Controller Green - good, Miners should be Blue though.  Did you burn 841 firmware for the Pi?

Not to mention its current Proof-of-Work is completely dependent on GPU manufacturers coming out with newer cards with more memory.  Risky at best.  (Just ask those folks with a few year old 2GB memory cards...)

Although I applaud both of your for switching to a pool that share transaction fees, I'm curious why you did not go to Kano.is?  Slushpool charges 2%, Kano charges 0.9%.  Variance is higher on Kano, but getting better by the day.  UI isn't as slick, but for me, profit is what mattered.  Kano community is great (just check the Kano thread in this section).  There is a fairly steep ramp-up for full payment, but that is just a delay, not a penalty (e.g. if you left, you would continue getting paid as you "ramped down" at the same rate you "ramp up".).

Great in theory, hard in practice due to the wire gauges involved in carrying the current to your miner.  Math:  13.5Th at 0.098w/GH = 1323 watts at 12V = 110.25 amps.  Would you really be willing to run 00 gauge wire to each miner and then break that down to (9) 6-pin connectors?

Yes.  The correct firmware was loaded.  That said, I have some time today, so will try that again.

Update:  Thanks for the sanity check.  Reflashed the SDHC cards and both came up fine.  Not sure how I got those flashed with the wrong version... its not like a full size SDHC card would fit in the Pi3, but somehow I manged.

Also chuckled when you said "a warehouse with a roof big enough".   Basic solar math, presuming you want to run on 13.5TH S9, with a real world consumption of around 1450 watts:

* (1) typical panel, lets say a 130W one, will produce about 4 hours of power a day at 80% of its rated rate (since the rated rate is optimal during noon soon in the tropics):  130*0.8*4=416 watts per panel per day
* Each miner therefore requires about 3.5 panels
* 1 megawatts of power = (690) 1450W miners = 2,145 panels at about a square meter each, so call it 21,450 square feet of roof space over a data center that doesn't need to be over 2000 square feet in size, likely larger.

But wait...  Nobody is going it invest that type of money to run miners for 4 hours a day, so lets add batteries and enough panels to run for a full 24 hours:  80% efficient charging batteries = 2145*6/0.8 = 16,100 panels.  So about $2M in panels.  Batteries to carry say, 3 days of clouds, will match that cost.  So now your at $4M.  Oh, but wait again... we only have enough panels to keep ourself running for 24 hours and now, after a cloudy day or three, we need to run AND charge those batteries.  Better double the array size (now $4M and 32,200 panels).  Plus battery chargers, inverters, etc.  Add at least another $2M.  

So, we have $6M in solar infrustructure to run 690 miner at something like $1300 each current pricing (sorry, Bitmain doesn't have 13.5's listed right now, which this math is based on); eg. ~$900K in miners.
At $0.10/kwh, and 1MW of required current, each hour cost $100.  So in about 7 years you will break even.  Except that your batteries will likely have been replaced once in that timeframe and likely be coming due for a second round, at $2M  a pop.  Good news, electronics are likely good for 10-15 years.

Oh, and since we now have 32,200 panels in our little setup, your going to need about 320,000 square feet of south facing roof.  That just a bit over 7 acres ON THE SOUTH SIDE.  So a 640,000 square foot warehouse of which you will be using around 2000 square feet.  Or perhaps you could find a 6000 square foot building and provide about 1% of your power for a mere $60,000 - but why bother.

Granted, you could grid tie, and eliminate the batteries and half the electronics, but then you need to deal with paying retail power rates and likely being credited wholesale power rates.   Not aware of many NEW grid tied solar contracts that allow for reversible meters like they use too.  They electric companies have wised up and paid off the right politicians to protect their turf.

Hi all.  Received my first batch of Avalons today.  I went to set up (3) strings:  String 1 was headed by a pi 1 with a single 821 on it.  String 2 was headed by another pi 1 with (5) 841s on it.  String 3 was headed by a Pi 3 with (5) 841s on it.  Once I found real data cables to use with the AUC3s (and found out just how many charging only cables I have!), the Pi 3 unit came up just fine.

The (2) Pi 1 units had AUC3's that were mostly blue but slow blinking green.  e.g.  They were more-or-less always in setup mode and resetting as soon as they came ready.  The Avalons behind those two Pis remained idle (yellow light).

