Well maybe today it went a little better, at least I spotted where the problem is!!!

We always talk about the SMT-486 Terminator motherboard, what I suspected, was not wrong at all, and having seen codes sometimes, made me think that somehow the board was ready to start, but something prevented it.

I had a doubt that some jumpers were not correct, there are really many, and it is not well explained what they do (?), but they can be considered correct, even if strangely between a DX-33 CPU is a DX4-100, the only different thing looks like the Voltage, I don't know maybe this card has an automatic recognition of the multi, or are the CPUs that are able to select it by default, I didn't look for info about it, so I keep the selections found in the manual for good, and I hope they didn't get confused before going to print.

The points that are most suspicious for me are the RAM banks and the CPU Socket, I tried to move the RAM from one bench to another, and I got nothing, no start sign, so I decide to try to remove and put the CPU back, here there is something positive, sometimes I see the code C0 —, this means that something started, but it failed to continue.

After several attempts, it gets to display a couple of 0b 0A codes, at that point I don't know, I was trying to understand if it was blocked, but shortly after surprise !!!

In the video there is the BIOS screen, which asks to press F1 or Delete (or F2?), I press F1 and it continues, of course without any drive connected, I can't do anything!

I restart and enter the BIOS, it's quite simple, but I would say adequate compared to other 486, and it's not essential, obviously its limits, are largely due to the type of PC, but from what I see, you can optimize some items to make it a little faster, in this regard, only after turning it off, I realized that the L2 cache is 0 KB, while on board there are 4 chips for a total of 256 KB, unfortunately I couldn't do much else, because after a few minutes of operation, it crashed, and not It was possible to restart it, I'll try again ️ as soon as it cools down, and for the record, the CPU was an Intel 80486 DX33, and the VGA was ISA Cirrus Logic.

Although I still haven't solved it, some of the problems that remain, it seems that with the use they get better, I got to do the Phil's benches, and not only those, I ran the usual diagnostics, in short, it worked without problems for quite a while!

Which motherboard? The SMT-486 Terminator II, with Intel 80486DX-33, the RAM was initially only 4 MB, but after seeing that some of Phil's benches did not work, probably because there was not enough memory, I replaced the RAM with another 8 MB EDO, later I wanted to add a second 8 MB EDO RAM, but unfortunately it no longer managed to start, here I think I will have to remove and reinsert the CPU once again ️, usually this works and in the luckiest cases, it works very well.

Well, but it's not really correct, because when the PC works, the L2 cache doesn't work, you think there's no problem with the chips or HW, maybe it depends on the jumpers, those of the capacity are correct, but those of the CPU instead are not clear, and correspond to the UMC CPUs, maybe you should try different settings, like those of the AMD CPUs, but unfortunately I don't have an AMD DX33, this makes me think of changing the CPU again and going to a 486 DX2, just to see if the L2 cache works, or not.

Then, once full functionality is reached, I should redo some benches, and start the various upgrades, yes because there is the possibility of putting more RAM, and adding a sound card, plus other cards, not many but because we only have three PCI slots, one already occupied by the Miro PCI with 1 MB S3 chip (video card), maybe here it would be the case to use an ISA sound card, and maybe a PCI card with USB ports, with this I saw recently, that it is possible to have three floppy players, two connected directly to the motherboard, the third 3.5" is Connected via adapter to the internal USB port, although for access and use, you need DOS drivers, with Windows 98SE instead no problem.

This FD3.5”-USB adapter, I think it is very useful in PCs that do not have it (the floppy), or even in motherboards that do not have the 34 PIN port for connecting to Floppy, it is also very convenient, if we have to move the files, but for this very reason recently, I bypassed the limits of the 1.44 MB disks, using the null-modem cable and the LapLink program, they also exist in USB version, certainly faster than the parallel and serial ones.

The current performance looking at the image of the bench that I put, is about 10 times lower than a P200, I think that the value with the L2 cache working, may be different but the considerable distance would remain, the only way to get closer, would be the change of CPU with a fast AMD, because unfortunately Intel stopped at 100 MHz, the competition instead went further, for example the 5x86-133 P75, reaches performance similar to a Pentium 75, on the other hand you notice that the 486 is almost twice as fast, in short, to reach the Pentium 200, you would need a 486 of about 350 MHz!!!

