Today the second Core2 Duo E4700 has arrived, the first is on the Linux PC overclock @3.0GHz, with that I am quite satisfied, that's why I have in mind a higher overclock, beyond 3.0 GHz, going from FSB 800 to 1066, since it starts from 2.6/800 I will get to 3.466/1066, this is the idea then the reality could be different.

However, having on the PC Office an E4600 which is a 2.4/800, maybe I try with that to raise the FSB to 1066, so the MB ABIT, it could be quite suitable for the purpose, the result could be an E4X00 at 3.2/1066, if I did not get the calculations wrong, they would be a good 800 MHz more, a nice jump I would say, and I think the current score of 5.9 will rise at least half a point (we hope !!!), maybe even the RAM score could rise by a few tenths, in short it promises well but we will see if it will be easy, or not, go up by FSB.

For this jump, there are on-Line various images to connect or isolate the points of contact, of the CPU with the LGA 775 socket, but I don't know if they are necessary, because I think they are needed, on PCs with limited BIOS, that is, without options to select different FSBs, typical would be Dell HP PCs etc...

I start experimenting on the PC Office, the increase of FSB, currently I have gone from about 200 to 233 MHz, actually those 233 correspond to about 933 MHz, this is interesting, because we are halfway, between 800 and 1066, at this moment the CPU frequency has risen from 2400 to 2800, the PC works quietly and does not show any problem, indeed as we will see, there is a small gain, perhaps not visible, but still there is.

The BIOS has several possibilities, you can enter the value of the FSB manually, that is, just write instead of 200, for example 266 to get the fateful goal, for the moment I'm at 233, I want to complete other benches, just to understand, if the performance improves only for CPU and RAM, or even for the rest of the HW.

I only made the bench of CPU-Z Ver.2. X, later I will do that of the Ver for W98, then I re-executed the performance index rating of Windows 7, there is an increase of 0.3 for both the CPU and RAM, it's not a lot but going up to 3.2 GHz, it could get close to 6.5, which is the maximum score I get (Game Graphics).

For today I'll stop at these benches, tomorrow I'll see if I can make more, for example it occurs to me that I haven't loaded Sandra yet ️, in short, there's still ️ some more tests to do.

As anticipated yesterday, I did some other benches, I noticed that there is a slowdown, in the 3D Mark 2000SE, perhaps as I was suggested, it could depend on the service pack, which may have slowed down the system, but it did not happen with the other programs, indeed some show an increase, as normal it was.

With SuperPI, I'm just above 20 seconds, and this doesn't seem bad at all, I'm curious to see how long the i5 takes at this point, maybe less than 10 seconds????!!!!

Finally the score of the CPU-Z 1.76, the values of the ST and MT grow, absolutely I don't know if they are interesting or not, but in the meantime I can see if still going up ️ up to 3.2 GHz, how much it goes up.

About the microscope and the Soyo 5BT5, the current situation is this, one of the cache chips has a missing pin, I have no idea why it is not there, maybe it had detached and in an attempt to get it back up it could be broken ???!!!???

The missing pin is the n.8, I don't know what it corresponds to, however currently I can only make the MB work by disabling the L2 cache on board (which would be 512KB), unfortunately there is no way to make it go to 256 KB, that is with a single chip, so either 0 or 512, not being able to use the 512 KB, I could use a CPU with integrated L2 cache, but I would have to find a rare K6-3 with VCore 2.2V, as far as I know there were but maybe they were for laptops, otherwise I would have to do a HW mod, or try other undocumented settings, to find some One that is about 2.0V, and at that point I could use an AMD K6-2+ 400, or possibly a 450 or a 500.

I concentrated my time ⏱ to make the microscope work, and finally there I did it, it works and I can focus many details, which normally you can't see, but yes there is a though, at the moment I can focus only if it is very close, so see I can see, but I don't have enough space to work, in short, if so, see photo, I can't do any repair, that is, I can but I have to do it by eye 👁, and then check the result, ultimately this was not the idea, I thought you could focus being at least 3-5 cm away, I'm currently just above the centimeter!!!!!

Here I am again, for a while I detach with the SW, I want to do some work on the HW, and the microscope could help me, not so much for the welds, but for the checks, I can see very well the welding points and the scratches on the tracks, here is just this until now it was quite problematic, looking carefully I was able to identify the interruptions, but I could hardly have seen the welds so well, and the crooked pins of the LGA sockets (the Intel ones), I make a brief summary of the work to be done:

MB SOYO 5BT5: chip cache pin repair;

MB ASUS 1156: straighten pin socket;

MB Miscellaneous: they have scratches in the back, interrupted tracks.

