Well now with the P3S 1266 it's starting to get more interesting 🧐, unlike what I thought, I haven't seen from Phil's benches, a clear leap in performance, but this is absolutely not true, because there are many other Windows programs to do bench, and for the moment I haven't used any.
I did a short research 🔍, and I found something that could give an idea 💡 more precise, of what 😨 a PC like this can do, it seems that even Intel has deliberately interrupted production, to avoid embarrassing comparisons 😳 with the Pentium 4 1.3GHz, although in fact the P3S 1266 is among the best performing, or even in some the fastest.
But I'll have to complete the system update, it's missing a lot of RAM and a disk to load Windows, more of course a CD or DVD player 📀
For the RAM maybe 🤔 considering that the PC is the end of 2000 or 2001, you could put about 1 GB 🇬🇧, with a disk of about 32 GB 🇬🇧 and Windows XP, but maybe it would be more correct to use a 20 GB disk 🇬🇧 and Windows ME, hoping that ME is not limited to 512 MB like the 98.

Well I'll disconnect for a moment from the S.370, lately I fixed a Gigabyte GA-K8NF9 Ultra motherboard, for AMD 939 CPU, it's a socket that I used for years on the old Internet PC, initially I had an Athlon 64 3200 CPU, but later I noticed a certain slowdown (very annoying), and in short I replaced it with an Athlon 64 X2 3800, but now the PC has been stopped for some time ✋, it would need a change of HD, but at that point 🧵 I would say that the CPU could be replaced, because that PC will no longer go on the Internet, I would remove it to put it on the Gigabyte.
But that's not all, I saw that the 3800, you can easily bring it higher than 2000 MHz, some without problems even the 2500-2600, and maybe 🤔 at standard voltages you can even get to 2800 MHz, this means that you can match the performance of the best CPUs, without spending 🤑 big amounts.
For those who don't remember the frequencies, they are not the figure printed on the CPU, the 3800 is a 2000, the fastest is a 4800 that goes to 2400, so we are beyond the frequency of the fastest A64X2.

I made a list of all the capacitors present in three motherboards, I have to change some, but they are not very common, the simplest solution, would be to use those that come closest, with slightly higher capacity, or you have to order them, assuming you find the spare parts exactly the same (same capacity and voltage), the boards are these:
Intel D865PERL mPGA478,
MSI MS6163 VER:1 MS-6163VA Slot1,
MSI K7T Turbo VER:3 MS-6330 S.462.
On the Intel motherboard, there are strange 820uF 6.3V, strange because they usually have a diameter of 10 mm, here instead the diameter is 8 mm., they are therefore narrower and taller 🫡 than those of 10; there are also 560uF 4V, I have seen these in 775 cards, but I don't think there are other 478 boards that use them.
The MSI S.462 has in addition to normal capacitors, large electrolytic capacitors around the CPU, they even reach 4700uF 6.3V, I have a couple of similar ones in an ABIT MB, here there are even four; those instead of normal size, have a strange value, there are two 2700uF 6.3V, a fairly rare capacity value, which I think can be replaced without major problems with normal 3300uF, rather than making a large stock of 2700uF, even if they find them equal and in small quantities, they could be used in the future for other motherboards (equal?).
For the MSI Slot1, everything is quite normal, except for a couple of capacitors, a 1500uF is a 1000uF both 10V, they have a diameter of 10 mm, instead of the usual 8 mm, here I think that if there are no height problems, you can put the 8 mm ones. 😋
A value very present in MSI motherboards, is the 1000uF 6.3V, the dimensions are 8 in diameter and 14 in height (maybe 🤔 something less), despite having a certain number of replacement capacitors, changing them could zero my stocks, it is something to evaluate 🤔 if you have urgency in finishing the repairs of other boards.

I start the repair work on an ASUS P4B266 REV.2.01, the problem is already visible in the VRM CPU area, the large 3300uF 6.3V electrolytic capacitors, are swollen and some brown stuff has leaked, they are all to be replaced, I have the same replacements, they are identical, I hope 🤞 that in addition to this problem, there is nothing else, and the motherboard is working again, when the work is finished.
I had two strange problems, one ☝️ in removing the heat sink, the CPU is glued and even if the lever was at the bottom (closed), it came out of the socket, the pins all seem straight and so now I have to disconnect it, to insert it into the socket with the lever raised.
The other is in removing the coupling of the CPU heatsink, two of the four stops, they did not want to lift, insisting I still managed to lift them, and I was able to remove it, to facilitate the work, and to avoid accidentally damaging it with the soldering iron.
Soon I should remove the swollen electrolytic capacitors, and replace them with the replacement ones, I hope 🤞 wires all smooth.

