exxos's Atari, Amiga & retro forum

IngoQ wrote: 09 Oct 2017 07:28 There is another possibility, that my CPU-socket construction, which exposes CPU lines vertically for about 5-8mm will be of fault here. When I can rule out the SIMMs I will have to investigate this. It turned out way to high anyways, so I will have to start over with that.

ahh yep, I was going to ask you about that. You are in somewhat unknown territory there.. Likely not going to be easy to diagnose. A few mm of inductance in the gnd connection is enough to cause chaos at 32mhz. I had the same issue on my V2 booster.

Have a look at March 10 2015 post..

http://www.exxosforum.co.uk/atari/last/16mhz/index.htm

.. the 2 big blue-ish pcb images. A thin connection between gnd points on the CPU was all it took to be unstable.

Only other thing is to make sure P100 resistor pack is 1.2K. Its near the front of the CPU/Blitter somewhere.

rubber_jonnie wrote: 08 Oct 2017 23:23 I did have some strange memory issues, but luckily I had a big bag o' ram from back in the old days, and found some suitable 8 chip SIMMs that worked fine.

I only found a bunch of EDO 72pin and S0-SIMMs, sadly :-(

rubber_jonnie wrote: 08 Oct 2017 23:23 In terms of the 32Mhz wire, I came pretty much direct from U405, across the top right of the SIMM sockets (You should be able to see in the picture), and direct to the Booster, using the wire supplied with the kit, but keeping it as short as possible.

I tried both routes (above and beyond SIMMs), results were different, but both bad. Is there something special about that wire? Mine came without and I used a thin stranded copper wire, roughly 0,05 mm².

exxos wrote: 09 Oct 2017 08:49 ahh yep, I was going to ask you about that. You are in somewhat unknown territory there.. Likely not going to be easy to diagnose. A few mm of inductance in the gnd connection is enough to cause chaos at 32mhz. I had the same issue on my V2 booster.

Yep, that's why I go with elimination. But in any case, my current solution kinda sucks. Here is how it looks like:
As you can see, it is way to high. Problem was, that that the PLCC plugs pin out is some fractions of mm off. I assumed it would be within 1.27mm grid, but it really is a tiny bit smaller. And in addition, the pins do not have the same size, my adapter only fits at the end of the legs, you cannot move them further up. So I ended up with the plug having some distance to the board, thus beeing way higher, than I wanted.

So I guess to have this fit, I will need to do a new adapter with bigger holes (if possible at all) and and the correct measures. But then it most likey won't fit within the funky footprint, so I might have to place it next to it, which would make the board bigger, the traces longer, etc...

Here is the current adapter:
Nevertheless, I am a little proud of it, it is my first PCB ever after all (not counting the ones I etched during my trainee time looong ago) :mrgreen:

It looks like a decent job if anything :) so many solder connections, you only need one tiny bad contacts somewhere then that pin becomes floating in causes chaos, which is exactly the same problems as the parity Simms have, as they have inputs which are not connected.

But of course, it might just be the extra length on the CPU signals is just enough to push you over the edge again. This really doubles up to the problems with the DMA circuit, as a tiny bit of ground bounce in the wrong place causes chaos everywhere. So even just a few millimetres of track length from the motherboard to the booster board could be enough to push you over the edge again.

Why can suggest is double check all your solder connections make sure you do not have any bad contacts anywhere.

You can of course try 16 MHz input, you do not have to use 32 MHz. You may find 16 MHz might be more stable..

exxos wrote: 09 Oct 2017 13:09 It looks like a decent job if anything :)

Thanks ;)

exxos wrote: 09 Oct 2017 13:09 Why can suggest is double check all your solder connections make sure you do not have any bad contacts anywhere.

Already double, tripple and quadruple checked, since some of the solder joints aren't accessible any more. If there are contact issues, I believe the could be between plug and socket. But again, I wait for the SIMMs to rule this out, or ideally solve it.

exxos wrote: 09 Oct 2017 13:09 You can of course try 16 MHz input, you do not have to use 32 MHz. You may find 16 MHz might be more stable..

How would I do that?

IngoQ wrote: 09 Oct 2017 15:15 How would I do that?

The yellow and black wire as shown in this image, is the 16 MHz point on the STE. Just use that instead of soldering it to the IC.

Thanks, another option to try.

Just received the shipping confirmation for the SIMMs :)

Nice idea for the PLCC socket idea, it looks really good, but the extra connections, plus the possiblility of iffy connections between the plug and socket could be part of the instability, especially since the on board socket is quite old and the connectors could have retracted sightly over the years.

Mine is soldered, and apart from the original RAM issues, it's been very stable. I can't say what the black wire is, it's what came with the kit, exxos can guide you here.

rubber_jonnie wrote: 09 Oct 2017 22:13 Nice idea for the PLCC socket idea, it looks really good, but the extra connections, plus the possiblility of iffy connections between the plug and socket could be part of the instability, especially since the on board socket is quite old and the connectors could have retracted sightly over the years.

Didn't had much of a choice, my STE had it's CPU directly surface mounted on the PCB, without a socket. So no through-holes to put header pins in to.

See the thread here:

https://www.exxosforum.co.uk/forum/viewt ... ?f=24&t=68

Yeah, I was reading, I take it that was a late model STE with no socket?