With the SMT-486 Terminator II and Intel 80486DX-33 motherboard, I make some slight progress, while before to get to the BIOS screen, you had to remove and reinsert the CPU, now even after a few hours, it manages to work even after it was turned off, it seems a not very important thing, but it is a very small progress, on the other hand, there is a stability not so good, sometimes it crashes quite soon, other times it works for a while, but at a certain point it freezes anyway!!!

Of course, that the Socket problem is not completely solved (?), or there is something else that causes problems, a suspicion is on the jumpers, maybe they are not set correctly, but this CPU still manages to start and work, another problem that I have not yet understood well, is that of the L2 cache on board not detected, maybe it is somehow connected to the jumper problem, or it could be a RTC problem, yes because I discovered that although the battery is charged, it does not keep the changes to the BIOS in memory, and after the shutdown in a short time Go back to the default settings, this is a third problem to solve!!!

All in all for me they are special satisfactions, having literally resurrected this MB 486 PCI was important, because it filled the gap that had been created, when the family created it exactly I don't know, I learned it quite recently, when by now the price of these motherboards had risen a lot, however that was a fairly similar motherboard (BEK P405), it had an extra RAM and PCI slot.

The current CPU, an Intel 80486DX-33 is temporary, the final one will be at least 100 MHz, the starting point would be an AMD DX4-100, but there are other CPUs that I could use, I could for example exchange it with the Intel DX4-100, or staying in AMD I would have these:

DX2-80, DX4-120, 5X86-133 P75

I wouldn't use Cyrix or ST equivalents here, although I don't have any of these other than the DX2-80. Unfortunately I don't have the Pentium Overdrive (for 486), which exists in two versions 63 and 83 MHz, it would be interesting to understand if the 83 MHz is comparable to a P75 (even in overclock at 83 MHz with FSB 55), but I imagine that the S.5 is a little faster with the same frequency.

I ran other benchmarks, and some diagnostics, one provided interesting information, that is, what SMT means, you can read it clearly, personally I would not have imagined it.

I continue to do tests on the SMT-486 Terminator II motherboard with the Intel 80486DX-33, first of all I would say, the problems currently to be solved are mainly three, the first that I would like to solve is the L2 cache not working, the second is the RTC not working, the third is the instability that sometimes occurs in a short time and others not.

Let's start from the idea that all the jumpers are correct, as I have already written it may not be so, because it is not clear what some do (?), a simple test I did, was to try to change the place to the Turbo SW, there is strangely a three PIN instead of the normal two, the jumper is currently on 1-2, moving it to 2-3 does not change anything except that on Speedsys a memory error appears (in the Cache performance box), I put it back to 1-2, since it has not changed anything.

I think it's a bit strange that a defective L2 cache doesn't block the PC at startup, very strange I would say, it's also true that looking on TRW, there are two versions of the BIOS, an older one is a more recent one, I have the one equivalent to the older one, the doubt arises in what the other differs, maybe there are fixes including problems of various kinds, and maybe even on this one of the L2 cache (?), it would be interesting to find info on this, I don't think it's easy to find them though.

For the RTC instead, it could be very simple if it had a DALLAS type chip (but it's not here), I would have to look for interruptions on tracks that arrive at the CR2032 battery holder, if it's charged it's not said that the 3V arrive on the other side (?).

For instability instead, it's quite complex, the most common things I should try would be these:

RAM CPU and Controller

The RAMs only for testing purposes, I would change them with others, of course that's not necessarily the problem, but if it's better fine, otherwise you have to hope you don't have others just as problematic, here's maybe a memtest or something similar should be performed.

Chapter CPU, I investigated the various types of 486 even if the same, in this case DX33, some may have differences due to subsequent Rev, but I don't think they distort the CPU to the point of requiring different jumpers, in short it would be easier to try a CPU like an iDX4-100 and see what happens!?!

Controller, here this would be a big problem! Since it is integrated on the motherboard, theoretically it could be disabled and switched to one on an ISA or PCI card, I don't even like the idea, even if in those years, some used SCSI controllers, for CD discs or other.

In the meantime I added an ISA sound card, a SB16 Vibra CT4170, I have to finish configuring, we need the drivers, and I need to add a couple of lines on the boot files, from what I saw it was sold together with a 32X CD player and a couple of speakers, it was part (it seems) of a multimedia kit, which you added to the PC later.

Other tests on the SMT-486 Terminator II motherboard with the Intel 80486DX-33, I looked for info on the RTC chip, and it is clear that there is no external chip, the one that could normally be found in a DALLAS type chip, was included in one of the ALi chips, exactly it is located inside the M1487 chip, and the pin that feeds it, is the 12 (VBAT), here you find the 3V from the CR2032 battery, all quite simple, at least on the card.

