• Please review our updated Terms and Rules here

I overclocked my IBM 5150...

Wonder if anyone has found examples of or personally attempted wiring the ram banks out of the equation 🤔
 
Not on an IBM 5150 motherboard.
I guess that is the advantage of a generic board

What components were different between the 16-64kb and 64-256kb?

I usually didn’t touch the motherboard and just changed dipp switches , one oddity that came inop to me I had to set to 576k despite 640kb being between the 64k board and 576k upgrade card.
 
What components were different between the 16-64kb and 64-256kb?
IBM 5150 motherboard - 64KB-256KB

Four banks of 64K-bit chips.

See the diagram at [here].

/CAS0 pulse (for motherboard RAM bank 0) generated by U47 when [A19=0, A18=0, A17=0, A16=0], which equates to address range 0 to 64K
/CAS1 pulse (for motherboard RAM bank 1) generated by U47 when [A19=0, A18=0, A17=0, A16=1], which equates to address range 64K to 128K
/CAS2 pulse (for motherboard RAM bank 2) generated by U47 when [A19=0, A18=0, A17=1, A16=0], which equates to address range 128K to 192K
/CAS3 pulse (for motherboard RAM bank 3) generated by U47 when [A19=0, A18=0, A17=1, A16=1], which equates to address range 192K to 256K

/RAS0 pulse (for motherboard RAM bank 0) generated by U65 when [A19=0, A18=0, A17=0, A16=0], which equates to address range 0 to 64K
/RAS1 pulse (for motherboard RAM bank 1) generated by U65 when [A19=0, A18=0, A17=0, A16=1], which equates to address range 64K to 128K
/RAS2 pulse (for motherboard RAM bank 2) generated by U65 when [A19=0, A18=0, A17=1, A16=0], which equates to address range 128K to 192K
/RAS3 pulse (for motherboard RAM bank 3) generated by U65 when [A19=0, A18=0, A17=1, A16=1], which equates to address range 192K to 256K

The exception to that is a motherboard RAM operation that is part of RAM refresh.
In RAM refresh:
- No /CAS pulses are generated, because the RAM refresh mechanism used in a 5150 is 'RAS only refresh'.
- /RAS generated for all four banks simultaneously, irrespective of address.


IBM 5150 motherboard - 16KB-64KB

Four banks of 16K-bit chips.

Per the 64KB-256KB motherboard only differing in that an additional two address bits (A15, A14) are used to achieve:

RAS+CAS for bank 0 when address is 0 to 16K
RAS+CAS for bank 1 when address is 16 to 32K
RAS+CAS for bank 2 when address is 32 to 48K
RAS+CAS for bank 3 when address is 48 to 64K


IBM 5160 motherboard - 64-256KB

Four banks of 64K-bit chips.

Motherboard RAM bank decoding is done by a ROM. That ROM is U44 in the diagram at [here].

The ROM contains four tables, and the table in use is selected by the jumpering of E2. On a 64-256KB motherboard, there are no jumpers on E2, which results in the first table being used, and that table assumes 64K-bit chips in all four RAM banks.


IBM 5160 motherboard - 256-640KB

Two banks of 256K-bit chips and two banks of 64K-bit chips.

Like for the 64-256KB motherboard, except that there is a jumper on pins 1 and 2 of E2, resulting in the second table in the ROM being used, that table assuming 256K-bit chips in banks 0 and 1, and assuming 64K-bit chips in banks 2 and 3.
 
I once had a system with 256k and bad chips in every bank.
I didn’t have spare 64k chips and desoldering then used 16k chips instead then installed a memory card and everything worked .
That does not sound right.

For example, at [here] is the manual of a particular XT clone motherboard, one that can take various sizes of RAM chips on the motherboard.
In the '2-2 How to Change the System Board Memory (256KB/640KB)' section of that manual (pages 12 and 13), jumper pads JP3 and JP4 need to be changed according to the RAM chip size.
 
Wonder if anyone has found examples of or personally attempted wiring the ram banks out of the equation
An analysis of the circuit diagram for the particular make-model of motherboard is expected to provide a solution.
The solution will be unsuitable for some people because the solution possibly involves things such as cutting tracks and adding resistors.
 
