Hello from Island VI in the Pacific!

[Amiga7500]

I'm joining here in the hope that I can find some good advice in maintaining and building up my collection of Amigas. I have 4 of them: an A1000 I originally purchased in 1986; an A3000 that I bought in about 1995 or so, another A3000 and an A500 which were both given to me in the 2000s. Add the numbers up to get 7500! The A3Ks and the A500 are mostly running, so I'm now concentrating on the 1000. I unboxed it in the last few months to find that the note that I had packaged with it last time I put it away (year: 2000) was still accurate: yellow screen and nothing else. I also observe now that the screen is not yellow all the time, it switches back and forth between yellow and blank seemingly randomly and with each condition lasting for a few to several seconds.
I'm a (retired) Electronics Design Engineer: the first Amiga was purchased using a "perk" from the Aerospace company I worked for when they decided to help employees buy PCs (had to add the CrossDos and a 5 1/4" disk to meet their requirements).
I used my Amigas for actual work, not games. I've designed a Zorro-based logic analyzer and written lots in Arrex and C for them.

[intric8]

Welcome, Amiga7500! Very cool backstory and really nice collection.

Do you have any internal or external upgrades in your A1000? (e.g. some sort of Fast RAM?) Or is it completely stock?

[Amiga7500]

I had tricked it out with a Kwikstart from DKB Software and an "INBOARD1000" from Spirit Technology. These both piggyback onto the MC68000 socket; the KwikStart over the INBOARD1000. Each of them also connect with their own pair of wires to locations on the daughterboard. The KwikStart installation kit also included a PAL and a 20 pin socket to replace the DAUGCAS PAL on the daughterboard which I must have done myself back in the day. (I even located the original PAL in my IC hoard!)

Is it best to continue with the story here or start a thread under Hardware? I have removed the PALs from the daughter board (I didn't mention it, perhaps it's obvious, but it is a North American/NTSC machine) and am scratching my head over what I've found.

[intric8]

Is it best to continue with the story here or start a thread under Hardware?

It doesn't really matter.

Whenever you get a yellow screen it usually seems to indicate something going on with either the CPU or Fast RAM. With the multiple things trying to use the CPU socket it's likely some combination of those upgrades or the connection itself. Since the mods you performed eons ago are likely affecting that CPU socket, if it were me I'd try to remove one mod at a time if it isn't too much of a PITA.

The Kwikstart is such a great little mod, I'd try in earnest to keep that one if it isn't the problem, though.

[Amiga7500]

Knowing that people have had problems with the PALs I started scoping their pins. What I saw concerned me: many of the signals were weak and/or missing. As an example I looked for the CAS and RAS signals right at the dynamic RAM chips on the daughterboard: nothing, including at power-up. Also, there were several lines that exhibited different signals depending on whether the screen was yellow or blank.

I own a Data I/O Universal programmer, so I decided to remove and check all PALs against whatever references I could find. I found these sources:
1) https://www.amigawiki.org/doku.php?id=e ... load#a1000 which has A1000_Pals.pdf, which is a scan of an old Commodore document with PALASM equations for the four PALs.
2) https://github.com/joethezombie/Amiga-1 ... al%20A1000 which has links to both Abel and Jedec files for each PAL.

What I saw in the readouts surprised me. ALL of the PALs showed completely different and apparently nonsensical fuse patterns compared to the .JED files from reference 2. An example: the DPALEN chip, shown as a partial side-by side comparison:

Code: Select all

                                                                                                
16L8-12/-15/-25                            DPALEN (-01) As captured by Data I/O 2900:              
Fuse map universal device programmer*      25-Feb-2025 kdb                                         
QF2048*                                    Almost no rows match! (only 8 do)  
L00000 00000000000000000000000000000000*        0   11111111111111111111111111111111   
L00032 00000000000000000000000000000000*       32   11111111111111111111111111111111   
L00064 00000000000000000000000000000000*       64   11111111111111111111111111111111   
L00096 00000000000000000000000000000000*       96   11111111111111111111111111111111   
L00128 00000000000000000000000000000000*      128   11111111111111111111111111111111   
L00160 00000000000000000000000000000000*      160   11111111111111111111111111111111   
L00192 00000000000000000000000000000000*      192   11111111111111111111111111111111   
L00224 00000000000000000000000000000000*      224   11111111111111111111111111111111   
L00256 11111111111111111111111111111111*      256   11111111111111111111111111111111   m
L00288 10101011100111111011011111110110*      288   11111111111111111111111111111111   
   . . . . .                                                                                  . . . . . 
L01568 01010111100111111111011111110110*     1568   11111111111111111111111111111111   
L01600 10101011100111110111011111110110*     1600   11111111111111111111111111111111   
L01632 11111111111111111111111111101110*     1632   11111111111111111111111111111111   
L01664 11111111111111111111111111101011*     1664   11111111111111111111111111111111   
L01696 00000000000000000000000000000000*     1696   11111111111111111111111111111111   
L01728 00000000000000000000000000000000*     1728   11111111111111111111111111111111   
L01760 00000000000000000000000000000000*     1760   11111111111111111111111111111111   
L01792 11111111111111111111011111111111*     1792   11111111111111111111111111111111   
L01824 10010111101111111111111111111111*     1824   11111111111111111111111111111111   
L01856 00000000000000000000000000000000*     1856   00000000000000000000000000000000   m
L01888 00000000000000000000000000000000*     1888   00000000000000000000000000000000   m
L01920 00000000000000000000000000000000*     1920   11111111111111111111111111111111   
L01952 00000000000000000000000000000000*     1952   11111111111111111111111111111111   
L01984 00000000000000000000000000000000*     1984   00000000000000000000000000000000   m
L02016 00000000000000000000000000000000*     2016   00000000000000000000000000000000   m

All four PALs seem to contain garbage. The rusty-looking condition of the pins on all the PALs leads me to think that moisture could have been absorbed in to the chips themselves: what else could account for such massive differences in the fuse readouts?

Just to gain confidence in the DataIO, I found an unprogrammed chip and read it: all rows were ones, still connected.

I've installed new 20-pin machined contact IC sockets for the DPALEN, DPALCAS, and the DAUGEN PALs. The DAUGCAS chip already had a socket due to the KwikStart installation. I still have to replace the beefed-up ground and power wires to these sites, though...I'll try to do a neater job.

I guess my next step is to drive some test signals int to the PALs directly and see what they produce. I'm willing and able to program more modern GALs: guess I'll need adapter boards from PLCC-20 to DIP-20...anybody got a handy source?