Here it would be a big problem, if in PIN 12 we had 3V and the clock resets every time it starts, this means that it is most likely broken, or the internal circuit is broken, so I hope I don't find any voltage, and I proceed to the measurement, all this but with the PC off, I don't think it's necessary to motivate this, I'm just saying to limit yourself to disconnect the AT power supply cable (or possibly ATX) from the socket!

First of all, you have to make sure that the PINs of the battery holder lead, here everything seems ok, I have 3V measuring from above and also from below, so it's okay, but there is the problem that from below you can't see where the track goes, from above it's covered by black plastic, in short you can't see anything!!!

So I'm going to measure whether or not there are 3V on the PIN 12 of the M1487, after trying and trying again, there's nothing!!! That's better this way!!!

This gives me hope, that the problem can be easily solved, the simplest solution is a bridge, from here to the + of the CR2032, but first I would like to make sure that the track is not interrupted, usually you think of a superficial scratch, but sometimes it is the junction between the upper and lower PCB, which is the cause of the problem, the inside of the hole is easily oxidized, and too much corrosion interrupts the passage, I saw in some YouTube videos, that they pass a thin wire inside and then weld it superficially.

When the work is finished you have to try the PC, to see if it loses the stored settings, and if the clock works well, that is, if it continues to work with the PC off, then we can check if the L2 cache is detected correctly, because for now it is 0 KB, this would be one less problem to solve, which if added to that of the RTC are two, so the random instability would remain, maybe with a stroke of luck that problem will also disappear.

In reality, however, the problems of the L2 cache, could be due to something else, the cache communicates through the other M1489 chip, but here it is not as simple, first we will have to make sure that the chips are all working, in total there are 5, 4 caches and 1 TAG 🏷, I don't think there is the Dirty chip, but I could be wrong, I should check what it is connected to, that is, if there are PINs of the M1489 for the Dirty chip, in short, easy to say, difficult to do?

For today I don't think I can do anything else, having little time ⏱, but in the next few days I should start, with the reconnection of PIN 12 to the CR2032 battery, I hope everything goes ok

I make a short move to a 423 motherboard, an MSI 845 Pro VER:1 (MS-6529), with 1.5 GHz P4 and SDRAM, the chipset is not the usual i850 with RDRAM, but the derived version for the use of normal RAM, this makes the motherboard cheaper, because it is not necessary to buy the expensive RDRAM, but it must be said that the PC will be slower, with the same PC compared to the 423 with i850.

The motherboard, in previous startup attempts, turns off immediately, typical behavior of something short, which triggers the power supply protection, unfortunately it is not very obvious where the problem is, I imagine it is in the VRM area, because I see something in a pair of Mosfet (?), the electrolytic capacitors are not swollen, this would already be a good sign, but it does not mean anything, they could if checked be out of tolerance, and therefore have caused a failure as a result, the only thing you should do is check the Mosfet, strange values could be a sign, which would indicate where the problem is.

I see three large signed 82L02S, plus three smaller ones with the abbreviation D0134, I didn't see if I can find some relative pdfs, which maybe contain useful indications, to check them they should be removed, those in my opinion more critical, are Q6 and Q9, visible near the plastic for the CPU hook-in, with an inductor in between near an electrolytic capacitor.

Of course it would help, to have a diagram of how they are connected, just to understand where the current comes from, and what should be in the output, but not having one, I just have to see if they are somewhere connected to the PINs of the power supply, maybe the +12V or even the +5V (?) is connected, I think I tried to disconnect the 4 PIN auxiliary connector (JPW1), and it didn't work, this would indicate that the short is on the +5V.

I concluded the research with the ATX 423 motherboard, MSI 845 Pro VER:1 (MS-6529), with 1.5 GHz P4 and SDRAM, I checked the connections of the three large Mosfets and the three small ones, basically the three small ones are connected in the same way with the big ones.

For simplicity, I make a brief description of what it is like, and what the small Mosfet is connected to, then of the three PINs, the PIN 2 (the one in the middle) is connected to the yellow wire of the power supply (+12V), and this connection scheme is the same for all three (large and small), PIN 3 connects directly to the large Mosfet in PIN 2, and they are all connected to each other by the big coils.

On the small Mosfets, I didn't find anything strange, or that could cause the doubt that there is a possible problem, while on the big Mosfets the discussion changes.

