Jump to content
HWBOT Community Forums

Recommended Posts

Posted

Hi all

 

I wasn't lucky with Abit for sA benchmarking - they never survived more than a few days (had several ones, ones I remember were fully modded from moose83 and stock one from Ananerbe). I also had no luck with Epox boards. Many other boards either don't have mounting holes near socket or miss good/custom bioses. The board that was almost flawless for me is Asus A7N8X-E Deluxe - it has a bunch of modded bioses (thanks Scotty!) and I never had bios corruption issues with it. But it has a major flaw that is killing this board for most of the benchers - no 12V cpu power connector, so it uses 5V rail for cpu vrm. I was using external vrm to avoid this problem but this approach also have certain problems and is not always available for many of us. Anyway, with NB mod, this board is able to hit >250MHz fsb with tight timings which makes it strong alternative for NF7/AN7.

Yesterday I made a simple mod to power cpu vrm from 12V. All you need is to remove/replace stock 6.3v capacitors and coil, solder 12V rail from PSU to the input of VRM:

IMG_2018-12-19_003932.thumb.jpg.ae690867344f0502284041b8e4fefcce.jpg

 

I have used 12V connector desoldered from the dead mobo and made a ground connection from connector as well:

IMG_2018-12-19_012339.thumb.jpg.b891b1b385e87e4a1b9692a1417f2067.jpg

IMG_2018-12-19_012352.thumb.jpg.6dd893fb27f5aa3d34e6bcb005ec3c3d.jpg

 

I have tested it to boot 200x11 yesterday and the mosfets are cool (and it should be quite easy to cover these with radiator). Further steps are NB/SB cooling, Vcore mod, Vnb mod, Vdimm mod which are well known. I will also resolder a 12V connector to a different location as the above shown version was just the fastest one for testing purposes. Although the board is 2-phase design, mosfets are powerful enough (I was benching stock board with 2700MHz 2V barton and the weak point was the 5V rail of my PSU) so current should not be an issue even with higher overclocks. Anyway, A7N8X-E should be WAY easier to find and 5V rail issues will no longer be a limit. Good luck with this nice board :)

 

 

  • Like 6
  • Thanks 6
  • Crew
Posted

Nice mod. Would be great to see voltage ripples with oscilloscope or at least compare overclock before and after. Have you checked temps of upper mosfets? They should be opened 2,4x times less resulting in more ripples. Could be tuned out with lower ESR caps and PWM overclocking (never heard of anyone doing this though :D).

Posted
2 hours ago, Antinomy said:

Would be great to see voltage ripples with oscilloscope

Will try to do that with cheap 200Khz osc. Should also go with SS this weekend as all other mods (except cap replacement) are now in place.

 

 

  • 4 weeks later...
Posted

Okay, I have tested this mod. Setup is as follows:

Corsair XMS3200LL, 2x256MB BH-5 based - capable of 268MHz with my FX57.

Random A-XP 2500+

A7N8X-E Deluxe (3.5Vdimm via hardmod, 1.9Vnb via hardmod, 2.1Vcpu via hardmod) - ceramics/tantals added to socket and nb (this was needed to enhance stability a bit), small radiators are on top of cpu mosfets as they got hot (not extreme temperatures though). Air cooling for NB and ram. Bios used is 1013-1T from Trats with dfi 4-19 romsips.

20020103-012431.thumb.png.9ed92856d15638a811dda9dac875f040.png20020103-023606.thumb.png.dee0d568b6353b304faad61e3d61d038.png20020103-024602.thumb.png.69d412bacd6674365db3f4044879d5c7.png

 

I wasn't yet able to get full stability above 260MHz FSB with 1T bios but the board still performs strong and I believe is on par with tweaked NF7/AN7 on same settings.

  • Like 3
  • 3 weeks later...
Posted (edited)

One thing to note is that the board has a built-in CPU OV protection of 2.0v's and will kick in if modded or not, at least that's been the history of it.

If this mod can truly negate the CPU OV protection that's a real win, every voltmod I've ever seen done before could never get around it. Speaking of such I still have a couple of these boards around, both can hit 240(+) on the bus with no voltmods done running the right chip but both are A7N8X Deluxe 2.0's, I don't currently own any of the -E boards.

BTW the ones I have are also a good alternative like the -E version, as long as it's a 2.0 board it should be OK.

And I'll just say good job TR! ?

Edited by Bones
  • Thanks 1
Posted

@sammurai will post pictures as soon as I get to the board

15 minutes ago, Bones said:

One thing to note is that the board has a built-in CPU OV protection of 2.0v's

So far, I had at least 2.15v measured. Probably just my board. Anyway, ovp-mod should not be hard to implement

Posted (edited)

Does this board have "boot from USB drive" option in bios?