The replacement of the electrolytic capacitors on the ASUS P4B266 REV.2.01, I finished it a little while ago, and all in all it was quite easy, of course 👌 some photos were not completely free, but working on it a little I released it, however now there is a test on the bench, as a function check.
Today it won't be possible to try it, on the bench I have several motherboards, for this reason, also because I already have a case ready 🏘, which I kept aside on purpose, I should first make room 🚀, or find another place to try this PC, so maybe 🤔 I'll mount everything on the case, and I'll try to start the full PC, instead of the single motherboard.
Of course I start with the bare minimum connected, that is, CPU RAM motherboard and provisional video card (ATI RAGE XL PCI 8MB), probably also the PCI post card, if all goes well 😌, I will move on to the almost final configuration, adding more RAM, a 64MB VGA AGP 8X GF4 MX440, for the sound card instead I will temporarily leave the integrated one, and I should add a PCI LAN card, and a USB 2 card. X PCI, because they are not present despite the card being prepared, only have the USB 1. X is always useful, but it is slow in transfer, finally if it is useful I could add a PCI SATA card, and finding one ☝️, the SPDIF bracket, with gold contacts.

Since I recently fixed the QDI Advance 10T motherboard, and that supports Tualatin, without any modification, I looked for another similar board that I have in repair, it is an MSI MS6368 VER:5, S.370 chipset VIA, N.B. VT8601T and S.B. VT82C686B, uATX format with three PCI slots and a CNR, unfortunately no AGP, but maybe you could add an ISA slot, the AGP is missing because the video is integrated as well as the audio, the RAM slots are only two.

The first job to do is to replace a damaged electrolytic capacitor, after that requires a fairly energetic cleaning, there is a layer of oily dirt (I don't know what it is, and it's in many areas of the MB, maybe a nice bath in soap and water, it would be done.

Dirty aside, it seems all in good condition, and I wouldn't mind making it work again, here I could use instead of the P3S, a 1200 MHz Celeron Tualatin, and try overclocking at 1600 MHz, raising the FSB from 100 to 133 MHz.

Finished cleaning the MSI MS-6368 motherboard, the result for now seems pretty good, maybe I should remove the BIOS chip, to see if there is still ️ some dirt underneath, for now it doesn't matter, I was interested in removing the superficial one, and at least the visible one, I removed it.

The work that remains to be done is to replace a damaged electrolytic capacitor at the top, it is a 1000uF 6.3V G-LUXON, the brand is what it is, but in this tab there is a strange mix between quality and non-quality capacitors, since the others are Rubycon and TEAPO, I proceed to remove the damaged one, from the next verification, I did well to change it, it measures 989.5uF and 1.54OHm, this last value is definitely higher than the approximately 0.2OHm of the new ones.

For the replacement I used a Rubycon 1000uF 6.3V, it's only slightly lower, but it's okay, it's one or two millimeters shorter.

So after checking and cleaning the BIOS chip socket, the card is ready for a boot test, here unfortunately not having the AGP, I will have to use the integrated VGA, but if there are no problems it will be fine, this card is the equivalent of an i810 with VIA chipset, I hope however that it is possible to disable the integrated VGA, so that you can use a PCI card type GF or equivalent ATI Radeon, I do not think that the 8MB PCI Rage XL is better than the integrated one, but this could be a problem only if you want to use The PC for GAME, for Home Office use instead the integrated video is already sufficient.

I started cleaning the ASROCK P4VT8 motherboard, the result is quite good, but there is still ️ some dirt to remove, especially in some places, although I have to say that most of the dirt has gone away, only in some areas it has remained, but maybe insisting it is possible that it will be removed.

Finished with the cleaning, I should start doing some repairs, you have to replace four swollen electrolytic capacitors, it's KZG from 3300uF 6.3V, these are quite common in motherboards, it may be a case but many motherboards use them, and often they are to be replaced, in this case all four need to be changed, but if some had been visually in excellent condition, I would have changed them anyway, because it is not said that checking them were ok.

Honestly, I don't know when I'll finish cleaning, but there are a couple of things to check, an SMD near the audio chip, which I think is partially missing, and the pins of a chip (probably I/O), which don't touch each other, are a little bent inwards.

