VOGONS


Looking for a 486 mb

Topic actions

First post, by Jan3Sobieski

User metadata
Rank Member
Rank
Member

Hello,

I'm a long time lurker, finally decided to post 😉

I'm looking for a 486 motherboard, for a baby AT case. I have purchased one on ebay a while ago but to my dismay, it only has settings for 3.3v,3.45v,3.6v and 4v. No setting for 5v for some reason. it's a TK8881 FA20 motherboard. Looking on google, there's a picture of the board out there somewhere, but it shows the board with jumpers to set it to 5v or 3.*v. whereas mine just has it "pre-soldered" at 3.*v (other jumpers set it to the other four voltages). So now I have two questions:

As I am pretty handy with the soldering iron, can I "re_solder" it and add the jumpers, or would that mess up the whole mb?

Edit: Deleting search for another MB. ( I actually have 3 of these 😉 I soldered the jumpers in and all motherboards work like a charm! Thanks everyone!

f8b91872174323.gif
411e7372174326.gif

Last edited by Jan3Sobieski on 2010-04-05, 19:58. Edited 1 time in total.

Reply 1 of 15, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

Your board looks exactly like the syl8884 motherboard I have. I'm pretty sure mine supports (or atleast the online manuals I find say so) both 3.3v and 5v chips.
Either your motherboard is realy missing some part so it realy doesn't support 5v chips...or 5v was perhaps purposefully disabled for whatever reason (not likely, but it can happen).

If you like, I could photograph my syl8884 motherboard to see if any part on your mobo is missing. If it's exactly the same, theres a good chance that soldering will help you get a 5v chip runing 😉

That, or you could try 4v and see if your cpu still works. Undervolting will not damage any hardware btw, it should be safe to try 😉

Edit:Unofficial homepage is here:
http://th2chips.freeservers.com/syl8884/

Reply 2 of 15, by Jan3Sobieski

User metadata
Rank Member
Rank
Member

Thanks for the reply!

If you can make a high res photo of your motherboard, that would be awesome. I've found it on google images, but the quality is quite poor. The differences I was able to distinguish are the chip in the bottom right corner, mine has a bigger VIA chip, whereas the syl has a smaller chip, also, in the top right corner (where the jumpers are supposed to be) my motherboard has an additional capacitor, whereas the syl doesn't. The only way to be sure is if you could make a high res photo so I could compare.

I already tried running it at 3.3v and 4v, and it ran fine, but i'd rather have it run at the voltage it's supposed to. I didn't do a lot of benchmarking but i know the system is unstable if it's not running at the proper voltage.

I located a TK-8881 FA20, just like mine on the internet and same story, exact same motherboard, even the exact same chips and capacitors, but this one has jumpers. The only thing i noticed is this one part close to the keyboard connector that looks like there's a red wire going from one spot to another. My motherboard doesn't have that, but again, this might be a crappy picture.

5e601d72434403.gif

Reply 3 of 15, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

I have a not too detailed pic of my syl8884, It's the middle left one.
DSC00201.jpg

I was 🤣 when I looked at my syl8884 and it seems I don't have those voltage jumpers either, I never noticed that before!

Seems we're in the same boat heh 😜

Though the few pics I did find on the internet seem to all have no jumpers over there. Perhaps all of these were made this way? I know this board was made at the end of the 486 reign and thus was very cheaply made. Soldering the jumpers to the newer lower voltage 486's would've saved a few cents on manufacturing costs.
If this is so, perhaps the soldering trick will work?

It seems you have the realy big cache chips on your motherboard. Those seem to be quite hard to find. Mine has the smaller cache chips.
Btw, this board supports EDO 😀

Edit: Could you please explain to me how you made the thumbnail of your pic? I'm a lil bit of a forumphoto newb 😜

Reply 4 of 15, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie

Have you considered the possibility that the board autodetects the CPU voltage? Many late 486 boards had this feature.

The only way to find out for sure is to measure the voltage at the soldered side of the CPU socket while the system is running - which can be a bit risky if you're not familiar with it.

