Sandy Bridge i5-2500k with 4-pin CPU Power Connector

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

fullofzen

Distinguished
Jan 25, 2011
280
0
18,810
In a few months, I'm planning to pick up a Asus P8P67 Pro and a i5-2500k. My existing power supply (PCP&C Silencer 470) has only a 4-pin CPU power connector. I have done several searches and many have asked whether this will do. What I haven't been able to find is someone saying "Yes, I'm using the 4-pin cpu connector with the 2500k and it's working just fine."

For its part, Asus' P8P67 manual suggests that either an 8-pin or 4-pin is acceptable, but it's just a picture showing how to do it in either scenario. There is no documentation.

To cut unwanted responses off at the path, understand that 470w is not a lot. However, since the i5-2500k is overkill for almost any game I'd want to play in the next few years I'm not looking to overclock immediately, so the Silencer 470 will last me for a while.

So the question: Is anyone currently powering a 2500k or 2600k (or any of the other Sandy Bridge i5s or i7s) with only the 4-pin CPU power connector?
 
Solution
Okay, NO I don't have a 2500 or 2600......

BUT, let me try to explain the electrical engineering principles behind the 4 wire vs 8 wire debate.....

If the power supply can produce the power (Amps...)

Then bottleneck is NOT the for 4 wires running from the power supply to the motherboard. The true bottle neck is the trace on the PCB itself.

Resistance is the culprit here and the heavy gauge wires from the power supply are VERY low in resistance. On the other hand, the flat, thin, metal traces on the circuit board are relatively very HIGH in resistance.

(this is true since the equation for resistance is heavily dependent upon the surface area of the conductor....)

So, the $5 four pin to eight pin adapter in NOT a patch or a...
One thing I will add is that after buying a new PSU (OCZ Fatal1ty 750W) and RMA'ing the motherboard I need far less volts for the same overclock.
Previously I needed ~1.45v to get 4.6ghz, now the same CPU is perfectly stable at 1.25v.
Whether this is a consequence of the smoother current I don't know but it's the same CPU and the same model motherboard so it would suggest that a proper 8 pin connector on a dedicated PSU rail is better than a shared single rail with a 4 pin connector when it comes to CPU voltage requirements.
 
I think it is more likely that your acknowledgely-cheap no-name power supply shorted temporarily and that's what fried your motherboard. I find it hard to believe that there would be a sudden failure of the motherboard lanes related to using a 4-pin connector. Maybe a failure over years with BSODs every so often, but a catastrophic failure? That doesn't make sense.

I'm going to say that that was probably your PSU itself causing trouble. There's no other way to explain it.
 
jb6684 explaination was excellent and true. But there is a worst culprit than the circuit board traces and the primary reason for adding the 2nd pair of 4 wires to the ATX connector. And that is the connector itself. The contact resistance between the Male/female pins is much higher than either the wires or the circuitboard traces and therefore the highest IsqR Loss - all in the form of Heat.
Quote from CtJackal: "...When I removed the 4 pin plug there were scorch marks actually inside the socket"
I've seen a couple of other posts indicating the the plastic around the connector has melted.

This resistance also has a problem with oxidation over time (oxidation increases the resistance). Another problem deals with the quality, either the MB connector or the PSU connector which results in a poor fit and therefore higher contact resistance. Side comment: Silver is a better conductor, but never used as it is the worst for Oxidation, Copper 2nd best conductor, but also has an oxidation problem, gold plated - Gold is a 3rd rate conductor, but has 2 advantages - Slower oxidation and it is "slippery" meaning that contact pressure can be increased and still allow matting for us non 800 Lb gorillias.

Before I carry on. jaquith Has a good point - it's how the MB is laid out, Does each trace go to a dedicated VRM, or are the traces Shorted, either at the VRMs, or at the Socket - prefered. Easy to tell Just ohms check the pins at the ATX connector, if all red wires read 0 ohms to each other, they're short at one end or both.

For CPUs under 120 W (10 amps on 12V line), excluding what jaquith indicated on Phases, the 4 pin is fine, 8 pin better as IR loss at the connector is lower. The adaptor is fine also. The advantage of true 8 pin connection is that there is "Slightly" higer voltage at the VRMs, they do not have to work as hard - Very small positive, but as jaquith pointed out - ANY improvement is better.

For CPUs > 120 Watts eight pins should be used, And here I will disagree with jb6684 "EXACTLY like a "true 8 pin connector"", Not true, Yes at the motherboard it will see a "TRUE" 8 pin connection, ie the total current thru each pin will be I(total)/4. However you have added injury to insult. You still have the 4 pin connection, not on MB but at the PSU connector (injury) and you have added another connector with the added contact resistance (the Insult).
Hope I don't fall in your catagory of "(anyone who tells you different has only the tiny fractions in the 4th or 5th decimal place to back them up....)"
 

TRENDING THREADS