For now I limit myself to identifying where the problem is, later if it is not too complicated, I move on to the repair.

Arrived today various motherboards, one is FM2 socket, of this format and Socket, I had nothing, I have to order a CPU, there are two different types, Athlon and AXX, AXX stands for A4-A10, while the Athlon there is both X2 and X4.

An ASUS for P4 478, it's quite interesting, the P4B266 is a model I thought was unobtainable, it matches CPUs with FSB 400 to DDR RAM, this way I can make a comparison between Willy, I have the 423 with RDRAM (PC800), the 478 with SDRAM (PC133), and this 478 card with DDR 266 or higher.

First, however, some electrolytic capacitors must be replaced, in the VRM CPU area, and then the operation must be verified.

The others are mostly all 775, some seem perfect, but on the ASROCK there are various swollen capacitors, even here it is necessary to replace them

The motherboard chapter with AMD FM2 CPU, it's pretty much closed!!!

Having the microscope available, I checked the back, because I noticed that there was something missing, there is actually more, two semi-detached SMDs, plus two or three probably missing, so I would say that it is clear that it is not easily repairable, because first you have to see what is missing, and then you have to understand what they are.?

I took some photos to show the magnifications, they are not very clear but they give the idea, unfortunately some SMDs are so miniaturized, that it is difficult even to put them back up.

I'd say it's a case of shelving it until an FM2 CPU turns up cheap or free, then seeing if it works anyway. A lot of those components can be "insurance" or the FCC test examiner likes to see a capacitor across there, and may not prevent it working.... though some will prevent it working... or some might let it work for certain CPUs and not others (FID, VID, multiplier lines)

BitWrangler wrote on 2024-03-16, 00:32:

I'd say it's a case of shelving it until an FM2 CPU turns up cheap or free, then seeing if it works anyway. A lot of those components can be "insurance" or the FCC test examiner likes to see a capacitor across there, and may not prevent it working.... though some will prevent it working... or some might let it work for certain CPUs and not others (FID, VID, multiplier lines)

Thanks so much friend, I hope you are right, and that the card works anyway, in fact the seller said it was working, and it was in the photo complete with RAM and CPU, unfortunately those were not for sale.
As a CPU I would have bought an A10, or an Athlon X4, but I noticed the problem, and left it alone.
However, those two semi-detached ones have to be fixed, and although this is a bit difficult, because they are tiny, it can be done, for the two or three missing ones, however we have to hope that they do not prevent the functioning, there would be a third possibility, which connect PCI-E slots, in that case everything should be fine as long as you don't use them.

Other tests with the microscope, I recovered the ASUS ROG Strix motherboard, in short it has numerous crooked LGA pins, if you want there are not many, at most a dozen, but some are also bent, and getting them back in place, it will not be easy at all.

The card after fixing the pins, may return to working and this depends on the crooked pins, whether or not they return to make contact with the CPU.

I have done further tests to increase the focal distance, well I have perhaps gained about a centimeter, but there is always a minimum distance between the board and the closest point of the microscope, about two centimeters more or less, maybe I would gain space, by removing the transparent part, and in fact it is possible to remove it, but I do not know what effect it has, that is, if it is used to convey the light of the LEDs, or it serves to help yourself to point the point to be enlarged.

I add photos of the details with crooked pins, some are the same but photographed from a different point.

I have aside a MB 775 with some pins to fix, I have not found where I put it, but I think I will try to straighten the pins in that one, just to gain experience, and not do damage in the best boards, certainly not said that if I can fix the 775, then automatically I will be able to fix the others, those pins should be quite delicate, and breaking one might be possible, but only if it is very bent, in cases of decentraling, it will just be enough to reposition it, the effort will be much less, here maybe it will be difficult to dose the force, but if they are not Particularly elastic, a slight displacement could put it back in its original position.

Apart from these problems with the pins, looking through the MB I took an ASUS P2B-F that I couldn't get to work, let's say everything seems ok, I'll try to do some checks, to see if it's just a problem of contacts at the Slot1, I think I'll take a look at it with the microscope, and then I'll try to spray some IPA, and do some tests with other CPUs, maybe one could make contact and awaken the MB.

