Hi cucgod,

Welcome as a Vogons member. 😀

A curious problem with this GA-5AA board. Measuring frequencies would indeed be helpful to know what’s going on.
Luckily the board still works with AGP graphics, so a DOS tool like CHKCPU.EXE can be used to measure the actual CPU frequency, the multiplier, and the FSB speed.

You can get CHKCPU via the first link in my signature or use the copy in Phil’s DOS Benchmark pack v1.6.
With the K6-2/400, boot to a real-mode DOS prompt, run Chkcpu /v and tell us what it detects.

Cheers, Jan

hi and thanks for your suggestion... which make good sense btw.

so i got the board out of my storage put an agp card in and fired it up:
and it was again posting with k6-2 400 ... WTF
now i wanted to enter BIOS and ... nothing
just sat there with "WAIT...."
i tried multiple times.. and even let it sit there for a few hours to boot anything or at least give an error about missing floppies etc... - nothing.
strg+alt+del not registering nor NUM

i measured the voltages for the cpu and they are stable and correct - so i think the caps are fine (which seems often to be a problem with this board?)

the next day:
i let the board connected to the PSU and fired it up.
cmos checksum bad (the battery should be quite new, but i will get a replacement nonetheless)
et voila, entering bios works, floppy a: not found error which is to be expected and booting win98 from cfcard just fine.
ran chkcpu and it shows the correct numbers 😒

maybe its just related to faulty bios? i mean i have seen weird behavior on my 486 machines, where a newbattery and clearing cmos fixed alot of glitches...

so my plan is:
replace the cmos battery just to be sure.

try lowering fsb to 66 and see if it works
try out pci vga cards to get a post screen
and then put everything back to the current state
just too see if i can trigger this weird behavior again

get analyzer card to check where it hangs if it still reproducable

in the longrun i wanted to try out an k6-III+ 400 1.6V on this board because it can do 2.0V..... which is not ideal but i think ppl did this without problems as long as theres a heatsink on?

cucgod wrote on 2026-02-10, 06:47:

maybe its just related to faulty bios? i mean i have seen weird behavior on my 486 machines, where a newbattery and clearing cmo […]
Show full quote

maybe its just related to faulty bios? i mean i have seen weird behavior on my 486 machines, where a newbattery and clearing cmos fixed alot of glitches...

so my plan is:
replace the cmos battery just to be sure.

try lowering fsb to 66 and see if it works
try out pci vga cards to get a post screen
and then put everything back to the current state
just too see if i can trigger this weird behavior again

get analyzer card to check where it hangs if it still reproducable

in the longrun i wanted to try out an k6-III+ 400 1.6V on this board because it can do 2.0V..... which is not ideal but i think ppl did this without problems as long as theres a heatsink on?

Well, the plot thickens…

When the board POSTs, a faulty BIOS is unlikely. These socket 7 BIOSes are compressed and contain a lot of checksum and integrity checks during bootup, before and after the decompression phase. If one of these checks fail, the BIOS will jump to Recovery mode directly, or simply hang early-on in the boot process.

A POST analyzer card is indeed a great tool and should tell you more when the fault recurs.

Using a 1.6V K6-III+/400 on 2.0V Vcore is fine, provided you use a heatsink with fan. These low voltage K6-2+/III+ models were designed for embedded fan-less applications, but are manufactured with the same 0.18 micron die structures as the faster 2.0V models and experience shows that they will run stable and faster at 2.0V Vcore, without breaking a sweat.

Note that when running a K6-2+/III+ on the GA-5AA, you need BIOS version F6 or F7b. For your 5AA Rev 2.2 board, the F6 BIOS is reported to work better than the F7b beta BIOS.

Jan

diagram Re: Help with a Gigabyte GA-5AA rev 2.2

hi,

quick update. i did have a little spare time to test it again.
booted up with k6-2 400 -> everything ok
switched to k6-3+ 400 1.6V and only changed SW5 on the motherboard to go down from 2.2V to 2.0V -> et voila POSTing as k6-iii 417 WITH PCI card (i grabbed one i havent used before)

jumper settings on board (i only had to turn SW5 to off for 2.0V)

hm after playing around with different cpus and voltages i think i shortened something... the voltage regulator only outputs 2.7 or 3.5v... 😒 so i guess i have to figure out whats going on here before further investigating my weird fsb issue...

cucgod wrote on 2026-02-17, 17:27:

hm after playing around with different cpus and voltages i think i shortened something... the voltage regulator only outputs 2.7 or 3.5v... 😒 so i guess i have to figure out whats going on here before further investigating my weird fsb issue...

ok at least i found the chip on the motherboard that seems to be outputting the voltage to the cpu on this board: https://www.alldatasheet.com/datasheet-pdf/vi … /HIP6008CB.html
even the table in this datasheet matches the voltage table on the motherboard... now if i only find out why there is only 2.7 or 3.5v available... i currently have no clue .. never been a hw guy nor have much clue about eletronics haha... but maybe someone has an idea... 😀

so i did a little measuring:
found pin 9 of the HIP6008CB so i can find vid0-3.
sw4->vid2
sw5->vid1
sw6->vid0
sw7->vid3

i also measured the switches and 4-7 is giving continuity when set to on. so they are working.

still poking around...

