Those timings are quite tight, although frequency is low.

tRAS = tCL + tRCD + tRP (or sometimes tCL + tRCD + 2) - first use conservative formula, if it works tighten tRAS by 1 and memtest again. 3-4-4-8 is not right.
tRC = tRAS + tRP

You can try conservative timings 3-3-3-9 and Command Rate 2T. If it has problems raise DRAM voltage by 0.1V. More shouldn't be needed. Some BIOSes may also allow configuring memory refresh, start by following chip spec first.

Most likely you will have to follow advice in the thread and recap the board.

Sorry, 3-4-4-11, not 8.

What keyboard are you using? USB or PS/2? Try a PS/2 keyboard. AF usually means power supply delivery/memory issues on Abit boards. Try a CPU with low TDP - a Duron for example.
It may be a stupid question, but have you checked for the board on an open test bench, not installed in a case?

PS2 always. With AF I tried various different BIOSes and 3 BIOS ROMs. Heat on SB seems to be excesive on AN7 with AF error code. All tests are being done on the table. Unfortunately I have only Bartona (3 unlocked and 2 mobile). Have you guys ever tried to replace SB (MCP-T)?
Those 2 Abit boards drive me insane…:(

SlowA wrote on 2026-08-07, 06:01:

I’ve put them as on user manual. This board looks like it came from the factory yesterday. I’ll focus on Asus now, Abit AN7 (Error code AF) and this NF7-S give me way too much problems right now. 🙁

Thx. Weird, if the jumpers are set, something else is broken. Maybe a trace or a smd part.
Postcode AF looks like a RAM problem to me. 🙁 Contacts in the dram slot are clean?

SlowA wrote on 2026-08-04, 15:53:

I have another A7N8X, but it is deluxe, so BIOSes are different 🙁

Several A7N8X are cross-flashable. FOr example A7N8X-X (256kb bios chip) will POST and work flawlessly with a A7N8X v2.0 or A7N8X v2.0 Deluxe bios (both 512kb). You only need to replace the bios chip, as the -X only has a 2mbit (256kb) chip. Even the -E Deluxe bios will work on regular v2.0 boards. What you can't mix is: -XE as this is basically an Asrock K7NF2-RAID board (those two are also cross-flashable). And the older non-ultra 400 boards (v1.xx) won't run with a v2.0 bios and vice versa. So if your other A7N8X Deluxe is v2.0, then the bios chip should work on the regular A7N8X v2.0.

Regarding your ABITs:
The NF7 is extremely picky when it comes to bad caps, especially the 6.3V 1500uF between the AGP and first PCI slot. That one is probably dead and will cause issues like instability or even a "dead" board.

And a general question regarding your settings:
If you're trying DDR400, is this with FSB200 (so 1:1) or are you trying to run the ram async to the FSB? Always keep ram on nforce 2 in sync (1:1) to the FSB. Everything else will be slow and cause instabilities. So FSB133 + DDR266, FSB166 + DDR333 and FSB200 + DDR400.

Ram settings aren't important, 3-4-4-8 will work. 2-2-2-5 or 2-2-2-11 will work. 2.5-3-3-8 will also work. You can't set 2T in bios (bios flash required for this) and running 2T is pointless on NF2. Stay at CPC on (1T).

Thanks guys. i will try to recap NF7-S because I cannot find another one. I will try to flash BIOS on Asus with deluxe, because it already had (strange!) 1012 deluxe flashed prior to my flash. As for AN7, I tried everything I know. I would Like to revive it so much.

Cant help with the abit.
If you have a thermal camera it's important to remember that SMD caps can also go bad.
Here's the asus:

Guys, what do you think of these Gemini responses regarding the Abit NF7-S (AN7 will be the donor board):

REPAIR 1: How to restore stability at DDR400 (200 MHz FSB)
The 166 MHz wall is caused by electrical noise (ripple) in the voltage regulation section of the memory and chipset. Even though the capacitors look flat visually, they have lost their capacity (dried out).
Complete replacement (Recap) of critical capacitors:
You must desolder and replace specific capacitors on the board.
Do not change everything randomly; focus on these groups instead:Around the RAM slots: Locate the electrolytic capacitors that filter VDDR (memory voltage). These are typically three or four capacitors rated at 1000µF / 6.3V (or 10V).Around the Northbridge: Locate the capacitors responsible for VDD (chipset voltage). They are placed right next to the chipset heatsink and the AGP slot.
Buy exclusively Low-ESR (low equivalent series resistance) capacitors from reputable brands like Panasonic (FR or FM series), Nichicon, or Rubycon.

REPAIR 2: How to fix the onboard audio
Since the sound has completely vanished across all frequencies, we must determine which component in the audio section (bottom-left corner of the board, near the rear jacks) has burned out.
Checking the Realtek chip (ALC650 or ALC655):Power on the board and carefully place your finger on the small, square Realtek chip.
Critical diagnosis: If this chip starts heating up rapidly or becomes burning hot within 30 seconds, it has an internal short circuit. It is blocking the sound and dragging down the voltage.

How to fix: You will need to desolder it using a hot air rework station. You can salvage a replacement chip (ALC650 or ALC655) from almost any other broken motherboard of that era. Soldering a new chip with hot air and flux is a routine task for any electronics or phone repair technician.
Replacing the audio capacitors:
Right next to the Realtek chip and the audio jacks, there are smaller electrolytic capacitors (usually 10µF, 22µF, or 100µF rated at 16V/25V). They serve to pass the analog audio signal to the outputs. If one of them ruptured under the increased voltage testing, the audio line is broken.