Then I can start removing and replacing the four electrolytic capacitors, and it could easily work, before cleaning it was written ok, so it is possible that it was removed working, but we will only understand this by trying to start.

A motherboard that in the past I tried to fix, is an ASROCK 775i65G, it is interesting because it supports CPU Core2 Duo and has the AGP slot, normally we will have had a PCI-E 16X, it is also true that the video is integrated, so it can work very well without an AGP or PCI video card, both audio and LAN are integrated, in short there is everything indispensable (except serial ports), just add CPU and RAM, to have a working PC (in this case not obvious),.

What's wrong?

Unfortunately some of the 775 PINs were crooked, despite the fact that I straightened them, they did not come back perfectly straight, some are slightly displaced, others slightly higher or lower, in short, wanting to avoid straightening them again ️, you could disconnect the entire socket (LGA 775), and replace it with a voucher recovered from a scrap motherboard, why is it possible?

Because here it is not soldered by bubbles, but from the back of the motherboard, the problem may be precisely this to find the same one, then of course it will take a lot of patience, in welding 775 PIN, if they were only two or three crooked PINs, maybe by insisting you could be able to fix them, here we are a little further, about ten, and it is not at all easy to do it without a microscope.

I also noticed a couple of potential problems, a Mosfet (?) With signs of overheating around, in the RAM area, which should be checked, and the VGA connector with the missing metal part, this may be a damage caused by someone, who maybe had stuck the cable, and forced it, to extract it comfortably.

Another motherboard that I would have repaired, is a 478 socket, of the P4M2-MVB AZZA, in uATX format with two DDR RAM slots, the chipset is VIA P4M266/A, there is integrated video and audio, then we have an AGP slot and three PCI, to add other features or for a video card, one of the PCI can be useful if we want a LAN or a modem, or other types of receiver cards like Radio or TV.

It has the usual defect of swollen electrolytic capacitors, one is the usual 3300uF 6.3V, but being of a non-quality brand, I would also change the other six, in total there are seven, there are then another thirteen swollen electrolytic capacitors, they are all 1000uF 6.3V, and they are scattered around the RAM to the chipset and the PCI slots, here practically all need to be changed, unfortunately I do not have a sufficient quantity to replace them all, they should be ordered, and at the moment it is not possible for me, maybe I could alternatively, recover them from some scrap card, but the best I could find are OSTs, So nothing that could go well, of course I would try them and check them with my ESR-meter, but they could reserve some surprises, that is, they are in bad condition, despite being visually fine.

Wow. I just want to tell you how much I like reading about your repairs. Looks like you do some good work! Keep them boards alive! I have a few of the boards you've worked on and have inspired me to do some recap work myself. Even if a board "doesn't need" it, they sure as hell work better with new capacitors. Most notably, the sound cards/on-board sound.

My favorite board got a re-cap and it was a pain, much time, but it works GREAT now. About 60 capacitors on that board; changed them ALL. First board ever re-capped, should've probably started with something with LESS capacitors, but I really wanted to USE this board and the VRM caps were JUICING after running it for an hour-ish when I first got it.

Major Jackyl wrote on 2025-01-23, 23:23:

Wow. I just want to tell you how much I like reading about your repairs. Looks like you do some good work! Keep them boards alive! I have a few of the boards you've worked on and have inspired me to do some recap work myself. Even if a board "doesn't need" it, they sure as hell work better with new capacitors. Most notably, the sound cards/on-board sound.

My favorite board got a re-cap and it was a pain, much time, but it works GREAT now. About 60 capacitors on that board; changed them ALL. First board ever re-capped, should've probably started with something with LESS capacitors, but I really wanted to USE this board and the VRM caps were JUICING after running it for an hour-ish when I first got it.

The attachment 20241110_140320.jpg is no longer available

Thank you very much 😊 friend
I'm glad to know that there are people who share their work on the old hardware, I do it as a hobby, and here I write only about the work I do on the motherboards, but also about the tests and the problems detected, lately I've dedicated a little more time, but I'm also trying to reorganize and put the various motherboards in order, it's not easy and it will still take some time.
Often I manage to make non-working motherboards work, simply by changing a few capacitors, it's not a simple job, it's usually those of the VRM that are swollen, but sometimes it's not like that, for the AZZA the work is a lot, and it takes a long time, but I'll try to start, maybe I can find 1000uF 6.3V replacement electrolytic capacitors, having to order them but I buy others to keep a sufficient supply.

