Yes is possible, once I had a motherboard (an OEM from Acer) that fried three or four modules until I figured out what's happening.

Look at the motherboard, specifically the area around the SIMM slots, and search for bulged capacitors, burned traces or things like that.

shiva2004 wrote:

Yes is possible, once I had a motherboard (an OEM from Acer) that fried three or four modules until I figured out what's happening.

Look at the motherboard, specifically the area around the SIMM slots, and search for bulged capacitors, burned traces or things like that.

I did check it out the best I could and this was only thing I was able to find:

The topmost capacitor does not seem as shiny as the others to me.
Anyone knows if that is just corrosion on top of the capacitor or if it is really broken?

It doesn't look good, that's for sure. What's the brand?
Also I don't know if it's the photo, but I don't like the look of the diodes just below it either.

shiva2004 wrote:

It doesn't look good, that's for sure. What's the brand?
Also I don't know if it's the photo, but I don't like the look of the diodes just below it either.

Diodes are probably ok. At least my multimeter gave me a beep with each of them using diode test mode.

It could be the board sends too much voltage to the simms and slowly kills them. It would be very unlikely that cap is to blame, it seems to me it's just a filtering one.

Could broken cap cause the post memory test to fail even if the simm is working?
I don't have any other motherboard that likes 72pin EDO modules, so I can't test them.

Nevermind about what I said about beeping when I tested diodes. It beeped because I forgot to take off the RTC battery and it was giving diodes voltage before testing. It showed correct values for each diode after I removed the battery.

I managed to find another motherboard that worked with EDO. It was actually a pentium motherboard that was actually older than the 486 motherboard.
I ran memtest86+ with the modules and I got hundreds of errors, so they must be broken. Now the main problem is how to find out if they were broken in the first place or did the 486 motherboard break them. Any ideas?

Baoran wrote:

I managed to find another motherboard that worked with EDO. It was actually a pentium motherboard that was actually older than the 486 motherboard.
I ran memtest86+ with the modules and I got hundreds of errors, so they must be broken. Now the main problem is how to find out if they were broken in the first place or did the 486 motherboard break them. Any ideas?

None that don't involve sacrificial RAM.

Should it make any difference if the sacrificial ram is fpm or edo? Would edo break more easily than fpm?
I could buy some 4Mb fpm ram for 3 euros from ebay.
Too bad about those broken ram because it was the first time I saw 50ns ones. I have only seen 60ns or 70ns before.

Maybe you can measure the voltage being supplied to the SIMM slots with your multimetre?
In case there is indeed any under- or over-volting going on.

In case there is indeed any under- or over-volting going on.

Undervolting can't kill hardware and SIMMs are powered directly from 5v line, so PSU may be involved.

Baoran wrote:

I managed to find another motherboard that worked with EDO. It was actually a pentium motherboard that was actually older than the 486 motherboard.
I ran memtest86+ with the modules and I got hundreds of errors, so they must be broken. Now the main problem is how to find out if they were broken in the first place or did the 486 motherboard break them. Any ideas?

Hang on... were you trying to run a 486 with EDO? Even though EDO is supposedly a backwards-compatible superset of Fast Page, it's very rare for a 486 to run reliably with EDO.

Conversely, maybe you're just mixing up your terms and it really is 72p FP, not EDO - and rated for a lower speed than the Pentium runs it at (eg 80ns or something).

Could you give exact specs of the SIMMs in question (or if unknown, the text on one of the memory chips on them)? And exactly which motherboards are you using?

dionb wrote:

Hang on... were you trying to run a 486 with EDO? Even though EDO is supposedly a backwards-compatible superset of Fast Page, it […]
Show full quote

Baoran wrote:

I managed to find another motherboard that worked with EDO. It was actually a pentium motherboard that was actually older than the 486 motherboard.
I ran memtest86+ with the modules and I got hundreds of errors, so they must be broken. Now the main problem is how to find out if they were broken in the first place or did the 486 motherboard break them. Any ideas?

Hang on... were you trying to run a 486 with EDO? Even though EDO is supposedly a backwards-compatible superset of Fast Page, it's very rare for a 486 to run reliably with EDO.

Conversely, maybe you're just mixing up your terms and it really is 72p FP, not EDO - and rated for a lower speed than the Pentium runs it at (eg 80ns or something).

Could you give exact specs of the SIMMs in question (or if unknown, the text on one of the memory chips on them)? And exactly which motherboards are you using?

This is the motherboard.

And when I got it, it came with two 72 pin 16Mb 50ns edo ram modules. It says HTEC TBS6416B4C50 on the chips of the modules.
At least I know now that they are bad because of all the errors I got with memtest86+ with the socket 5 motherboard. Only thing I can't be sure of is if the motherboard broke them or if they were already bad when I got them.

The Serpent Rider wrote:

In case there is indeed any under- or over-volting going on.

Undervolting can't kill hardware and SIMMs are powered directly from 5v line, so PSU may be involved.

