Intel Bad Axe 2 (D975XBX2/KR)
Oct 28, 2022 13:50:05 GMT -5
Macsbeach98, ShrimpBrime, and 4 more like this
Post by MachineLearning on Oct 28, 2022 13:50:05 GMT -5
I present to you: the Intel D975XBX2KR, AKA the Bad Axe 2. 
As you can see, this is primarily an electrolytic capacitor ordeal. Capacitor brands used here include:
- Rubycon
- Nichicon
- Nippon Chemicon
- SamXon
- Fujitsu
- Sanyo
I didn't notice any manufacturing date listed on the PCB itself, nor a layer count. I can safely assume that it's either a 4- or 6-layer design, probably 4. In all likelihood, this board's OEM is Foxconn (or maybe Pegatron).
The northbridge's production code is Wk39/2007, and the southbridge's is Wk29/2007. This leads me to believe that it was produced in Q4 2007, and probably sold in H1 2008 - very late production! i975X would have been over 2 years old at that point. Its revision reflects this, being the latest I know of - 509 - as well as the -KR designation (which I'm still unclear as to what exactly it means).
The board has similar markings to nVidia reference motherboard PCBs, in that there are graduated markings which organize the board into segments: alphabetically on the left and right, and numerically on the top and bottom (as oriented above).
Example:
EVGA 750i SLI FTW for comparison:
Audio is handled by the Sigmatel STAC9274, branded as IDT.
Networking is, naturally, with Intel, my preferred choice for both WAN & LAN. This is the 1Gbit/s 130nm Intel 82573L.
The VRM controller is the Analog Devices ADP3189, a true 5-phase controller operating in the aforementioned configuration.
VRM cooling is done by those five small blue aluminum heatsinks surrounding the LGA775 socket area. They were stuck on using thermal adhesive at the factory, and since I don't have any of my own, they will stay as-is for the forseeable future. Therefore, I cannot tell which MOSFETs lay beneath, but I can figure out the number of them.
This is the same phase in both pictures. 2 high-side, 1 low-side MOSFET. You can see the mounting pads for the second low-side capacitor in the second photo.
The NB & SB are both cooled by aluminum heatsinks, which attach to the motherboard via soldered metal hooks. I believe that both heatsinks are using single-side adhesive fiberglass-reinforced phase-changing thermal pads ( lol
). It takes some more serious stuff than isopropyl alcohol, so I cleaned the actual chips as much as I could and applied some new MX-2. Unfortunately, what's left on the aluminum is not coming off without gouging it out.
Now... overclocking.
479.66MHz on D975XBX2KR
--- #2 for i975X overall!
CPU-Z Validation
This required 1.7v vMCH, 1.4v vFSB, and PCIe clock of 140MHz. The board is actually capable of over 150MHz PCIe when in-OS (SetFSB PLL SLG505YC56DT), and increasing it seemed to make higher FSBs go a bit easier. Without increasing PCIe, the maximum FSB is about 3-5MHz lower. In order to get any higher, I either need:
- New Conroe samples
- Wolfdale; Supported processors only include 65nm - no 45nm CPUs are supported. This requires a BIOS mod, which I currently don't have the skills to do - and I've read that this BIOS is unmoddable (whether or not that's actually true)
- vMCH volt mod. The resources for this are currently lost to me, all dead links.
Before the board came in, I read 124 pages of this Xtremesystems thread so I could be as familiar as possible with it (I will finish the rest, don't worry ). One person even made it to 500MHz FSB (phase-change) on this board, and I noticed another running @ 1200MHz DDR2. There's also a lot of repeated or known information, but it's the absolute best real-life-user thread that I could find on this motherboard. And I'm glad I read it, because I would have torn all of my hair out otherwise. There are a ton of quirks, too many for me to list them all. I bolded the most important issues IMO, alongside other observations:
- A Watchdog Timer that can only be disabled by going into a special diagnostic BIOS (enabled=system goes on the fritz when straying from stock)
- The inability to change multipliers on non-Extreme CPUs, in any direction(!!!)
- Failed POST recovery (i.e. during OC) is not in Intel's lexicon
- Only can adjust tCL-tRCD-tRP-tRAS in-BIOS
- vCore vDroop @ 1.4v: -25mV set @ idle, another -9mV under load (very good actually!); Advanced Power Slope set to Enabled
- vMCH vDroop -20mV set (not bad, at least it's consistent)
- Flaky onboard USBs (unknown if due to age), using VLI USB3.0 AIC & PS/2
- MCH needs >1.6v past 460MHz
- CLR_CMOS solder points on the board, yet not installed(??), shown below
- After a legitimate CMOS clear (=battery removal or death) or CPU swap, default BIOS settings are 800MHz @ SPD timings & 1.84v. This will not boot with most kits!
- A result of the previous two issues is another problem: the only surefire way to get into BIOS after failed OC is to switch the jumper into Diagnostic mode.
- vCore maximum is 1.6v without vMods as a limitation of the VRM voltage controller
- vMCH maximum is 1.7v without vMods (lost to time AFAIK
)
- Single BIOS chip soldered to the motherboard
I want to express my huge love for Intel's documentation. It's quite possibly second to none. The manual for this board is 78 pages of beautiful, painstaking detail, and that's just English (except for regulatory translations toward the end). The only thing you could really ask for is a Boardviewer file. Whoever makes the documentation at Intel loves their job, and I love them... lol.