Of the boards I showed earlier, the only one I could easily repair is the ASROCK P4VT8, I removed the swollen electrolytic capacitors, and I just have to weld those of the Nichicon from 3300uF 6.3V, the original ones were KZGs, in total there were four to be replaced, from the verification with my ESR70, they are all "In Circuit/Leaky", and the ESR is 2.0 2.1 2.4 3.8 OHm 🕉 for each capacitor, it is evident that the ESR is high and that the capacitors are now to be thrown away.

Besides this, what work is there to do?

Some points despite the cleanliness, are dirty, I think we need a toothbrush with a little hard bristles, but I don't know if it's worth insisting, another problem is a chipped SMD capacitor, the top is missing, but it's possible that it doesn't create problems, because it's connected to a pin of the audio chip, for the moment I wouldn't waste time, I checked the pins of the Super I/O chip, some were very close and I tried to space them, now the spacing would be sufficient, but I would like to check under the microscope, if in some places they are too much Neighbors.

Looking at the retroweb, I discovered a couple of negative things, one is quite worrying, it concerns the supply of energy that is inadequate, for this reason they advise contacting them to find a solution, the other instead concerns the support of the CPUs, only Willy and NW, nothing Prescott, these problems instead are not on the next P4VT8+, in fact in that one I have a Prescott of about 3 GHz, in addition to having two other cards to repair in the future.

Sometimes the repair of a motherboard is interrupted, because you are short of ideas, or you think you have found an irremediable failure, but this could be true up to a certain point, with the passage of time, you can come into possession of knowledge that can make the repair possible, because it is true that some chips cannot be changed easily, but it is also true that with the right equipment, it is possible to replace the BGA chips, which would be those welded with micro tin balls, this would be the insurmountable obstacle, for those who do not This type of work, however with the right technique, can be done.

Years ago I had made some progress, with a LuckyStar 6ABX2, an ATX Slot1, in which to put up to three SDRAM PC100, and it is well equipped with expansion slots, there is an AGP five PCI and three ISA (of which one is shared).

Unfortunately I was not able to solve some possible problems, and despite the effort and the precious help received, the motherboard never showed any code in the post card, this made us imagine that the problem was in the SB chip, but it could be something else that created the problem there, so in conclusion, it is necessary to take the card back between the legs, and do some checks, which then I did not know could be done.

I usually write down in a block the motherboards, and the main features, in short, everything that can be useful to perform a test, or the repairs to be performed before trying them.

But it can happen that for some motherboards, nothing has been written, and that therefore it takes a lot of time ⏱, before finding them, and starting to work on them.

Among these forgotten cards, there is an ASUS P5KPL-VM REV uATX. 2.01G, with an Intel 775 Pentium Dual Core E2160, 1.8GHz/1M/800 SLA3H, and Intel G31 chipset, for RAM there are a couple of DDR2 DIMM sockets, for expansion (addition cards) there are: Two PCI-E (one 16X and one 1X), and Two PCI.

You need to check one or two tracks in the back, they are a little scratched, but possibly if they were broken, I could repair them, but it is not necessarily necessary.

The only job to do is to replace a bulging electrolytic capacitor, near the BIOS chip, it's strangely a Rubycon, strange because I don't usually see condensers of this brand swollen, and it's an 820uF 6.3V, tomorrow I'll proceed and change it, apart from these two things, I don't think you need anything else to try it.

The uATX ASUS P5KPL-VM REV. 2.01G, with an Intel 775 Pentium Dual Core E2160, 1.8GHz/1M/800 SLA3H, and Intel G31 chipset, was repaired a little while ago, all that remains is to look at the back in the CPU area, there is a scratch in a couple of tracks, but it is not clear if they are interrupted, for this reason you have to look better, probably the microscope can remove the doubt, whether or not they are interrupted, in this case further work will be needed to fix them.

The swollen capacitor was changed with the same, same capacity and voltage, the brand is different (Panasonic), the one I removed, was measured, I think it was still ️ at the limit, but I preferred to change it, although honestly I don't know if it was basically for the operation of the board, it seems that a PIN comes directly from the ATX power supply, and I would say that it would definitely have created problems in the future, if it didn't work now.

Years ago I tried to repair the LuckyStar 6ABX2V VER: 1.2 motherboard, despite the fact that I solved some problems, I seemed close to the solution, and maybe I had approached (?), however a fundamental problem that remained is in the +12V line, two of the three FANS do not work, only FAN3 goes, which is directly connected to the +12V of the power supply, and also reaches the PIN9 of the ISA slot, I don't think there is +12V in the PCI or AGP slot, in case I'll check.

