ZX81 strange keyboard problem?

[1024MAK]

I can not change line so it is impossible to run commands.


Here is a picture of what came up when i wrote

10 print "hello world"

And what happened when you pressed the [NEW LINE] (enter/return) key?

Mark

[sirmorris]

Something is wrong with the addressing logic used to select the character data. The ULA uses the character code in the dfile as part of the calculated address of the character's bitmap data in ROM. It's consistently wrong, not random.

The ULA looks in the d-file and sees '1' but produces '-'. 'P' is unaffected. As are L,O,R and space.

Code: Select all

'1' = 1Dh,  00011101b

'-' = 16h,  00010110b

'P' = 35h,  00110101b

So to turn '1' into '-' you remove bit 3. Bit 3 is clear in 'P' and therefore it remains unaffected.

Because the computer is working this problem with A3 has to be internal to the ULA. If A3 were damaged on the processor side then seemingly random opcodes would be executed and the computer would not work.

[Andy Rea]

.

Because the computer is working this problem with A3 has to be internal to the ULA.

Or it is not connected / dry jointed at the ula

We gotta be optimistic that the ula ain't dead

Regards

[Andy Rea]

I'm.more convinced on the open circuit idea now, i. The word PRINT the wrong letters à3 is high but in the word HELLO the wrong letters a3 is low this pattern will reapeat every 8 characters as à3 is following the refresh register as it counts down across the screen...

Regards

[sirmorris]

Or it is not connected / dry jointed at the ula

Yes that's a possibility but then other functions that rely on a3 would be broken - kb input for example.

I didn't carry on to check my theorem so I missed the fact that the other wrong letters had a3 pulled high. Shame on me

[1024MAK]

I think an internal ULA fault is the most likely. However, a poor connection could result in some functions working, while others don't work. As the impedance of the electrical circuit may affect the voltage level.

Mark

[sirmorris]

Code: Select all

E A   E  A   E     A
- -   -  -   -     -
1 -   1d 16  1101  0110
0 =   1c 14  1100  0100
p p   35 35        0101
r r   37 37        0111
i a   2e 26  1110  0110
n n   33 33        0011
t l   39 31  1001  0001
h 9   2d 25  1101  0101
e 6   2a 22  1010  0010
l l   31 31        0001
o o   34 34        0100
w o   3c 34  1100  0100
d 5   29 21  1001  0001

Bit 3 forced to zero in every case. Well that killed 10 minutes

[sirmorris]

I didn't carry on to check my theorem so I missed the fact that the other wrong letters had a3 pulled high. Shame on me

Double shame for doubting myself

[Andy Rea]

Oops sorry everyone think my brain is addled Guess this is what happens when one neglects the humble zeddy for so long

[Andy Rea]

Oh one more thing, just to be sure I'm completely bonkers and haven't a clue as to what's going on... we all going bit 3 of the character code translates into bit 6 ( A6' on the ula ) because we have the line counter at A0' A1' A2'.

It is very odd that it's only presenting a logo. Low during character generation and not all the time. But i guess it is looking like the ula is frazzled.

Regards.

P.s. do they do certificates in crazy