As you can see, this is primarily an electrolytic capacitor ordeal. Capacitor brands used here include:
- Rubycon
- Nichicon
- Nippon Chemicon
- SamXon
- Fujitsu
- Sanyo
I didn't notice any manufacturing date listed on the PCB itself, nor a layer count. I can safely assume that it's either a 4- or 6-layer design, probably 4. In all likelihood, this board's OEM is Foxconn (or maybe Pegatron).
The northbridge's production code is Wk39/2007, and the southbridge's is Wk29/2007. This leads me to believe that it was produced in Q4 2007, and probably sold in H1 2008 - very late production! i975X would have been over 2 years old at that point. Its revision reflects this, being the latest I know of - 509 - as well as the -KR designation (which I'm still unclear as to what exactly it means).
The board has similar markings to nVidia reference motherboard PCBs, in that there are graduated markings which organize the board into segments: alphabetically on the left and right, and numerically on the top and bottom (as oriented above).
Example:
EVGA 750i SLI FTW for comparison:
Audio is handled by the Sigmatel STAC9274, branded as IDT.
Networking is, naturally, with Intel, my preferred choice for both WAN & LAN. This is the 1Gbit/s 130nm Intel 82573L.
The VRM controller is the Analog Devices ADP3189, a true 5-phase controller operating in the aforementioned configuration.
VRM cooling is done by those five small blue aluminum heatsinks surrounding the LGA775 socket area. They were stuck on using thermal adhesive at the factory, and since I don't have any of my own, they will stay as-is for the forseeable future. Therefore, I cannot tell which MOSFETs lay beneath, but I can figure out the number of them.
This is the same phase in both pictures. 2 high-side, 1 low-side MOSFET. You can see the mounting pads for the second low-side capacitor in the second photo.
The NB & SB are both cooled by aluminum heatsinks, which attach to the motherboard via soldered metal hooks. I believe that both heatsinks are using single-side adhesive fiberglass-reinforced phase-changing thermal pads ( lol
). It takes some more serious stuff than isopropyl alcohol, so I cleaned the actual chips as much as I could and applied some new MX-2. Unfortunately, what's left on the aluminum is not coming off without gouging it out.Now... overclocking.
479.66MHz on D975XBX2KR
--- #2 for i975X overall!
CPU-Z Validation
This required 1.7v vMCH, 1.4v vFSB, and PCIe clock of 140MHz. The board is actually capable of over 150MHz PCIe when in-OS (SetFSB PLL SLG505YC56DT), and increasing it seemed to make higher FSBs go a bit easier. Without increasing PCIe, the maximum FSB is about 3-5MHz lower. In order to get any higher, I either need:
- New Conroe samples
- Wolfdale; Supported processors only include 65nm - no 45nm CPUs are supported. This requires a BIOS mod, which I currently don't have the skills to do - and I've read that this BIOS is unmoddable (whether or not that's actually true)
- vMCH volt mod. The resources for this are currently lost to me, all dead links.
Before the board came in, I read 124 pages of this Xtremesystems thread so I could be as familiar as possible with it (I will finish the rest, don't worry
). One person even made it to 500MHz FSB (phase-change) on this board, and I noticed another running @ 1200MHz DDR2. There's also a lot of repeated or known information, but it's the absolute best real-life-user thread that I could find on this motherboard. And I'm glad I read it, because I would have torn all of my hair out otherwise. There are a ton of quirks, too many for me to list them all. I bolded the most important issues IMO, alongside other observations:- A Watchdog Timer that can only be disabled by going into a special diagnostic BIOS (enabled=system goes on the fritz when straying from stock)
- The inability to change multipliers on non-Extreme CPUs, in any direction(!!!)
- Failed POST recovery (i.e. during OC) is not in Intel's lexicon
- Only can adjust tCL-tRCD-tRP-tRAS in-BIOS
- vCore vDroop @ 1.4v: -25mV set @ idle, another -9mV under load (very good actually!); Advanced Power Slope set to Enabled
- vMCH vDroop -20mV set (not bad, at least it's consistent)
- Flaky onboard USBs (unknown if due to age), using VLI USB3.0 AIC & PS/2
- MCH needs >1.6v past 460MHz
- CLR_CMOS solder points on the board, yet not installed(??), shown below
- After a legitimate CMOS clear (=battery removal or death) or CPU swap, default BIOS settings are 800MHz @ SPD timings & 1.84v. This will not boot with most kits!
- A result of the previous two issues is another problem: the only surefire way to get into BIOS after failed OC is to switch the jumper into Diagnostic mode.
- vCore maximum is 1.6v without vMods as a limitation of the VRM voltage controller
- vMCH maximum is 1.7v without vMods (lost to time AFAIK
)- Single BIOS chip soldered to the motherboard
I want to express my huge love for Intel's documentation. It's quite possibly second to none. The manual for this board is 78 pages of beautiful, painstaking detail, and that's just English (except for regulatory translations toward the end). The only thing you could really ask for is a Boardviewer file. Whoever makes the documentation at Intel loves their job, and I love them... lol.
