- Paras-front.jpg (45.22 KiB) Viewed 70712 times
- Paras-rear.jpg (39.2 KiB) Viewed 70712 times
The "case" is hacked (yes, with a hacksaw) from the Adapteva originals.
The Akasa 80mm 12v case fan is powered by a spare 12v PSU wall-wart, suitably/safely re-terminated to the fan plug.
The blue heatsinks are mentioned , with careful modification of fin spread to allow them to stand closer together on the Zync and Epiphany. They are not (yet) clamped on, but the paste seems sticky enough as long as I don't touch.
The thermal paste is "Antec Formula 7 Nano Diamond" - so these are running Nano on Nano!!
{

I run the Parallellas headless, connecting from Windows via (Personal Edition). This allows me to run concurrent ssh, X, and VNC sessions on one or more systems. The 'Multiexec' feature allows commands - such as xtemp, or shutdown - to be kicked off at the same time. Really quite neat if you have a cluster - even if it's only a cluster of two!
Controlling the fan speed (via a Zalman FanMate) cuts the noise of the flat-out, so-called Akasa "Ultra Quiet" fan. The graphics below show Para01 (top) and Para02 (bottom) running xtemp, with the FanMate turned to low, high, halfway, and back to high speed. It's a statement of the bleedin' obvious that the more air flow the better, if your ears can take it. So those who use a 12v fan, but run from a 5v supply are probably loosing significant degrees of cooling.
- xtemp2.jpg (25.03 KiB) Viewed 70712 times
Para01 shows a temperature range of 50C/57C but annoyingly, at the same time, xtemp on Para02 shows 38C/44C !! I mean, that's great, and what I'm aiming for - but why the difference between the two boards? So far, have swapped boards up/down, but temp differences are same. No idea, but I'll be running more tests...
Pad vs Paste: On a previous run on Para01 with the supplied heatsink thermal pad, xtemp showed 56C/59C (full speed fan). The nano paste wins!!
HDMI vs Headless boot: The tests above were run with the HDMI devicetree/FPGA files. Re-running the xtemp test with the headless devicetree/FPGA files showed a further drop of 1-2 degrees C. Same difference between the boards though

Could do better - but currently running at sub-50C and sub-40C
