot loader, SD file system, and networking facilities. The other memory is the SD card which is where upgraded admin firmware and user files are stored.

Before flipping the PCB over, let’s look at it from sideways perspectives. Front first: the IR module is implemented on a “daughter card” for system manufacturing flexibility, with the main board therefore supporting both IR and non-IR product variants:

Now for the back:

And the two sides:

Finally, let’s see what’s underneath. Not much, as usual, aside from a bunch of visible traces:

I’d lost touch with Universal Devices and its CEO, Michael Kohanim, in recent years, so I was pleasantly surprised to unintentionally come across them several times in recent months. First off, the guy who bought (and then returned to me for refund) the SmartLabs 2412S PLMs I mentioned back in July told me that he had been in communication with the very same Michael Kohanim, who helped him with some remote debugging in striving to figure out whether they still worked (and apparently even found him a functional PLM after mine didn’t pan out).

And then I heard through press coverage that Kohanim had actually put in a bid to buy and resurrect SmartLabs after that company abruptly shut down in late April. Universal Devices had always “punched well above its weight”, from my experience; LinkedIn reports that there are only four full-time employees. It’s nice to hear that a decade after my last in-depth interactions with them, they’re still alive and kicking; I wish them the best going forward. And now it’s over to you, dear readers, for 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.


Leave a comment