That's the only thing I miss on NF7 and always use AN7 to install windows, then bench on it or transfer the ssd to NF7.

To me, most of the NF2 board are very close in terms of performance when modded (voltmod and bios). I don't find a big difference between AN7, NF7 v2, DFI Ultra-B, Soltek FRN-2/3, Epox 8RDA3+. Haven't tested any Asus, MSI or Gigabyte boards, though.

Edited by I.nfraR.ed
Posted
25 minutes ago, I.nfraR.ed said:

Does this board have "boot from USB drive" option in bios? 

Yes, boots fine with flash drives prepared in Winsetup2USB to install w2k/wxp/2k3 if boot device set to USB-HDD.

  • Thanks 1
Posted (edited)
17 minutes ago, TerraRaptor said:

Yes, boots fine with flash drives prepared in Winsetup2USB to install w2k/wxp/2k3 if boot device set to USB-HDD.

Thanks! That's a nice option, hate that NF7 does not support it and have to always switch boards.

So it's a very nice alternative, but requires more modding.

You can't easily find socket A boards and CPUs anymore. 3-4 years ago ebay was full with s.A parts, mostly from Chinese sellers. I regret not buying more CPU models that are now gone.

Not that it's a big problem - I have CPU models for months of benching, but still my collection is very far from complete...

PS: Here in Bulgaria most of these CPUs (and many more vintage parts) went through gold diggers :(

Edited by I.nfraR.ed
Posted
1 hour ago, I.nfraR.ed said:

To me, most of the NF2 board are very close in terms of performance when modded (voltmod and bios). I don't find a big difference between AN7, NF7 v2, DFI Ultra-B, Soltek FRN-2/3, Epox 8RDA3+. Haven't tested any Asus, MSI or Gigabyte boards, though.

Back then and out of the box ... DFI , ABIT and EPOX were "king of the hill"

Asus, MSI and Gigabyte boards , were never that good (mainly cause of lack of bios options and o/c specs).

 

44 minutes ago, I.nfraR.ed said:

PS: Here in Bulgaria most of these CPUs (and many more vintage parts) went through gold diggers :(

Unfortunately , not only in Bulgaria.

 

  • 6 months later...
Posted (edited)

Thank you for this amazing mod. I also did the Vdd mod and couldn't even get the Board stable at 208Mhz FSB. However after making the 12V Mod, i'm now able to boot straight into win at 245Mhz FSB and with 1.88V Vdd. Amazing!

I used 2x 10k for Vcore and Vmem and 2x 1mOhm for Vdd. I also didn't add the 4pin connector, but just soldered a wire from the VRM to the 20pin ATX connector. And last but not least: I also added two Panasonic FM 3300uF 16V instead of the removed three 3300uF 6,3v caps on the Cpu VRM.

I also got a question:

How far can i raise the Vmem without blowing the Ram VRM up? I went with 3.05V max. so far.

Edited by Tzk
Posted
2 hours ago, Tzk said:

How far can i raise the Vmem without blowing the Ram VRM up?

I used up to 3.8v with 2*256Mb bh5 for 1-2hour runs. You should care about RAM not vrm here imho.

  • Thanks 1
Posted (edited)

Nice to know. I'm running 2x256mb Infineon 6A so the sticks should handle the high(er) Vdimm without issues. Also good to know that you can raise the Vdimm past 3.3V, so hardmodding the Vdimm to the 3.3V rail doesn't help.

And here's how i did the mod. I'm using 10k fixed + 10k VR for Vdimm and Vcore. And 1m fixed + 1m VR for Vdd. That results in 3.05V VDimm max and 1.89V Vdd max so far.

EDIT: Changed the Vdimm resistor to 3kOhm + 10k VR. Still not enough Vdimm for the 6A to hit DDR500... I'm currently trying to get 245Mhz FSB stable.

Result with this Board: https://hwbot.org/submission/4217023_tzk_superpi___1m_athlon_xp_m_2600_(barton)_38sec_109ms

_DSC9106_1.jpg_DSC9107_1.jpg

_DSC9109_1.jpg_DSC9112_1.jpg

_DSC9114_1.jpg_DSC9117.JPG.c23417de249763d276b5b77b8a5a44e9.JPG

 

 

Edited by Tzk
  • Like 1
Posted

So, here's my final result for my board + mods on custom watercooling:

https://hwbot.org/submission/4217965_tzk_superpi___32m_athlon_xp_m_2600_(barton)_32min_9sec_31ms?recalculate=true