Reply 5 of 15, by Jan3Sobieski

User metadata
Rank Member
Rank
Member

Tetrium, thank you for providing the picture, but it's too small for me to actually make out all the differences. Can you get a higher resolution one somehow? I upload my pictures to imagebam. They provide you with a thumbnail when you link to it.

5u3. I considered that possibility, unfortunately i don't know where exactly do i measure the voltage. If i knew, i would try. Is there a technical drawing of a socket 3 anywhere which shows it? I'll try to google for that, but if you know the answer, please post.

Reply 6 of 15, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie

There is a lot of contradicting and even plain false information on the net, but these two sources look rather reliable:

Upgrading and repairing PCs See pages 89 and 127 for relevant info.

Jan Steunebrinks site on 5x86 CPU upgrades also has some info on Socket3 voltage detection.

Hope this helps...

Reply 7 of 15, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

Woops, I did mention I would get a more detailed pic of the mobo in a day or 2, but it somehow got lost. I'll take a better pic of the mobo, as detailed as my camera can handle

Reply 8 of 15, by Jan3Sobieski

User metadata
Rank Member
Rank
Member

That would be awesome Tetrium.

Well, after some initial testing, it seems like the motherboard doesn't automatically choose the voltage. The voltage is whatever the remaining jumpers set it to. 3.3v, 3.45v, 3.6v and 4v. However, checking the voltages exactly shows that they vary (obviously) but setting it to 4v actually produces ~4.5v I have to take into consideration that nothing else was draining power from the motherboard, maybe a video card would put a little stress and the voltage would drop to ~4v.

So now the question remains, can soldering in additional jumpers will make the board go to 5v?

Reply 9 of 15, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
Jan3Sobieski wrote:

That would be awesome Tetrium.

Well, after some initial testing, it seems like the motherboard doesn't automatically choose the voltage. The voltage is whatever the remaining jumpers set it to. 3.3v, 3.45v, 3.6v and 4v. However, checking the voltages exactly shows that they vary (obviously) but setting it to 4v actually produces ~4.5v I have to take into consideration that nothing else was draining power from the motherboard, maybe a video card would put a little stress and the voltage would drop to ~4v.

So now the question remains, can soldering in additional jumpers will make the board go to 5v?

The voltages rely completely on the (quality of the) powersupply 😉
-If- you experience weird voltage drops or jumps with a good power supply then there could be something wrong with the capaciters which is very unlikely since this board was made years before the capaciter plague.

My educated guess is this:
If the jumpers were omitted just to save on manufacturing costs, theres a good chance that soldering it to 5v will in fact work.
Perhaps you could try to set it to 5v with a temporary fix (maybe put a little bit of tape between the solder dots and connect using silver paint?) and test with an expendable processor?
I don't know if theres any way to actually be for sure it in fact will be 5v and not something totally else, only newer mainboards have volgate sensor readings in the BIOS

Reply 10 of 15, by h-a-l-9000

User metadata
Rank DOSBox Author
Rank
DOSBox Author

I've made scans of the mainboard.

http://archiv.to/GET/FILE4BA37D90859B2

The jumper block bypasses the transistor that is used for voltage regulation so the 5V can go to the CPU directly.

Did you have a CPU in the socket when measuring the ~4.5v? Without a load behind the regulator the voltage can go up a bit.

1+1=10

Reply 11 of 15, by Jan3Sobieski

User metadata
Rank Member
Rank
Member

h-a-l-9000: I only measured the voltage in the socket itself, without the cpu. Once the cpu was in, and i measured from the back side, it was a lot more accurate.

And now i have some news.
Thank you tetrium and 5u3 for responding, and thank you h-a-l-9000 for posting such good resolutions photos. I was a little skeptical about soldering the pins in because of the part where the keyboard connector is. Your motherboard has some sort of resistor ( ?? ) there, and mine had nothing. I was worried that something else was needed for it to run at 5v. Nevertheless, i had to try and i soldered the pins in, connected the jumpers and voila! I'm getting 5.2v without the processor and ~5v when it's in.

