archive.worldofdragon.org

Hi All,

New member here - glad I found this information trove on the venerable Dragon, and hopefully somebody might be able to point me in the right direction.

30 odd years after my first Dragon computer, I've finally taken it out of the attic and fired it up. Its a Dragon 64 Issue 2A, which did work (30 years ago), but sadly is now not behaving.
Initial issues were with a complete lack of display, which I would was due to the PSU/Display kludge/wire which had broken away from the PCB (the little green link wire on the underside).

Sorted that, and got garbage on screen, so went checking RAM chips, and found one with no DQ out signal with logic probe - all others were good - and it was IC15. The remainder showed nice pulsing on DQ in, address lines and DQ out. I replaced this chip with a socket and an MB6264A-15, which I believe is a direct replacement. Thinking I would have the D64 operational after this repair, I was disappointed to find not - rather at this point I have a strange screen. On it I have rows of zeros - arranged as follows:-

4 rows of 4x8 Zeros top half of screen
4 rows of 4x8 Zeros bottom half of screen.

[img][/img]



The second rows have some gibberis displayed, which does not change on reset or power on/off. All voltages look ok, and I tested continuity on the new IC with the existing RAM chips and all was well.
Obviously removing either the 6809 or the 74LS783 will give gibberish, but I've no spares of either to test with at present. However, interestingly enough removal of IC37 (128K ROM 2) does not change the screen display. In other words, the display remains the same with this chip in or out. Not the same if ROM 1 is out. So I'm not sure does this point to a ROM issue or an address decoder issue.

Has anybody come across this problem before, or any idea where I could go from here to troubleshoot further. I've a scope and logic probe, but the D64 diagrams are a little hard to read so sometimes unsure of specific pins while trying to test.

Any help anybody can give would be greatly appreciated.

Kind regards
Ger

Hi Ger, welcome to the forum!

That looks like a Dragon that is nearly working. The reset routine got as far as clearing the screen before the wheels fell off.

The difference between a space and '0' is a single bit: D4. So it could be another faulty DRAM IC. As you say, the schematic is not very clear, but it looks like D4 comes from IC32.

Might be worth having a look for shorts on D4 first before replacing the DRAM.

Hi Stew,

I thought I supmitted this already, but it didn't appear on the forum....

Many thanks for your reply - and it appears you are certainly correct. Rather than just removing IC32 I piggy-backed a replacement on, and the D64 is now much better - I get the Dragon Data logo, and Basic Interpreter message, but some letters are muddled, i.e. typing "R" gives "B" etc. The motoron "MOTOBON" fires the tape relay. So I'll replace this IC later tonight.

I'd be delighted if you could explain a little further how you came to determine from my screenshot that IC32 might be dodgy - I get your comments that it feed D4, but how you came to conclude D4 might be the issue eludes me. In addition I'm a little confused as I tested all the RAM chips with the probe and their DQ in and out, together with their respective address lines seemed to be pulsing away fine. Am I missing something on my RAM test methodology, or is there a better way. By this reckoning, I may have more dodgy RAM chips as I cannot now count on my original test method, as IC32 appeared fine by my probe.

In any case, sincere thanks for your assistance thus far.

Kind regards
Ger

FWIW, modifying XRoar code, here's a screenshot of what happens when writes don't include D4. Familiar!

Maybe I should generate this for each bit in turn - might be an interesting reference (edit: scratch that, it's pretty predictable actually)

Specifically, that's only writes; reads then return the "uninitialised" pattern as observed. So maybe even if you're seeing the Din line change, there's an internal fault with that connection in the chip.

...if you could explain a little further...

Great to hear you could be close to up and running!

Your screenshot shows a screen that is at least partially cleared, so it looked like the CPU managed to start running the ROM reset routine, set up the PIAs, set the video mode and video base address, then write to the correct memory locations to clear the screen. i.e. pretty much everything was working.

The screen is cleared by writing hex 60 to each location, but '0' characters are represented by hex 70. In binary it looks like this:

60 : 01100000
70 : 01110000

So it looks like bit 4 (D4) is sometimes set when it shouldn't be. If this fault was on the CPU data bus then the machine would almost certainly fail to get as far into the reset routine as your screenshot was showing. Either it would read the wrong values from ROM and crash or write the wrong values to the hardware and look different. That leaves the memory as the most likely location of the fault and D4 is handled by IC32.

The overall pattern on screen is another clue that the problem is memory related, and is nicely demonstrated by sixxie's simulation. (Uninitialised DRAMs contain a characteristic checkerboard pattern)

As the fault appears to be a failure to write rather than a failure to read, you would still get activity on IC32 output as it's just outputting the initial contents. A logic probe or scope won't tell you the whole story. The only way to fully test a memory IC is with a complicated series of writes and reads.

I hope that makes sense!

Stew/Sixxie,

Thank you both again for your comments - yes it's beginning to make a little bit more sense now. I have not tried emulated Dragons before so I will certainly take a look at this.

I'll proceed to change out the offending RAM chip asap, and will revert and let you know how it went once complete.

Thanks again,
Ger

Hi,
Just to let you know my D64 is operational again fully - replacing IC32 did the trick.
Thanks to Stew and Sixxie for their assistance and information.

Kind regards
Ger