Overclock.net banner
1 - 20 of 761 Posts

·
Premium Member
Joined
·
9,160 Posts
Discussion Starter · #1 ·
The VRM (Voltage Regulator Module) or VRD (Voltage Regulator Down) is a circuit of a Z97 motherboard that supplies the main input voltage to the CPU. It does this by converting the 12V ATX supply to the desired voltage with the PWM method and the use of a controller, drivers, mosfets, chokes and capacitors. The quality and ability of this circuit is crucial to the overclockability of a processor.

Quick and dirty how-to read the table:
  • Controller: You want this to be digital. It's higher precision and saves power. It may also report additional data on HWInfo.
  • Controller Phases: "Real" CPU phases. The more the true phases the steadier the supply. In most cases 4 or lower isn't considered great.
  • Doubler: If the board doubles the phases it gets higher current capacity, but it's better to have e.g. 8 real than 4 × 2 (given that the rest are equal).
  • Amps per 6.5 Watts: This is a normalized current a MOSFET package can carry based on 6.5W Power Loss, a realistic scenario. Aim for this to be high enough. e.g. 200W power delivery on 2.0VIN will require 200/2= 100Amps split on the total phases.
  • Total phase count: Shown by the amount of Inductors(Chokes).
  • RAM phase count: Independent from CPU VRD, supplying power to RAM. It's often important on extreme overclocking to be higher than 1.
(to see the image enlarged: Right click->open link in new tab.)


Note:
  • Also look at the updated Sin's list: http://sinhardware.com/index.php/vrm-list The original reason of the existence of this work was the lack of the updated list.
  • This is not official (I mainly tried to find a way to quickly evaluate good boards), I don't try to replace Sin (neither on lists or on knowledge), and part of the info is incomplete (but I do put an "?" on any info that isn't confirmed by sources based on photographic evidence).

Literature:

Common caveats:
  • The improvement from 6 to 8 true phases is lower than the improvement from 4 to 6 true phases (look at Sin's Hardware literature above for details)
  • Some gigabyte boards have mosfets that are package-limited to 25Amps but because of that external limit they have only 4W heat loss (at that current) instead of 6.5W the normalized list may imply.
  • You can multiply the number of total phases with the current capacity of low-side mosfets for a rough estimation of boards' total current ability.

Sources follow (whole of thread)
=================================

Gigabyte thread, and Sin posts in it: http://www.overclock.net/t/1487772/gigabyte-z97x-discussion-help-and-owners-club-soc-and-gaming-editions-also-included/0_100

Just discovered:

http://wingphoenix.pixnet.net/blog/post/40835563-asus-rog-maximus-vii-ranger-%E8%A9%95%E6%B8%AC

ASUS ROG Maximus VII Ranger:



ASUS Z97-A:

http://www.modders-inc.com/asus-z97-a-motherboard-review/2/
Quote:
As with all ASUS motherboards, the ASUS Z97-A power delivery is a digital design with a DIGI+ VRM EPU controller underneath the heatsink labeled ASP1252 with four drivers and eight pairs of NXP 6030DLB/4030DLA LF-PAKs . There are also two phases for the memory with two pairs of NXP 6030DLB/4030DLA LF-PAKs controlled by a Richtek 8120 instead of a DIGI+ controller unlike the higher end ASUS motherboards. This means that memory overclocking will not be available through ASUS AI Suite, only through the UEFI. There also won't be phase controls for the DRAM in the UEFI. The chokes are 1.2uF ferrite inductors while all the capacitors used are solid Japanese capacitors with a 5K rating at 105C. Compared to other motherboards with a 2000-hour rating at 105C (which last 200,000 hours at 65C typical), the 5K @ 105C rating translates to 500,000 hours at 65C.
 

·
Premium Member
Joined
·
9,160 Posts
Discussion Starter · #5 ·
  • The ASP1251 controller was probably a disguised IR3563A. Silly ASUS marketing.
  • The Intel voltage spec that controllers must adhere to is called VR12.5. That narrows down the google search for them enormously.

ASP1252 on new ASUS boards is possibly a disguised IR3563B, which adheres to VR12.5, IR3563A didn't.

later edit: it's not since 1252 has 10 pins instead of the 12 of the IR.
 

·
Premium Member
Joined
·
9,160 Posts

·
Registered
Joined
·
50 Posts
MSI Z97 MPOWER MAX AC





Quote:
Originally Posted by Modders Inc.
The power delivery design has also changed compared to the Z87 MPower MAX. Phase count is down from 20 to 12 utilizing a 6-phase Intersil ISL6388 VRD12.5 controller with six ISL6611A doublers at the back. ON-Semiconductor NTMFS4C08N (high-side) and NTMFS4C05N (low-side) Power MOSFETs are used instead of NXP PH7030AL/PH3830DL LF-PAKs that were used on the Z87 MPower MAX.

