Since there is an open discussion some time ago, I avoid opening a new one, for the same reason.
I bought a lot of cards, including an ABIT KG7-Lite, the card seems to be in excellent condition, but I noticed that several electrolytic capacitors are missing, mostly in the CPU area or nearby, for now I am interested in knowing which ones are missing, and if possible what capacity and voltage they have (each), then I would like to know if it is useful to add the two RAM banks missing in the Lite version, that is, if adding them, you can put more RAM (total), which currently would be 2X1GB, and could become 4X1GB, since the two versions KG7 and KG7RAID, if I understand correctly they are the same MB, but with additional equipment.

If all of the passives are populated around the RAM slots, they'll *generally* work. I've added RAM slots back to boards that were missing them, and I've not had a failure yet.

The only problem you may run into is that the BIOS is sometimes configured to ignore those slots when counting memory. So you can end up with a rather amusing POST screen that states you have 0 MB of RAM installed, but it does indeed work, and the OS can see it.

I would make sure you can get the motherboard working before you go about trying to add RAM slots though. At 184 pins per slot on DDR, it's a super monotonous and time consuming task. You'll definitely need a hot air tool to preheat the solder VIAs and a desoldering gun. Using a soldering iron and solder wick/solder sucker will likely result in board damage.

As for the capacitors, EC29 and EC32 (capacitors right above the AGP slot) are 4700 uF 6.3v. They're probably 12 or 12.5x25 or 30mm in size.

The five capacitors directly under the CPU socket are 3300 uF 6.3v, as is the one diagonally to the CPU socket (EC16). You may want to substitute these for polymer capacitors with the lowest profile you can find, because they're right in the path of the CPU exhaust. That board has a really crappy layout.

The last capacitor, I'm not sure of. It's the small circular pad between the North Bridge and the VRM coils. In other pictures, the capacitor looks to be the same as the ones between the PCI slots. I'd guess 820 or 1000 uF at 16v. It's probably 8x11 or 8x12 in size.

Falco wrote on 2024-11-18, 21:57:

If all of the passives are populated around the RAM slots, they'll *generally* work. I've added RAM slots back to boards that we […]
Show full quote

If all of the passives are populated around the RAM slots, they'll *generally* work. I've added RAM slots back to boards that were missing them, and I've not had a failure yet.

The only problem you may run into is that the BIOS is sometimes configured to ignore those slots when counting memory. So you can end up with a rather amusing POST screen that states you have 0 MB of RAM installed, but it does indeed work, and the OS can see it.

I would make sure you can get the motherboard working before you go about trying to add RAM slots though. At 184 pins per slot on DDR, it's a super monotonous and time consuming task. You'll definitely need a hot air tool to preheat the solder VIAs and a desoldering gun. Using a soldering iron and solder wick/solder sucker will likely result in board damage.

As for the capacitors, EC29 and EC32 (capacitors right above the AGP slot) are 4700 uF 6.3v. They're probably 12 or 12.5x25 or 30mm in size.

The five capacitors directly under the CPU socket are 3300 uF 6.3v, as is the one diagonally to the CPU socket (EC16). You may want to substitute these for polymer capacitors with the lowest profile you can find, because they're right in the path of the CPU exhaust. That board has a really crappy layout.

The last capacitor, I'm not sure of. It's the small circular pad between the North Bridge and the VRM coils. In other pictures, the capacitor looks to be the same as the ones between the PCI slots. I'd guess 820 or 1000 uF at 16v. It's probably 8x11 or 8x12 in size.

Thank you very much ☺️ friend

I'll check if I have any available, I definitely don't have the 4700uF ones, I usually use Rubycon or Panasonic electrolytics, but I also have Nikicon or other similar ones, it depends on the capacity and voltage.

For the missing RAM sockets, it is in fact better to do the work, after the card is working, if it had problems due to swollen capacitors, probably replacing the missing ones, it should make the board work again.

But for this, it will take some time, I hope to find the bigger capacitors, and then they arrive soon.

Mouser ships worldwide, and Digikey does too I think, though they're going to be pretty steep on the shipping. Not sure what supply houses exist in Europe.

Falco wrote on 2024-11-19, 00:17:

Mouser ships worldwide, and Digikey does too I think, though they're going to be pretty steep on the shipping. Not sure what supply houses exist in Europe.

Well I'm looking and evaluating which ones to take, I wanted to tell you Thanks ☺️, I didn't realize that the one near the NB was missing, and from a closer observation, I saw two more missing, one EC8 (near USB), the other EC45 (in the middle between USB2 and FDC1), for the values of these two, EC8 should be 1000uF 10V (TEAPO Black), it seems that ABIT only used TEAPO brand capacitors, I have some recovered from MB SA6 and SE6, but I prefer to keep them aside and use something else.
I don't know if I can, but I'll try to make a diagram with the values of each individual capacitor, maybe 🤔 in the future someone could use it.

Those 1000 uF Teapos can be downvolted to 6.3v if necessary. It looks like they're on the +5v rail and don't really need to be 10v, that's just whatever the manufacturer could get on the cheap in the back alley dumpsters of Huaqiangbei.