Let's move on to the big ones then, PIN 3 is connected directly to the Black wire of the power supply, and this seems to be possible, later we'll see that it's just like that, I don't know where PIN 1 goes, but I don't think it's important to follow its path, because I'm looking for a short (!), so with the diode test of the digital tester, I try to measure on PINs 2 and 3, strangely exchanging the tips, the result is always the same, and the bell rings!

The measured value is almost 30, I don't know exactly, because the probe is not perfect, but after several attempts, it could really be between above 20 and 30, so if the tracks were connected directly, the value would definitely be less than 10 (maybe 1), so it remains to be seen if internally, those pins 2 and 3 are somehow connected, I look for the PDF and find one, here it is clear that there is a problem!

Between PIN 2 and 3, in fact there is a diode, which should pass the current in only one direction, probably one or all three Mosfets are faulty in the internal diode, since the current passes in both directions!

The biggest problem is for me to replace them, first you need the replacement ones, then removing the broken ones might not be too complex, I would cut the PINs 1 and 3, and I think it is possible to melt the tin that holds it attached, without damaging the motherboard, I have never done anything like this, but in the future I could try.

Last thing to add, this motherboard seems to come from an Olivetti PC, if I remember correctly an M7000, if it is indeed so, I have the Case 🏘 ATX, but inside I mounted a different motherboard, however with the repair is finished, and a working PC, I would put it back in the Olivetti PC.

I didn't put aside the SMT-486 Terminator II motherboard, since I ran out of benches with the i486DX33, I tried to solve the instability problem, even though that CPU doesn't need any heatsink, I put one (without fan), but it didn't make a difference, whether it was there or not, so it shouldn't be due to overheating of the CPU, but something else.

In the past it happened that a defective CPU, apparently works and in certain situations it freezes, I don't know if this could be a consequence of something else, for example of electrolyte capacitors not in perfect condition, but it is possible that the two problems, adding up create instability.

For this reason, I can't rule out that the cause is the CPU or the motherboard, it could also be the RAM, but for this there is the memtest, and for now I would say to exclude RAM problems.

If it was a CPU problem, it disappears by changing it, unfortunately I don't have another 486DX CPU, I absolutely don't want to try a 486SX, because I don't like it, it's a lame CPU or anyway a defective DX, at that time the FPU was not normally used, except for spreadsheets and graphics programs, so I take an AMD 486DX2-66 (my first DX2), I change a jumper that must be closed, and I try to start.

After a few failed attempts, I try to remove and reinsert again, and it starts!

The CPU is recognized at the right speed, unfortunately the L2 cache continues to be 0 KB !!!.

Started the PC, by pressing F1, because I still have ️ the problem that the BIOS is memorized, I proceed with the usual Phil's bench, there is a difference but the score is not double, as you might imagine, it is definitely faster than before, but I think it can be even faster, I should try with the RAM by setting Fast, for now I go by default, without changing anything.

During operation with the DX2-66, it worked without any blockage, completing all the benches, even the heaviest ones, in truth they last a few minutes, so it is not said that any problems have disappeared completely, but at least for now it seems to be going very well.

So, with this AMD 486DX2-66 CPU, the SMT-486 Terminator II PC, seems to be doing much better, I would say that from what I see, the instability problem has disappeared, the PC always starts at the first shot, something that rarely happened before, in short, it goes very well, so the first two problems remain to be solved, that is, the L2 cache not detected, and the clock and the BIOS settings that lose memory.

For the watch and BIOS settings, I haven't done anything yet, but I'm quite confident that I can solve it, the simple solution is a long bridge that directly connects the CR2032 battery to the M1437 chipset (PIN 12), but if I can I would avoid doing it, limiting myself to finding the point where there is the interruption, and then restoring the connection.

For the L2 cache, on the other hand, it could be a similar problem, that is, some too oxidized track, which has been interrupted, finding it could be complicated, because they go under the hooves, and then you don't understand where they arrive, I think it would still be better to have it working, in some moments using programs could be essential to speed up the calculations, maybe there is a minimal difference, like for example 10%, but it would still be time saved.

To check the operation, I went to the BIOS and changed the speed of the RAM to Fast, the rest was already by default on the lowest values, so maybe in the end it didn't make a big difference, indeed in some cases it is tiny, but there is an increase in the speed of the RAM, as you can see from the bench, for the moment I put the images of a couple, later I will see if I can group all the values, so that I can highlight the improvements with the various DOS programs.