The inductors are 0.47uF "Super Ferrite" chokes and the surrounding capacitors on the CPU area use the same Tantalum-core Hi-c caps used on MSI's high-end motherboards which are convenient for LN2 overclockers because their low-profile design makes it very easy to insulate (with Dragon Skin or kneaded eraser) the surrounding CPU area. The rest of the capacitors are solid aluminum-core Low-ESR "Dark Cap" capacitors.
Source: http://www.modders-inc.com/msi-z97-mpower-max-ac-motherboard-review/2/
 

·
Premium Member
Joined
·
9,160 Posts
Discussion Starter · #13 ·

Gigabyte is cryptic on some models, but we can derive by combining from what we know externally, that "Precise digital CPU power design" means an IR digital controller.

Those are probably only the SOCs, the UD5H, the UD7 TH and the Gaming G1. Ironically, the Gaming G1 doesn't have it on the description but it's probably the beefiest.
 

·
Premium Member
Joined
·
9,160 Posts
Discussion Starter · #14 ·
Conflicting review on the Gigabyte Z97X-G1 Gaming WIFI. This one reports it's on IR3563B rather than IR3580 of the other review. They are either both right, someone is wrong, or the BK model differs.

http://www.overclockear.com/reviews/2231-3-gigabyte-z97x-gaming-g1-wifi-review.html google translate



Can anyone make the chip?

Wait, he probably did dun goofed, because that does look more like a 3580 than a IR3563B.

BTW, how a IR3563B looks like,



A 3580 posted by SonDa5 in Gigabyte's club,

 

·
Premium Member
Joined
·
9,160 Posts
Discussion Starter · #15 ·

·
Premium Member
Joined
·
9,160 Posts
Discussion Starter · #16 ·
I added "/Per Power Dissipation" for MOSFET Amps because it's considered good to have your Amps without having to go huge on Power loss/cooling in order to do so.

e.g. IR3553 40A/10.5W easy, CSD87350Q5D 40A/6.5W easier, PH6030AL/PH4030AL 79A/55W, 100A/69W not as easy.

no bump table update: ISL6388 reported to be on the MSI Z97 GAMING 7 http://pclab.pl/news57516.html
 

·
Premium Member
Joined
·
9,160 Posts
Discussion Starter · #17 ·
The quest for finding out what the heck is ASUS doing with renaming IR chips:

The old ASP1251 that was reported by a few sources to be really an IR3563A, next to a IR3563A:



The ASP1252 that I suspected it's an IR3563B:



Error. It's 10 vs 12 pins. It's not the same.

What about the ASP1257?



An IR3580 perhaps?



Or an IR3563B?

What are the alternatives?

An ISL6388 perhaps that MSI got?



Aha, 10 pins.

That might might mean the 1252 is actually an ISL. With the 1257 maybe being an IR 3563B or 3580

What does the ISL6379 used on Asrocks look like?



Bummer. It's 12 pin. ASP1257 chip can be that.

Alternatives to IR for digital controllers don't appear easily. The CHL8328 from older boards is 14 pin so it's out of the question.

But we derived:

IR Controllers are 12 pin so ASP1252 is not an IR. What ASP1252 or ASP1257 actually are is probably inconclusive with the information at hand.

note: The ISL controllers tend to be on 6 or 6x2 phase cards so they might be restricted to that. Hard to google their specs. I may continue tomorrow.
 

·
Premium Member
Joined
·
9,160 Posts
Discussion Starter · #18 ·
We've hit a potential Diamond in terms of low cost over high quality in an Asrock news thread. The Z97M OC Formula has the potential to have an IR 8 phase controller on the most advanced MOSFETs in the business, the 40A over 6.5W dissipation ones from Texas Instruments that are better even that the SOC Force's of GB because of their different dissipation stats. The board's info is not confirmed so I'll just put it with question marks based on the official marketing.
 

·
Registered
Joined
·
52 Posts
Quote:
Originally Posted by fateswarm View Post

We've hit a potential Diamond in terms of low cost over high quality in an Asrock news thread. The Z97M OC Formula has the potential to have an IR 8 phase controller on the most advanced MOSFETs in the business, the 40A over 6.5W dissipation ones from Texas Instruments that are better even that the SOC Force's of GB because of their different dissipation stats. The board's info is not confirmed so I'll just put it with question marks based on the official marketing.
They already used that IT Nexfet in Z87 Formula and also used in Asus Rog mobos Z87/97, and for the PWM i am sure that its a ISL but can be wrong hard to tell from the photos i seen.
 

·
Premium Member
Joined
·
9,160 Posts
Discussion Starter · #20 ·
Quote:
Originally Posted by Noway77 View Post

They already used that IT Nexfet in Z87 Formula and also used in Asus Rog mobos Z87/97, and for the PWM i am sure that its a ISL but can be wrong hard to tell from the photos i seen.
Correct on the Z87 MOSFETs. Sin's list is a great source for this. The controller of the Z87 was an IR according to his list, and it's my main source of the suspicion, though of course it's not confirmed.

edit: Important clarification: the list was on the EATX.
 
1 - 20 of 761 Posts
Top