Bitcoin Forum

So the question is....
Does anyone see better performance by slightly over volting the DC input? 

On my adjustable voltage PSU's I used to tune then so I actually get 12.0V-12.1V into the miner to counter the voltage drop if the wire runs. 

Just wondering if it does any good to maybe provide 12.2? or 12.4? 

Same question for the opposite the PSU's that output exactly 12.0V on the breakout board are actually providing less then 12.0V to the miner, does this have any negative effect?

Is one way hard or easier on the VRMs?

Any constructive input is appreciated.

TIA.

Unless your using 100-200 feet of extension cords,you should have very little voltage drop.A 3-6 foot cord will have negligible voltage drop.

A little "boost" may be ok,depending on the device's sensitivity.

Just keep a close eye on em & measure (volt,wattage,temp & performance) for a period of time to be sure your not harming them  ;)

What kind of cable are you using?  I see VERY noticeable voltage drops with even 24-36 inches.  That is with 16AWG Copper.

And I have never seen a 100 foot PCI-E cable before, from what I know I do not even think 12V is usable if you tried a 100 foot PCI-E cable.

you guys are talking ac and dc.

@ op what psu do you have that is adjustable dc? 

this would be good to use with antminer s-5's

Higher voltage from the PSU will mean, for the same power draw on the miner, less current draw through your cabling and therefore less loss (resistive losses are current squared). However, higher voltage from the PSU means a lower duty cycle in a VRM miner which means increased current ripple (with losses in parasitic resistance on the board) and decreased overall efficiency. Higher voltage in a string miner means higher node voltages, and therefore higher current consumption. If anything, you probably want to lower the PSU's output voltage. I even saw a measurable difference in efficiency with my S1 testing between identical clocks and voltages with an 11.6V and a 12.0V input.

16AWG copper has 4.014 ohms/kfoot, so a 100 foot 6-pin cable would expect about 0.27 ohms. Powering an S5 at 600W you'd see 150W per cable, so 12.5A per cable, expected wire loss of 42W PER CABLE. That's big enough you'd have to factor it into delivery calculations since the endpoint voltage is less, and the current is that much more so the power loss is actually greater. Gross.

I thought you were talkin bout wall to PSU....if your seeing that big a drop on your PCIE cables,you need better cables or connectors or even a better PSU  ::)

Sidehack covered it  ;)

You are correct I forgot to specify DC, I figured since I refereed to 12V it was obvious but my fault.  I am talking DC.

The PSU is HP DPS-800GB can adjust from 12.05V-14.0V with the breakout board. The boards were made by Jack Sausa, they have a small potentiometer on the board for adjust.

'
Thanks.  That was what I needed, and of course I do not run 100' cables even if I needed on for that reason.  I use 36" or shorter 16AWG Copper wires.

Yeah, as much of my stuff as can be run with 18" gets those cables. Less than 0.7% wire loss typical, not including the PCIe contact resistance.

All depends what do you plug at that regulated +12V. If miner have string design then you will increase core voltage of each chip and power draw  will also increase. If miner have "classic" design with DC/DC converter then increasing +12V will reduce current draw (power draw will increase just little bit). In first case you will be able to overclock miner higher, in second you will be able to use thinner cables and cheaper low quality PCI-E plugs.

Also keep in mind that as the cabling ages it can pickup corrosion and deterioration that will slowly increase the resistance in the wire. No, nothing over night but, as the months rack up, in this hostile Texas air, occasional tune ups will be beneficial. Unless you live in Stink-A-Denna, then you might want to check things hourly ::)