Commodore PC 20 III "Bad ram at 8075D = 00h, expected = 00h"

[dippaolo]

Error during startup:

"Bad RAM at 8075D = 00h, expected = 00h."

Which is the ram memory? Where to look for the cause?

badram.jpg

[modem7]

First, be aware that I am unfamiliar with the Commodore PC 20-III.
Maybe someone who is will 'chime in'.

From information on the Internet, I see:
* The PC 20-III is simply a PC 10-III with a hard drive added. ASSUMPTION: From that, the two motherboards are the same.
* 640 KB of RAM

"Bad RAM at 8075D = ...

"Bad RAM at 8075D" equates to address 526173, approximately 514 K.

That fits with the displayed "512 Kbytes OK"; that is, at least 512 KB of the fitted 640 KB is okay.

From what I make of the technical information online:

RAM bank 0
------------
256 KB sized, addressed 0 to 256K.
Eight 41256 type RAM chips, each providing 1 bit.
No parity bit.
U312 = bit 7, U311 = bit 6, U310 = bit 5, U309 = bit 4, U308 = bit 3, U307 = bit 2, U306 = bit 1, U305 = bit 0

RAM bank 1
------------
256 KB sized, addressed 256K to 512K.
Eight 41256 type RAM chips, each providing 1 bit.
No parity bit.
U320 = bit 7, U319 = bit 6, U318 = bit 5, U317 = bit 4, U316 = bit 3, U315 = bit 2, U314 = bit 1, U313 = bit 0

RAM bank 2
------------
64 KB sized, addressed 512K to 576K.
Two 41464 type RAM chips, each providing 4 bits.
No parity bit.
U321 = bits 7/6/5/4, U322 = bits 3/2/1/0

RAM bank 3
------------
64 KB sized, addressed 576K to 640K.
Two 41464 type RAM chips, each providing 4 bits.
No parity bit.
U323 = bits 7/6/5/4, U324 = bits 3/2/1/0

... = 00h, expected = 00h."

When I first saw that, I thought to myself, "That will be just like the POST in the IBM PC; if no data bits are reported as faulty, it means that it is the parity bit that has failed."

But when I looked at the circuit diagams for the PC10-III/PC20-III motherboard, I could not find RAM for parity. The motherboard's FE2010 chip supports motherboard parity, but I see that the FE2010's PTYIN pin (pin 32)(FE2010 datasheet: "Parity bit from parity RAM") is grounded. The FE2010's PTYOUT pin (pin 33)(FE2010 datasheet: "Parity bit to parity RAM") is routed through driver U108, but I cannot see it after that. So, because the FE2010's PTYIN pin is grounded, the POST in the PC10-III/PC20-III motherboard must be disabling parity checking by the FE2010, via bit 4 of the FE2010's control register.

With no RAM parity bits, the "= 00h, expected = 00h" puzzles me. If the POST wrote 00h to address 526173 and then read back 00h, what mechanism informed the POST that there was a "Bad RAM" problem at that address?

[dippaolo]

FINE.

I ordered the MSM41464-10 RAM (+socket dil1 to start.

I will replace the RAM. I will write what the test showed.

[dippaolo]

Ram replaced. Bad ram is still there.

I read the diagram again. I checked the connection for 2 hours.

I found one more error.

U325 -> U109

NDACK0.jpg

I fixed the connection. Works correctly. I have 640Kb RAM ... YUPI

Thank you all