K7S5A recap

[Yakusa]

Hi.
Board have:
7x2200uF/6,3V
3x1800uF/6,3V

Can i use straight 10x 2200uf like Panasonic EEU-FR0J222B?

This is my weird retro project.

Maybe any ideas about poly cap replacement?
APSC6R3ELL222MJB5T
https://www.digikey.be/product-detai...198-ND/3998084

[ChaosLegionnaire]

Re: K7S5A recap

where are those caps located on the board? are they on the cpu vrm? if so, looking at pictures of this board on the web, the vrm design seems to use a low switching frequency, so no poly-modding. the vrm may not like it. stick to the panny fr caps.

measure the original caps. make sure the replacement caps can fit there. the 2200uF 6.3v panny fr caps are 10x20mm in size. if u get the wrong size caps, they wont fit because the caps on the cpu vrm are located side by side with no room if u get the size wrong. if the caps are too tall, they may interfere with the installation of the cpu heatsink as well.

[eccerr0r]

Re: K7S5A recap

hmm... my ole' K7S5A 5.0 is still in cold storage, haven't used in probably at least half decade after using it for about that much at least.
Mine has all OST caps for the big ones, all 3.3mF/6.3V RLX. Uses an old KA7500C.
Rest of the caps are G-Luxon.

None have leaked as of yet.

[Yakusa]

Re: K7S5A recap

@ChaosLegionnaire it looks like this
good info about non poly ones. But question is if i can use all 2200 instead of 2200 + 1800 ones?

[ChaosLegionnaire]

Re: K7S5A recap

i dont see why not if they fit. 2200 may be slightly out of the +20% tolerance band but an extra 40uF out wont make a difference.

[eccerr0r]

Re: K7S5A recap

Interesting, mine has the KA7500 populated in the bottom 16-pin DIP but it's depop in this picture. And mine uses 3m3 capacitors instead of all those 1m8/2m2 capacitors.

[eccerr0r]

Re: K7S5A recap

Oops. Mine is the K7S5A Pro, which probably explains why they look different. Don't know why that KA7500 placement is there but depop however...

[momaka]

Re: K7S5A recap

I have a K7S5A here as well.

All the caps next to the CPU on mine (both high side and low side) are G-Luxon LZ, 6.3V, 2200 uF.

Panasonic FR 2200 uF should be absolutely fine for everything. I replaced two of the G-Luxon's on mine with Nichicon HZ, 6.3V, 2200 uF - a bit overkill, but only used two caps just because I didn't feel like doing a full recap and the original G-Luxon caps were, surprisingly still reading normal. So a polymod might work if you use at least 1800 uF polymers... but I frankly don't think it's worth it on this board, as it uses low-speed switching for the CPU VRM.

By the way, are the caps on your board bad or is this just a preventative recap?
If you have problems and the caps are not the issue, try the board with SDRAM instead of DDR. For some reason, mine won't post with DDR RAM at all (gets stuck on BIOS code "D3" on my POST card.)

Also, look up "HoneyX" BIOS and "CheapoMan" BIOS for this board - should unlock some extra options in BIOS, if you are interested.

[ala_borbe]

Re: K7S5A recap

hey im suprised to see this thread... im in need of info here as well... i got K7S5A v3.1 board in bad shape, caps ware ok, but it was filthy as hell... and it had sempron???? cpu in it...

after cleaning i put athlon 2400 (133fsb) in it (known good) and it turned on... yeee... but it only worked with 100/100 (fsb/mem),
no post with 133/133, which is strange cos in know the board is compatible with 133 and cpu is native 133...
after that i googled a bit and found out its problem with DDR... i god knows i tried at least 10 modules, form pc2100 to pc3200...
tried all moded/OC bioses.. all the same...
tried better psu (this one is known working but i said what a hell...), all the same
after that i put some SDR and turned on, set fsb to 133/133 and it woked WTF???

ok i set to reinstall windows 98... it froze at one moment, but after restart it finished install... driver install went ok....
but upon putting AGP graphic card and installing driver board would work for some time, like 1min or so and then froze...
i tested a lot of different graphic cards from 1X AGP to 8X AGP... all the same problem...

all temps, voltages reported by mobo are ok, but it freezes...
reading from the net i found out that KA7500 regulates voltage for chipset+AGP+MEM... but i dont have spare to change

sorry for long post, but what would you suggest to do?
i was thinking about recapping but i dont have caps now...
should i only recap around KA7500 or the whole board? caps seem ok, not bulged... can i use 3300 6.3v?
i think i have some panasonic, low esr somewhere...