In recent times, I have accumulated various cards to try and/or repair, let's say that I prefer to start with the most dated ones, or from some particular cards, my preferences are, without a particular order, the Socket 3 7 370 462, but also the 754 939 478, on the next ones, even if today they are now dated (up to about 20 years), I work with them occasionally, while I have almost nothing of a decade (more or less), the reasons are mainly that the most recent MBs are not for sale at bargain prices, and last but not least the main problem, is that the miniaturization of the components, It makes it impossible for me to work with it, at least with the simple tools 🛠 I have.

Lately I have repaired three 486 motherboards, a couple in truth did not need repairs, they were just to be awakened, and for each of these it was possible to make them work!

In particular with the latest 486 motherboard, a PCI SMT Terminator II, I was able to verify the gap between a simple i486DX and a DX2, it may seem insignificant, but it makes a big difference, some software manage to run decently, others instead would need a further boost, currently I am with a DX2-66, but later I will see to switch to a DX4-100, or possibly to something faster, I could get up to 5X86-P75 133 MHz, but honestly I don't think it's necessary to go, since we are already almost at the same level as a P75.

I noticed, however, that at that time, cards with fake cache chips were for sale, and deceptive BIOS, that is, starting the PC you read L2 cache of XXX KB, but your PC was not fast enough, this may have made many owners of 486 believe that if you wanted more speed, you had to switch to something newer.

In reality today we know that the main cause was the fake chips, and they were often accompanied by BIOS, very simplified and not optimized, all things that are visible, using diagnostic programs, which at that time were not very widespread (I guess).

However, the 486 PCI represents the latest evolution of these motherboards, usually with CPUs of at least 100 MHz, while for the previous typical VLB it was the DX2-66.

I go back to one of the motherboards to be repaired, some time ago I worked on some interesting motherboards, including a Socket 462/A motherboard, this is in particular a board of an above average level, it is a Gigabyte GA-7VAXP REV:1.1, it has some peculiarities, including a dual BIOS chip (welded!), four IDE 40PIN ports, three optional 1394 ports (FireWire), finally some strange sockets for me, for Secure Digital - Memory Stick - Smart Card Reader.

I have already done some work, I removed the swollen electrolytic capacitors in the VRM-CPU area, two of 1200uF 6.3V and four of 3300uF 6.3V, then I removed the PINs of the ATX socket, because some were burned and the plastic part was missing, I recovered all the pieces to be replaced, but I had to stop because I noticed two missing SMDs, of one of the two there was a little piece left, of the other instead I would not know if it was really there, or that was just a predisposition.

I'm thinking by the end of the week, to look online for fairly clear images, just to understand if two are missing, or just one, and then I should start the welding work, for the one that is definitely missing (BC284), it's clearly an SMD capacitor, I don't know the original values, and therefore the replacement will be the same as the ones next to it, both in size and color, I'll use a used component taken from a scrap motherboard, I hope it's of a fairly common type, and that it's not too complicated to find other replacements.

I don't think it's the top of the S.462, but maybe at that time it was a nice card, looking at the settings of the jumpers (screens), I notice that there are 3 FSBs:

100 MHz (ON OFF) | 133 MHz (OFF OFF) | 166 MHz (OFF ON), in reality the switch allows a fourth selection, which would be (ON ON), but I don't know if it corresponds to 200 MHz?, or no frequency, in this regard it will be necessary to consult the manual.

In any case, however it is possible to use CPUs with FSB 166 (333), such as the Barton Athlon XP, or others similar, and this allows you to match even with about 2 GHz, the performance of a P4 of about 3 GHz, then I don't know if this only applies to integers, or also take into account decimals, however after support for FSB 200 (400), Socket 462/A was abandoned in favor of 754, which was however limited by single-core CPUs, and unlike 462, there were no MB with two Sockets, which in the case of 462, required The use of two Athlon MP.

Today, I went back to 486 PCI, SMT Terminator II with AMD 486 DX2-66, to make other benches, in reality they are always the usual, but I found a third setting for memories, by default there is Normal, then I tried Fast, now it's Fastest's turn!

So I redid Phil's bench, and some others, consider that this setting does not change the CPU speed at all, but manages to speed up the Bandwidth of the memories, to see how much, let's start with Speedsys (n):

The score is 24.88, the same as the one with Fast memories, slightly higher than the 24.87 obtained in Normal.

