Covox

[nollkolltroll]

First iteration of a Covox interface for the ZX81 is working now.
For those that missed it, it's a type of simple 8-bit DAC using resistors, so really cheap.
Next step will be to get my EPROM-programmer up and running again.

Lessons learned:
- I should NEVER solder anything when having a fever. I had to redo everything later.
- There is no IN/OUT in ZX81 basic, but PEEK/POKE is. Doing a memory-mapped device is easier to test.
- When doing a memory-mapped device, don't try to write to ROM-area and expect the data to be what you want.
- A 74x574 does NOT have an active low clock signal. I've used these quite a lot (a long time ago), but somehow I did not remember this...
- ULA:s can be slightly broken. Mine could not read 2 columns of the keyboard, and the keyboard was not faulty. The leads are even shorter now.

[siggi]

- When doing a memory-mapped device, don't try to write to ROM-area and expect the data to be what you want.

Did you pull the /ROMCS line to HIGH during write access to your hardware located in the rom address area? That would avoid data bus conflicts during writing data to your hardware.
Typical circuit is:

Code: Select all

                |\  |
                |  \|  
CS-DAC ---------|   |-------------- /ROMCS
                |  /|
                |/  |

where CS-DAC is your acitive HIGH chip select of your DAC during write cycles at rom addresses. The diode could be a 1N4148

Caution! The BASIC writes to the first 5 rom addresses 0000-0004, so don't use them for your hardware address decoder.

HTH Siggi

[PokeMon]

- ULA:s can be slightly broken. Mine could not read 2 columns of the keyboard, and the keyboard was not faulty. The leads are even shorter now.

This could be an issue with the power supply when connecting something additional over the expansion bus.
If the voltage drops a little bit - OR - if the address bus get too much load, keys tend to not respond due to a cheap and simple but not perfect keyboard attachement via the simple diode logic. People always use drivers for the data bus but forget to use them for the address bus too, where the address logic is connected. Replacing the 1N4148 diodes with schottky of type BAT46 or similar at the keyboard connector could help also. By the way, they can be soldered in parallel helping to drop down the voltage at the kbd inputs at the ULA.

[nollkolltroll]

Siggi: that explains a lot! Did not know that basic wrote to ROM. I tried to use those low addresses, now I know why the values kept changing.
I didn't try to pull /ROMCS high. I just wanted to get things working memory-mapped in basic before changing to port-mapped.

PokeMon: I really think the ULA is broken, tried it in 2 different ZX81:s with the same results. Plain unexpanded, 3 different expansion memories, nothing mattered.

[1024MAK]

Also keep in mind that due to the minimum decoding used by Sinclair, the ROM (and the echoes of the ROM in the unused part of the memory map) may produce outputs when written to. This alone will mess up anything else connected to the data bus in the same memory range.

Mark

[olofsen]

By the way, why does the BASIC write to ROM?

[1024MAK]

By the way, why does the BASIC write to ROM?

Because the first five bytes of the ROM are written to by the Basic as the result of an error in the maths.

See here

Mark

[PokeMon]

By the way, why does the BASIC write to ROM?

This is a bug in the firmware or "BIOS".
I think there are only a few candidates of BASIC instructions doing it. I found for example PRINT does try to write to the ROM. I found this during testing with an EEPROM instead of a ROM.

[olofsen]

It is the "skip constants" subroutine at 0x1a2d!

[sirmorris]

I have heard a more charitable explanation - that writing to somewhere 'non-changable' was another clever optimisation which saved a few bytes in the ROM. I can't remember where I heard this, though.