Amiga hardware info, help and support with a focus (but not limited to) North American NTSC experiences. Open to all.
User avatar
TommyE

by TommyE posted Mon Mar 25, 2019 3:54 am

Hello again.

I dug out an old A2000A memory board with 512KB on it and soldered in the pinconnectors.

Plugged it in, I get everything from yellow to red to mostly black screens when powering on.

@Christian, if you are able to share the 80ns hack you mentioned I'd be interested in trying that.

Maybe I'll burn out an diag rom and see if it can see the memory that I believe should now be existing on $c000000...


Tommy
User avatar
TommyE

by TommyE posted Mon Mar 25, 2019 10:39 am

IMG_6712.JPG
IMG_6712.JPG
OK.

So now the board works :)

I now have 512KB extra memory on the A1000 coming from the A2000A memory board.

I inspected the soldering thru a microscope, and there it was, the smallest braid of copper ever, shorting to address lines.

Removed it, and it booted.

The thing now is that the machine is very unstable and throws gurus at me all the time.

I can use it for everything from 1 minute to 5 and then it gurus.

So I would like to look into the 80ns trick mentioned.

Regards
Tommy
Attachments
IMG_6711.JPG
User avatar
Christian

by Christian posted Sat Mar 30, 2019 10:19 pm

Hi Tommy

I’ll dig the hack up in the next several days and will reply once I find it.
Christian
User avatar
Christian

by Christian posted Sun Apr 07, 2019 2:48 pm

I just did the modification on my A2000A RAM expansion board and the upper 512K memory bank (starting at c80000) started having memory test errors.

The upper bank (socketed) had 41256-15 DRAMs while the lower bank (soldered in) had 41257-15 DRAMs (probably faster than labeled since those also support Nibble Mode).

Anyway, I replaced the upper bank DRAMs with 41256-10 (100ns instead of 150ns) and the memory tests fine now.

The performance (Sysinfo 2.73) increased from 0.86 times the speed of a B2000 to 1.00.

See if yours already has the modification (see below). If so, you may have to reverse it if the memory on the card is too slow.

Modification to remove the 1 waitstate:

Change the patch wire that goes from

pin 9 on U16 to pin 12 on U13

to

pin 9 on U16 to pin 17 on U13.

A good way to test unstable c00000 memory is either to add real fast mem (which will have a higher priority, keeping OS and programs out of c00000) or to use Kickstart 1.0/1.1 which don’t automatically add c00000 memory.

Then use GVPMemTest or Microbotics MBRTest-2 (don’t know if that runs under 1.0/1.1).
Attachments
BEDAA525-3388-4162-883D-BDC8281EDD44.jpeg
After
89DD4060-5F2D-4790-A488-53B5DDF4E4B3.jpeg
After
B2144433-FA30-4652-8637-61415A3B4B8B.jpeg
After
C89EBAFD-E66A-47AE-91C1-D6954A481318.jpeg
Before
AB5714AB-E086-4999-B14E-6F34787A33CB.jpeg
Before
A6CA99EF-E78E-4C04-83F9-BCB5BB2DFD7C.jpeg
Before
User avatar
TommyE

by TommyE posted Tue Apr 09, 2019 3:11 am

Christian wrote:I just did the modification on my A2000A RAM expansion board and the upper 512K memory bank (starting at c80000) started having memory test errors.

The upper bank (socketed) had 41256-15 DRAMs while the lower bank (soldered in) had 41257-15 DRAMs (probably faster than labeled since those also support Nibble Mode).

Anyway, I replaced the upper bank DRAMs with 41256-10 (100ns instead of 150ns) and the memory tests fine now.

The performance (Sysinfo 2.73) increased from 0.86 times the speed of a B2000 to 1.00.

See if yours already has the modification (see below). If so, you may have to reverse it if the memory on the card is too slow.

Modification to remove the 1 waitstate:

Change the patch wire that goes from

pin 9 on U16 to pin 12 on U13

to

pin 9 on U16 to pin 17 on U13.

A good way to test unstable c00000 memory is either to add real fast mem (which will have a higher priority, keeping OS and programs out of c00000) or to use Kickstart 1.0/1.1 which don’t automatically add c00000 memory.

Then use GVPMemTest or Microbotics MBRTest-2 (don’t know if that runs under 1.0/1.1).




Hello Christian

Thanks for the information

I tried the hack, but all it did for me in the A1000 was to identify the board as 1MB instead of 512KB as mine is.
When undoing the hack, the machine went back to seeing 512KB.

Was looking for some chips to expand the card to 1MB, but could only find 10chips in my drawers. Maybe I´ll order a few more to try the hack with a fully populated board.

I can also add that hack did show an increase in speed in sysinfo, but the machine was unusable due to it thinking it had 1MB ram extra...

Tommy

Who is online

Users browsing this forum: No registered users and 3 guests