850w for an s4???
Hate to tell ya this but my s4's running stock speed and Vcore draw between 1,250 and 1,375w from the wall as reported by my UPS's.
My S2's report 1,039w again @ stock speeds.

Certainly though is hefty overkill for the S2 as they 'only' pull a hair over 1kw. If you have >208v for feed then the HP 1200w CS (Common Slot) supplies would be a better match. Just have to do some soldering with those tp attach the PCIe power leads as breakout boards are not easy to find.

My 2-cents is to get an IBM 2kw PSU (208-240v only) and breakout board to replace it. http://www.gekkoscience.com/products/IBM2K_board.html is a good source for the breakouts and PSU's or check eBay for the PSU's. Together they should run around $75-80 total+ shipping.
I got 4 of the beasts for when my S4's croak...

^^ Same query: When are more going on sale? I've got a ton of BTC burning a hole in my wallet waiting to be spent...
Cheers!

Hmmm... I'd think that clock and Vcore needed will pretty much follow what the S3+ or better yet the C1 takes. In fact - your freq and Vcore are exactly what at least 1 of my C1's is set to. Plug power should follow as well. More to the point is that 2.7THs for 160 chips is exactly the same per-chip as the C1 @ 1.1THs gives with its 64 chips.

Since it seems that the system can be happy with multiple supplies and I have a few spare HP 1200w server supplies on hand this bears investigating methinks... 2 strapped together is pretty much the same size as the one the s4 ships with. For now thinking 1 sitting external to run like yours.

Love the bit about "CUSTOMER IDENTITY PROTECTION IS ONE THING BUT WE ARE NOT WE'RE TO LET PEOPLE GET ROBBED BY OTHER PEOPLE."

Thass right! The only robbers Authorized by AMT is Joshua Zipkin and company!

Legally I would think being defined as a 'Consumer' depends on the Plaintiffs:  Mainly, are they registered as a corp or other commercial entity? Is the cost of the miner(s) expensable to that entity? If not then they are for personal use and fall under the Consumer Protection laws.

Just because miners generate income is irrelevant. If you install solar/wind/hydro/whatever alternative energy to your home and 'sell back' any excess to the power company that gear is also generating income. And, said equipment is considered non-commercial ergo - personal and falling under CPL.

re: clear liquid. It is most likely the carrier for the thermal compound. Typically it is either a silicon oil or Xylitol (a sugar). Was perhaps not thoroughly mixed?

As for 'not much thermal compound' - don't know your particular circumstance but thermal compound is to be used SPARINGLY to only fill in minor surface irregularities. Filling large gaps uses an entirely different composition than typical thermal grease/compound. From what I see in the pics it looks like there are matching areas where the compound peeled off between the boards and sinks and thickness looks right.

As for the sinks touching the caps - as long as the caps don't get dented , no problem.

The judge has DENIED Josh's move for dismissal. The Class action is moving forward
https://bitcointalk.org/index.php?topic=584891.msg9946540#msg9946540

The Vcore regulators are typical flyback-bucking regulators that usually can provide anything that is less than 50% of the supply voltage. To some extent the wider the difference the better with the only real trade-off being design target efficiency as the permitted bands for in/out are increased.

Questions are, are the low volt-in limit monitors in pert-near all regulator chips taken advantage of (preventing running that low in) and is the regulator circuit setup to handle those conditions if lower supply voltage is allowed?

If I was designing the Vcore regulator circuit for S3+ and S4's, yes would lock out on supply under-volt and for initial design targets rather doubt that last part.

S5's are announced on Bitmains site Looks damn interesting.

Speaking of interesting: From point-3 in the S5 specs they say, "When better power efficiency is needed in the future due to higher network difficulties, you may want to buy some special PSUs 9V DC with more than 10A output, which will allow you to have a 0.2J/GH mining efficiency, but at lower hashing speed. "

Assuming the S3+'s and S4's do not have fault limits monitoring the +12v supply, could this apply to them as well? (using a 9v supply)
I assume you just highly undervolt which normally increases Vcore regulator losses or pushes it out-of-band. Lower supply gives lower in/out delta.