I started a job on an S.462, it is an ASUS card (already my favorite) A7S333, in memory it seems to me there is an Athlon XP 2400+ of about 2GHz, with a 512 MB DDR RAM bank from Hyundai, unfortunately it did not have a CPU fan, initially I had used a fairly small temporary one, today I replaced it with a normal size, it should be a 60 X 60, now it remains to find an AGP video card that goes well, I think something like GF MX, the slot present in the MB has no internal notch, and allows the use of any AGP VGA, but here I will be on models neither too old nor too new, so I would say that a look at the manual, can allow you to restrict the choice to certain cards, maybe I could even find the models that ASUS suggests to use.

As a video card for the ASUS A7S333, I have enough choice, I did not find a color card similar to the MB, so having several green color I chose one, hoping that it will be fine, of course it is definitely a cheap model, but for a PC that has to be used sporadically, and without heavy commitments, I think the ATi Radeon 7000 with 32 MB DDR, and dual VGA and DVI-D output is fine, possibly if you do not recommend it I can use other ATi cards, such as 7500 or 9200, and if necessary with more than 32 MB, let's say 64 or 128 MB easily.

KT333 chipset was more DX8 cards in era, like 8500 and Gf4200 so not sure if you are going for a close year match for higher end spec or doing Win98, which those are fine for. 9200 is a year later but more like an 8500. CPU speed is pushing more into DX9 card times. I don't know what your plan is though, making the most out of of GPU in earlier games or making high end spec for 2002 which would run 2001/2002 games, or pushing the platform for the best it will do, which is more like needing a 9800Pro or FX5950. Out of what you've listed there for that board though, I'd say 9200 is probably best. 7500 I'd put with a thunderbird 1100-1400 or tualatin and 7000 with a Duron/sdram or PIII but that's me.

BitWrangler wrote on 2024-03-17, 15:57:

KT333 chipset was more DX8 cards in era, like 8500 and Gf4200 so not sure if you are going for a close year match for higher end spec or doing Win98, which those are fine for. 9200 is a year later but more like an 8500. CPU speed is pushing more into DX9 card times. I don't know what your plan is though, making the most out of of GPU in earlier games or making high end spec for 2002 which would run 2001/2002 games, or pushing the platform for the best it will do, which is more like needing a 9800Pro or FX5950. Out of what you've listed there for that board though, I'd say 9200 is probably best. 7500 I'd put with a thunderbird 1100-1400 or tualatin and 7000 with a Duron/sdram or PIII but that's me.

Thank you very much friend, unfortunately even though I had several PCs in that period, I don't remember which VGA I had, perhaps probably a GF2MX, that was initially on the P3 800, and later moved on to the P4 2000/512, subsequently I purchased an ATI 9250 if I'm not mistaken, however I will follow your precious instructions, change the video card, and put a Radeon 9000, which also has more RAM, probably at least 64 MB, but it should easily have 128 MB, I will later try to run some benches, such as 3D Mark, and I will try to install Win98SE, and maybe even 2000 or XP.

Searching among the other video cards I found a Radeon 9000 AGP with 64 MB DDR, I hope it is enough for a PC like this, if I wanted I would have other cards but I don't know what they are, so they could be interesting models, but I don't believe in them, in any case this one 9000 is visually very beautiful, it seems almost new, it has two video outputs like the other, VGA and DVI, and there is also a video output, we will see the scores with 3D Mark when I start the PC, I hope they are not painful, but using Windows 98 SE, I don't think it would be useful to expand the RAM beyond the 512 MB already present, different story with 2000 or XP, if you want 1 GB would it be fine?

I'll go back to the VGA topic for a moment, so since I don't have a 9200, I have an alternative 9000, which according to some is quite similar, so it should practically do more or less the same things as the 9200, and for me it could be fine.
Unfortunately I have other cards, but they don't have any indication or label of what they are, to understand I would have to put them on a MB and run a diagnostic, and it could be that one is a 9200, or something similar, also they have the heatsink without a fan glued on, so it is not possible to remove it and read the writing on the chip.

By now the microscope allows me to see much better, and if I notice something that is not perfect, I can actually see if there is any problem, in particular I have some cards with LGA 775 sockets, you know the pins are quite delicate, and they tend to misalign or bend, if handled in a summary way or without paying due attention, personally I have never ruined it, but I get the motherboards of those who instead, they improvise capable of a CPU change, in reality failed, and motherboard no longer working !!!

So, having now several 775 cards, I check one that I have never tried, an ASUS P5Q PRO, the socket observed without the aid of any magnification tool, it is perfect except a slightly out of place pin, in the image you can barely see it, but it should be fixed, now I have not seen very well, if it is slightly raised, or sunken, just in case it were sunken, it would be to be lifted as soon as.