General Stuff(tm)

[stephen_usher]

Now I have the start of the pulse correctly tied down I need to discover how to end it correctly. Here are the two signals, the upper one is the one I'm generating and the lower is the one generated by the custom logic chip.

IMG_2656.jpg

As you can see, my pulse is too long by about 40ns, i.e. 2 clock cycles and I need to find a way to terminate at the correct time using just the GAL.

The signals I have available (other than the 20MHz clock) are these, the signal I'm trying to replicate is the bottom one:

logic-analysis-scc1-write.jpg

Anyone have any ideas? I'm not sure if I do using simple logic.

[stephen_usher]

I wonder if the only way to get a precise 240ns pulse will be to have an external 555 timer, if that can go down to that length of time. Basically triggering it from the GAL and feeding the output back into the GAL so that it can select which output to direct it to.

This is unless someone knows how to generate a pulse that's a non-multiple of a clock pulse long using a 16V8.

Any ideas?

The alternative would be to somehow use the output from pins 15 and 27 (see LA output above) but I can't see how to do this in simple logic. Again, any ideas?

[stephen_usher]

Thinking about this further, I think the timing is a red herring. Tracking pin 21 should be good enough looking at the (what I think to be) address line activity. I may have to instrument the SCC chip itself to see when the data for writing is active. It could be that I'm missing something there.

Of course, my extrapolation from the SCC1 write line may not be correct for the other lines but without any working system to compare against it's almost impossible to work this out.

Oh, and to assist this I'm working on the disassembly of the ROM. It turns out that Ghidra ( https://ghidra-sre.org/ ) is an amazing tool. It even makes a decent job at decompiling the assembler to C code.

P.S. For Ghidra to work, when you import the ROM image you have to know the start address.
P.P.S. Ghidra works with *MANY* architectures and processors, not just m68k. There's x86, there's Z80, 6502, 6809... A VERY long list. That's given me an idea... I could look at the Apricot F Series ROM...

[stephen_usher]

I've put the Sun away again until I have some more information about what the signals should be.

Instead I have my TT keyboard apart to see if I can find out why a lot of the time it thinks that the right mouse button is pressed.

Well, well... It looks like the internal pull-up in the keyboard controller on the line connected to pin 9 is iffy. Sometimes it does register +5V on the line but after the keyboard has been on a few seconds it starts dropping.. and dropping... and hovers around 2.3V. This is the only line which goes directly to the chip from the port. I wonder if the Tom+ dongle has tried to pull too much from the line and damaged the controller.

Putting a 10K ohm resistor between +5V and the line pulls it up to the correct voltage and allows it to work again.

Has anyone else seen this fault before?

P.S. The Tom does seem to have a significant power draw on the line. Adding a 120 ohm resistor seems to be a better pull-up for the line. After all it's a normal switch and not TTL logic pulling it down so a low value isn't problematic and it will take the load off the chip's pull-up. The keyboard processor still won't start properly if the Tom's plugged in though.

[stephen_usher]

After the conversation with @rubber_jonnie this afternoon I decided in the end to open up the C64 and take a look inside and fit some heatsinks to the PIA and SID chip, to help them stay alive for longer.

It seems that I have a machine built in mid 1983, from the latest chip date codes. It has a ceramic VIC-2 chip! The board serial number if 2709, meaning that it was probably the 2709th board to come off the UK production line.

IMG_2659.jpg

It also has an odd RAM chip in there from a manufacturer whose logo I don't recognise at all.

IMG_2660.jpg

Anyway, I tried twiddling the potentiometer next to the VIC-2 chip to see if that made any difference to the jail bars... It did but not in any good ways, so I put it back to its original setting.

[bigbloke]

That's a wacky bunch of sdram chips, 6x Mitsubishi, 1x with early Micron logo and 1x Fujitsu part! As for the video noise, I'd probably replace all the caps (not just the electrolytics) in that area if you know the values. Even ceramics from that age will likely have drifted too far from design values - not so important for decoupling, but affects filters and timing circuits. I like seeing inside all these old machines, especially if they haven't previously been messed with!

[stephen_usher]

Oh, it's early Micron. That's a strange ceramic packaged RAM chip too isn't it?

[bigbloke]

The RAM timings must be very comfortable to mix and match the memories in one bank so much. I'd never risk mixing device families like that in my designs! I used to use all sorts of packages back in the 80's - my favourite were the SDRAMs in ceramic packages with gold lids. You could take the lids off and shine bright lights at the silicon and get a really crappy camera with it. (Camera somewhat exagerating the shadow/light one bit nature of it, and I never got any sort of focus in my experiments back then!) But still cool, when CCD cameras cost loads, and needed a lot of electronics to get working!

[stephen_usher]

Sounds fun, if expensive at the time.

They're all 150ns chips. I think the RAM timings are such that even 200ns would be OK, though I could be wrong.

[rubber_jonnie]

After the conversation with @rubber_jonnie this afternoon I decided in the end to open up the C64 and take a look inside and fit some heatsinks to the PIA and SID chip, to help them stay alive for longer.

It seems that I have a machine built in mid 1983, from the latest chip date codes. It has a ceramic VIC-2 chip! The board serial number if 2709, meaning that it was probably the 2709th board to come off the UK production line.


IMG_2659.jpg


It also has an odd RAM chip in there from a manufacturer whose logo I don't recognise at all.


IMG_2660.jpg


Anyway, I tried twiddling the potentiometer next to the VIC-2 chip to see if that made any difference to the jail bars... It did but not in any good ways, so I put it back to its original setting.

Wow, that ceramic VIC-2 is a thing of beauty!!!

As for the RAM chips, I reckon they were short that day and sent Keith from parts down to the local shops!! Nice ceramic RAM IC too though.