Teapo-Cheapo garbage.

Falco wrote on 2024-11-19, 19:59:

Those 1000 uF Teapos can be downvolted to 6.3v if necessary. It looks like they're on the +5v rail and don't really need to be 10v, that's just whatever the manufacturer could get on the cheap in the back alley dumpsters of Huaqiangbei.

Teapo-Cheapo garbage.

OK thank you very much ☺️
You should need the USB ports, so in fact the voltage of 6.3V should be more than enough, or they wanted to make sure that in case of overvoltages, they held 😀
In any case, having both the 10 V and 6.3V ones, I would choose between the 10V ones, because the 6.3V ones go a lot on almost all MB, that's perhaps the real reason is that they ran out of stocks of 6.3V.

It's not the voltage rating that caused them to fail, it's the fact that they're trashy Chinese/Taiwanese capacitors made during the capacitor plague era with bad electrolyte formulas. Then there were the counterfeit name brand caps made by same said people.

Falco wrote on 2024-11-19, 21:22:

It's not the voltage rating that caused them to fail, it's the fact that they're trashy Chinese/Taiwanese capacitors made during the capacitor plague era with bad electrolyte formulas. Then there were the counterfeit name brand caps made by same said people.

Yes, this problem of capacitors, maybe if for others it was a problem, for me it was a nice opportunity, I replaced capacitors in several motherboards, and some sold them out because they didn't work, I usually put identical capacitors, that is, same values and same sizes, but in some cases, when it was not possible to find the same ones, I put slightly higher voltage, let's say some time ago, I bought several pieces of the most common, I still have a good assortment, which except in special cases, allows me to change capacitors in 95% of the MB.

PC@LIVE wrote on 2024-11-19, 16:31:

I don't know if I can, but I'll try to make a diagram with the values of each individual capacitor, maybe 🤔 in the future someone could use it.

You just made me remember I actually did make a diagram of the caps on a KG7 a long time ago. I just found it.
I don't think this was ever really finished and some of the little "remarks" on the diagram could be wrong. I don't remember this stuff anymore.
There's even a bunch of random capacitor "icons" on the left side that I was copy-pasting - so I obviously never really finished this.
It was a long time ago when I was recapping 1 or 2 of these boards, and thought I might see more, when I did this.

Interesting to note that this board does not use the ATX 3.3V supply at all. It's not connected. 3.3V is regulated onboard from the 12V supply. I don't know why. If you changed the connector, you could run this board from an AT power supply. 🤷

The CPU power comes from 5V as usual.
What I labeled as "Vmem" might also power the chipset. If I believe what I wrote back then, apparently this comes from 5V also.

shamino wrote on 2024-11-23, 06:17:

You just made me remember I actually did make a diagram of the caps on a KG7 a long time ago. I just found it. I don't think th […]
Show full quote

PC@LIVE wrote on 2024-11-19, 16:31:

I don't know if I can, but I'll try to make a diagram with the values of each individual capacitor, maybe 🤔 in the future someone could use it.

You just made me remember I actually did make a diagram of the caps on a KG7 a long time ago. I just found it.
I don't think this was ever really finished and some of the little "remarks" on the diagram could be wrong. I don't remember this stuff anymore.
There's even a bunch of random capacitor "icons" on the left side that I was copy-pasting - so I obviously never really finished this.
It was a long time ago when I was recapping 1 or 2 of these boards, and thought I might see more, when I did this.

The attachment diagram.png is no longer available

Interesting to note that this board does not use the ATX 3.3V supply at all. It's not connected. 3.3V is regulated onboard from the 12V supply. I don't know why. If you changed the connector, you could run this board from an AT power supply. 🤷

The CPU power comes from 5V as usual.
What I labeled as "Vmem" might also power the chipset. If I believe what I wrote back then, apparently this comes from 5V also.

Thank you very much for providing me with this image, and for the related info, I hope 🤞 it can also be useful to other users, who have the same motherboard.

Today I received the replacement electrolytic capacitors, I had ordered the 4700uF 6.3V ones, while I already have the others, 3300uF 6.3V and 1000uF 10V, the latter are slightly larger, if you want you could use the 6.3V ones, which have the same size, but I prefer to keep them for other cards, they are very common on several MB.
I prepared the MB, freeing the holes from the pond, it was not difficult 😩, apart from a couple that there was some tin left, and they required a patient work, to eliminate the excess tin.
Tomorrow if I can, I'll complete the work, welding the new ones, and if I have enough time, I'll start gathering what I need to try the MB, I'll probably start with a Duron, to then switch to an Athlon XP.

Unfortunately 😣 after replacing the missing electrolytic capacitors, I found in the back of the motherboard, several scratches on the tracks, some are interrupted, and many of these go direct to the CPU socket, in short I thought I had finished, but I have a lot of work to do (still).
But that's not all, I noticed that the welds of a group of four electrolytic capacitors, are in very bad condition, and the same goes for those nearby.
If and when I can fix the tracks, could I change those four 16V, which are in the +12V line, even if they seem to be in good condition, could it be a short time left and will they create problems in the future?

shamino wrote on 2024-11-23, 06:17:

You just made me remember I actually did make a diagram of the caps on a KG7 a long time ago. I just found it. I don't think th […]
Show full quote

PC@LIVE wrote on 2024-11-19, 16:31:

I don't know if I can, but I'll try to make a diagram with the values of each individual capacitor, maybe 🤔 in the future someone could use it.