I ended up using 1,92V Vdd, 3,4V Vdimm and 1,85V Vcore. I tried these resistance ranges, maybe it's of help to anyone modding his/her A7N8X:

VDimm:

10kOhm resistor + 10kOhm VR = max 3,05V

5kOhm resistor + 10k VR = max 3,22V

3,3kOhm resistor + 10k VR = max 3,4V


Vdd:

1mOhm resistor + 1mOhm VR  = max 1,88V

500kOhm resistor + 1mOhm VR  = 2V++ possible, i stuck to 1,92V as more didn't help with FSB

  • Like 2
  • 1 month later...
Posted

I'm going to make a break until I get more cpus with this board but here is the latest I've done - 269 1:1 2-2-2-11 (which is actually the absolute limit of my ram even with s939)

https://hwbot.org/submission/4242478_terraraptor_superpi___1m_sempron_2200_(thoroughbred)_39sec_247ms

 

To break above 258-260MHz I change 2 registers in b0d0f3 with wpcredit - offset 66 & 68 set to C7. Voltages are Vcpu 1.85v, Vnb 2.05v, Vdimm 3.8v. The system is not stable at these clocks and it took me several attempts. 265-266Mhz are much better.  Another important notice - once you ran into instability you should power off completely, boot once with fsb 100 and then back to high fsb (I boot with 255 and then raise it with clockgen to the desired value with 1.5-2Mhz steps). Booting once with low fsb is essential - seems like chipset resets something inside with new strap.

I have tried to reproduce this result with NF7-S v2 using exact same hardware but no success - system is unstable above 261.5Mhz. 

 

Good luck all!

  • Like 1
  • Thanks 5
Posted

I've got a second A7N8X, this time a Deluxe v2.0, and i can't seem to make it past 245Mhz on both boards. I've tried up to 2.0V Vdd and 3.4V VDimm (plenty for Winbonds and 245Mhz) and still Memtest86+ locks up in seconds when running test #7.

So far i only did the 12V conversion, replaced the cpu caps (now 3300uF, 16V ultra low ESR), VDimm and Vdd mod. @TerraRaptor mentioned added ceramic caps on the cpu. Does it help with stability? If so which are needed and where did you add them?

I'm currently running the 1008 bios by trats with cpc enabled (1T). Anything else i'm missing?

Posted

I would solder some tantals into socket/on the back of the socket if missing . Also, I've added ceramics on every big electrolytic cap. However, this helps a little to be true. 

Try Trats 1013-1T.

Do your boards lock up beyond 245mhz or just get unstable? What is your max fsb validation so far?

  • Thanks 1
Posted (edited)

I'm already running Trats 1008mod3 (the newest bios for the A7N8X Deluxe v2.0). The 2nd board just hits a wall. I can get 240Mhz stable with just 1.82V Vdd but it just won't run stable at 245Mhz, regardless of Vdd. I tried up to 2.05V and different Rams. TCCD or Winbond didn't matter. I haven't tried validating it yet, but it freezes during POST at 250Mhz. It's rocksolid at 240Mhz though. I've thought about trying another bios just to make sure it's not holding me back...

I'll retest the first board and report back. If both boards show similar behaviour, then it might indeed be the BPL or romsips inside the bios.

Just to clarify: I'm currently on aircooling.

 

EDIT: Just switched to my other A7N8X (OEM "4MBO", 1008mod3-1T by Trats), same behaviour. 240Mhz is Memtest86+ stable, test #7 freezes at 245Mhz regardless of voltage. I'd say its not the board but a timing (romsip or BPL) which is holding me back. I got two additional boards (2x A7N8X-X) and may test them with trats 1010-1T bios, maybe it's the combination of 3.04BPL and dfi 6/19 romsips which just maxes out at 245Mhz.

I'll probably try the 1008mod2a by Trats which got different romsips. Maybe it's running better than the 1008mod3. I'll keep you posted ;)

EDIT2:

I think i found the difference... The A7N8X-E 1013-1T bios uses 3.17 BPL and DFI 6/19 Romsips. My A7N8X Deluxe (1008mod3-1T bios) and A7N8X-X (1010xmod2-1T bios) both use 3.04 BPL and 6/19 romsips. This might be the issue why i'm failing to make it past 245MHz ;)

Edited by Tzk
Posted

Time for a doublepost i guess...

I've now modded the 1008mod3 bios by Trats and tried BPL3.04 (stock by Trats), BPL3.17 and BPL3.19. All seem to crash at 245Mhz in Memtest test #7. Next up i'll replace the romsips with the ones Merlin uses on the Taipan bios and the Mantarays XT D26 by Tictac. If that won't do i'll try another cpu and memory sticks (only tested 2 cpus and 3 sticks so far).

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...