Sound chips don't usually die.
IF yours was dead, it would be the first I've ever heard of.
Power is the most likely issue.

Caps always die. Count them and their ratings.

"REPAIR 1" sounds coherent to me.

SlowA wrote on 2026-08-08, 07:42:

i will try to recap NF7-S because I cannot find another one.

This is the way if you feel the mb is worth saving. And that is pretty much the only way. Or you can of course remove the caps, measure capacitance+ESR and see if the caps are the problem. This way you can save some money, if it turns out that the caps are still good.

But as for the earlier posts about the caps, I really wish that capacitor ”diagnostics” based on how they look alone would stop, or more like change how it should be done. It should work like this: leaky or bulged caps are a sign that they are bad, but the opposite isn’t true. If cap looks normal, it doesn’t mean that the cap is good.

I’ve changed hundreds and hundreds of electrilytic capacitors on motherboards, PSUs and other stuff and I often tend to measure some of the old caps I remove to see what their condition is. I’ve seen so many caps that look perfectly fine, but are still completely out of spec that it isn’t even funny anymore. They aren’t an outlier or an oddity, it is a significant percentage of old capacitors.

As for AN7, I tried everything I know. I would Like to revive it so much.

I had both NF7-S V2 and AN7 back in the day and AN7 sucked, NF7-S V2 was amazing. I returned AN7 after couple of days when I couldn’t get it running stable with two ram sticks stock clocks. It is possible that it was early BIOS problem as I got the board immediately after it was released. On paper it was a winner and I expected it to improve on NF7-S boards, but it was a complete opposite. Dude in the computer store said that I wasn’t the first that returned it, so I got the impression that it didn’t got a very good start generally. It never managed to reach the fame or legendary status of NF7-S V2 and probably wasn’t a sales success for abit. For that reason it is much rarer motherboard compared to NF7-S V2, so I guess that is at least something.

I would gladly take one to testing nowadays to see if it still blows so badly, or did things get better with perhaps improved bios.

Thank you for the info. I will definitely recap NF7-S just be sure that I tried everything. AN7 is dead, I guess.
I'll try next week to put the BIOS chip from the A7N8X Deluxe 2.0 into the A7N8X. I am still waiting for the PCI POST card...
Still, I cannot understand the problem with sound. I never heard someone had a hardware problem with sound on Nforce2.

CharlieFoxtrot wrote on 2026-08-09, 07:42:

... leaky or bulged caps are a sign that they are bad, but the opposite isn’t true. If cap looks normal, it doesn’t mean that the cap is good ...

^ Words of wisdom. 👍

CharlieFoxtrot wrote on 2026-08-09, 07:42:

... I’ve changed hundreds and hundreds of electrilytic capacitors on motherboards, PSUs and other stuff and I often tend to measure some of the old caps I remove to see what their condition is. I’ve seen so many caps that look perfectly fine, but are still completely out of spec that it isn’t even funny anymore. They aren’t an outlier or an oddity, it is a significant percentage of old capacitors ...

From your experience, what's the approximate percentage of caps that get completely out of spec without showing any external visual signs of it?

TELVM wrote on 2026-08-09, 15:44:

From your experience, what's the approximate percentage of caps that get completely out of spec without showing any external visual signs of it?

It is difficult to say, my observation just is that they are not uncommon. First, I often make more or less full shotgun replacments, that is I change everything or at least large majority of capacitors. I don’t still often measure every cap I remove, often I check caps that I think are most prone to failing and important, such as VRM caps or main filter caps on PSU secondary side.

If I’d had to make an estimate, of all the failed caps I have encountered, I’d say it is at least around 10% that have failed without any visual signs. Most likely it still higher as I have bias for low esr stuff for my measurements, which are more likely prone to swelling/leaking than general use caps, which gradually just dry out or at least the chemistry fails without pressure build up. One thing to note, often the capacitance might still be in spec, but ESR isn’t. Sometimes these caps are completely dead, I can’t get any kind of measurement. I’ve even measured leaked caps that still show good capacitance, but ESR has shot up. These caps are wildly unstable, although one parameter still seems to be fine. So your basic component tester which can’t measure ESR might not even show any issues.

By the way, if this is an MCP-T south bridge with SoundStorm, the audio processing is inside it. Realtek simply acts as a DAC.

shevalier wrote on Yesterday, 10:29:

By the way, if this is an MCP-T south bridge with SoundStorm, the audio processing is inside it. Realtek simply acts as a DAC.

Yes, Both NF7-S and AN7 have Soundstorm. NF7 without S doesn’t have it.

shevalier wrote on Yesterday, 10:29:

By the way, if this is an MCP-T south bridge with SoundStorm, the audio processing is inside it. Realtek simply acts as a DAC.

That is correct, that is why I am not sure what could be the problem with sound. Maybe I will reball MCP-T, too, just to be sure.

Lol
One function isn't working so you leap directly to reballing the hardest chip of all without even poking around at the dac, or doing anything else.
It's probably a dud cap on that circuit too.
But go ahead and reball it 🤣

I hope recap will fix my problems... Today I tried once more every option with VDD (1.7V) and VDIMM (2.9V), it simply refused to go over 173MHz no matter what timings. I used brand-new sets of genuine BH5.