I like your idea.  I have cleaned the ASICs using a 1" soft bristle paint brush cut off at 3/4" and acetone.  I sometimes use auto brake cleaner.  I do use compressed air to blow off the liquified paste. 

I like your idea because thermal conductive pads on the ASICs don't mess up the pins or provide a material dust in the air flowing by can adhere to.

It's Saturday morning and last night I had decided to rip it apart and repaste.  I decided this because an S3+ board ran hot a couple of days and then even tho I belatedly improved its cooling, the board went belly up - entirely.  But I got up earlier than usual.  I turned the C1 off but then decided to go back to bed and deal with it after breakfast so turned it back on.  The third chain had been running hot.  Before turning it back on I rearranged the cabling somewhat.  Now, during breakfast I glance over and it's been back up for well over 2 hours and not only has the hashrate remained good but temperature on chain 3 is just right.  So, I suspect a connector, cable or supply.  Each of the 4 supplies, one to each board and each supply having two cables, has its own DVM at the supply.  So the problem isn't the supply for that board not providing properly I think but maybe a crack at the solder joint on the board or bad connector.  I make my own cables and the last batch of PCIE-6 connectors I got in were a mixed batch - two different quality, gauge, wires.  I've been splicing these to 14 gauge, 15 amp wire.  I have some doubts about placing the DVMs on the supplies next to the voltage adjustment pot.  I may move them to the PCIE-6 end or splice in a thin wire pair at the PCIE-6 and run it back to the supply leaving the DVM on the supply but reading the voltage at the PCIE-6.  I might do better extracting the pins from the connectors and affixing pins directly on the 14 gauge.  But for now, if my having disturbed the wiring before turning it back on has corrected the problem and that chain 3 doesn't heat, then I'll leave it be until it goes bad and go about getting pins for the connectors.

Seems to be dropping again.  Down to 930GH/s.  May have to rip it apart.  The good C1 has temps 34,35,34,35 while this is 32,35,38,35.  It's running  hotter.  Cooler water going in but higher temps.  That 38 board is what I see.

I haven't ripped down a C1 yet.  Started but quit.  You use paste on the side to the waterblock?

Just like what happened to one of my S3+ boards although the other is hashing at 227.49GH/s.  I haven't yet found a solution.

If you could put an oscilloscope on pin 13 of either ribbon cable....

Well, when you have them both up try pinging both, then after disconnecting the one, ping the other.

You ARE giving them different addresses, yes?

Put the "bad" controller board in a miner, fire up while holding the reset button for a full 90 seconds, then shutdown, wait almost a minute perhaps and fire up.  Then tell us how you did.

Maybe it's a routing problem, e.g. drop the eth connection and drop the route.

Odd.  I've been running my two C1's at 243.75/0675,960-985GH/s so they'd last longer.  One has dropped a hundred GH/s to 848GH/s repeatedly the last few days.  Raised the voltage from 0675 to 0700 and again have 977.44GH/s(avg) after an hour.  It had been running so well at 0675 and the only change was that I moved the miner from the bottom shelf of a 4 shelf cart to the top.  Dust buildup rate while on the bottom shelf was too much to ask.  Low pressure air cleaning of internal dust while it was shut down didn't return it to its former rate at 0675.

Actually I find it interesting.  Not much going on while we wait for an S6 or C2.

Remember the days when docsis modems provided a cable connection then for the user a USB port and an RJ45 port without informing the user it's one or the other not both?

One might ask how many devices does he want to connect to the internet.  All via ethernet cable might be easiest.  If he has 4 high speed switch ports and a single uplink port on the comcast and if he has 4 or fewer devices to connect then just use the comcast router.

If he has more devices than high speed switch ports then add the netgear router.

Understand a home router has one uplink port and an internal downlink port (192.168.1.1) but the internal downlink port is attached to a multiple port high speed switch.  The uplink port will have a different network designation than the downlink switch ports.  If your comcast box is giving your  netgear box an uplink address of .1.2 then your netgear downlink switch ports must have a different address like in the 192.168.2.0 network.  If you then put your laptop into a netgear downlink port and the S3 into the netgear as well, they might be 192.168.2.2 and 192.168.2.3 while the invisible internal downlink port has 192.168.2.1.

I make my top router have the highest network and the lowest the 192.168.1.0 net.

Too bad this linux doesn't supply nmap in the default distribution.

I find a 90 second depression of the reset dependable.

If you look at the first line, it has http://stratum+tcp://xxxxx  it's wrong.  You use either http:// or stratum+tcp:// not both.

My first guess would be the ribbon cables but if that there would be pools.

So, what you are showing is immediately after flashing while the antminer pool is still the default?

Can you show a shot of your Overview?  ...and of your pools page without the passwords?

I remember more clearly now.  The paperclip fix to get a computer power supply up and running may have had this glitch - the current ramp up from a cold start was okay - but if the power supply was shut off and on again in a short period of time there could be a current surge driving the voltage up blowing caps - if the current ramp up was dictated by a fixed current charging a capacitor, giving a linear ramp, and the capacitor doesn't properly discharge on turn-off if the motherboard isn't in place, but that adapter with a fixed jumper and fixed resistor works allowing discharge on turn-off.

May I ask what voltage the fried caps are suppose to see?  And these are in series across the voltage supplied to each ASIC?  So, the 12vdc from the power supply is divided equally across the ASICs?  This is what I took away from the series ASICs description.

Are those who are seeing the blown caps using the power supply adapter having a resistor or just using the paperclip jumper?  I ask because KnC had a similar problem of blown caps and I theorized the motherboard actually on the plug allowed a different rise time on the voltages, e.g. without the mb and a paperclip instead, a ramp for the max current didn't ramp instead jumped to full with the result of a turn-on voltage spike - sometimes.

I've had success with a full 90 second reset on some miners.