More interesting are the values of:

Memory Bandwidth that goes from 76.02 MB/s to 179.40 MB/s, a big jump compared to 109.84 MB/s in Fast.

Data Cache L1 (8 KB):

From 60.95 MB/s we get to 61.78 MB/s, slightly better than the 61.40 MB/s in Fast.

Finally the Memory Throughput, which goes up from 36.71 MB/s to 47.08 MB/s, 40.64 MB/s in Fast.

Let's move on to Phil's other benches:

1) 47.6 - 50.0 - 52.6

2) 46.6 - 48.8 - 50.5

3) 31.2 - 32.9 - 34.0

4) 9.3 - 9.6 - 9.7

5) 10.0 - 11.0 - 11.6

6) 4.1 - 4.5 - 4.7

A) 63.51 - 72.66 - 80.48

B) 24.18 - 25.91 - 27.13

C) 6.9 - 7.5 - 7.8 (equal to e))

D) 3.3 - 3.5 - 3.6

So seeing these results, I would say that optimizing the PC is possible, maybe in my case the 60 ns EDO RAMs help, they are definitely better than other normal RAMs, however overall, the difference is greater in lighter benches, some for example those c) d) e), they are not suitable for such slow PCs, maybe a P200 would need to have a greater number of FPS.

I don't know if it's possible to do better than that, that is, by modifying other things, maybe yes, of course I currently use a mechanical HD, maybe you could load everything in a RAM Disk, but we must not forget, that I currently go without Cache L2, if I could fix it, could there be improvements in some cases up to 10-20%?

Here I wanted to add a couple of things on the latest motherboards, the S.462/A of the Gigabyte (GA-7VAXP), I found confirmation that the position of the switches both on ON, corresponds to FSB 200 (400), on TRW as further confirmation, there is a PDF with the supported CPUs, including a pair of Athlon XP (3000 and 3200) FSB 400, so I think that at the end of the work, I could experiment with an overclocking of a Barton 2500+, raising the FSB from 333 to 400, theoretically there should be no particular problems, but sometimes the success depends on the motherboard, but I imagine that the Gigabyte is Almost at the same level as other boards, maybe a step below the ABIT or similar, however we will see it as soon as I have finished welding the components, currently I have difficulty with the two SMD (or maybe one?), I can't find images of that detail (ATX connector), which confirm the presence of one of the SMD capacitors, could there be no ???

For the 486 PCI Terminator II motherboard, the transition to 486DX2-66, was a nice step forward, now the second problem to be solved, is that of the clock 🕰 that resets, and the BIOS that loses the saved settings, the problem is solvable, but I would like to locate exactly where there is an interrupted track, to do this though, I will have to disassemble the PC from the bench, and at that point I should also deal with the third problem, the L2 cache not detected (0 KB), likely that it is a contact of some socket, I will revive them with a professional contact spray, unfortunately I do not have Way to try the individual chips, so I can't rule out that they work.

With the Fastest setting of the memory, I saw that the PC works better, and according to some it could be done better, raising the FSB from 33 to 40 MHz, in practice I would get a 486DX2-80, honestly I'm not interested in overclocking, because I have some DX2-80 and I can put on one, but first as I said, I would like to be able to make the L2 cache work, but I would avoid switching to DX2-80, rather I would go straight to DX4-100, or even 5X86-P75 133 MHz.

Among the strange 486 CPUs, there is the AMD 486DX4-90, a 3X 30 MHz, at least looking at the manual, I don't see any way to select FSB 30, it was a strange frequency I think supported by a few, and those who own it, will probably have led it to work at 100 MHz.

I checked the ASUS CUV4X REV.1.03 motherboard that I have, there is a Pentium 733/256/133 and 128 MB of RAM, the card was tested in June 2024, the BIOS is from 07/25/2000 Revision 1006, so I ran a couple of benches, which I summarize here:

Speedsys 4.70

847.63

Memory speed index 395.78

Cache L1 (16 KB): 2799.33 MB/s

Cache L2 (256 KB): 1573.90 MB/s

Memory throughput: 398.24 MB/s

I did another benchmark with QTPRO, let's see the results:

431618 Dhrystones

233095K Whetstones

The results are obtained with the 8MB ATI RAGE XL PCI video card, I don't remember if I put an AGP video card later, so it's not said that the results can be slightly better, or the same (?)

