VOGONS

Search

Common searches

Half of the memory slots not working - 386 board

Topic actions

First post, by hasat

Posted on 
User metadata
Rank Newbie
Rank
Newbie

Hello, I am out of ideas, what could be wrong with this board. Only one memory bank (bank 0) working. These are slots 1-4. Memory in slots 5-8 are not detected. Used 8 identical 1 MB modules with three chips on it, tried randomly changing their position in slots and nothing - still detected only 4 MB of RAM.
Board is DTK PEM-4036YB

Here are results of memory slots probing:
DATA0-7 connected to an identical slot in second bank
CAS connected together in one bank
CASP connected together in one bank
RAS connected together in one bank
ADR0-11 each line connected to each slot. ADR11 not connected
WE connected to an identical slot in second bank
DP connected to an identical slot in second bank
QP connected to an identical slot in second bank

My opinion is that something is wrong with CAS and RAS signals, because their are independent for each bank. Their are connected from chipset (SL82C461 chip) through 74F244 buffer (20 pin SMD chips between memory slot and BIOS chip) to slots.
Or maybe bank 1 is somewhat disabled in BIOS. I tried program AMISETUP and not found any hidden options in BIOS related to this.
I replaced Dallas chip with dead battery, board now remembering it's setting, but nothing changed.

Thanks for any tips how to resolve this problem!

Attachments

  • DSC_0024.jpg
    Filename
    DSC_0024.jpg
    File size
    1.38 MiB
    Views
    645 views
    File license
    Fair use/fair dealing exception

Reply 1 of 8, by majestyk

Posted on 
User metadata
Rank Oldbie
Rank
Oldbie

If all contacts, traces and solderings are o.k., I would test / replace the buffers first. Measuring the resistance between every single input/output and ground with a ohmmeter (red tip to ground, 10R or 100R range) can show if there are any defective buffers.
You should also check the solderings of each single leg of the memory controller AND periphery controller with a pointy needle or cutter blade.

If there are no problems found, a defective SL82C361 (memory controller) or (probably more likely) C362 (periphery controller) has to be taken into consideration.

Btw. The population of the Cache-banks seems quite odd. Have you checked with the manual?

Reply 2 of 8, by hasat

Posted on 
User metadata
Rank Newbie
Rank
Newbie

Cache is detected and populated correctly for 128 kB by "PEM-0036Y Settings Sheet" from ultimateretro.net. Also silkscreen on board correspond to this. I measured resistance by your instructions and it's very high (5,6 MOhm) in all I/O in all three buffer chips.

Reply 3 of 8, by weedeewee

Posted on 
User metadata
Rank l33t
Rank
l33t

Just noticed, you only mention CAS CASP and RAS.
intel 32bit will have four CAS lines to address each byte of the 32 bit word separately .

Right to repair is fundamental. You own it, you're allowed to fix it.
How To Ask Questions The Smart Way
Do not ask Why !
https://www.vogonswiki.com/index.php/Serial_port

Reply 4 of 8, by hasat

Posted on 
User metadata
Rank Newbie
Rank
Newbie

Correct me if I'm wrong:
One CAS and RAS signal for each bank
Here are 2 banks, each four 8bit slots = 32bit for 386DX

= 2x CAS and 2x RAS signal from chipset to memory

Reply 6 of 8, by hasat

Posted on 
User metadata
Rank Newbie
Rank
Newbie

Good tip! When i got this board, BIOS chip don't have it's original sticker, but this silver piece of tape over window. Probably someone manipulate with BIOS. Must find original BIOS and compare with my current.

Reply 7 of 8, by hasat

Posted on 
User metadata
Rank Newbie
Rank
Newbie

Important update: it was wrong BIOS! I purchased Xgecu TL866II Plus programmer, tried some other BIOS and after three attempts finally find correct one! It's directly from DTK,(not Award or AMI), not much settings for chipset, but it works.

https://hwmuseum.pp.ua/motherboard/2053.html

Go to top of page Go to top of page
Back to General Old Hardware
Close window
Close window