[mariushm]

Re: K7S5A recap

The K76S5A is a shitty board in the first place... poor layout , poor bios , poor everything

The board won't benefit from super low series like Panasonic FR, but you can use them if that's what you have.

Looking at Yakusa's picture a few posts above

The 4 1800uF 6.3v caps tell you the VRM takes 5v from the psu and converts it to lower voltage.
So if the psu outputs a clean 5v, then these capacitors don't matter that much. You can use higher capacitance here, won't matter.
You can also use 10v or 16v rated capacitors if they're cheaper, as long as it's more than 5v.

The six 2200uF 6.3v caps above are in 3 groups of 2 caps.. basically the VRM has 3 phases, and each phase has 2 capacitors in parallel, in order to reduce ESR.

They used 2200uF 6.3v most likely because they were the tallest capacitors they could squeeze there and not affect heatsinks (vertical height is limited near the socket)

You could probably replace them with 1200uF or 1500uF 4v polymer capacitors if you want to. Can probably replace them with FM/FR series 1800uF 6.3v / 10v as FM/FR series probably have better ESR in the first place compared to whatever OST or "who cares" brand ECS used.

For example, tme.eu has kemet 1500uF / 6.3v with 14 mOhm ESR: https://www.tme.eu/en/details/a750ks...acitors/kemet/
4$ for 10 ... could easily replace all 10 capacitors with these instead of FR or FM series

[ala_borbe]

Re: K7S5A recap

oh cmon, the board is not that bad... plus i got it for free... it has decent performance, should be compatible, that nice universal AGP slot that can run my savage 4 and voodoo 3 with no problem. i test a lot of old vga cards, some go to collection, some for sale. main highlight is that AGP and both SDR/DDR support

now for topics...
i saw that cpu voltage was stable most of the time, plus i had problems with agp cards and ddr.... so i did not touch that vrm...
instead i replaced 1500uf/6.3v... two near chipset, for near memory slots, two next to agp slot....
put in athlon 2000 palomino core and 2x512ddr
it booted, set to 133
now running aida stress test for 20min.... so this comes as solved
for me guilty ware 8 caps

gona do come oc and some games to make sure its rock solid

tnx for advices

[momaka]

Re: K7S5A recap

oh cmon, the board is not that bad...

Agreed.
It was actually quite a popular board back in the day. And quite reliable, if capacitor issues are to be taken aside.

i saw that cpu voltage was stable most of the time, plus i had problems with agp cards and ddr.... so i did not touch that vrm...
instead i replaced 1500uf/6.3v... two near chipset, for near memory slots, two next to agp slot....
put in athlon 2000 palomino core and 2x512ddr
it booted, set to 133
now running aida stress test for 20min.... so this comes as solved
for me guilty ware 8 caps

Thanks for posting back your results and keep us updated if anything changes.
So it looks like it was the other caps that were the cause. That's quite strange, as this board uses linear regulators everywhere else. I figured none of those caps would matter too much if they went a bit higher ESR, due to them being used in a linear regulator circuits. But I guess not. I'll see if I can recap my board this or next week and report here as well if I have success with mine working again with DDR memory. Seems that's the only issue so far. It gets stuck on code "D3" with any DDR sticks (and I tried both newer PC2700 sticks and older "compatible" / low-density PC2100 sticks.)

gona do come oc and some games to make sure its rock solid

Awesome!
What games do you have in mind?

The 4 1800uF 6.3v caps tell you the VRM takes 5v from the psu and converts it to lower voltage.