You just made me remember I actually did make a diagram of the caps on a KG7 a long time ago. I just found it.
I don't think this was ever really finished and some of the little "remarks" on the diagram could be wrong. I don't remember this stuff anymore.
There's even a bunch of random capacitor "icons" on the left side that I was copy-pasting - so I obviously never really finished this.
It was a long time ago when I was recapping 1 or 2 of these boards, and thought I might see more, when I did this.

The attachment diagram.png is no longer available

Interesting to note that this board does not use the ATX 3.3V supply at all. It's not connected. 3.3V is regulated onboard from the 12V supply. I don't know why. If you changed the connector, you could run this board from an AT power supply. 🤷

The CPU power comes from 5V as usual.
What I labeled as "Vmem" might also power the chipset. If I believe what I wrote back then, apparently this comes from 5V also.

Ok I'm not super knowledgeable about this but I see in your diagram for vcore you have noted: Vcore = 3300uF 6.3 Nichicon HM(M) or 4x 4700 @ 12mm pads and 2x 2200uF at 10 mm pads

I have seen boards with both cap setups, are they both equally good or ???

I have even seen someone that recapped them with poly caps that seem to be 2200uF https://www.retrohardware-reviews.de/review-abit-kg7-raid/

Seems all a bit (Abit =p) confusing to me.

Well looking at the link that Grem Five suggested, I decided to complete the work on the KG7-LITE, in particular I will have to repair the interrupted tracks, it will be very difficult 😩 to do, but I think it's worth it, it will take time and luck 🍀, but I hope there is nothing else to fix.
And since there are, I will free the holes of the two missing RAM banks, to insert a couple of DDR sockets (recovery), in this way I could get to 4 GB 🇬🇧 of total RAM, which is really a lot, for a PC like that.

Grem Five wrote on 2024-12-13, 06:10:

Ok I'm not super knowledgeable about this but I see in your diagram for vcore you have noted: Vcore = 3300uF 6.3 Nichicon HM(M) […]
Show full quote

Ok I'm not super knowledgeable about this but I see in your diagram for vcore you have noted: Vcore = 3300uF 6.3 Nichicon HM(M) or 4x 4700 @ 12mm pads and 2x 2200uF at 10 mm pads

I have seen boards with both cap setups, are they both equally good or ???

I have even seen someone that recapped them with poly caps that seem to be 2200uF https://www.retrohardware-reviews.de/review-abit-kg7-raid/

Seems all a bit (Abit =p) confusing to me.

If the caps are otherwise the same, then there's probably not much difference between either way of doing it.
With the caps used on Vcore output to the CPU, the most important parameter is the ESR (equivalent series resistance). The total ESR of parallel capacitors follows the parallel resistor formula:
1/ESRtotal = 1/ESR1 + 1/ESR2 + 1/ESR3 ...
if each cap is the same, then you can simplify it to just dividing the ESR of one cap by the total number of caps.

I would just use whatever setup is convenient with caps that are suitable (low ESR types, not general purpose). I don't know what ESR the originals were - did they really come with Nichicon HM? I thought ABit didn't start using the Japanese brands until later. I don't remember why I put that on the diagram.
If the ESR is low enough then you don't even have to match the original total capacitance. That's why people who swap to polymer caps tend to use smaller values - matching the original capacitance with polys is expensive and usually not necessary.
Usually the original capacitance is overkill but that's what was required to get the ESR they wanted from cheaper electrolytic caps. It's conceivable they might have even gone overkill on the ESR because they didn't trust the specs on the brands they were using.

I still have to get my hands on my board in a few days but I found this on hardforum.

Its from 2021 post where the guy says he hasnt powered it on for 20 years and the caps look fine. The example I'm getting all 6 caps near the socket have the tops popped but they seem to be the same ones as his example. I have read the Nichicon HM suffered from the plague which is odd as his look perfect but the ones on my board are popped. Guess Abit couldnt win, some of their boards have Temco subpar caps and when they use a better brand ones they suffer from the plague. <the cap in the backround of the one shot says HM, I will find out when I have my board>

I have some Rubycon MBZ 3300uF 6.3 (that checked good) I harvested from some Dell BTX boards I have around here so I can replace and test the board quickly when I get my hands on it to make sure it works before spending $$ on new caps. Heck they might stay there as they spec out the same as the HM ones. <Nothing more permanent then a temporary solution>

Grem Five wrote on 2024-12-14, 01:13:

I have read the Nichicon HM suffered from the plague which is odd as his look perfect but the ones on my board are popped.

My understanding is that Nichicon HM had a manufacturing defect until sometime in 2005. I think some batches were good (like yours, apparently) and others were defective for whatever reason. When those caps started bursting, major clients like Dell and Apple got upset, and Nichicon found something wrong with their manufacturing process and fixed it. HM caps after that time didn't have the rampant failures anymore.