Welp one just died.  Wont even init.  Solid green, no init pulse, wont run at 150, just sits at 0.  I think I killed the chip.  RIP little bugger.  Waiting on MM Test Leads to ensure core voltage etc.

Thoughts about soldering the power leads direct to the +5v/Grnd leads of the USB connector instead of running it off of another port through another cable?

Drupie

vapourminer, or any other pi-heads out there, I searched the forum here and elsewhere and can't find an answer.

On my either 2B or 3b (I'm pretty sure 3B) I run two copies of cgminer concurrently - one runs 4 sticks at 425, and the other runs three sticks at about 350 or 375, whatever I can get them cranked to.

POWER for the pi comes from an unpowered (port on-off switchable) hub connected to a powered 2.0 hub, as does the on board pi port go to the powered 2.0 hub.  THe 2.0 uplinks to 2xSuperbpag 3.0 running sticks in alternate slots, staggered for maximum airflow, with four fans going on.  Two fans are run from the *2.0* hub, and the other two from the secondary 3.0 hub (where the three slower sticks sit).

Here's the issue:

If I run both simultaneously, I can almost always count on loss of network connectivity within 24 hours.  I can restore it by unplugging and replugging the network cable, and there's some talk of ifup/ifdown commands to refresh the DHCP lease, or setting the IP static.  I Also added eligius to the hosts file to see if that helped, but it didn't.  I lose the ability to talk to eligius and im down until I notice it.

HOWEVER - if I just run the 4 sticks at 425 its rock solid for days. (the other three are plugged into their superbpag but the hardware refresh is 0'd out or they are zombied on my active cgminer)

Any ideas?

Thanks,

drupie

Just a shoutout and hat tip to SideHack - Still running strong!  Obliged!

Right, got it now!  Boards - not my thing.  Thanks HD!

Frankenhub....



IS ALIVE!




One of these days the damned pictures will post gosh durnit - Haven't hooked her up to my Pi yet, but the +5VDC has been cut on the inbound line, and is only powered by the soldered leads.  Continuity testing confirms that all joints are isolated and continuous with original +5VDC input solder point, and no continuity between any other non-power solder joints, there are no bridges.

lmgtfy eh? hehe yes i felt about as dumb when i read the reply - touche'.  My concern at this point would be the twin black thick wires as compared to the singular thinner power lead.  I can MM the +5 and Ground pins with minimal difficulty.

<--- rightfully a noob!

Anyone familiar with the USB 3.0 pins and how to ramp the DCV?  USB 2 has only four wires, but my little 2.0 unpowered bit the dust (tiny wires, no physical support, they broke) so I bought a new 3.0 unpowered so that the ports themselves could handle higher amps - but I crack her open and while the connections are kept in place by hot glue (yay) - there are like 11 pin holes on the bottom, and purple yellow blue wires too, two thick black wires for ground it appears.  Anyone have a hint as to where to move from here?  Is it as straight forward as the others, cut the red to prevent backfeed, solder new leads to red and ground, then run individual wires to hot pins for the ports? 






Still no pics?  Forever n00b. (sadpandaface)

Thanks for the explanation!

Will across to its neighbor do, or do I have to go down to the board?  its a large gap, I don't have hot air, and to heat that much mass i'd need a big tip at a moderate/high temp no?  Its a Weller adjustable pen, I'd give you the model and tip info but I can't lay hands on it now, but i have 0.25mm tips up to honkers at least 1.5 or 2mm across if not larger, lots of mass to them.  Temp pot (analog) goes from like 550/600 to 850/900 I believe.  Probably would need a new tip, oxidized to hell and back it would be by now.

I might could melt the down side and push the cap down closer to the board on the up side.  Its already touching (if not soldered) to its neighbor, and continuity is good (strong constant beep, 0 on the meter), resistance shows the same between middle and left (both on their pads) as compared to middle to right (one up one down) at .9 ohm (on 200 ohm scale).

If down to the pad, flood it with enough solder to fill the entire gap?  Or try the push down mentioned earlier?

Ty.

Drew

Is that a big deal?  Should I bridge it down?

(and I have no idea why imgur isn't doing anything other than a link - noob)

<jaw hits floor>

I was more referring to the concepts of how to determine the amps per port rating, and my suspected yet unconfirmed notion that the power regulation devices within a given hub work usually in pairs, so that you can safely draw 1.8a out of one port in a pair of ports on a USB 3.0 hub.  (and that a 9 port with 3 power switches is probably regulating 2.7a over three ports)

In order not to flood, although it may be applicable, I sent Morguk 4 hubs that I found on Amazon, one of which is rated 5 stars by a miner.

