#94 Apogee Symphony Thunderbridge teardown

As mentioned in the previous post, that Symphony I/O came bundled with an Apogee Symphony Thunderbridge PC-32 to Thunderbolt adapter. Just this on its own seems to sell for more than i paid for the entire package... But let's see what makes this bugger tick, shall we?

 

Not much upfront, apart from indicator LEDs for the two connections.

A little bit more going on in the back - DC power input, wordclock output (apparently "for diagnostic only"), the two PC-32 connections, and the two Thunderbolt connections.

Despite the deceptively(?) heavy casing, the inside isn't as cramped as one may or may not expect.

The heavy lifting is done by a Xilinx / AMD Spartan-6 XC6SLX45T FPGA. The firmware is stored in a Winbond W25Q64FW (i think; it's branded with some "TB11" version number) 64Mbit / 8 Mbyte flash chip, while the core is powered by a Microchip MCP1826 1.2V LDO.

The main buck converter is a Monolithic Power Systems MP28258DD-A - dang it took "some" digging to track down that "ACF" marking... Not sure what part that "VA32"-marked U11 is, but it's clearly some linear regulator, with an input and an output capacitor. U21 seems to be an OnSemi FDC638APZ P-channel MOSFET (acting as a load switch).

Over to the side is a Texas Instruments TPS73725 2.5V 1A low-dropout regulator, and U54 is likely some voltage supervisor / reset controller IC.

Now then, on to the much more involved part of this, the Thunderbolt interface. It's a separate board, likely because it requires presumably 6-8 layers (at least), while the "main" board probably gets away with "just" four. This connects into a slot on the main board, and it may well be a PCI-Express interface. Not sure which generation (that will likely be in the main controller datasheet), but likely a x4 link nonetheless.

Couldn't find anything going by the markings on this "main brain", but a bit of digging on Wikipedia (package size and thus generation) and Google finally got me to a slideshow - this is very likely an Intel CV82523EF or CV82524EF Thunderbolt 1 controller.

These first-generation unit sure need(ed) a whole lot of support circuitry... Parade PS8301 DisplayPort buffer / re-driver (figured out only from some Macbook schematics); NXP CBTL06DP211 DisplayPort 2:1 multiplexer; Parade PS8321 Displayport 2:1 multiplexer; Parade PS161 DisplayPort-to-HDMI/DVI converter. I can only wonder how (nevermind why) the Thunderbolt data gets repackaged or treated as video data, but then again, it's all just a bunch of differential data links, and the silicon doesn't care what the "video data" contains. 

 

On the other side reside several pairs of chips, seemingly one set for each Thunderbolt port, one way or another - NXP LPC1114F ARM Cortex-M0 microcontrollers; Pericom / Diodes PI3VeDP212 DisplayPort 2:1 multiplexers/demultiplexers; and next to the Thunderbolt connectors themselves, Texas Instruments TPS22980 power multiplexers. 

Near the card-edge connector is a Pericom / Diodes PI6C557 spread-spectrum PCIe clock generator. And since it's in frame, might as well move on to the power circutry as well - to start with, a Semtech SC194A buck converter.

On the other side of the board is a Monolithic Power Systems MP28253 buck converter, and a Diodes DMP2022LS P-channel MOSFET. 

Near the above-mentioned support circuitry for the Thunderbolt controller, is yet another Monolithic Power Systems MP28253.

 

And that's more or less it, really. After hooking up everything for testing, it took a while to blindly find the setting on the Symphony I/O to set it to "Apogee mode", but eventually i found it and this Thunderbridge works fine too! Now, just to get a power brick for this, and a new OLED display for the I/O, and then... Any takers?