Correct.
It's an oldschool s462 board. And like many of that era, it uses the 5V rail to generate CPU V_core.

The six 2200uF 6.3v caps above are in 3 groups of 2 caps.. basically the VRM has 3 phases, and each phase has 2 capacitors in parallel, in order to reduce ESR.

Incorrect on that one.

Although this board does appear to have 3 phases (as evident from the 3 inductors), the CPU VRM for this motherboard is NOT a 3-phase design. The K7S5A uses a single-phase, single-ended buck VRM design for the CPU power. All three upper MOSFETs of the CPU VRM are driven at the same time from the KA7500 PWM chip (though not directly - each MOSFET is driven from a buffer circuit implemented via 2 small SOT-23 transistors next to each MOSFET.) Also, this motherboard's CPU VRM does NOT have lower MOSFETs. It uses free-wheeling schottky diodes instead (a very inefficient and simple design.)

They used 2200uF 6.3v most likely because they were the tallest capacitors they could squeeze there and not affect heatsinks (vertical height is limited near the socket)

Not at all.
Those capacitors are way out of the way of any heatsink, so there really is no restriction on the height.

You could probably replace them with 1200uF or 1500uF 4v polymer capacitors if you want to.

While 1200-1500 uF polymers will likely work, I do NOT recommend it at all.

Again, this motherboard uses a very primitive CPU VRM design with a single-phase driven from a KA7500 PWM chip. (One could even argue this is just a tiny step-up from a 555 timer driven circuit. ) Because the KA7500 PWM IC has very low switching frequency (less than 50 KHz, likely, if even that), ESR does not matter as much as capacitance does in this circuit. So keeping the output capacitors to what they were originally is recommended. Going higher in capacitance won't hurt. But going lower in capacitance (for example, due to using polymers) may actually increase noise and lower over-clocking abilities (not that this board is known to have any, but still. )

So stick with 2200 uF or higher capacitance on the VRM output / CPU V_core. The ESR is almost irrelevant at the switching frequencies that the KA7500 uses... so any entry-level low ESR caps should work fine too. Though, Panasonic FR and FM are always a nice welcome upgrade. Rubycon ZLH are great too.

[Uranium-235]

Re: K7S5A recap

Brings back memories. Loved my S5A pared with a 1.33ghz T-bird & 512M DDR 266, also a geforce 3 ti200

used it for years, still have it. Caps did eventually bloat

Fun fact: Frys released a bundle with a K7S5A and a Duron 800. ALL the S5A's did not include the bios to support it. They had to do a lot of returns

[momaka]

Re: K7S5A recap

Fun fact: Frys released a bundle with a K7S5A and a Duron 800. ALL the S5A's did not include the bios to support it. They had to do a lot of returns

Really, no Duron 800 support?
I don't know if mine still has stock BIOS or not, but it's running fine with a 1.4 GHz Duron Applebred (Thoroughbred B). BIOS detects it as "Duron unknown", but applied speed and voltage are correct.

Speaking of which, I really should [re]enable the L2 cache on it some day. 96 KB of L2 - you've gotta be kidding me! Can you imagine this was the CPU in my first own daily use PC? At the time, I thought it was decent... until later I realized it was low-end even for its time. But it got the job done, though. So many good memories on that PC.

[Dan81]

Re: K7S5A recap

IIRC I still have a KA7500 Skt462 here, a ECS K7VZA though - KT133A.

I recapped mine before powering on, and flashed a Chaintech 7AJA2 BIOS on it as I had shutdown delays with the original BIOS. Still works fine to this day.

[momaka]

Re: K7S5A recap

I'll see if I can recap my board this or next week and report here as well if I have success with mine working again with DDR memory. Seems that's the only issue so far. It gets stuck on code "D3" with any DDR sticks

Well, I just did that today. I didn't really do a full recap, but I recapped all of the important rails (minus DDR Vtt and STB3.3V - both of which are linearly-regulated with enough caps in parallel that I don't think they'd be a problem).