So *my* understanding on how to find amps per port.  Take the input, which on Amazon UK you have inputs at 12v by #a, as well as 5v by #a.  Considering that, we'll assume 2A.  12*2=24Watts.  Watts divided by volts = amps.  So 24/5=4.8 total amps for the system.  Then divide that by the number of ports to get average amps per port at 5v. 

Note that USB runs at 5v, so the voltage of the input needs to be converted to 5v.  So if you're looking at an input of 5v*2a you have 2a for all ports, but 12v*2a is 4.8a for all ports - so the input is important.

NOW number of ports: the fewer the ports vs the higher total amps is better.  Some chipsets will actually amp limit if you draw over a certain level, but take into consideration that the device only takes the amps that it needs.  You can have 10a on a port, but if the stick only need 1.4 it will only use 1.4.  But the same stick in a 0.9a will only get 0.9a or, depending on circuitry it will deliver more till it reaches a certain point and then cuts power to prevent overheat internal to the hub.

When it comes to routing the electricity to the ports, its more expensive to regulate every single last port.  You can cut your costs in half on that aspect by regulating two ports on the same line.  So that means that branch would get 2x(per port amps), ie 0.9x2=1.8A per branch - so what that means is that the hardware only watches to ensure that that branch doesn't draw over 1.8a, so as long as you use one stick per branch you can have all the amps through that single port.

Ive no idea how to split up 7 ports, but on the 9 port mentioned earlier (it says 10 port but the 10th is a charging only port) there were three on/off switches - which leads me to believe that its a 3 branch, and while you may only have .9a per port, you actually have 2.7a per branch.  You can overclock one stick, doubtful two, so you'd only be able to use three sticks there (one per branch).  However there is probably enough to drive a fan each as well on those branches - they don't use much amperage, and you could have one fan per stick and be safe and cool.

OTOH the hubs provided to Morguk are 10 ports, and I would be they're split in pairs, so that means 5 sticks. 

But sidehack would probably be able to correct me on any of this.  I only have approximate knowledge of many things.

Questioned Item:

http://www.ebay.co.uk/itm/ORICO-DCP-48W-7-Port-Universal-USB-Charging-Station-Hub-with-7-x-2-4A-Charger-/191295371185?hash=item2c8a17b7b1:g:Mt8AAOSwqu9VMOm7

I don't see any USB 2.0 or 3.0 uplink port.  There is only the four pin round connector, which I'm thinking is for power!  This may have the word hub in it, but its not a hub, its only a charging station.  Has power, but doesn't connect to computers.

I'd wager a tuppence anyway.

What Biodom said.

Now, with the placement of the on/off switches, it looks like its splitting the power to three banks, so per bank you have 1.8a so I'd try putting one stick per button - this limits you to only 3 sticks.

I've run 4@400 on the Spag - the hub gets rather warm but not uncomfortable physically (mentally that's another issue) and of course the heatsinks are hot but not burning with two usb fans from the 2.0 hub.  I'd think you can run 2x4@400 on 2 Spags so long as they're not stacked and have room for air..

When Run 2 comes in, and 812 is repaired, i'll be doing just that...

Now, as for 8@425.... cooling would DEFINITELY be an issue!

I don't remember what the Win32 version does, however the linux compile the only software switch you need worry about is the frequency with --compac-freq ### where ### is a number between 100 and 500 .

As there is no working switch for voltage in cgminer/compac, you have to adjust the voltage with the potentiometer on the front of the stick, with the assistance of a digital multimeter.

The highest voltage you will probably get is around 800mV and the highest stable frequency you can get is around 400-420mHz if you have enough power from your hub (hence the need for a USB3.0 hub - has more power per port, but you can use a y-cable with a USB2.0 hub)

Note that the more power and frequency you use, the hotter the stick(s) will get - SO USE ACTIVE COOLING!  if you are running less mHz, you need less mV, so if you want to run at say 300mHz, set the potentiometer to 800mV, start the program, and adjust the potentiometer downwards until you start to get errors, then turn it back up slowly until the errors go away.

There is a mathematical equation, but its only hypothetically correct and, given manufacturing variances of all products combined (not just the compacs) and my abysmal failure at math, I just start it high and tweak it till its right.

Oh, if I didn't say it before, USE ACTIVE COOLING!!  Fans!  Get some!



En français, gracieuseté de Google.

For Clarification...
Morguk, Biodom told me, who was apparently told by that other kind member.  Information flows nicely

Anyone tracking highest share completed by a COMPAC?  Is 22.5M anywhere on the list?



(ok so that embed didn't wotk - suggestions?)