Overall I would say that the motherboard has a certain backward compatibility, unlike the contemporary motherboard with i815, you can even install an ISA card, which was not possible on the boards with i81X chipset, even if some actually had it (ISA slot), for the sound board you have to add (ISA or PCI card), we notice that the motherboard was prepared for integrated audio, but here it is not there, the other PCI slots can be used for SATA RAID accelerator modem LAN cards or whatever you want, but for the type of PC I would not exaggerate, and I would limit myself To sound card and LAN, so that you can possibly connect another PC if you need to pour some files.

Lately I've had various 486s between my legs, I've managed to make them work, they're a little bit peculiar PCs, not the classic VLB, but of these classics, I have some in the collection, and one would be in repair, but in practice it's a motherboard badly placed (loss of acid battery), which I could fix thanks to a second board equals scrap, it's a job not impossible but quite demanding, I've always postponed it, because I don't have enough time to complete it.

The motherboard is a ZIDA 4DVS, theoretically it should be a good board, having the SiS chipset, it does not have support for CPUs from 3.3V or other voltages up to 4.0V, so only 5V.

The motherboard is prepared for the voltage regulator, beyond that I think you need to add other components, like some resistors, maybe not too complicated to do, also missing the BIOS chip, you can use a compatible chip and program it with the latest version, in this way it would be possible to use, any 486 or even the 5x86.

Being able to make this motherboard work, could be really complex, in addition to the numerous tracks to be fixed, there is the unknown of further failures, that is, if the problem of the tracks is the only one, or is there something else wrong.?

Of course, however, having a lot of free time, it would be a nice pastime, surely years ago the motherboard would have gone directly into the scrap, today instead if you are able to repair it yourself, it could be convenient to fix it, for two reasons, the satisfaction of making it work, and why not, the economic savings, today motherboards so working, they even cost too much (in my opinion).

When the repair is successful, some L2 and TAG 🏷 cache chips should be recovered, unfortunately they have been removed, so the board is devoid of all the chips (8+1), recently I saw some circuits made for faster and more spacious cache chips, if you don't need to have them like the originals, it would be a possible alternative, but I don't know if they are convenient.?

Speaking of 486 VLB, among the ones I have working, there is the ZIDA Z4DVP (or 4DVP), which was my first 486 DX2-66 VLB, I used it initially as the main PC, but within a few months, I put my paws on a BEK P405 PCI with AMD DX4-100, and then it became the spare PC, until in the early 2000s the PC power supply broke, later after replacing it with another second-hand one, it crashed in the first BIOS screen, after a few years I discovered that it was the L2 cache that blocked it, deactivating it from the BIOS, the PC works but it was a little slower.

Over the years I tried to solve this problem, but to this day without success, I think I tried other chips, with the same result, when it worked, I managed to double the capacity from 128 KB to 256 KB, adding another 4+1 chips.

The limits of this motherboard are essentially the 30-PIN RAM, which however could be expanded up to 20 MB or even 32 MB, using four pieces of 4 MB each, or even eight of 4 MB each.

Another doubt, there could be about the validity of the Opti 82C495XLC chipset, flanked by the IMP82C206P chip, I don't know if both are at the same level as the other chipsets, that is, if the PC in its best performance is fast enough, or if it doesn't shine and is less fast.

Later several years later I replaced the CPU with a Cyrix 486DX2-66, because it does not require a fan, and the PC is quieter, the original processor, the AMD 486DX2-66 is currently installed on the Terminator II, strangely even in that motherboard I have the same problem as the L2 cache, the only difference is that here you have to deactivate the BIOS from the L2 cache, while on the Terminator II not!

Looking at the cache circuit, I came to the conclusion, that you could suspect the U36 (74F08N), it's the one above the empty TAG 🏷 bench, the rest doesn't seem to be broken, but unfortunately the cache and TAG chips, despite being removable, I have no way to test them, to rule out that there is a defective one.

Maybe somehow, the U26 chip (74F125N) is part of the circuit, from what I see some of the pins go to the JP3 jumper, which is used for the capacity of the L2 cache, in case of problems it could mistakenly detect a wrong capacity, unfortunately this is also directly welded, so it is a bit complex to remove it, assuming it is possible to check its operation.

On another 486 motherboard under repair, signed ATC-1411B with CPU UMC U5SX 486-33F (welded), I have sockets for cache chips L2 and TAG 🏷, I don't know if the chips were originally there, but if anything it would be a problem to be addressed later, for now the motherboard is being repaired, I have removed the oxide by scratching it away, to remove the one on the contacts of the slots or sockets, I will try a professional deoxidizing spray for contacts, I will initially try it on some easily accessible points, so as to understand how effective it is.