Ok, update time… desoldered the bank 0 ram, put sockets in, loaded it up with new ram, and…. computer not booting properly.

Here’s where I’m stuck… my monitor (5151) isn’t turning on at all now/not getting power it seems, board kicks on and isa slots are getting power (leds on)… I hear the speaker kick on as well, but that’s it… I’m curious if a pad lifted during the process, but I didn’t see any issue during my soldering… any ideas?
 
Yup, tried that… same result. Also took out the pc-sprint and put the 8284A chip back in to take that option out of the loop…
 
I once had a system with 256k and bad chips in every bank.

I didn’t have spare 64k chips and desoldering then used 16k chips instead then installed a memory card and everything worked .

Not sure what controls density but hopefully it’s one of the few situations where it detects on its own.
It has to be said that many RAM tests does not detect if you use 64k chips in a 256k bank. Everything will appear work fine... until DOS (or wathever is ruinning) uses more than the first 64KB of that bank.

An analysis of the circuit diagram for the particular make-model of motherboard is expected to provide a solution.
The solution will be unsuitable for some people because the solution possibly involves things such as cutting tracks and adding resistors.
For the 5160, two more modes are available other than the mentioned 64k-256k and 256k-640k modes. One of these use only 128k per bank (nice if you have 256k RAM chips with failing bits in the upper half), the other uses only the first 16k of each chip (originally compatible with a rare DRAM-chip from Intel). This latter mode also needs additional jumper-changes at E3 and E4, changes that IBM never documented themselves.

If the board is a clone board, then your situation can be anything. If it isn't printed on the PCB silkscreen, you may have to trace up the circuit and decode the mapping circuit yourself.
 
Have you made sure all flux is cleaned of the board, that stuff can be slightly conductive.
Also how do your psu voltages look?
 
Reflowed the connections and didn’t make a difference… checked continuity on the back traces and everything checked out, no cross-joins either… still need to check the front traces and also check voltages… I kind of suspect a power issue as well but not sure. Seems odd, but starting to feel like that might be something…
 
Perhaps try double checking the dip switch settings... Is there anything not set exactly to your current configuration?
 
Didn’t change any of the dip switches, I should have tested more thoroughly at first and started with one bank at a time, but oh well… I’ll figure it out eventually!
 
Also the variable capacitor next to the crystal. Although only supposedly used for tweeking cga output, can also be a factor when it comes to motherboard startup. If you touch it don't go crazy!!! Just a few thousands of an inch in rotation can be a sufficient change.
 
Here’s where I’m stuck… my monitor (5151) isn’t turning on at all now/not getting power it seems,
At [here], see the following two sections:
- SYMPTOM: No video, but a stable raster shows if brightness turned up
- SYMPTOM: No display (not even a raster)

One of those two symptoms will apply to your 5151.

If you can't resolve the issue based on the information in those two sections, create a new thread for your 5151.
 
Ok, update time… desoldered the bank 0 ram, put sockets in, loaded it up with new ram, and…. computer not booting properly.
Reading earlier posts, this is substituting the µPD4164C-15 chips with just-acquired KM4164B-10's. Working before, but not after the change.

board kicks on and isa slots are getting power (leds on)… I hear the speaker kick on as well, but that’s it… I’m curious if a pad lifted during the process, but I didn’t see any issue during my soldering… any ideas?
A click from the speaker is somewhat normal when the motherboard gets power. The POST in the IBM BIOS ROM does not click the speaker.

Did the seller of the KM4164B-10 indicate that the chips were tested? If not, maybe one (or more) is faulty. The combination of IBM BIOS ROM (for 5160) and a faulty RAM chip in bank 0, produces a 'motherboard appears to be dead' symptom. Try swapping the KM4164B-10's with the mystery chips (that must be 4164 class) that you have in the other banks.

Beyond that, try Ruud's Diagnostic ROM at [here].
 
I've had faster DRAM chips (80ns) be unstable in a 4MHz machine before. That was not due to faulty chips, but rather the fast transients causing issues with how the PCB was laid out.
 
@modem7 Do you have a photo or schematic that shows the top traces? I can't seem to find a clear photo where the traces aren't hidden under the chips to be able to test for continuity...
 
Back
Top