Topic: BITMAIN‘s Liquid Cooled Miner C1: It is not only cool, but it is cool - page 7. (Read 18339 times)

I'd disagree that a 16AWG cable with two plugs is almost as good as two separate 18AWG cables on a modular supply. Just looking at wire resistance it's close, the 16AWG will be 26% higher than two 18AWG and is obviously better than two plugs on a single 18AWG. On 17A of draw to a blade with 600mm of cable to the first connector, you'd be looking at 1.53W of loss and 90mV drop in just the cable with 16AWG vs 2.41W loss and 142mV drop with single 18AWG or 1.2W loss and 71mV drop with dual 18AWG. Not a huge difference, though one should keep in mind the contact resistance spec of the Minifit Jr connector can actually be as high as a 600mm run of 16AWG (7.9mOhms for the wire, while the contact spec is 10mOhm max initial and 30mOhm max after 30 cycles)

There's a couple disadvantages of running the two plugs on one cable though.

• Two separate 18AWG cables will have almost twice the surface area to dissipate heat from vs one 16AWG. That's why you see ampacity in a cable go up only 56% when you move up three gauges even though the resistance halves. Flat cables help with this, of course.
• Two plugs on a 16AWG cable pulls the entire cable current through the Minifit Jr on the PSU side. At 200W per cable this isn't terrible, but it's still a significant difference. At 17A per board that's 5.67A per pin vs half that with separate modular cables.
• Worse fault tolerance. If you're running one cable and one pin/wire is compromised for whatever reason, you can then be pulling the full 17A through just two pins, which is over spec. If you're running two cables and one pin/wire gets compromised, you're now pulling the 17A over five pins which is still fine.

I would say two plugs on a 16AWG cable is closer to two plugs on an 18AWG than two separate 18AWG cables even if just the wire resistance seems to show they're comparable. A lot of these issues aren't a problem two plugs on a fixed 16AWG cable though. If you find yourself with a supply that has two fixed dual plug cables and four modular dual plug plug cables, use both plugs on the fixed cable and a single plug on each of the modular ones to make up your eight plugs.

Not that the RM1000 won't work, but unfortunately not all 1000W PSUs are 16AWG on the PCIe cables. For instance the otherwise excellent rated Cooler Master V1000 uses 18AWG on its four dual plug PCIe cables. In most cases even dual plugs on 18AWG won't cause an issue, but you're still more likely to melt a plug at the PSU with a modular dual plug cable than with separate cables.

I may have to wait for the 2nd batch or at least until some people post feedback on the first few arrivals before I order.   Also is the liquid cooling kit link not working for everyone else?

i have to think about all these new developments.

The RM1000 has 8 PCI-Es on 4 branches of 16AWG cabling. Its almost as good as 8 individual 18 AWG branches.

You might even want to look into one of the server breakout boards so you can get eight full cables. You probably need to get up into the 1300W+ range for ATX PSUs before you can find one with 8 individual cables, so with a smaller PSU you'd be looking at running with two plugs on a single cable. Depending on the cabling and ambient, that might be a little marginal.

About 780w for the miner, 50W for the watercooling, requires 8 PCI-E. Still recommended to get a 1000W PSU though to get the best efficiency.

So the big question (and since it is early in the morning and I am skimming through these posts and I may have probably missed it........so all apologies) - what type of PSU should be used for the C1 and how much wattage should it have? I have seen all of the other details, but unlike Bitmain's other offerings, they aren't crystal clear on the C1's PSU requirements. Talk to me like I'm completely stupid and have just crawled out from under a rock

This was my concern I wonder if the 800watts is really 880

I received .

Appears S4 USD payment only at this moment.

That's because no one has one. They are the same as an S3 though, so if they're running at 1TH/s they're probably clocked at 250MHz.

Can anyone confirm what the frequency on these chips are?

I'd like to know if there is headroom for overclocking. Trying to figure out if getting 1100gh/s or 1200gh/s will end up eating 1000w+


I have asked a few times in the past 

Wow nice - Bitmain has a "Pay with USD" option now upon checkout

No PSU included

I'm confused ... is not necessary a PSU to start mining or is not included in price?

you assume that all three C1 waterblocks are equal. Theres a high chance that there could be slight machining diferences, or slight barbs from the machining, that could cause slightly higher resistance in some units. Its not particularly likely, but it is still possible. (easy way to adjust for this if it was an issue is to put a small vavle on each loop of the coolant to compensate any flow differences

Great news :-) So I can wait for my sockwarmers to come.

No need for experimenting noob,

Fluid dynamic dictates since all C1 are equal in restriction, all of them will have equal flowrate.... gee  

"To the moon" would get a totally new meaning....

Of course parallel and not serial. The available distributor I have leads from 1/4" to 3x10mm. Although the distance between the three exits is only some millimeters, I'm not sure if this will affect the flow to the third massively. It will take a bit of experimenting.

Are you kidding...

ZiG