exxos's Atari, Amiga & retro forum

stephen_usher wrote: 30 May 2021 12:01 Without the correct signals from IC2 the output from the video ULA will be nonsense. Check there first.

I agree, and I'll be looking at that next.

Didn't really look properly at the 6845 much this afternoon, as I got distracted installing the TurboSPI in the good machine, fixing the faulty shift lock LED on the same, and finally finishing off the keyboard rebuild on the bad machine, No2. I'd had 3 key switches soaking in contact cleaner for 2 days and it revived them nicely, so it was a good time to get the No2 kbd finished.

However I did poke around a bit, and it seems that there is no H or V sync output on pins 39 & 40. If I drop the 6845 into the good machine, I get H & V sync. Also, the clock into and from the 6845 looks wrong.

Unfortunately the Fisher Price scope doesn't measure 1Mhz, but it does show the waveform, and it looks about half the frequency of the working machine. I'm looking at a way to actually get it measured properly, but if the 6845 works in the good machine, it suggests that maybe the 1Mhz clock in the bad machine is just that, bad.

I do now have a signal generator and it will also measure, so trying to figure that out. Off to read the documentation now, so hopefully I can find out what the clock actually is.

What I do know is that no H or V sync is a bit of a problem.

If the 1MHz clock was bad then the keyboard and lots of other things wouldn't work.

I'm wondering if the /CE and CLK lines are working correctly. I've not found on the schematic where /CE (known as CRTC on the schematic) is being generated. the CLK (pin 21) if from the ULA pin 28. Without CLK, E and /CE working then the 6845 won't work at all, and hence no h-sync and v-sync.

The 1MHz clock is running the E (enable) line, which clocks the data in and out of the chip, i.e. gives the chip access whilst the 6502 isn't accessing the bus on alternate 1MHz ticks.

There's also the "Display Enable" DEN line which is being generated by IC15, 74LS273 pin 3. That would be easy to check.

Actually, looking at the chip block diagram, horizontal sync is generated directly from the CLK pin, which is generated directly by the ULA pin 28. So if there's no h-sync then it's likely this that's the issue.

stephen_usher wrote: 30 May 2021 20:07 If the 1MHz clock was bad then the keyboard and lots of other things wouldn't work.

I'm wondering if the /CE and CLK lines are working correctly. I've not found on the schematic where /CE (known as CRTC on the schematic) is being generated. the CLK (pin 21) if from the ULA pin 28. Without CLK, E and /CE working then the 6845 won't work at all, and hence no h-sync and v-sync.

The 1MHz clock is running the E (enable) line, which clocks the data in and out of the chip, i.e. gives the chip access whilst the 6502 isn't accessing the bus on alternate 1MHz ticks.

There's also the "Display Enable" DEN line which is being generated by IC15, 74LS273 pin 3. That would be easy to check.

Actually, looking at the chip block diagram, horizontal sync is generated directly from the CLK pin, which is generated directly by the ULA pin 28. So if there's no h-sync then it's likely this that's the issue.

I had thought other things wouldn't work if 1Khz was off too, but I have just pulled out my most recent purchase, a signal generator that goes up to 24Mhz, but will also take an input and measure frequencies.

All the frequencies at IC6 are good, so 16, 8, 4, 2 & 1Mhz. So basically, the Fisher Price scope was leading me astray a little. No more.

Just checking and the 6845 CLK pin, 21, there is 1Mhz there, as is there at pin 28 of IC6. Pin 23 of the 6845 has 1Mhz as expected.Pin 18, DEN is pulsing as tested with a logic probe. Pin 2, RST, active low, has +5v.

What I don't see is anything on pin 25,CRTC is missing. No frequency of any sort, and according to the logic probe, it's high, but is active low. However it looks the same as the good machine.

Currently I see 50Hz on pin 40 of the 6845, which is vsync. Hsync has nothing at all, and has 15.6Khz on the good machine.

Dropping the 6845 from the 'bad' machine into the good machine and it works just fine, so whatever the 6845 needs to generate hsync is missing on the bad machine.

So next I need to understand how hsync is derived in the 6845 and what it isn't getting to prevent it being generated.

Sounds as if the chip isn’t being initialised. The h-sync is controlled by a set of registers and according to the data sheet this needs to be written into the chip at boot time.

See http://bitsavers.trailing-edge.com/com ... s/6845.pdf

So, either it’s not getting the right data (data lines) or it’s not listening to the CPU (address decode logic).

/CRTC seems to be the line used in the address decode, coming from IC26 pin 12. I guess that you would only see this go low briefly when the system powers on or when you change video mode as it’s only when the registers are read or written to that it’s used.

Yes, I think you're right, so I'll focus on that area next. I'll also see if changing the screen mode affects CRTC.

@stephen_usher

I have been poking around for the last hour and a half comparing the 6845 pin activity on the good machine to the bad, and aside from the hsync being zero, was struggling to see any difference.

Just as I was about to wrap up, I decided to bend the hsync leg out and put the 6845 back in and measure the frequency on the leg. Lo and behold it's reading 15.61Khz, just like the working machine. Since we already know the 6845 works in the other machine, this suggests that the inputs are good to the 6845, but somewhere the hsync line is being pulled down, else why would it give the correct hsync frequency with the leg out?

So I'm off to trace hsync, a quick look shows it goes to the SA5050 via IC33, it hits IC41 with vsync to combine to give csync and also pops up in the modulator circuit for UHF out.

Unfortunately I need to go and do some non retro stuff now, but I feel like it's progress.

So, the potential culprits are:

IC33, IC37, IC41 & IC44.

They'll be getting some special attention next so I can find out what is pulling down the hsync pin when it's connected to the board.

Follow the schematic!!

Turns out I have spares of all the ICs bar IC41, a 74LS02, so that's ordered just in case.

It's not like you won't find it used elsewhere, so it can be a parts bin item if I don't use it.

I've decided too that I'll enable colour composite on both machines once the No2 machine is fixed and showing video again, since it's only adding a capacitor and a jumper, and both already have the capacitor :)

rubber_jonnie wrote: 01 Jun 2021 10:09 Turns out I have spares of all the ICs bar IC41, a 74LS02, so that's ordered just in case.

It's not like you won't find it used elsewhere, so it can be a parts bin item if I don't use it.

I've decided too that I'll enable colour composite on both machines once the No2 machine is fixed and showing video again, since it's only adding a capacitor and a jumper, nd both already have the capacitor :)

Usually it's just the jumper.