Hi cucgod,

Going over your previous replies, it is clear that the occasional CPU speed issues were due to the FSB jumping between 100 and 105MHz.
The pictures with the dipswitch and jumper settings show a correct setting for x4 and 100MHz. But looking at JP14, it seems not to make a good contact.

Checking the user manual, an intermittent contact of JP14 would perfectly explain the FSB jumping between 100 and 105MHz. Maybe this jumper was hit when the cooler clip snapped.

About your voltage regulator issue, first a friendly warning:
DO NOT use your nice K6-III+ CPU until the Vcore issue is resolved!! These K6plus CPUs will get damaged when the Vcore is 2.5V or higher!
A more robust dual-voltage CPU like the Pentium MMX is better suited for troubleshooting this circuit.

The HIP6008CB is indeed the Vcore controller, but the heavy work is done by the FDP6030L Power FET (Q8). This transistor is also the component where you can conveniently measure Vcore.

Set your multimeter to Voltage DC and attach the black lead to a ground point like one of the middle 4 pins of the AT power connector. With the red lead, you have to connect to the Drain of Q8. That is the FDP6030L's top metal lip that is soldered to the board.

I hope these steps help with troubleshooting the Vcore issue, and let us know what you measure with the various dipswitch settings.

Jan

hey jan!

i postponed the fsb issue until i can safely use the board again but i will consider your tip 😀

actually when the short happened i had either the k6-2 or the k6-3+ in the board - i cant remember... the power went out. even after turning it on again, it immediately turned itself off. dunno what that means but it made me reconsider turning it on again 😀

i pulled everything from the motherboard and all my current measurements are without cpu or any cards - i hope this does not effect the voltage reading.

as i was not sure where to measure voltages at first, i looked up the socket7 pins and used VCC2 (pin B2) to measure what i guess would be the Vcore voltage.
and then i figured it would probably be one of those 2 mosfets on the boards which would power this pin (having seen lots of youtube videos about board repairs 😀)

my Q8 is a nec k2984 (its hardly readable) but as it worked before i guess its a similar one to your mentioned fdp6030L.
thanks for your explanantion, i indeed had problems understanding what "drain" is...

so my readings are as follows (i will go down the voltage list starting by 2.0V (skipping auto) step by step, everytime turning the board off, changing dip switches and turning on, then measure)
2.0V-3.1V -> 5.05V
3.2V+3.3V -> it fluctuates between 5.05V and 5.06V
3.4V+3.5V -> 5.06V

not sure what todo with that...

cucgod wrote on 2026-02-18, 18:38:

all my current measurements are without cpu...

This can be complicated. There are many threads here about problems related to doing this. There are also other sources you can refer to, like alphabet videos.

Basically, Socket 7 can use many different CPUs, including single voltage CPUs as well as dual voltage CPUs. Different manufacturers have implemented this feature in various ways over the relatively long life of the socket standard. Some implementations use manual jumpers, auto detection of CPU pins, shorting of supplies through the CPU, separate VRM modules, BIOS settings, etc. On top of this, some power supplies will not regulate well with no load. Perfectly functioning boards have confused people with "false" readings. Also, damaged boards have proven difficult to diagnose without a better understanding.

nickles rust wrote on 2026-02-18, 18:57:

cucgod wrote on 2026-02-18, 18:38:

all my current measurements are without cpu...

This can be complicated. There are many threads here about problems related to doing this. There are also other sources you can refer to, like alphabet videos.

Basically, Socket 7 can use many different CPUs, including single voltage CPUs as well as dual voltage CPUs. Different manufacturers have implemented this feature in various ways over the relatively long life of the socket standard. Some implementations use manual jumpers, auto detection of CPU pins, shorting of supplies through the CPU, separate VRM modules, BIOS settings, etc. On top of this, some power supplies will not regulate well with no load. Perfectly functioning boards have confused people with "false" readings. Also, damaged boards have proven difficult to diagnose without a better understanding.

ok thank you for your input.
i grabbed a p1-200 mmx and set the board to 66mhz*3 and also to 2.8V vcore as stated on the board.
sw1 sw2 sw3 sw4 sw5 sw6 sw7 sw8
dip: off off on off off off on off

JP1/2/14 all to 2-3

the system is posting as p1 200 mmx

when i measure Q8: drain gives me 4.98V and the pin going to the coil gives me 2.91V

now setting it to 2.0V with sw4-7 set to off, it still posts and gives a reading on Q8: drain 5.02V pin to coil: 2.1V

so it seems to be 0.1V too high on both those settings

ok getting curious i switched to 2.2V on the board: sw4,sw6,sw7 off sw5 on and it gives me 2.31V with the p1 200 mmx

now trying out my amd k6-2 with this voltage - rest of the settings untouched. the system does not power on. so i guess i fried the k6-2... ? (which suxx but thank god it was not the k6-3 😀)

cucgod wrote on 2026-02-19, 08:59:

...and the pin going to the coil gives me 2.91V

That sounds like the switch output to the inductor in a buck converter. It will have a crazy fluctuating AC signal. The other end of the inductor should be connected to a bank of capacitors. Bulk capacitors are generally a good place to measure DC voltages in a running circuit, as these are meant to have a stable voltage.

en.wikipedia.org/wiki/Buck_converter

I'm glad you've got the board working as expected. Thanks for posting your results.