Partially not working: VGA test pattern ok but Weird screen within Apple //e

[iolo]

I recently got a PCB(retrotink version) and do soldering by myself. But it didn't work correctly.

So, I made my own version with prototyping PCB.
It still doesn't work. Same result.(no PCB problem)

It works with RENDER_TEST_PATTERN test pattern.(No VGA DAC problem)

But, within My Apple //e enhanced, it shows weird screen.

This is the result when I run following basic program:

10 N=0
20 FOR I = 1024 TO 2047
30 POKE I,N
30 N=N+1:IF N>255 THEN N=0
50 NEXT I

https://youtube.com/shorts/NP9JtHIC-q0?feature=share

As you see, something changes but incorrect address.(abus pio?)

Text mode only. No GR, HGR, PR#3 and so on. Maybe, it can't read soft-switches too.

I double&triple checked all wires and registerance and votages and so on.

I'd tried to debug it, It's toooooo fast with log.
(I saw some address and data bus was captured but I can't check whole logs)

What can I do further?
Any suggestions or advice would be appreciated.

[markadev]

Hey there!

From those symptoms it does seem like the interface to the Apple II bus is not working right. If you have an oscilloscope then probing pin 31 on the Pico must show a clean regular clock signal, which is the Apple bus clock that the Pico firmware uses to time the bus reads. If this signal is messy then you can get garbled output like you are seeing.

I'm not 100% familiar with the retrotink designed boards but it looks like they use a resistor instead of a logic gate for that clock signal to translate from 5V to the Pico's 3.3V. Personally, I have not had good luck with high speed (over 200KHz) signal integrity when large resistors were on the signal line which is why I recommended checking that clock signal with a scope.

One other thing to check could be the 74LVC245 chips themselves. Other logic families like 74LS will not work and if you got your chips from a less reputable source then there's a small chance they could be relabeled or counterfeit. You could probably tell if they're good or not by putting an oscilloscope probe on the output side (pins 2-9) and verifying that the voltages are reasonable 3.3V TTL logic levels with clean transitions.

[iolo]

It's working!

I just modified a number of abus.pio
into

    set PINS, 0b101  [10]                ; enable AddrLo tranceiver and delay for transceiver propagation delay

and finally it works!
Thanks for your considerations!

---

For information,

My debugging log

1. bus snooping partially working. data bus ok. address bus? something work or not.
2. always text mode. softswitches are on bus too.
3. modify abus.c to add logging to apple ii text mode screen

uint16_t ptr = 0x400;

void putlogchar(uint8_t c) {
  main_memory[ptr] = c;
  if (++ptr > 0x800) ptr = 0x400;
}

void putloghex(uint8_t c) {
  putlogchar(c < 10 ? c + '0' : c - 10 + 'A');
}

void putlog(uint16_t addr, uint8_t data) {
  ptr = 0x400;
  putloghex((addr >> 12) & 0x0f);
  putloghex((addr >> 8) & 0x0f);
  putloghex((addr >> 4) & 0x0f);
  putloghex(addr & 0x0f);
  putlogchar(' ');
  putloghex((data >> 4) & 0x0f);
  putloghex(data & 0x0f);
  putlogchar(' ');
}

...

static void shadow_memory(bool is_write, uint_fast16_t address, uint32_t value) {
...
    case 0x2000 >> 10 ... 0x3c00 >> 10:
// *****************************************
// when access $2000.., print snooped address and data to $400, those will be shown on apple ii text mode screen. 
      putlog(address, value);
// *****************************************
        // hires page 1
        reset_detect_state = 0;
        if(!is_write)
            break;

8. On Apple II, I poked 20xx: yy, 2020 yy was shown instead of 20xx yy. Not address low byte but Address high byte again.

9. My guess: 245 didn't enabled in time by some unknown reason(low current? invalid impedance? poor soldering? but both pcb? anyway)
10. increase delay 5 to 10 (I don't know what it means exactly)
11. It works!!!

That's all folks.