Under the heatsink there is also another IO connector. This one is a 20-pin Hirose DF52 connector. This unfortunately does have to be crimped and assembled. It's a job I'm not looking forward to. The strength of this board is not the IO. There are better boards if you want that. It's in the fact that I might be able to communicate with it over PCIe. I bought these connectors anyway, but I'm in no rush to assemble them for now. Product Link
=
JTAG
With a nicely labelled 6-pin JTAG port at least I didn't have to start randomly probing or guessing at test points. To be honest if it didn't have this I probably wouldn't have taken it project on.
Power and ground
This was pretty simple. There is a small 2-pin connector for the fan (now disconnected) and a quick continuity check with a multimeter shows that two of the JTAG connector pins are the same. The supply is not for powering the device, but so that the JTAG knows what levels the device is using. That leave four more unlabelled pins to work out. It didn't seem to match any obvious standards. Time for some brute force.
TCK, TDO, TDI and TMS
So there are 24 combinations possible for the remaining 4 pins. It wouldn't be insurmountable to try them all, but there has to be a slightly better way. If it's one thing that computers are good at it's blindly following instructions. JTAGenum to the rescue! By the way, this is a great article on reverse engineering that steered me towards JTAGenum. Worth a read.