After scraping the oxide, it was possible to see the copper underneath, in some places, it is not visible because it is no longer there, so I will have to fix one or more tracks, so as to guarantee the connection, but there are a couple of missing SMDs, one is an SMD capacitor, the other instead is a XXX OHM resistor 🕉, here to understand the value is impossible, or it is the same as the other near, or it has a different value, and without a clear image of another same motherboard, it is impossible (at least for me) to understand the correct value.

Unlike the ZIDA, here the cache chip circuit seems simpler, inside the TAG socket 🏷, it says 62256, I imagine it can indicate which chip to use, in the four sockets for the cache chips, it says 61102, I think it also indicates here the type of chip that can be used.

For the cache chips the sockets are 32 PIN, for that TAG 🏷 28, but it is possible that you can use the DIP28 cache chips, inserting them into the 32 sockets.

This motherboard is quite similar to the PINE PT-2068.1, unlike that, there is the predisposition for cache chips, for the rest, both the size and the equipment, seem the same.

The difficulty to repair this motherboard is not as low as you might think, in addition to the work to be completed at the top, there is a nice scratch on the tracks in the back, at least one seems interrupted, a complicated one from the minimum size, and the proximity of other tracks.

Overall it would be interesting, it could even be equipped with a voltage reducer, there is room for a linear regulator, this allows the use of 3-4V CPUs, the board is then prepared for a socket for normal 486 CPUs, I think it has 168 PINs, but in case you need to add some jumpers for the selection, some can be seen exactly where they go, this could clarify, what functions they really have.

Another job to be done is the replacement of the 72-PIN RAM socket, on the one hand the clip-stop lock is completely absent, on the other there is only the clip (without a stop), but I strongly advise against it, if you are not able to do it yourself, you should rely on more skilled people with this kind of work, personally I find it not so complex to remove the damaged part, rather I find it extremely complex to remove a 72 PIN socket, from a scrap motherboard, another possible difficulty, could be to free the holes without damaging the tracks.

So, if I found a detailed image of the area where you see R71, I would be able to read what numbers that resistor has, I could continue with the research, and continue the repair work, unfortunately this type of motherboards, I don't think it's so common to find, but at least initially, this board if put back into operation, it would be a typical PC of the early 90s pre VLB, and according to what I saw on the PINE PT-2068.1, the UMC CPU should be faster than an equivalent Intel CPU.

So, since I had referred to the PINE PT-2068.1, I was looking for motherboards similar to the ATrend ATC-1411B, so that I could identify the resistance value, and by chance I found an ATC-1411A motherboard that looks the same as the PINE PT-2068.1, but it's not over, I found others similar, on TRW:

Https://theretroweb.com/motherboards/s/tk-u5s-tk-v03a

Https://theretroweb.com/motherboards/s/ ... -atc-1411a

Https://theretroweb.com/motherboards/s/powertech-mb457

Let's say that apart from one that has two SIMM 72 PIN sockets, the others at least visually seem to me the same card, proposed with different brands and models, from what I understand, this card is from 1994, and it was probably a very cheap card, which unlike the PCCHIPS, did not have any L2 cache chip, while in the PCCHIPS the chips were fake, and the BIOS was a liar!!!

The evolution of this L2 cache-free card, is version B of the ATrend ATC-1411A, which has four sockets for L2 cache chips, probably the performance of this one, should be a little faster, and some had the linear regulator to support 3-4V CPUs.

Going back to the PINE-like motherboards PT-2068.1, I saw that there are some BIOS, which could be compatible (?), usually the BIOS is AWARD, but I found a file for AMI BIOS, I don't know if it's of the Windows type, at that time in fact, we switched from the Color one to the Windows-like one, which after a few years was abandoned.

It would be interesting, to compare the various BIOS of the other motherboards, it could emerge that maybe it is an identical BIOS, except in the identification line of the MB, this might be useful, if you encounter any problem with your motherboard, or if maybe the other has more selection possibilities, for example for memories or other.

I also saw that in one there are manuals, I haven't had time to check yet, but they could be the same as that of the PT-2068.1, I mean at the jumper level and related settings.

From the names of the brands marketed, there are in addition to PCCHIPS, where however the MXXX model is not specified, there is LuckyStar TK and even QDI, a sign that the motherboard was built, probably not too economically, maybe even with a certain quality, I don't think that a brand like QDI, agreed to brand a low quality motherboard (then this is not said that...).