KNC just posted pics of their case, PCB samples coming next week \o/

[Tehfiend]

https://www.kncminer.com/news/news-28

KNC is looking good so far IMO. Can't wait to see the PCB samples next week...

[DeathAndTaxes]

Why 5V external power?  Most ATX PSU are designed to deliver most of their power on the 12V rail(s).

[hchc]

I'm not invested in knc but the pic gave me a chuckle..."box and a bunch of fans"
I'm not implying anything else

[J35st3r]

Why 5V external power?  Most ATX PSU are designed to deliver most of their power on the 12V rail(s).

It has to be a typo, these beauties need hundreds of watts (I did suggest elsewhere that 48V would be a sensible PSU voltage, but those are hard to come by outside of server rack systems).

[dogie]

5V HAHAHAHAHA enjoy that.

[MrTeal]

Why 5V external power?  Most ATX PSU are designed to deliver most of their power on the 12V rail(s).

That was the first thing that I thought when I saw those pictures. Sure 5V to ~1V will be a little more efficient, but that's just unloading the problem onto the end user who as to come up with a source of 30A of 5V.
It must be a mistake.

[DeathAndTaxes]

Why 5V external power?  Most ATX PSU are designed to deliver most of their power on the 12V rail(s).

That was the first thing that I thought when I saw those pictures. Sure 5V to ~1V will be a little more efficient, but that's just unloading the problem onto the end user who as to come up with a source of 30A of 5V.
It must be a mistake.

Yeah ultimately it is 120VAC/240VAC in and ~1VDC out. The individual steps (or number of steps) doesn't really matter.  What matters is the overall efficiency from the wall to the chip.  ATX PSU are very efficient at what they do.  It is unlikely any design (from wall to chip) will be more efficient than using a 12V rail on ATX PSU.  I had even looked into dedicated 12V PSU (i.e ones used for industrial equipment not PCs) and while they are sometimes cheaper they aren't as efficient (efficiency tends to be high 80%).

Making the user try to find a PSU capable of up to 200 amps on 5V (Jupiter pulls 1000W) is going to immediately kill of the most accessible, best priced, and high efficiency options ... ATX (computer) PSU.

Like you said, lets hope it is a typo (it really uses 12V not 5V at board level) and not a horribly stupid design choice.

[augustocroppo]

Like you said, lets hope it is a typo (it really uses 12V not 5V at board level) and not a horribly stupid design choice.

If that is not a typo, this would be enough to power up the unit?

http://www.ebay.co.uk/itm/1pc-Switching-Power-Supply-SE-600-5-5V-100A-600W-AC85-264Vin-247x127x63-MeanWell-/130891366233#vi-content

[J35st3r]

Making the user try to find a PSU capable of up to 200 amps on 5V (Jupiter pulls 1000W) is going to immediately kill of the most accessible, best priced, and high efficiency options ... ATX (computer) PSU.

Not to mention the fire risk (thinking of the guy who melted his Avalon PDU). Even 12 volts is 80 amps at 1000W. Make sure you plug all of those connectors in before firing it up (I wonder if there may be a consumer equipment safety problem here. I would hate for it to fail to get a CE/UL sticker).

[J35st3r]

Like you said, lets hope it is a typo (it really uses 12V not 5V at board level) and not a horribly stupid design choice.

If that is not a typo, this would be enough to power up the unit?

http://www.ebay.co.uk/itm/1pc-Switching-Power-Supply-SE-600-5-5V-100A-600W-AC85-264Vin-247x127x63-MeanWell-/130891366233#vi-content

Note that those (like most industrial PSUs), have screw terminal connections. Not consumer friendly.

100A at 5V is just 500W. And its input is 12V so we still have the multiple stepdown problem with efficiency scratch that, damn eyesight 

[augustocroppo]

Note that those (like most industrial PSUs), have screw terminal connections. Not consumer friendly.

100A at 5V is just 500W. And its input is 12V so we still have the multiple stepdown problem with efficiency.

KnCminer terms and conditions specify that they are selling a product for business, not end users. So that could justify why there is necessary a PSU not so "consumer friendly".

I will Google more to see if I can discover a 5V 100A PSU without connectors.

[bbxx]

those cases will not survive shipment.

wtf is this lol.

it is ugly, not durable, looking cheap and disgusting.

terrible design.

it is such a shame...

[J35st3r]

The bad news is that PSU highlight the problem exactly.  It's efficiency 120VAC -> 5VDC is 82%.  Now the chips don't run at 5V (or 3.3V or 12V) so there has to be another converter on board.  Most like 5VDC down to ~1VDC.  Efficiency there may be 90%.

If you look back at previous newsletters it shows a picture of the quad chip with 0.9V DC/DC power converters. Almost certainly 12V input (its not a problem for a single stage buck converter). They also mention hosting, with explicit ATX power supplies. so the 5V is almost certainly a typo.

[polarhei]

Four trimmed Beings like 4 gPU on board, close to the CAD design and even better, good job. Some people may ask if the actual design can handle the heat as the being, 1000W, the 8-pin and the 6-pin can answer the most as ORSoc may take part in some serious designs like GPU.

The actual picture is not enough, where is the board?

[greenbtc]

Note that those (like most industrial PSUs), have screw terminal connections. Not consumer friendly.

100A at 5V is just 500W. And its input is 12V so we still have the multiple stepdown problem with efficiency.

KnCminer terms and conditions specify that they are selling a product for business, not end users. So that could justify why there is necessary a PSU not so "consumer friendly".

I will Google more to see if I can discover a 5V 100A PSU without connectors.

I would think it would have to be a typo. I've reached out to KnC to confirm, as this would be a major design flaw even if they were targeting the "business" market.

[MrTeal]

That Meanwell would work, but as has been said it's very inefficient relative to a good ATX supply. You're also going to have ~6x the losses in your cabling from the unit to the onboard DC/DC that you would have at 12V as well.
It is also going to be cooled by a single 40mm or 50mm fan howling like a banshee. You will not want to be in the room (or house) with a couple of those guys running.

[Xialla]

if they are targeting business market, I'm fine with it, but in this case PSU outside miner doesn't make any sense.

[af_newbie]

Why 5V external power?  Most ATX PSU are designed to deliver most of their power on the 12V rail(s).

+1

Someone did not think this through.  It should have multiple banks of PCI-e 8 pin connectors, run in parallel.

And these people will deliver 28nm ASIC?
Hard to believe.

[DPoS]

Why 5V external power?  Most ATX PSU are designed to deliver most of their power on the 12V rail(s).

+1

Someone did not think this through.  It should have multiple banks of PCI-e 8 pin connectors, run in parallel.

And these people will deliver 28nm ASIC?
Hard to believe.

The small slot at the front is to accommodate the multiple 5V power connections from the external PSU

[CoinHoarder]

Nice to see a little progress, but like everyone else I am not happy to see it will use 5v. I have a lot of ATX PSUs laying around, it would be nice to be able to use them.

I do not think it was a typo... its kind of hard to accidentally type 5 instead of 12..

Would something like this require much redesigning to switch it over to 12v... as in, if everyone requests they use 12v, can they make the switch easily? I know nothing about PCB design.