So going back to the two non-working FANS, which are FAN1 and FAN2, the situation is that of the drawing, pin 2 is connected to the S of a Mosfet N-Ch (Q1 signed 702), to an SMD capacitor (C2), and between the two FANS.

Theoretically there should be +12V, on pin 2 of the FANS, in reality they are not there, I don't know if on the G there is voltage, there should be (?) More than 12V?, while the D could be grounded (?), but it looks like it's connected to a diode.

Even if I don't think that by solving the FAN problem, the card will work again, it's still something that will have to be solved, because the working socket, it's very far from the CPU, for the rest I'll try to do a boot test, I don't expect miracles, but we'll see if I can display something.?!?

I put some pictures of my Soyo SY-5BT (5BT5), it should be a Rev.2. X, because it has the jumper for 2.2V, the pins are 12 instead of 10, the only difference is the addition of the jumper and relative resistance, for the rest I think it is the same, except for the frequency generator chip, here there could be a couple of frequencies beyond 66, with PCI at about 33, so if those selections work, you could even put a 500 MHz (FSB 83) without the PCI going beyond 41 MHz, the same goes for 450 MHz (FSB 75), I tried these settings with a normal Pentium and they don't work, but it could be Is it possible that with a Cyrix or a K6-2-3 instead they work???

The problem I have is the cache chip with missing pin, I have already tried to discover the chip, but I could not find the internal part of the pin, so I would have decided to leave it like this, and switch to a K6 CPU with integrated L2 cache, I have a K6-2+ 400 MHz 1.6V available, and I would like to use that, since the minimum VCORE is 2.2V, I was thinking of adding a resistance in the jumper, to make it go below 2.0V, or alternatively maybe you could use a diode.

I will do some tests, to measure the voltages on the mosfets, but initially I would use a K6-2 CXT, if I see that it is possible to go below 2.0V, I will switch to K6-2+, but for support, I think I should update the BIOS to the latest version, which should be 1B7 if I'm not mistaken (?).

In the next few days I should start doing some tests, the alternative would be to look for undocumented settings of the VCORE, maybe there could be both 2.0V and 2.1V, and even the 2.4V, unfortunately in this MB, is not like other similar ones, with jumpers that increase by 0.1V 0.2V 0.4V 0.8V.

Maybe it will even work like that. I'd install 32 MB of RAM, boot DOS. See if it's working with cachechk, memtest86 or other software of that kind. See if there are any errors.
If it doesn't, I'd keep digging more. There should be a pin in there. If there's nothing and it doesn't work, only then I'd have the peace of mind to remove the poor thing. Finding a replacement won't be that hard. Soldering it will require some steady hands.

I'd be interested to know what's the best way to reduce the vcore too.

tauro wrote on 2025-01-27, 23:14:

Maybe it will even work like that. I'd install 32 MB of RAM, boot DOS. See if it's working with cachechk, memtest86 or other software of that kind. See if there are any errors.
If it doesn't, I'd keep digging more. There should be a pin in there. If there's nothing and it doesn't work, only then I'd have the peace of mind to remove the poor thing. Finding a replacement won't be that hard. Soldering it will require some steady hands.

I'd be interested to know what's the best way to reduce the vcore too.

Hello and welcome 🤗
So, the card only works with the L2 cache disabled by BIOS, the missing pin is no.8, I don't know what it is, but I think it's connected to something (chipset?), unfortunately the card arrived without that pin, I don't know how or why they removed it, from what I know the cache only works with both chips, removing this I don't think it's possible to make only 256KB of cache work, and I didn't worry too much, because I think I find the hidden part of the pin, and reconnecting it with a bridge 🌉, the 512KB cache would work.
I also thought about replacing it, but first I would have to find the replacement one, it's a job I've never done, and so I don't know if I would succeed.
As RAM I currently have a 32 MB SDRAM (or maybe 🤔 64 MB), I will know for sure when I start it in the next few days.
Regarding the VCORE, there are two possibilities, both must be tested, the first is to add a resistance, and see what changes, I don't know if it will go down 0.5V or 0.05V, but if it only goes down by 0.05V, I increase the value of the resistance, if I find one that lowers 0.5V, for me it's fine; the second is to try undocumented jumper settings, there should be a couple to try, but I could find values that are not very useful for much like 2.5V or 2.6V, although maybe a 2.5V would be good 😌 for a K6-III or some K6-2 with VCORE 2.4V.