exxos's Atari, Amiga & retro forum

Well, not getting very far with the Sun 3/80.

I think my GAL encoding isn't working correctly. One of the read lines is swapped with the write, but changing the sense in the description makes nothing work. No idea why. SER1WR and SER1RD seem swapped.

Maybe making progress, sort of.

It seems that one of the lines out of the chip which I thought was dead is only mostly dead. If the line is not loaded, other than with the oscilloscope, there is some signal getting out. It's "floating" down to about 2.2V with square signals down to zero. Any attempt to pull it up, even with megaohm resistors, kills the signal.


What this has told me is that the read signal is totally unlike that of the write signal in terms of activation, so my original assumption for regenerating the signal is completely wrong. I'm going to have to test every signal on the chip to see which ones (if any) uniquely identify when the signal is activated.

Also, my original write signal was happening about 80ns too early, due to one of the signals I was using (which I think is UART_ENABLE from the Sun 3/60 schematic). This was causing a crash, but ANDing this with the I/O enable (assumed) signal brought this into line.

Nuts! It's actually the signal from the pin next to it bleeding through the chip. Oh well.

The signal I'm generating still has slightly wrong timing, going low about 20ns too early and going back to high about 40ns too late. That's half a clock cycle at the beginning and almost a whole one at the end.

If I could work out how to program the GAL in register mode I could probably use the system clock to tie the signal generation better.

So, who knows CPLD well enough to teach me how to do registered mode on a 16V8?

Also, I need to be able to trigger an output (low) when triggered by a set of inputs on the following rising clock and then de-assert the line 3.5 clock cycles later.

*tricky!*

The code you have looks fine to me ? Have you got the gal on breadboard to verify it is the gal at fault ? Might the GAL be faulty ?

With a clock on pin1 you just add the .D extension to make clocked delays.

Yes, it's on a breadboard and it's outputting the correct sort of signals, given the inputs.

OK, I'll try that clocked outputs next time. I've had to pack things away so I can revert the room into laundry mode.

I may not have had the space for the 3/80 investigation but I did for de-RIFAing and testing an early BBC Micro, part of the haul from last weekend. This is the last of the machines to fully test.


Some interesting things to note on early machines. Firstly the "crackle" on the surface of the machine far, far rougher. Secondly the logo is completely different.


Add to that, you don't have to feed the power supply cord through a hole to remove the PSU and there's a steel grill underneath the PSU.


Otherwise it's like later machines. Indeed, although it's early this machine already has an Issue 4 motherboard, the most common revision.

Oh, and how long has it been since you saw a swing door floppy drive?

stephen_usher wrote: 20 Mar 2022 15:40 I may not have had the space for the 3/80 investigation but I did for de-RIFAing and testing an early BBC Micro, part of the haul from last weekend. This is the last of the machines to fully test.

Nice, the BBCs are one of the greats of the era IMHO, just wish I could have had one as a youngster, but the cost was just too high. To be fair my parents did get me a ZX81 then Spectrum, so they did well by me and it did set me off down a long road in IT.

The BBC was way out of our price bracket too, at £399. I started the same way as you.