Now for the bad news... ?
I ran speedsys before i soldered the pins in and measured the performance with 8mb of ram at various voltages. I'm not entirely sure why, but it seems i was getting the highest scores when my cpu ( i486dx2-50 ) was running at 3.3v! I then tried it again and i tried all the other voltages twice and was getting a score of around 12.3, some voltages had slightly lower, maybe 12.2. I just ran speedsys with the cpu running at 5v, and on the first run i got 11.8! I then changed the cache setting to write-through instead of write-back and got 12.1
What's the deal? Are my other settings in the bios not configured correctly for the cpu?

Is there a list of best award bios settings for this and the dx2-66 processors? There's all bunch of stuff there and i honestly don't remember what they all do.

Reply 12 of 15, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie

It is normal that the SpeedSys CPU benchmark doesn't come up with the exact same score on each run. However, with a DX2-50 you should get a score about 20. Lower scores could be caused by simple things, like turbo button turned off.

As for the BIOS settings: I'm not sure anyone knows the available settings for this board by heart, but if you post a photo of the settings (or write them down), we'll be glad to help.

Reply 13 of 15, by Jan3Sobieski

User metadata
Rank Member
Rank
Member

I'm gonna try and figure out the turbo thing. the motherboard has a turbo led header and a turbo switch header, however, my case only has a turbo led, no switch. I'm not sure how to connect that. At boot, the bios indicates the processor at 50 mhz, and the turbo switch header is open, so again, i'm not sure.

Here are the screenshots from the bios. If someone can shed some light on what are the optimal settings for an i486dx2-50 and i486dx2-66, that would be great, also, if i wanted to put a cyrix chip or an amd chip in, what should be changed, if anything?

I already noticed i have to enable EDO DRAM install option setting for the mb to use edo ram, otherwise it just gets stuck before it reads the os.

46c02072983995.gif 15ad9c72983997.gif 620c7d72983999.gif e7ef4772984000.gif 8e3ba072984002.gif

Edit: I found an old turbo switch/led, hooked it up. by default, turbo is on, so whatever benchmarks i performed, it was with turbo. If i turn it to off, the system goes to a crawl, it bould barely go through memory test. the cpu score was at something like 3.5

Reply 14 of 15, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie

The only CPU-related setting is the one called L1 Cache Update Scheme. To find out if your CPU supports WB, look at these identification pages: Intel, AMD.

Most of the values are already set to the fastest option, but you could try and change these:

Early Cache Write Mode: Enabled
Slow Refresh (1/4) Freq: Enabled (No, this is not a typo 😉)
PCI Posted Memory Write: Enabled
Burst Copy-Back Option: Enabled
IBC DEVSEL# Decoding: Fast

Your system may become unstable with these settings enabled.

Reply 15 of 15, by Jan3Sobieski

User metadata
Rank Member
Rank
Member

I don't know why, but for some reason, speedsys now shows a score of 18.88. I tried running it several times, same score. It's wierd. I haven't changed anything...or maybe i did and now i don't remember what it was?

I am having a problem though and i can't determine what it is. The computer will just randomly freeze. Sometimes i boot into dos, load norton commander and trying any options within the program (like changing left/right views) freezes the computer. I was trying to figure out whether or not i have fake cache, but haven't found a reliable way to check. When i disable external cache in bios, i still get the same score in speedsys. I ran memspd and it says that "This machine seems to have one code cache!?" I'm not sure what that means. Running Cachechk shows 8kb of L1 and 256kb of L2 cache which seems to point that it's actually there, is that right? or does cachechk only check what the bios reports? When i disabled external cache in the bios, cachechk said that there is no L2 cache and i might have fake chips. But that's expected if i disable it.

I'm not sure what to think about it. Like i said, running speedsys with external cache enabled or disabled doesn't change it's score, neither does it change it in sysinfo. Is there another benchmark i can run that will show a difference with it enabled vs diabled (provided they're not fake)?