That said, my K7S5A still won't accept DDR RAM! Oh, and POST code where it gets stuck is NOT "d3" as I mentioned above, but rather "dE". When that happens, the board sits on it for about 3-4 seconds, then resets (my POST card shows BUS activity), waits momentarily on "d3" for about a second, then gets stuck on "dE" again... and ad-infinitum. Also, when the board resets, the Caps Lock key unlocks and I can toggle it until motherboard freezes on "dE" again. So the motherboard is definitely trying to POST with the DDR sticks.

But SDRAM? -No problem! The board detected even a pair of high-density PC133 512 MB sticks.

You might say, maybe it's because I didn't do a full recap that I am still having issues. But I doubt that, because every single original capacitor I pulled from the board tested spot-on - even the really small 16V/100u and 25v/22u caps.

What I noticed, though, is that my DDR Vdd voltage is a bit low with DDR RAM: about 2.49V. While DDR RAM is rated for 2.5V, I rarely see motherboards giving anything less than 2.55V (and many Biostar motherboards even do 2.7V by default! ) Same goes for RAM Vtt voltage: it's a bit low at barely 1.2V. Normally, I see it on other boards at 1.25 to 1.3V (i.e. half of 2.55 or 2.6V) Makes me wonder if some component in the RAM detection circuit has drifted out of value. Given that the board uses a 1084 regulator for the RAM, I'm thinking I might try to manually drive the DDR voltage to see what happens. On the other hand, it's a cool classic board, so maybe not worth the risk. Might get some cheap China 512 MB sticks of SDRAM for it instead and call it a day.

Anyways, I'll try pulling the pics from my camera this week and posting a diagram with all of the voltage rails here. That should also serve as a starting point for anyone trying to troubleshoot this board with issues other than caps, too.

[Uranium-235]

Re: K7S5A recap

Really, no Duron 800 support?
I don't know if mine still has stock BIOS or not, but it's running fine with a 1.4 GHz Duron Applebred (Thoroughbred B). BIOS detects it as "Duron unknown", but applied speed and voltage are correct.

Speaking of which, I really should [re]enable the L2 cache on it some day. 96 KB of L2 - you've gotta be kidding me! Can you imagine this was the CPU in my first own daily use PC? At the time, I thought it was decent... until later I realized it was low-end even for its time. But it got the job done, though. So many good memories on that PC.

It has bios support for it later, but a lot of people didn't even know what it means to flash bios when they're building computers for the first time.

I played Tribes 2 on this thing for so many years. 'Twas the reason I got so many low grades in college

[momaka]

Re: K7S5A recap

I played Tribes 2 on this thing for so many years. 'Twas the reason I got so many low grades in college

Heh. Totally relate-able. For me it was mostly Half-Life 2 Deathmatch with CS Source on the side in high school... on that Duron 1.4 GHz CPU (with a Radeon 9200 SE / slow edition). I didn't get [too many] bad grades (as HS is a joke, really), but my sleep did suffer somewhat. And any poor performance was usually a result of my lack of sleep. So yeah...

[ala_borbe]

Re: K7S5A recap

sorry for resurecting old thread but i had same problem as @momaka with his board, not posting with DDR memory

latest version of bios does not work with some revisions of motherboard!
i flashed bios one eversion prior to latest and board posted with DDR as well as SDR

[momaka]

Re: K7S5A recap

sorry for resurecting old thread but i had same problem as @momaka with his board, not posting with DDR memory

latest version of bios does not work with some revisions of motherboard!
i flashed bios one eversion prior to latest and board posted with DDR as well as SDR

Hey, that's really good to know! Thanks for updating your progress.
Mine is still sitting on the shelf here just the way I left it last time I posted about it here - shame, I know. :\ This really gives me hope for mine, though. I need to finish recapping it one of these days so I can make something with it again. It should work great for a Windows 98 PC.

BTW, which BIOS version did you end up using exactly? I have several downloaded BIOSes for this board (some unofficial ones too that supposedly claim to unlock some OC options), so I'm still wondering which one I should try/use.