mobo phase power question..............

[yaplol]

ASUS Z97-PRO:
12 phases power
MSI Z97-GAMING7:
12 phases power
Gigabyte Z97X-GAMING7:
8 phases power

what's the phases power effect on mobo ?

 

[Jetster]

http://www.overclock.net/t/891696/a-short-power-phase-design-explanation

Someone correct me if I'm wrong but I believe Haswell handles power better then previous chips.

 

[a_ump]

yea only 8 phases. it must have better power efficiency. i remember like 24+X phases on older boards.

 

[GorbazTheDragon]

To an extent you can tell the quality of a VRM by the amount of phases, but it is a very inconsistent measure. There are dozens of other big factors and hundreds of smaller ones that will have quite a big effect on how well the VRM performs in efficiency, stability, etc.

By multiplying the amount of current each mosfet in a VRM can handle by the total amount of phases and you get the theoretical current limit of the VRM. So for example the Z97X-SOC Force has an 8 phase VRM with IR3553 power stages (Driver+mosfets). The IR3553 is rated at 40A maximum current, so the VRM has a theoretical maximum capacity of 320A. Now, realistically you are not going to draw much more than 100A in the most extreme scenarios, but having the headroom gives better stability in most cases and often also better efficiency.

Really you can't judge a VRM by the amount of phases it has, you need to take into account the mosfet types, the PWM, the configuration of the components and such.

The IR355x is probably the best mosfet out there for VRMs, they were heavily used by gigabyte on the 8 series motherboards.

Out of those three boards you mentioned the best VRM is the Asus followed by the MSi and lastly the Gigabyte. But the gigabyte one is a small board so it is hard to cram more phases in.

 

[MartinNixon0422]

http://www.overclock.net/t/891696/a-short-power-phase-design-explanation

Someone correct me if I'm wrong but I believe Haswell handles power better then previous chips.

Yes, you r right
but for overclocking, more power phases is good, because the load on each phase is reduced. that mean they generate less heat, so makes them less efficient and stable

 

[GorbazTheDragon]

Not necessarily.

An 8 phase board with IR3550s will be far better than an 8 phase doubled (usually marketed as 16 phase) board with SiRA12s... The IR3550 has far better current throughput (almost 250%) than the SiRA12.

Also, an 8 phase board with IR3550s will probably be better than a 6 phase doubled (12 marketing phases) board with the same mosfets. In this case because the effective switching frequency of the 6 phase board is lower than the 8 phase so the current output will be much more stable despite having a theoretical higher maximum throughput.

On top of that you would have to factor in differences in bulk capacitance, the capacitor choice, inductor choice, etc.

You can not judge a VRM solely on the amount of phases it has.

 

[Deleted member 24505]

Yes, you r right
but for overclocking, more power phases is good, because the load on each phase is reduced. that mean they generate less heat, so makes them more efficient and stable

Corrected

 

[yaplol]

so... more phases power is good for overclock?

 

[silkstone]

so... more phases power is good for overclock?

Sometimes

 

[GorbazTheDragon]

so... more phases power is good for overclock?

As I said it depends on other components in the VRM

 

[EarthDog]

For the most part, it really does not matter (on ambient cooling) how many phases you have on intel rigs (AMD octo's a COMPLETELY different story). A simple 4/6 phase (with heatsinks on it) will take Haswell into the low/mid 4's with no issues. If you are custom looping it, you may want to rest on the side of caution and go 8. As Gorbax said though, there are more than just the number of phases to count that matters.... but again for most users and especially with Haswell's addition of the FIVR, a 6 phase heatsinked power bits are plenty.

 

[Vario]

With the prior generation of Gigabyte, the phase count was lower because they used a higher quality phase than much of the competition. Not sure if thats still the case.

 

[EarthDog]

On some boards that was true... not all.

For example, their OC-Force used the higher quality bits etc, but the OC did not... at least lesser capacity on some parts IIRC...

there is also the 'doubling' concept as well, but that is for a more advanced discussion me thinks.

 

[GorbazTheDragon]

With the prior generation of Gigabyte, the phase count was lower because they used a higher quality phase than much of the competition. Not sure if thats still the case.

The Z87 boards from gigabyte nearly all used IR355x series power stages, so 4 phase power designs were completely adequate. My H87-D3H has 4 IR3553s, which is complete overkill for a board which doesn't support overclocking at all. You could easily run an 8 core xeon of one of those VRMs. Then look at the Z87x-UD4H... That has 16 IR3553s, which at 40A each means a maximum current of 640A, over 6 times what you can expect for heavy air overclocking...

The newer Gigabyte ones (Z97) are using a different design, not IR power stages, instead using separate drivers and mosfets. The current limit of the mosfets is 25A, so 4 phase designs are not capable of pushing quad cores past 4.5GHz. Take for example the Z97x-SLI. 6 would be a bare minimum with 25A mosfets and 8 would be where you can be within comfortable limits on air and water cooling.

The only gigabyte Z97 boards that use IR power stages are the SOC and SOC Force.

 

[eidairaman1]

You say overkill. I say durability.

 

[GorbazTheDragon]

No that's true, the IR power stages are by far the best for VRMs and IMO Gigabyte really went backwards with the Z97 boards.

 

[EarthDog]

No that's true, the IR power stages are by far the best for VRMs and IMO Gigabyte really went backwards with the Z97 boards.

+1. Absolutely.