I moved the AUC3s from those Pis over to the Pi 3 and confirmed the AUC3s were fine, the cables were fine, and that the miners were fine.  Currently running all (11) Avalons off that single Pi 3.

Curious if anyone knows what the problem was driving the AUC3s from the Pi 1s?  I thought perhaps insufficient power supplies to drive both the Pi 1 and the AUC3, so tried a 2.4 amp one - but had no change.

Any other ideas?

The widely accepted rule for S9s is "If they are not broke, don't fix them".  IF, AND ONLY IF, you are having problems, you might want to try upgrading the firmware.  For Auto-Freq S9s (just log in and see if the card speeds vary) the November 17 firmware is the latest.  Sometimes it helps, sometimes your better off with the April firmware.  You choice is more limited if you have static speed S9s (from over a year ago).

No firmware change will affect your performance on the pool, beyond possibly helping the miner perform better.  Maybe.

Just another tidbit... I've had (4) machines now lose their configurations when downgrading from 1171 to the 650 fixed firmware.  No big deal, beyond the confusion figuring out that they had new DCHP assigned addresses.

Update:  It appears that by controlling the card speeds I've managed to get a couple of "dead" cards back online.  By "dead" I mean those that appear under miner status with an expected Ideal rating, but a 0 RT rating.  Obviously, having (3) cards hashing a bit slower is better than (2) cards hashing at full speed and one dead.

Note on Update:  Not sure how much this has to do with the fixed frequency option, or simply as Fanatic26 has repeatedly pointed out:  The 1117 firmware sucks.

Presuming his wiring and breaker are both rated at 20 amps, that DUAL 15-amp receptacle is fine and code.  Each outlet in that receptacle is rated at 15 amps.  This is exactly what I do in my data center (except I used nice brown ones because they looked different, making it easy to tell, and I found them cheaper at Platt).

For what its worth:  I've swapped several controller cards over the past couple of months with new ones I purchased from Bitmain.  All the new ones start out with a set frequency of 450.  Each time I've thought this was cool, because it would confirm that all my hash cards were accessible and had basic functionality.

Based on the above post, I just finished taking one of my "underperformers" (a 13.6TH machine rated at 14TH) and reverting it to fixed 650 firmware, then bumped it from 650 to 662.  One of the 3 cards is clocking a fair number of hardware errors at that speed, but overall the box is running at 13.93-13.98TH, an improvement.

About to revert it back just to see if it reverts to 13.6 (you know, basic scientific testing principles:  any phenomena needs to be shown to be repeatable at least 3 times, ideally at least 5 times, to draw a cause-and-effect conclusion).

Update:  Still playing, but having problems getting it to revert to auto-frequency - even with the November autofreq firmware.  The good news is that with the 1117 firmware, but fixed frequency, its still hashing at near 14Th and without hardware errors.  Weird.

Patents are only as strong as the financial muscle of the legal teams used to defend them.

Well... theoretically he could get by with the 4" hole if he could move the air through it quick enough.  Thinking along the lines of one of those gas turbines that power the old Batmobile...

btw - I figured 300 cfm per S9, once you combine the effect of the 2 high speed fans, so he would need to move 5400 cfm through a hole with a cross section of about 12.5 sq inches.  Doesn't that mean he would need the air to move at about 5400 feet per minute, abit over 60 mph?  That would be amusing to see.  Its only slightly below hurricane wind speeds.

Ouch!  That hurts to even look at!

They are power hogs relative to current tech.

I tend to agree, but difficulty is also up.  One needs to consider the cybercoin equivalent to "inflation adjusted".  Perhaps "Difficulty adjusted"?  Depending on when you pick, that would turn your $1200 into something like $4800.  So we would still be in OK shape, just not wonderful (or in December "giddy") shape.

Sorry to say, but don't expect to get much for those.  With coupons, I could buy V9s (relabeled S7s by all accounts) for $16 each plus shipping.

And hopefully get some relatively good deals on new miners as the manufactures adjust pricing due to the lower demand...

Of course, "relatively" is very relative.  Is it a good deal to pay $2000 for a miner when BTC is at $7000 vs. paying $4000 if BTC goes back above $14K?  I find contemplating such things a short path to making my head hurt.

Excellent point.  Simply unplug power to them.

I've heard 115C.