Alioth's Journal: [Z80] I have a toast rack!

Well, I put it under the 'Z80' heading, because this will be a bit of rambling JE which will cover some of that.

Firstly, on the 3.3v problem, Ashtead mentioned that 74HCT or LS could be used as level shifters, given that the 3.3v outputs from the 3.3v logic could drive 5v TTL inputs or HCT inputs. By coincidence, today I was looking at the data sheet for Microchip's new single IC ethernet controller (a device to make it easier to make microcontroller circuits connect to ethernet) - which also mentioned doing the same for their device.

Secondly, I received my first toastrack 128K Spectrum. I've been wanting one since 1986! (It's the last machine Sinclair made before selling the company to Amstrad, who went on to make the Spectrum +2 and +3). It's called the toast rack because of the heatsink on the right hand side of the machine, which vaguely resembles a toast rack (for very very thin toast). So today's Pentium space heaters were not the first personal computers to need chip cooling by a long shot :-)

It even works, too. These old machines are a bit hit and miss, I've repaired a rubber 48K Spectrum already (bad 4116 RAM, horrible devices that need no fewer than *four* power supply pins - +5v, -5v, 12v and ground - and are sensitive to the order the power is applied - -5v has to be applied first and removed last otherwise the chip is likely to expire!). I also have a dead Spectrum+ to repair. I've already replaced the lower 16K of RAM (the 4116s), but it didn't fix it.

What I intend to make is a Spectrum diagnostic board at some point for Spectrums which are dead in such a way the Z80 CPU is still working. You can nail the ROMCS line high and keep the Speccy's ROM paged out, so on a PCB, I'll put a Flash ROM with some test routines, and an LCD interface (so you don't need to use the computer's ULA or frame buffer to display output) to simplify fixing these machines. Mainly, perform a RAM test and then test the ULA. Once RAM is tested, a program can be written to the Spectrum's RAM (which can't be paged out), then the onboard ROM paged back in, so that the program now in RAM can do a checksum on the ROM.

The other thing I want to do with the Speccy (perhaps a Winter 2007 project) is to make an ethernet card for the machine using the Microchip product mentioned earlier and an Atmel microcontroller (so the Speccy doesn't have to run the TCP/IP stack on its own CPU. There are still people who like developing stuff for the Speccy - you can now buy a complete hard disk interface for the machine (the DivIDE interface) which works for ATA disks and compact flash cards. I think an ethernet board could go down well (I'm sort of surprised no one's done one yet). I'm a bit terrified of frying the ULA though since that's one part on a Speccy that's irreplacable (except from another machine) and the ULA is pretty fragile. Which is another project someone needs to do - FPGA replacement for the ULA.

Spectrum development

[Tet]

There are still people who like developing stuff for the Speccy

Hmmm. I'm not one of them. I just never liked the Z80. In contrast, the 6502/6510 was a dream to program, and I'd happily develop for the C64 or the Beeb again, even today. But I never got on with writing code on the Spectrum. That said, I appreciate that you're not just talking about software here.

On a related note, my 25 year anniversary T-shirt shipped yesterday :-)

Re:

[Alioth]

CPU preferences are often what we got to experience first :-) The BBC Micro was nice because it had a built-in assembler. But unfortunately I never got on with 6502 since I was already used to (spoiled by :-)) the luxuries of register pairing, shadow registers, the LDIR/OTIR instructions and family on the Z80, even if it did mean I had to wait for GENS to load. The only way I coped on the BBC Micro was to write self-modifying code and you can imagine the horror of that! But even so, back in the day I never