exxos's Atari, Amiga & retro forum

*mumble* *mumble* clocks *mumble*

stephen_usher wrote: 27 Jan 2024 18:09 *mumble* *mumble* clocks *mumble*

Really, don't get me started :(

This journey will be longer than expected IMHO...

I hope you will find out soon what's wrong with your current set up.

HigashiJun wrote: 28 Jan 2024 01:44 This journey will be longer than expected IMHO...

I hope you will find out soon what's wrong with your current set up.

Me too :dizzy:

Isn't it supposed to be an 8MHz crystal?

stephen_usher wrote: 28 Jan 2024 14:04 Isn't it supposed to be an 8MHz crystal?


UK101-clock.jpg

Not on the OSI, and I have confirmed the crystal frequency with somebody else who built one:


There are other schematics that show 3.7Mhz, but you could argue it's 3.9Mhz and the schematic is terrible quality.

The service manual shows this:

That 74LS86 is acting as a clock doubler isn't it?

The 4MHz output is on U58 pin 3 and that's pushed through one of the XOR gates on U79 before going into a second gate on U79 and XORed with the centre frequency of the crystal oscillation circuit, which will be ahead of the output from U58, creating a second pulse for every pulse from the crystal. Not regularly timed but it's an oscillation, so generating an 8MHz-ish clock.

I'm guessing that they're relying upon exact delay timings from the gates in the chips. If you don't have the precise types and manufacturers then the timings will change and it won't work. Very sketchy.

stephen_usher wrote: 28 Jan 2024 17:23 That 74LS86 is acting as a clock doubler isn't it?

The 4MHz output is on U58 pin 3 and that's pushed through one of the XOR gates on U79 before going into a second gate on U79 and XORed with the centre frequency of the crystal oscillation circuit, which will be ahead of the output from U58, creating a second pulse for every pulse from the crystal. Not regularly timed but it's an oscillation, so generating an 8MHz-ish clock.

I'm guessing that they're relying upon exact delay timings from the gates in the chips. If you don't have the precise types and manufacturers then the timings will change and it won't work. Very sketchy.

What you're saying makes sense, but from the point of view of what's gone before, I was at one point getting 1Mhz to the CPU from the clock circuit, but it is now just delivering 500Khz, suggesting perhaps something that should be happening at U79 isn't.

It is designed to run at 1Mhz, but if the clocks are crappy across the board, even if the PHI0 clock is good, it doesn't follow that others are around the board.

I'll go and check the clocks at U58, U79 and U29, see what's what.

It could be that the timing with the chips you have are marginal, especially if the ones you're using are too fast. I'm guessing that they're hoping the delay is half a clock cycle. Too fast and there's no time to create the extra pulse. Was an 74LS part specified or just a plain 74?

stephen_usher wrote: 28 Jan 2024 19:08 It could be that the timing with the chips you have are marginal, especially if the ones you're using are too fast. I'm guessing that they're hoping the delay is half a clock cycle. Too fast and there's no time to create the extra pulse. Was an 74LS part specified or just a plain 74?

U58 is a straight 7400 and that's what I'm using, and I tried swapping it out but no change. U79 is not behaving though, I checked earlier and I have 4Mhz where I was expecting, but at pin 6 of U79 there is nothing at all.

U79 tests fine in my TL 866, but that doesn't mean it's going to work in this scenario if it's marginal. I'd need to check, but I think an LS is specified.

I thought I had spares, but I don't so will check if it's an an LS or not and order some more.

Pin 12 of the LS93 should be on about 500Khz and it's half that, so more evidence we the 4Mhz input isn't right and it should be 8, then pin 11 should be 1Mhz and not 500Khz.