Ordered one of these 1/7, received today in good shape.
Funny to see on tracking it left Hong Kong at 9pm, arriving in USA 9am the same day.

No damage inside the unit, no loose cables, no broken caps, no bent corners.  I took it apart to inspect it, there were no "fuse" labels silk screened on the ASIC boards, however there ARE 2 fuses in parallel at that location, they are marked P.  A Google search shows that these are most likely 32volt 3amp fuses.  There were some loose/cross-threaded screws holding the ASIC boards on the heatsinks and a loose wifi antenna to adjust and tighten.  I added some cardboard and masking tape around the intake and exhaust fans to direct more air through the heatsinks, also taped the gap between the two heatsinks to make a sealed tunnel.  Along the edge where the heatsinks are screwed to case I applied some thermal paste to try and increase heat dissipation.  The fans start at full screaming loud speed at power up, but quickly quiet down to a normal "gaming computer with lots of fans" level.  Reported temps stay in the mid 30s C.  One of the fan cables connected to the TP-LINK board has had 12v and ground cut off at the connector, leaving the speed sense and PWM wires connected.  The red and black fan power wires are soldered to the power board, presumably the TP-LINK cannot provide enough current to run 2 the large high speed fans.  Both fans still slow to a lower noise level after booting up.

Btmine posted in an earlier post that the unit could run on a single power supply, this is not true.  I tried with a single EVGA 1000watt 80+platinum power supply (12v 83amp single rail), with the 24 pin ATX motherboard connector in ATX_24P and nothing in the second ATX_24P_2 connector.  The unit would power up but only use the 2nd blade, running at ~100GHS.  Connecting a 2nd 750watt power supply to the ATX_24P_2 connector powered up both blades, it is now running at ~220GHS.  If I were to splice up a 24 pin ATX motherboard power connector splitter and connect both connectors to a single power supply it would probably work, perhaps I will try it later.

1400Mhz seems to be the limit for these boards, at first they ran for a couple of hours at 1500Mhz but then became unstable, the unit seemed to be rebooting and spinning the fans up to full speed then back to normal over and over, speeds dropped to less than 150GHS.  1600Mhz was even worse, many of the match_work_count units on the api log page were hashing at very low levels.  Everything was normal after switching back to 1400Mhz (~210-220GHS).