exxos's Atari, Amiga & retro forum

exxos wrote: 07 Jun 2020 11:09 I'm waiting on @PhilC to get back to me with testing internal and external blitter fixes currently.

Okay, no worries. We're not saving lives with these machines. ;)

Also note you need to likely do DMA fixes as well. https://www.exxosforum.co.uk/forum/viewt ... =17&t=1585

Yeah, I do. That's been somewhere on my mental list of things to do here. The main reason this was my "shelved" STE in the first place was because it's got the "bad" DMA chip. Emphasis on the quotation marks there of course. ;)

Reset fix:


DMA pullups:


And as long as I was doing things, ghosting fix:


...which has left me right where I started. System won't boot, has sync for a little while, loses it, comes back, loses it, comes back, gone. Rinse, repeat.

I don't think that it deciding no longer to boot has anything to do with these fixes (and it did successfully boot after the reset fix and after the DMA fix). It's just that whatever is causing that problem just hasn't been fixed in the first place.

I will take a break from this one for a little while and come back later. ;)

You must have a bad solder joint somewhere.. Only thing to do is press on various parts of the board while pressing reset and see what either triggers a boot or crash..

exxos wrote: 07 Jun 2020 18:01 You must have a bad solder joint somewhere.. Only thing to do is press on various parts of the board while pressing reset and see what either triggers a boot or crash..

What I fear is I've got a crack somewhere, but yeah, I'll keep sleuthing.

Edit: The really weird thing here is it's not exactly crashing, in any normal way. It's just... going away. CPU still has clock and HALT is high. But the diag program will sometimes just stop mid-menu, without any errors, sometimes shows errors, sometimes nothing at all. Usually nothing at all, once the system has warmed up. And same with the video output... It's there when the system is cold, but gone if it's been turned on any time recently.

Just to make things even more interesting. I convinced it to boot again, and now this is what GEMBench is doing:


It's actually running fast (looks identical to what I expect to see with the booster on), but something is wrong with the timing. :twisted:

Edit: After several failed reboots, GEMBench successfully runs again, and now the timing is back to normal. Yes, there is still sleuthing to do. Just kind of poking around at the moment. :)

That looks like the blitter isn't issuing dtack to the CPU.

exxos wrote: 07 Jun 2020 21:05 That looks like the blitter isn't issuing dtack to the CPU.

Seems a lot of different things are not happening. Occasionally the error I'm getting out of the diag software is "Bus error not detected", indicating that the MCU (same chip as the Blitter in this hardware of course) isn't successfully talking to the CPU as the system is coming up. It's all very strange, and seems rather random.

I am comfortable with the soldering on the Glue. I've done it multiple times, with multiple Glues, and inspected with a microscope. (Not to mention I was seeing these same kind of symptoms withe factory Glue.) I have reflowed the booster. I've reflowed the TOS switcher. I've used different RAM. I've even socketed and swapped out the MFP, all to no avail.

I noticed earlier, when the system had not booted successfully, that the CPU was getting an 8 MHz clock, which I am assuming is normal until the system boots. I just checked the clock pin now and found 32 MHz on it...and measuring it made the system crash.

I also noticed something rather odd, which I have not yet compared to my working STE (it's gotten rather late tonight, so I'm not going to be doing much more poking around tonight). This is what I see on GND on the shifter:


That's 16 MHz, fwiw. I don't know if it's normal to see that much AC in the ground. Seems like probably not. I'll compare with the working STE tomorrow.

Last night I was pondering what could be causing a perfectly functioning Glue to stop producing sync, and it dawned on me: The STE was in reset. And that made me realise the one thing I soldered and hadn't reflowed yet, and that's the tap on YM2149 to feed reset to the ube switch. Even though that was just a tap on the side of a leg on the YM2149, sure enough, when I reflowed that, the problem went away, and the system will now reliably boot.

However, it is not stable at 32 MHz. 8 MHz seems perfectly solid, but 32 MHz crashes YAART eventually, and GEMBench (from the exxos diag cart) won't reliably run either (usually won't even successfully start). I wonder at this point if I should just wait for @PhilC to finish his integrated blitter compatibility work and see if it addresses this problem.

Can you post a image of the board ? I've got that many versions of it its hard to keep track whos got what..

exxos wrote: 08 Jun 2020 19:07 Can you post a image of the board ? I've got that many versions of it its hard to keep track whos got what..

Here you go.