I did check the psu with multimeter before and both 5V and 12V were ok when measured from a molex connector.

Baoran wrote:

This is the motherboard.

[...]

That doesn't look like Socket 5 to me...

And when I got it, it came with two 72 pin 16Mb 50ns edo ram modules. It says HTEC TBS6416B4C50 on the chips of the modules.
At least I know now that they are bad because of all the errors I got with memtest86+ with the socket 5 motherboard. Only thing I can't be sure of is if the motherboard broke them or if they were already bad when I got them.

No, you don't know they are bad. You know that you get errors in combination with this motherboard and its current settings. That could be beacuse they are dead/dying or because they - or the memory controller, or the onboard cache - are being run out of spec/capability.

Of course, 50ns is very, very fast, but there's something fishy about these chips which leads me to doubt the rating. I've never heard about "HTEC", but the rest of the chip code sounds familiar. "TB" is the vendor code for M-Tec, but it makes no sense beyond that:

TB = M-Tec
S = SDRAM (!?! EDO would have been "E")
64 = 64Mbit density
16 = x16 organisation
B4 = package type
C = die revision C
50 = good question, if it's SDRAM this makes no sense at all, but SDRAM in a 72p SIMM makes no sense either. So either this is wrong or the S is wrong or the whole chip code is nonsense.

Judging by this code, if this was a SIMM it should have two chips to get to 32 data lines, which would indeed come to 128Mb, so 16MB (I'm assuming your "16Mb" is a typo), but then it's an SDRAM SIMM, not EDO. Tbh, I'd expect 64Mx16 SDRAM chips in 64MB DIMMs, and then with latency between 7 and 10ns.

So this chip code screams "FAKE!" as it stands. That most definitely doesn't mean it's not broken, but I'd expect it to rather be some bargain basement 70ns or worse stuff relabled as 50ns.

That is LuckyStar LS486E, Revision D. Not EDO compatible.

I have that exact board and Mine works perfectly with EDO ram, it has I think the latest version of the SIS chipset that supports EDO.

http://www.elhvb.com/webhq/models/486pci/ls486ed.htm

dionb wrote:

That doesn't look like Socket 5 to me... […]
Show full quote

Baoran wrote:

This is the motherboard.

[...]

That doesn't look like Socket 5 to me...

And when I got it, it came with two 72 pin 16Mb 50ns edo ram modules. It says HTEC TBS6416B4C50 on the chips of the modules.
At least I know now that they are bad because of all the errors I got with memtest86+ with the socket 5 motherboard. Only thing I can't be sure of is if the motherboard broke them or if they were already bad when I got them.

No, you don't know they are bad. You know that you get errors in combination with this motherboard and its current settings. That could be beacuse they are dead/dying or because they - or the memory controller, or the onboard cache - are being run out of spec/capability.

Of course, 50ns is very, very fast, but there's something fishy about these chips which leads me to doubt the rating. I've never heard about "HTEC", but the rest of the chip code sounds familiar. "TB" is the vendor code for M-Tec, but it makes no sense beyond that:

TB = M-Tec
S = SDRAM (!?! EDO would have been "E")
64 = 64Mbit density
16 = x16 organisation
B4 = package type
C = die revision C
50 = good question, if it's SDRAM this makes no sense at all, but SDRAM in a 72p SIMM makes no sense either. So either this is wrong or the S is wrong or the whole chip code is nonsense.

Judging by this code, if this was a SIMM it should have two chips to get to 32 data lines, which would indeed come to 128Mb, so 16MB (I'm assuming your "16Mb" is a typo), but then it's an SDRAM SIMM, not EDO. Tbh, I'd expect 64Mx16 SDRAM chips in 64MB DIMMs, and then with latency between 7 and 10ns.

So this chip code screams "FAKE!" as it stands. That most definitely doesn't mean it's not broken, but I'd expect it to rather be some bargain basement 70ns or worse stuff relabled as 50ns.

That picture was the original socket 3 motherboard that I suspected that broke the memory modules and gives memory test fail error at post. The socket 5 motherboard was a pentium motherboard that I used to try to verify if the edo ram modules that came with the socket 3 motherboard are really bad or not. Even if they passed the post memory test on the socket 5 motherboard, I got hundreds of errors in memtest86+. I would be really happy if there was nothing wrong with the socket 3 motherboard and it doesn't break any other ram I might try to use with it.
Not sure how I could test if the ram is actually fake or not. They have 3 of those Htec chips on both sides of the module and the text is very faint on most chips.

You can't test whether it's a fake, you can just test different settings to determine its true capabilities, regardless of what is claimed by the inscription.

Given the photograph I'm pretty sure the inscription is fake:
- lousy print quality
- non-existent brand name similar to existing one
- nonsensical chip code

One interesting point: this is a parity SIMM (or at least it pretends to be). Do you have parity on or off in BIOS on the boards? And which So5 board are we talking about here?