Anyone tried serious UV/UC with a lower supply voltage?

NO!!! Way higher price than Bitmain's real site. Bitmain's current price for the s4 is $1,200 USD + shipping https://bitmaintech.com/productDetail.htm?pid=00020140926075515819D846Og240603

 No idea who owns the one in the link you gave....

Both sorta right. Vcore stays the same regardless of line input as it is set inside the miners.
Generally yes the overall efficiency may go up 1-2 percent depending on the actual PSU design but that's all. But.. the higher line *in theory* will allow you to pull more from the PSU so you have a bit more power available for OC/OV.

Main limiting factor in PSU's and their final ratings is how much current the input stage can handle. To put out rated power @100-125v the input stage has to be able to handle x amount of current. Raise the line to 200-240 and current is cut in half. Input stage loves that, now the limiting factor becomes the rest of the PSU circuitry.


Bottom line, unless a PSU is specifically marked and rated for different power vs line rating (Like the HP CS server PSU's) DO NOT assume you can safely sustain higher power out by using a high line in.

For new or upgraded installations using 208-220v is the only way to go. Wire sizes are smaller for x-amount of power per plug and costs for plugs is lower because you can use fewer of them vs 110v. Also I've found that dual rated supplies when ran off of 220v handle powerline dips better to keep miners alive. The Universal input function give a bit more reserve time for the line to come back up.

dunna know about minimum order. I always seem to have no problem spending a tidy sum there... like a heap of BSP fittings for the 3/8 tubing, a flow meter, tubing, QC's.....

Is not just a matter of aluminum safe. Automotive coolants are designed for, well, cars/trucks and the operating conditions their engines place on the coolant - and what the engines require from the coolant. In short, additives (generally silicates) are added which form a film on the cooling passages. Great for protecting an engine with large surfaces and cooling passages. Terrible for precision cooling needs.

Check koolance.com for CPU grade coolants or just search for 'CPU liquid cooling'. You'll find lots of suppliers.

Truth be told, the biggest thing CPU coolants provide aside from colors is algicides/antimicrobials to cut down on things wanting to grow in the nice warm coolant. If you keep an eye on it from time to time using either DI or distilled water is just fine assuming there is no copper or brass in the system.

An excellent source for everything dealing with liquid cooling is Koolance http://koolance.com/ They have coolant, tubing, much better pumps and reservoirs, and those nifty 'dripless' quick connect Doggie shows.

Personally, I will be using 3/8" (8mm) ID tubing vs the 1/4" (6mm) ID because I will be using a longer run between the radiator and miner. Like 2 meters each way.

for shits'grins just tried AMT's site https://advancedminers.com/
"Server not found"
Finally gone or has he moved to other digs?

The poster was no doubt referring to the 8-pin EPS +12v cables used on A1/A2 miners based on Innosilicon's boards eg. 'Dragons'.

They are designed for supplemental power to CPU's with 4 +12v and 4 returns. http://www.moddiy.com/pages/Power-Supply-Connectors-%26-Pinouts.html

Those are not the 6+2 PCIe GPU variants with the 2 extra returns in them.

And yes, I'd be interested in a few of them as well...

To quote from that link "At 60 Hz in copper, the skin depth is about 8.5 mm."
That is 0.335" or around the same as the diameter of 4ga. cable making it entirely a non-issue in this case which is my point.

Skin effect has nothing to do with 50-60Hz power and is not a significant effect until several kHz.
As for capacity of stranded vs solid, depends on how it us um, gauged (pun intended). AWG when applied to electrical wire is not a hard number, there is a fair bit of tolerance in the unit when applied to stranded wire. Comparing capacity of wires exactly the same diameter in inches or mm and depending on strand thickness (changes fill factor) yes solid technically can carry more current from DC up to several kHz. That is why in many parts of the world and for larger cables the actual equivalent copper cross section of a wire or cable is used for ratings.